observational learning case study

• Bipolar Disorder
• Therapy Center
• When To See a Therapist
• Types of Therapy
• Best Online Therapy
• Best Couples Therapy
• Best Family Therapy
• Managing Stress
• Sleep and Dreaming
• Understanding Emotions
• Self-Improvement
• Healthy Relationships
• Student Resources
• Personality Types
• Guided Meditations
• Verywell Mind Insights
• 2024 Verywell Mind 25
• Mental Health in the Classroom
• Editorial Process
• Meet Our Review Board
• Crisis Support

How Observational Learning Affects Behavior

Kendra Cherry, MS, is a psychosocial rehabilitation specialist, psychology educator, and author of the "Everything Psychology Book."

• Influential Factors
• Pros and Cons

Observational learning describes the process of learning by watching others, retaining the information, and then later replicating the behaviors that were observed.

There are a number of learning theories, such as classical conditioning and operant conditioning , that emphasize how direct experience, reinforcement, or punishment can lead to learning. However, a great deal of learning happens indirectly.

For example, think about how a child may watch adults waving at one another and then imitates these actions later on. A tremendous amount of learning happens through this process. In psychology , this is referred to as observational learning.

Observational learning is sometimes called shaping, modeling, and vicarious reinforcement. While it can take place at any point in life, it tends to be the most common during childhood.

It also plays an important role in the socialization process. Children learn how to behave and respond to others by observing how their parent(s) and/or caregivers interact with other people.

Verywell / Brianna Gilmartin

History of Observational Learning

Psychologist Albert Bandura is the researcher most often associated with learning through observation. He and others have demonstrated that we are naturally inclined to engage in observational learning.

Studies suggest that imitation with social understanding tends to begin around 2 years old, but will vary depending on the specific child. In the past, research has claimed that newborns are capable of imitation, but this likely isn't true, as newborns often react to stimuli in a way that may seem like imitation, but isn't.

Basic Principles of Social Learning Theory

If you've ever made faces at a toddler and watched them try to mimic your movements, then you may have witnessed how observational learning can be such an influential force. Bandura's social learning theory stresses the power of observational learning.

Bobo Doll Experiment

Bandura's Bobo doll experiment is one of the most famous examples of observational learning. In the Bobo doll experiment , Bandura demonstrated that young children may imitate the aggressive actions of an adult model. Children observed a film where an adult repeatedly hit a large, inflatable balloon doll and then had the opportunity to play with the same doll later on.

Children were more likely to imitate the adult's violent actions when the adult either received no consequences or when the adult was rewarded. Children who saw the adult being punished for this aggressive behavior were less likely to imitate them.

Observational Learning Examples

The following are instances that demonstrate observational learning has occurred.

• A child watches their parent folding the laundry. They later pick up some clothing and imitate folding the clothes.
• A young couple goes on a date to an Asian restaurant. They watch other diners in the restaurant eating with chopsticks and copy their actions to learn how to use these utensils.
• A child watches a classmate get in trouble for hitting another child. They learn from observing this interaction that they should not hit others.
• A group of children play hide-and-seek. One child joins the group and is not sure what to do. After observing the other children play, they quickly learn the basic rules and join in.

Stages of Observational Learning

There are four stages of observational learning that need to occur for meaningful learning to take place. Keep in mind, this is different than simply copying someone else's behavior. Instead, observational learning may incorporate a social and/or motivational component that influences whether the observer will choose to engage in or avoid a certain behavior.

For an observer to learn, they must be in the right mindset to do so. This means having the energy to learn, remaining focused on what the model is engaging in, and being able to observe the model for enough time to grasp what they are doing.

How the model is perceived can impact the observer's level of attention. Models who are seen being rewarded for their behavior, models who are attractive, and models who are viewed as similar to the observer tend to command more focus from the observer.

If the observer was able to focus on the model's behavior, the next step is being able to remember what was viewed. If the observer is not able to recall the model's behavior, they may need to go back to the first stage again.

Reproduction

If the observer is able to focus and retains the information, the next stage in observational learning is trying to replicate it. It's important to note that every individual will have their own unique capacity when it comes to imitating certain behaviors, meaning that even with perfect focus and recall, some behaviors may not be easily copied.

In order for the observer to engage in this new behavior, they will need some sort of motivation . Even if the observer is able to imitate the model, if they lack the drive to do so, they will likely not follow through with this new learned behavior.

Motivation may increase if the observer watched the model receive a reward for engaging in a certain behavior and the observer believes they will also receive some reward if they imitate said behavior. Motivation may decrease if the observer had knowledge of or witnessed the model being punished for a certain behavior.

Influences on Observational Learning

According to Bandura's research, there are a number of factors that increase the likelihood that a behavior will be imitated. We are more likely to imitate:

• People we perceive as warm and nurturing
• People who receive rewards for their behavior
• People who are in an authoritative position in our lives
• People who are similar to us in age, sex, and interests
• People we admire or who are of a higher social status
• When we have been rewarded for imitating the behavior in the past
• When we lack confidence in our own knowledge or abilities
• When the situation is confusing, ambiguous, or unfamiliar

Pros and Cons of Observational Learning

Observational learning has the potential to teach and reinforce or decrease certain behaviors based on a variety of factors. Particularly prevalent in childhood, observational learning can be a key part of how we learn new skills and learn to avoid consequences.

However, there has also been concern about how this type of learning can lead to negative outcomes and behaviors. Some studies, inspired by Bandura's research, focused on the effects observational learning may have on children and teenagers.

For example, previous research drew a direct connection between playing certain violent video games and an increase in aggression in the short term. However, later research that focused on the short- and long-term impact video games may have on players has shown no direct connections between video game playing and violent behavior.

Similarly, research looking at sexual media exposure and teenagers' sexual behavior found that, in general, there wasn't a connection between watching explicit content and having sex within the following year.

Another study indicated that if teenagers age 14 and 15 of the same sex consumed sexual media together and/or if parents restricted the amount of sexual content watched, the likelihood of having sex was lower. The likelihood of sexual intercourse increased when opposite-sex peers consumed sexual content together.

Research indicates that when it comes to observational learning, individuals don't just imitate what they see and that context matters. This may include who the model is, who the observer is with, and parental involvement.

Uses for Observational Learning

Observational learning can be used in the real world in a number of different ways. Some examples include:

• Learning new behaviors : Observational learning is often used as a real-world tool for teaching people new skills. This can include children watching their parents perform a task or students observing a teacher engage in a demonstration.
• Strengthening skills : Observational learning is also a key way to reinforce and strengthen behaviors. For example, if a study sees another student getting a reward for raising their hand in class, they will be more likely to also raise their hand the next time they want to ask a question.
• Minimizing negative behaviors : Observational learning also plays an important role in reducing undesirable or negative behaviors. For example, if you see a coworker get reprimanded for failing to finish a task on time, it means that you may be more likely to finish your work more quickly.

A Word From Verywell

Observational learning can be a powerful learning tool. When we think about the concept of learning, we often talk about direct instruction or methods that rely on reinforcement and punishment . But, a great deal of learning takes place much more subtly and relies on watching the people around us and modeling their actions. This learning method can be applied in a wide range of settings including job training, education, counseling, and psychotherapy .

Jones SS. The development of imitation in infancy.   Philosophical Transactions of the Royal Society B: Biological Sciences . 2009;364(1528):2325-2335. doi:10.1098/rstb.2009.0045

Bandura A.  Social Learning Theory . Prentice Hall; 1977.

Kühn S, Kugler DT, Schmalen K, Weichenberger M, Witt C, Gallinat J. Does playing violent video games cause aggression? A longitudinal intervention study.  Mol Psychiatry . 2019;24(8):1220-1234. doi:10.1038/s41380-018-0031-7

Gottfried JA, Vaala SE, Bleakley A, Hennessy M, Jordan A. Does the effect of exposure to TV sex on adolescent sexual behavior vary by genre? Communication Research . 2013;40(1):73-95. doi:10.1177/0093650211415399

Parkes A, Wight D, Hunt K, Henderson M, Sargent J. Are sexual media exposure, parental restrictions on media use and co-viewing TV and DVDs with parents and friends associated with teenagers’ early sexual behaviour?   Journal of Adolescence . 2013;36(6):1121-1133. doi:10.1016/j.adolescence.2013.08.019

The impact of interactive violence on children . U.S. Senate Hearing 106-1096. March 21, 2000.

Anderson CA, Dill KE. Video games and aggressive thoughts, feelings, and behavior in the laboratory and in life .  J Pers Soc Psychol. 2000;78(4):772-790. doi:10.1037/0022-3514.78.4.772

Collins RL, Elliott MN, Berry SH, et al. Watching sex on television predicts adolescent initiation of sexual behavior . Pediatrics . 2004;114(3):e280-9. dloi:10.1542/peds.2003-1065-L

By Kendra Cherry, MSEd Kendra Cherry, MS, is a psychosocial rehabilitation specialist, psychology educator, and author of the "Everything Psychology Book."

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 13 March 2024

Observational reinforcement learning in children and young adults

• Julia M. Rodriguez Buritica   ORCID: orcid.org/0000-0003-4232-2238 1 , 2   nAff10 ,
• Ben Eppinger 1 , 3 , 4 , 5 ,
• Hauke R. Heekeren 1 , 6 ,
• Eveline A. Crone 7 , 8 , 9 &
• Anna C. K. van Duijvenvoorde   ORCID: orcid.org/0000-0001-9213-8522 8 , 9  

npj Science of Learning volume  9 , Article number:  18 ( 2024 ) Cite this article

3 Altmetric

Metrics details

• Social neuroscience

Observational learning is essential for the acquisition of new behavior in educational practices and daily life and serves as an important mechanism for human cognitive and social-emotional development. However, we know little about its underlying neurocomputational mechanisms from a developmental perspective. In this study we used model-based fMRI to investigate differences in observational learning and individual learning between children and younger adults. Prediction errors (PE), the difference between experienced and predicted outcomes, related positively to striatal and ventral medial prefrontal cortex activation during individual learning and showed no age-related differences. PE-related activation during observational learning was more pronounced when outcomes were worse than predicted. Particularly, negative PE-coding in the dorsal medial prefrontal cortex was stronger in adults compared to children and was associated with improved observational learning in children and adults. The current findings pave the way to better understand observational learning challenges across development and educational settings.

Similar content being viewed by others

A methodological perspective on learning in the developing brain

Heterogeneity in strategy use during arbitration between experiential and observational learning

The rational use of causal inference to guide reinforcement learning strengthens with age

Introduction.

Numerous findings indicated that others have a strong impact on learning and decision-making 1 , 2 , 3 , 4 , 5 , 6 . For instance, we may be swayed by the opinions of others to adjust our norms on acceptable behavior. Alternatively, others could be considered a source of information, as they can provide us with valuable information about our environment. Learning from observing others’s behaviors and outcomes may have benefits in dangerous or novel environments, in which observational learning allows us to learn from the actions and outcomes of others without having to engage in these (potentially hazardous) behaviors ourselves. Social situations lend themselves well for observational learning. That is, learning from others is ubiquitous on playgrounds, in schools and other social environments, in which we have the opportunity to observe others behaviors and subsequent outcomes without necessarily participating ourselves. However, compared to learning from own experiences, the developmental mechanisms underlying observational learning are poorly understood.

Several studies have examined how learning from own outcomes changes with age 7 , 8 . Many studies observed that adults typically outperform children during instrumental learning. That is, adults learn faster than younger ages and choose the most rewarding option more often. This is thought to be related to a developmental improvement in cognitive control, including behaviors such as sustained attention and working memory, which would benefit learning. It is yet unclear how a social observational context influences this adult advantage. That is, children are shown to be highly sensitive to the example of others, quickly copying behavior, particulary of their own peers 9 . On the other hand, a previous observational learning study showed that children may not process and use information of others as efficiently as young adults in their learning and decision making 10 . These age-related differences in observational learning have been related to differences in temporal processing of observed outcomes using electro-encephalogram (EEG) 9 , 10 . Children, ages 8–10, showed larger electrophysiological responses when observing peers as compared to adults 5 , but compared to adults their electrophysiological responses did not change according to their learning 6 , and they could not yet benefit in accuracy from observed information as much as adults. What remains unresolved in the current literature are the functional and computational processes supporting observational learning as compared to individual learning across development. In this study, we therefore applied a model-based neuroimaging approach to observational learning across development, by combining reinforcement learning (RL) modeling and functional magnetic resonance imaging (fMRI) in a children’s and adult age group.

Computational models of reinforcement learning haves been successfully applied to understand reinforcement learning in adults and children 5 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 . In these models, learning is driven by prediction errors, which reflect the mismatch between expected, Q a (t), and received outcomes, r(t), per trial (t). Whenever an outcome is better (worse) than expected, the model will generate a positive (negative) prediction error. Prediction errors are thought of as important learning signals, and are shown to scale with activity of midbrain dopamine neurons 19 , 20 , 21 , 22 . The weight in which prediction errors drive learning behavior is quantified by learning rates. High learning rates allow prediction errors to quickly change the value of choice options. Low learning rates result in slower updating and therefore a long-term integration of outcomes in the value of choice options. In addition, RL models typically include a temperature parameter that specifies how precisely one’s choices discriminate between the value of choice options, and thus controls the specificity in choice behavior. Previous research involving reinforcement learning models has yielded mixed results regarding age-related differences in positive and negative learning rates 8 . Recent reviews, however, indicate that choice specificity tends to consistently increase with advancing age 7 , 8 .

In contrast to individual learning situations, observational learning is a richer learning environment. For instance, behavior can be updated twice: First, by an experiential prediction-error (as in standard reinforcement learning) and second by an observational prediction-error. Where the first is generated by one’s own outcomes, the latter is generated by the outcomes of an observed other. Neuroimaging studies examining the underlying neural mechanisms have suggested neural systems may be partly overlapping and partly specific for learning from social and non-social outcomes 5 , 16 , 23 , 24 . That is, whereas prediction errors in various learning paradigms have been represented in brain regions such as the ventral striatum and ventral medial prefrontal cortex (vmPFC) 11 , 25 , 26 , 27 , prediction errors in other brain regions may be more specialized for social learning. For instance, social learning has been related specifically to the dorsal medial PFC (dmPFC) 28 , 29 and anterior cingulate cortex (ACC) 11 , 12 with the latter representing other-related reward-processing, and information on the consequences of actions of others 24 , 30 . In addition, social learning has been related to brain regions involved in mentalizing and modeling of others, such as the dmPFC, temporal-parietal junction (TPJ), and posterior superior temporal sulcus (pSTS) 23 . Recent studies highlighted the role of mentalizing particularly in situations in which individuals are confronted with opposing preferences and in which the mental state (goals, preferences, beliefs, or intentions) of another person needs to be inferred 14 , 28 , 31 . This mentalizing network is also thought to track trial-level updating during strategic social interactions 12 , 32 .

Until now, few studies have examined the computational and neural correlates of age-related differences in social and non-social reinforcement learning. We build on this previous work and extend this with the current study: Here, we aim to examine age-related differences in the neural correlates of observational and individual prediction errors. Including both observational and individual learning allows us to examine age-related differences in both learning situations, as well as to compare these learning situations directly. To do so, we used functional magnetic resonance imaging (fMRI) and a computational modeling approach. Thirty children (8–10-year-olds, 18 female) and 30 young adults (18–20-year-olds, 16 female) participated in this study. For the youngest ages we focused on middle childhood, including children between 8 and 10 years. In this way we captured an important phase of age-related differences in controlling responses towards positive and negative feedback 33 , and we built on previous EEG work on observational learning 9 , 10 .

Before the start of the experiment, participants met an age-matched and same-sex peer who was introduced as the observed other player in the task. Participants performed a probabilistic observational learning paradigm 9 , 10 , 11 (see Fig. 1 ). In this task, participants made repeated choices between pairs (i.e., 8 in total per pair) of stimuli, where one stimulus was associated with a high probability of reward (80% gains, 20% losses) and the other was associated with a low probability of reward (20% gains, 80% losses). Before participants could choose, they observed the other player choosing between the same two stimuli. We manipulated the amount of observable information of the other player across two learning conditions with each three runs (each with 2 stimulus pairs per learning condition; thus 32 trials per run): In the individual learning condition participants received no information about the behavior of the other individual; in the observational learning condition they could observe the other player’s actions and outcomes before making their own choice. With this setup, we can better understand how observed information of others is used for one’s own learning. The four stimulus pairs (two per condition) per run were presented intermixed. To better understand and explain age-related differences in learning in this task, we used a double update Q-learning algorithm 11 that captures learning from both other’s and own outcomes. Note that just in the observational learning condition, Q-values could be updated in the observational stage as in the individual learning no outcome was displayed (see Fig. 1 ; and Methods). Here, we included two independent learning rates (α pos , α neg ), separately for each valence (positive, negative) of the outcome for both learning conditions. A higher learning rate means a quicker updating of expected values and thereby faster learning. For both stages and conditions, the probability of choosing option a from a stimulus pair (ab) was computed using an inverse softmax function 34 (see Methods for further details), were we estimate the inverse temperature ( β )) indicating the specificity of the subject to differences in Q-values. Since β is the inverse temperature, this means that higher values indicate a less deviations from optimal choice behavior (i.e., a greater choice specificity). For the fMRI analyses we used trial-level calculated prediction errors when presented with own (action phase in IL condition; Fig. 1 ) and others (observational phase in OL condition; Fig. 1 ) outcomes derived from the median of the parameter estimates (i.e., α pos , α neg and β ) per age group (Fig. 2 ). Using a model-based parametric fMRI approach, we examined which brain regions correlated with prediction error activation.

Example of the trial procedure for the observation learning and individual learning condition.

Given our previous developmental work, we expected that both age-groups benefitted from the additional observational information, although adults are expected to outperform children in both conditions and are expected to learn faster than children 10 . Based on work in adults, we expect that RL-models can be used to describe individual and observational learning across development, although during observational learning choice values will be updated twice 11 . Given previous mixed findings on age-related differences in learning rates, we had no clear predictions here, but expected that the inverse temperature parameter is sensitive to age-related differences 8 . Based on previous developmental work, prediction-error activation of own and observed outcomes is expected to be related to striatal and medial prefrontal cortex activation across development 9 , 10 . Observational learning is expected to relate to the medial PFC including the ACC and dmPFC 5 , 11 , 12 , 23 , 24 , 30 , and mentalizing regions such as the TPJ 14 , 28 , 31 . Finally, whereas both age groups may benefit from observational compared to individual learning, we expected greater neural differentiation between own and observed outcomes in adults compared to children 10 , 35 , reflective of greater efficiency of learning which is increased in observational situations. Taken together, we include a computational approach to examine the neural associations and age-related differences in observational versus individual learning between children and adults.

Performance differences in observational and individual learning

One participant (child) did not finish the learning session in the scanner and was not included in further analyses (see Methods). Both age groups selected the correct option (rewarded 80% of the time) more frequently than the incorrect option (rewarded 20% of the time) and showed accuracies above chance level in both learning conditions (four t- tests against chance level per condition and per age group; all p ’s <0.001). Performance was correlated between the observational and individual learning condition (Persons’ r  = 0.46, p  < 0.001), indicating that individuals who performed well in the observational condition, also performed well in the individual learning condition. Descriptively, 47 (24 children and 23 adults) out of 59 subjects performed more accurately in the observational than the individual condition, suggesting that most participants benefitted from the additional observable information.

To test age group and condition effects on learning we analyzed choice behavior (averaged across runs) using a mixed-effects general linear model (controlling for intelligence) with the between subject predictor age group (children, adults) and the within subject predictors learning condition (individual, observational) and trial number (1:8), as well as all interaction terms (see Methods). The results of this analysis (see Supplementary Table 1 for full model output and Fig. 2a for visualization) showed significant main effects of learning condition (β = 0.12, t  = 5.2, p  < 0.001) and trial number (β = 0.04, t  = 3.0, p  = 0.005), as well as a significant learning condition x trial number interaction (β = −0.04, t  = -2.2, p  = 0.026). Post-hoc comparisons per condition revealed that participants improved across trials in the individual learning condition (β = 0.06, t  = 6.3, p  < 0.001), while performance remained stable in the observational condition (see Fig. 2a ; β = 0.004, t  = 0.43, p  = 0.671). This suggests that learning from observing others resulted in relatively high-performance levels early in the learning process, while individual learning showed a more gradual improvement across trials. We also found age-related differences: including a significant main effect of age group (β = 0.1, t  = 3.7, p  < 0.001), as well as a significant age group x trial number interaction, β = 0.04, t  = 2.2, p  = 0.031. Post-hoc comparisons per age group revealed that adults improved across trials (β = 0.05, t  = 4.6, p  < 0.001), while children showed limited improvement (see Fig. 2a ; β = 0.18, t  = 1.75, p  = 0.08). This indicates that adults generally demonstrated greater accuracy improvement across trials compared to children. No significant age by learning condition interactions was observed ( p  = .9). Thus, altough adults overall outperformed children, the age groups benefitted to a similar degree from the additional observable information (see Fig. 2a ).

To ensure that differences in learning conditions were not solely due to varying information levels provided to participants (i.e., more information in observational than individual learning condition), we conducted an additional mixed-model analysis. This analysis focused on an equal amount of information in each learning condition, comparing trials 1–4 in the observational learning condition (including 1/3/5/7 observed/received outcomes) to trial 2,4,6, and 8 in the individual learning condition (including 1/3/5/7 received outcomes). The results align with our main analysis, indicating better performance in the observational learning condition compared to the individual learning condition (main effect condition: β = 0.06, t = 5.03, p  < 0.001), across both age groups (age x condition interaction: p  = 0.8).

a Proportion correct choice (Pc). Pc displayed separately for the two age groups (adults, children) and learning conditions (individual (IL), observational (OL)). Data were averaged into four bins (across eight trials). Error bars reflect the SEM. Grey lines reflect individual learning differences between IL and OL condition. b Parameter estimates of the best fitting model across age groups. Learning rate (i.e., alpha pos and alpha neg) and inverse temperature (i.e., beta) per age group and condition (individual learning (IL) and observational learning (OL)).

Computational parameter differences in individual and observational learning

Next, we examined age and learning condition effects on computational parameters (i.e., α pos , α neg and β ; see Methods). Higher learning rates indicate that recent choice outcomes have a stronger effect on future choices than less recent choice outcomes. The inverse temperature indicates participants’ sensitivity to differences in these choice values. Here, the higher the beta parameter, the less stochastic choice behavior was.

Robust mixed-effects analyses were used to test main and interaction effects of age group, condition (IL, OL), and valence (Pos, Neg; only for learning rates; see Fig. 2b and Supplementary Table 2 ). Findings showed that positive and negative learning rates differed significantly across learning conditions (β = 0.75, t  = 3.4, p  < 0.001, main effect of valence) with both learning rates were higher for positive (median = 0.78) than negative outcomes (median = 0.2). No significant main effect of condition, age, or any interactions with age and condition were observed (all p’ s > 0.7). A similar analysis on the inverse temperature parameter (see Fig. 2b and Supplementary Table 3 ) showed that participants followed choice values more optimally in the observational (median = 2.14) than individual learning (median = 1.03) condition (main effect condition, β = 0.95, t  = 7.4, p  < 0.001). A main effect of age group showed that children (median = 1.13) were less optimal in their choice behavior than adults (median = 2.18; β = 0.91, t  = 4.5, p  < 0.001). No age x condition effect was observed ( p  = 0.9).

Age group and condition differences in prediction-error activation

We analyzed neural activation related to prediction error coding (entered as a signed continuous trial-by-trial predictor) when participants received their own outcomes (Fig. 1 , individual learning condition action phase) and others’ outcomes (Fig. 1 , observational learning condition observational phase). We initially examined whether there were brain regions that responded differently to prediction errors in individual learning and observational learning conditions. Results showed that individual compared to observational prediction errors were more strongly related to activation in the vmPFC ( p FWE’s < 0.05), the left lateral PFC ( p FWE’s < 0.05), the bilateral striatum, and bilateral parietal cortex ( p FWE’s < 0.001; see Fig. 3 ; and Supplementary Table 4 ). No brain regions correlated stronger to observational than individual prediction errors (all p FWE’s > 0.05).

Results are displayed at Family-Wise Error (FWE) cluster-corrected p  < 0.05, with an initial cluster forming threshold of p  < 0.001. For visualization we extracted the beta-values from the whole-brain condition effect from regions of interest. Since our activation spanned multiple subcortical anatomical regions, the functional activation from the subcortical cluster is overlaid with a nucleus accumbens (ventral striatum) anatomical mask.

Next, we examined with a whole-brain ANOVA whether this difference in prediction error activation between individual and observational learning conditions was sensitivity to age group differences. An age group × learning condition interaction was observed in the left TPJ/inferior parietal cortex ( p FWE < 0.05; learning condition (IL > OL) × age (adults > children) contrast). Follow-up tests per age group showed that the TPJ/ inferior parietal cortex differently responded to individual and observational prediction errors in adults compared to children: That is, prediction errors related to increased TPJ activation when other’s outcomes were worse than expected and to increased activation when own outcomes were better than expected in adults ( t (29) = 5.07, p  < 0.001). This differentiation was not observed in in children ( p  = 0.15; see Fig. 4 ; and Supplementary Table 5 ).

Results are displayed at Family-Wise Error (FWE) cluster-corrected p  < 0.05, with an initial cluster forming threshold of p  < 0.001. For visualization we extracted the beta-values from the whole brain interaction effect of learning condition × age. Significant differences of follow-up tests are marked here (*** p  < 0.001).

Age effects in observational and individual prediction error activation

We first used a whole-brain F -test ( p FWE < 0.05). to examine both positive correlations with prediction error activation (i.e., regions where activation increased with larger positive prediction errors) and negative correlations with prediction error activation (i.e., regions where activation increased with larger negative prediction errors) within each learning condition. The results revealed that in the observational learning condition, prediction errors were associated with activation in the right lateral PFC, right inferior parietal, and right insula (see Fig. 5a and Supplementary Table 6 ). Similarly, in the individual learning condition, a comparable whole-brain analysis indicated that prediction errors were positively correlated with activation in the ventral medial prefrontal cortex, striatum, and parietal cortex (see Fig. 5b and Supplementary Table 7 ).

a observational PE activation and b individual PE activation, displayed based on a whole-brain F -test (pFWE <0.05) Observational PE activation showed negative correlations with brain activation (i.e., greater negative PEs resulted in larger activation, see bar charts of extracted beta-values Supplementary Fig. 1 ). Individual PE activation showed positive correlations with brain activation (i.e., greater positive PEs resulted in larger activation, see bar charts of extracted beta-values Supplementary Fig. 1 ).

Since our focus was on age differences, we then compared age groups in prediction error-associated brain activation for the observational and individual learning conditions. A whole-brain t -test comparing children versus adults revealed that observational prediction errors were more strongly represented in adults than in children in the dmPFC, dorsolateral PFC, right inferior parietal cortex, and right insula (see Fig. 6 and Supplementary Table 8 ). As can be seen in Fig. 6 , more negative prediction errors resulting in stronger PE-related activation. We also compared age groups in the IL condition, but no significant age differences were observed.

This figure displays age-group differences in observational PEs with a whole-brain t -test at qFDR  < 0.05 with a primary voxel-wise threshold of p  < 0.001. For visualization we extracted the beta-values from the whole-brain effects per age-group. Error bars reflect the SEM.

Brain-behavioral relations: relating learning performance to prediction error activation

Lastly, we investigated the extent to which age differences in learning behavior were associated with prediction-error-related brain activation. To achieve this, we extracted parameter estimates from clusters responsive to both individual and observational learning conditions (as shown in Fig. 5 ), as well as clusters indicating an age-by-condition interaction (i.e., TPJ; as shown in Fig. 4 ). We employed a linear regression analysis for each region of interest to examine the relationship between prediction error activation and learning accuracy in both individual learning and observational learning conditions. Additionally, we included interactions with age groups to assess whether this relationship varied across different age groups (further details can be found in the Methods section and Supplementary Table 9 ). To account for multiple linear regressions, we applied an FDR correction across all behavior-brain regression analyses.

Learning accuracy in the observational learning condition was found to be correlated with observation prediction-error activation in the dmPFC (β = −1.12, t = −3.79, p  < 0.001; see Fig. 7 ) and the right dlPFC ( p  = 0.037; see Supplementary Fig. 2a ), but not with the right inferior parietal cortex or the right insula ( p ’s > 0.596; see Supplementary Table 9 for further details). Only the brain-behavior relationship in the dmPFC (not dlPFC) remained significant after multiple comparison correction. This association between dmPFC activation and task performance was consistent across age groups (performance × age group: p  = 0.558). Thus, improved learning in the observational learning condition was linked to a stronger prediction-error response in the dmPFC (see Fig. 7 ) for both children and adults. It is important to note that individual learning performance was included as a covariate in this regression analysis, indicating that this association appears to be specific to observational learning performance (Supplementary Table 9 ).

Scatter plot showing that more negative PE-related activation in the dmPFC was related to better performance during observational learning for both age groups. Shaded areas reflect 95% confidence intervals.

Learning accuracy in the individual learning condition, after controlling for observational learning performance, was associated with individual prediction error activation in the left parietal cortex (β = 0.98, p  = .007; see Supplementary Fig. 2b ). However, this relationship did not survive multiple comparison correction. No significant relationships were observed between individual learning accuracy and activation in the vmPFC or striatal regions, nor in the right inferior parietal/TPJ ( p ’s > 0.076; Supplementary Table 9 ).

In this study we examined the behavioral and neural mechanisms underlying observational and individual reinforcement learning in 8–10-year-old children and young adults. Overall, we found that adults compared to children showed faster learning, better performance, and they were more optimally following the value of choice options across learning conditions (see Fig. 2 ). As expected, both age groups benefitted from observing other’s choices and outcomes. However, in contrast to our expectations, adults and children benefited to a similar degree from observing others behavior during learning and learning in observational versus individual conditions did not vary across development. To better understand the computational and neurobiological processes underlying learning we used reinforcement learning (RL) model in combination with fMRI. We observed that behavioral updating (as indicated by learning rates derived from the RL model fitting) did not differ across condition, and age, and both conditions and age groups showed higher learning rates from positive compared to negative feedback in both learning conditions. Choice behavior in children was generally less value-driven than in adults, and therefore showed more random, or potentially more exploratory, choice behavior. In addition, choice behavior was more value-driven in the observational than individual learning. Model-based parametric fMRI analyses complemented these behavioral insights. Observational and individual prediction errors were reflected in partly distinct (vmPFC, striatum), and (for adults) in partly overlapping brain regions, such as the temporal-parietal junction. Age-related differences were observed in observational prediction error coding in the dorsal medial prefrontal cortex, dorsal-lateral prefrontal cortex, parietal cortex, and insula cortex; regions closely related to cognitive control, social cognition, and social learning 24 , 30 . Moreover, only observational prediction error signals in the dmPFC correlated with observational compared to individual learning performance in both children and adults. This finding underlines the relevance of the dmPFC when learning from others.

Our behavioral results supported previously observed age-related differences in instrumental learning. In line with other developmental studies adults outperformed children in both individual and observational learning conditions. Children learned slower than adults 10 , 36 , 37 , 38 , 39 , 40 and their choice behavior was more stochastic and less value-driven 41 , 42 . This decrease in stochasticity across age is one of the most consistent findings observed in reinforcement learning 7 , 8 . Potentially, this may relate to an increase in maintaining sustained attention, or improved working memory, accelerated by a developmental improvement in cognitive control. Alternatively, children may be more stochastic because they are more explorative than adults. Theoretical work suggested that higher exploration provides children with more learning opportunities and could allow them to quickly discover changes in environments 43 . With the current task design, stochasticity and exploration cannot be easily distinguished. Whether age-related differences in stochasticity are valuable for exploration in social- and non-social learning environments, should be addressed in future studies. Interestingly, our behavioral findings indicate that both children and adults can learn from observing other’s outcomes: All ages increase their updating after positive and negative feedback during observational learning, and all age groups benefit in performance in observational compared to individual learning. The adults’ advantage in reinforcement learning may therefore be reflected in more value-driven choice behavior.

Our brain-based results complement and extend these behavioral findings. First, our results showed age-related differences in the temporal-parietal region. Unexpectedly, this region was sensitive to observational and individual prediction errors in adults (although differently signed), but not in children. In adults, prediction errors correlated positively with brain activation in the TPJ during individual learning (indicating greater positive prediction errors resulted in greater activation), and negatively with brain activation during observational learning (indicating greater negative prediction errors resulted in greater activation). This supports findings supports prior findings that the TPJ is associated with other’s and one’s own associations 44 , self-other distinctions 45 and that it can be linked in a valence-specific way to social and self-related prediction errors. However, in the current study prediction error related activation in the TPJ did not relate to age-related differences in behavioral performance. Future research will therefore need to replicate and extend these findings on the role of the TPJ in instrumental learning.

Second, our findings showed that observational prediction errors were related to neural activation in frontal regions including the dorsal medial PFC and dorsolateral PFC. and that that these relations were stronger for adults compared to children. Our finding that the medial prefrontal region is related to observational prediction error signaling in adults and children concurs with a broader framework, which links the ACC and dmPFC to a variety of social learning and decision-making skills 24 , 30 , such as outcome prediction error for confederate’s advice 31 , mentalizing 32 , 46 , and egocentric and allocentric outcomes of social decisions 24 , 47 . It has also been suggested that the role of different sub-regions of the mPFC may be more specific than previously recognized, and a more rostral part of the anterior cingulate cortex gyrus and dorsomedial prefrontal regions could be particularly specialized for observational prediction errors 5 , 23 , 24 , 28 , 48 , 49 . In addition, we observed that prediction-error activation in dmPFC related to performance in the observational learning condition in both children and adults (see Fig. 7 ). This may indicate an important role of the dmPFC for behavioral improvements in observational learning across age, although this does not necessarily relate to the observed age-related improvement in instrumental learning. Interestingly, prediction-error activation in the observational learning condition mostly increased when outcomes for others were worse than expected (see also 11 ). This finding may be interpreted in multiple ways. Possibly, the current setup led to an experienced competitiveness between the observer and observed individual, although it was explicitly instructed that participants were neither in competition nor dependent on the behavior of the observed other (e.g. 11 , 50 ,). Alternatively, seeing others lose may simply be a stronger learning signal. These directional effects were present across frontoparietal regions, indicating that these regions may be jointly involved in social comparison processes.

Finally, individual learning situations resulted in specific prediction-error related activation in the vmPFC, the left lateral PFC, the bilateral striatum, and bilateral parietal cortex (see Fig. 3 ) 10 , 35 , 51 . This supports previous findings and highlights there may be limited age-related differences per se in prediction error activation 27 , 40 . Whether adolescence is a period of heightened activation in reward-related learning environments as suggested by other studies 52 , 53 , 54 , is something that future research using samples with a wider, and more continuous, age range should disentangle.

Although our results confirmed several age-related findings from previous studies on individual and observational learning 9 , 10 , some of the results were unexpected. For instance, in contrast to previous findings on observational learning 10 , we did not find age-related performance differences between learning conditions. It seems viable that differences to results of previous studies arise because of the administered paradigm to make the task amenable for fMRI. For instance, in the previous EEG-studies the task was more complex in terms of learning conditions (i.e., another third condition was included) and more difficult as the timing was faster (i.e., no jitters were included). Moreover, when administering a social and non-social learning conditions an important question is to what extent findings are specific to social learning or reflective of more general learning processes 48 . In our experimental setup children and adults learn indirectly from others. To create a social learning setting, we informed participants that they could learn from another participant they met before the experiment, and they observe the other’s photo and name during the experiment. Varying the age of the other participant, in a previous study 9 , using the same social learning setting and task, we demonstrated that similarity in age (same-aged child vs. young adult) between the observed player and the observer influenced both behavioral and neural responses in 8-10 year olds. Building on these findings and because participants judged others as being similar to themselves (see Methods) we argue that it is likely that the participants perceived the observational learning conditions as social learning conditions rather than conditions in which they simply received only more information. Based on these considerations we would like to a few recommendations for future developmental studies on observational learning:

First, future studies should take the social context into account in which children learn “indirect” observational learning (e.g., observing others’ choices and outcomes in absence of the other player) in comparison to direct observational learning (e.g., “directly” observing others’ choices and outcomes in presence of the other player). Secondly, it is important to compare active individual learning to observational learning, where participants learn purely passively from observed information (i.e., without the ability to evaluate this information immediately after the observational phase on a trial-by-trial basis, as was done in this study). This allows for a more direct comparison of both learning processes based on a similar amount of information, which is not achievable in the current task design. Finally, we observed that our age groups differed in fluid intelligence. Children showed higher age-normed intelligence scores than adults, which may have influenced age-related differences in performance and neural activation. Although controlling for intelligence in our behavioral analyses did not change our main findings, future studies may address to what extent task difficulties and intelligence relate to age-related differences in observational learning. For instance, learning condition differences associated with prediction error activation might be related to greater task difficulty and higher cognitive load when learning from own outcomes. This could be linked to higher neural activation as task difficulty increases 55 , 56 , 57 .

To summarize, our findings show that learning from the outcomes of others, particularly when outcomes were worse than expected, was related to neural activation in the dmPFC, dlPFC and temporal-parietal activation which was more pronounced in adults than children. Here, only the dmPFC related to performance in observational learning for both children and adults. These findings confirm and extend the functional role of the medial PFC to a social observational learning context and specify its functional relevance for social learning for children and adults.

Participants

Thirty 18–20-year-old adults and 30 8–10-year-old children participated in the study. Data of one child was excluded due to inability to complete the task. For occasional occurring head motion (Framewise displacements >0.5) volumes with motion were flagged and were not included in regressors of interest but instead modeled by nuisance regressors (i.e., censored), number of censored volumes regressors varied between 2–12 (<10% of volumes). The final sample consisted of 30 adults (Age Mean (SD) = 19.45 (0.86); 16 female) and 29 children (Age Mean (SD) = 9.71 (0.89); 18 female). All participants were right-handed, had normal or corrected-to-normal vision, were screened for MRI contra-indications, and had no neurological or psychological disorders. Prior to the experiment we obtained informed written consent from the participants and both parents (in case of children). The study was approved by the Ethics Committee of the Leiden University Medical Center (LUMC). All anatomical MRI scans were reviewed and cleared by a radiologist from the radiology department of the LUMC. No anomalous findings were reported.

Subjects participated in one experimental session in which we assessed psychometric covariate measures not specific to the current study, and observational learning performance inside the MR scanner. Participants were recruited through local advertisements and received a compensation. Participants’ intelligence (IQ) was estimated with the subsets ‘similarities’ and ‘block design’ of the Wechsler Intelligence Scale for Children, third edition (WISC-III 58 ). For both age groups, estimated IQs were in the normal to high range (Adults: Mean IQ (SD) = 106.67 (8.82), Range = 87.5–122.5; Children: Mean IQ (SD) = 112.76 (11.96), Range = 87.5–132.5). IQ did differ between age groups: Children showed higher IQ scores than adults ( F (1, 57) = 4.952, p  = 0.03, η p 2  = 0.08). IQ is controlled for in all behavioral analyses.

Experimental design

We used a probabilistic reward-based observational learning paradigm 9 , 11 (see Fig. 1 ; controlled by PsychToolBox-3 59 ). Participants were asked to choose one out of two abstract stimuli 60 . One stimulus was associated with a high probability of receiving reward (80% gains, 20% losses) and one associated with a low reward probability (20% gains, 80% losses). Before participants could choose, they observed an age- and gender-matched peer (who they met before starting the task) choosing between the same two abstract stimuli. Participants were told that the other player had already performed the task and that they could observe the recorded choices. In order to assess the credibility of our social manipulation we assessed participants perception of the other player at the end of the experiment. Almost all participants (93%) reported that they paid attention to the other player and on a 7-point Likert-scale (see Supplementary Fig. 3 ), that watching the other helped them for learning and that they judged the other as highly similar to themselves, reliable and believable ( p ’s < 0.001; one-sampled Wilcoxon tests for non-parametric data against 3.5). Age groups did not differ in their rating (unpaired Wilcoxon tests for non-parametric data; p ’s > 0.06). However, unbeknownst to participants the observed choices were computer generated using a RL model (see Supplementary Fig. 4 ). The computer-controlled behavior of the model players was associated with the objective percentage of probabilistic positive or negative outcomes associated with each of the stimuli (see supplementary files 9 , 10 ). The amount of observable information of the other player was manipulated across two learning conditions: Observational learning (observing both, the other player’s actions and outcomes; short OL) and individual learning (observing neither actions nor outcomes of the other player; short IL). In each condition, the trials followed the general structure of an observational phase that was followed by an action phase. That is, in the OL condition the participant would be first presented with a fixation cross for a variable amount of time per trial (not shown in Fig. 1 ). This jitter varied exponentially from 1 s to 8 s and was followed by a picture of the other person (i.e., precue of 1 s) and the presentation of a stimulus pair. By pressing a button with the ring finger of the right hand the other’s choices were revealed. Responses had to be given within a 2 s window and indicated by a white selection frame (i.e., 2 s – response time; see Fig. 1 ), which was followed by a 1 s outcome display representing the outcome of the other’s choice. Then, the action phase started with a fixation cross (i.e., jitter of 1–8 s) and then participants viewed his/her own picture (1 s), after which the same stimulus pair was presented. Participants could choose either the left or the right stimulus by pressing a button with the index or middle finger of the right hand. Again, responses had to be given within a 2 s window, which was followed by a 1 s outcome display. If no response was given within 2 s, in either the observational or the action phase of any condition the words “too slow” were presented on the screen. This happened rarely for adults ( M trials = 2.41, SD  = 2.27) and children ( M trials = 6.89, SD  = 5.54).

The IL condition followed the same timing and structure as the OL condition and participants also pressed a button when stimuli were presented in the observational phase, yet no choice (i.e., in this case both possible choices were surrounded by a white selection frame) or outcome information of the other were presented. Each condition was associated with two unique stimulus pairs per run and the order of conditions was mixed. In total, every pair was presented for eight trials per run (resulting in 32 trials and four abstract stimulus pairs per run). The four stimulus pairs (two per condition) per run were presented intermixed. Participants played in total three runs of approximately nine minutes each with 48 trials per learning condition (resulting in 96 trials with 16 abstract stimulus pairs in total). For five participants one run was excluded in further analyses because of high occasional motion (>10% of censored volumes per run). Participants were instructed to earn as many points as possible (as indicated by receiving a positive outcome signal) but were also informed that it was not possible to gain points on every trial, clarifying the probabilistic nature of the task. Before performing the task in the MRI, participants practiced the task for one run length.

Reinforcement learning models

Learning action-outcome-contingencies can be computationally captured using RL models 18 . During RL learning the discrepancy between expected outcome on trial (t), Q a (t), and the actually received outcome on trial (t), r(t), is called prediction error (PE):

If outcomes are better (worse) than expected, the model will generate a positive (negative) PE, which is used to increase (decrease) the predicted value, Q a ( t ), associated with the chosen option a in the current trial t . Positive and negative outcomes received in each trial are used to update the predicted values of both options a (Eq. ( 2 )) and b (Eq. ( 3 )) 61 , 62 :

Thus, for the value of the unchosen option b (Eq. ( 3 )) the counterfactural outcome is taken into account. The impact of the PE’s on forming new expectations is scaled by the learning rate α . A high learning rate (~1) indicates that a new experience (i.e., PE) has a stronger impact on future predictions whereas a low learning rate (~0) means that a PE only weakly influences the expected value and thereby choice behavior. Based on previous developmental studies, we included two independent learning rates for positive (α pos ) and negative outcomes (α neg ) to describe the learning behavior in the individual condition across development 40 , 63 .

For the OL condition, we extended the RL-algorithm used in the IL condition to describe social influences during learning: Learning from other’s choices and outcomes (see supplementary files for model comparisons and model recovery) was best captured by using a dual-update model, one updating phase for the observational (Eq. ( 4 )) and one for action stage (Eq. ( 5 )) as described in 64 .

Similarly, to the RL models used in the individual condition we included two independent learning rates (α pos , α neg ) in the OL condition. A higher learning rate thus means a quicker updating of expected values and thereby faster learning. For both stages and conditions, the probability of choosing option a from a stimulus pair (ab) was computed using a softmax function (Eq. ( 6 )) 34 :

The probability of selecting option a is influenced by the expected value Q of option a in trial t divided by the sum of the expected values of all possible options ( a and b ). The β parameter in this equation reflects the sensitivity of the subject to differences in expected value. Here, a lower β parameter indicates more stochastic responding.

For model comparisons we evaluated a set of alternative RL-algorithms (i.e., baseline RL-model with one learning rate (α), model with separate learning rates for positive (α pos ) and negative outcomes and (α neg ) 40 , 65 and a model with separate learning rates for observational and action stage; see supplementary files, Supplementary Table 10 and Supplementary Fig. 5 ). All learning rates (α, α pos and α neg ) and the noise parameter ( β ) per condition were individually estimated by fitting the model predictions to participants’ choices (see supplementary files for further details on the model fitting procedure). The β parameters were fit with constraints between [0 5]. The α parameters were constrained between [0 1]. For model selection purposes, we computed the Bayesian information criterion (BIC) across all subjects for the different models, where lower BIC values indicate better fit (see ref. 66 ). For both learning conditions (i.e., IL and OL) the best fitting model across all subjects included two independent learning rates (α pos , α neg ) and the noise parameter ( β ) (see supplementary files for further details and S2 for an overview of all model comparisons and parameter estimates per model). For the model-based fMRI analyses we used the median parameter estimates per age group (Fig. 2 ) of the best fitting model per learning condition to calculate trial-by-trial PEs 11 , 62 , 67 , 68 , 69 . In imaging analyses PEs were scaled and mean-centered.

To explore the validity of the RL models and the model selection procedure, we performed 1) model and 2) parameter recovery analyses (see supplementary files and Supplementary Fig. 6 for further details). As part of quality control, we further performed simulations using the individual parameter estimates for each subject for the best fitting model. The simulations indicated that the best fitting model per condition was able to capture learning on a trial-by-trial level in each age group (see supplementary files for further details and Supplementary Fig. 7 ).

Behavioral data analysis

Choice behavior was analyzed using a mixed-effects generalized linear model as implemented in the lme4 package in R 70 . Accuracy (proportion of optimal choice) was averaged across runs and modeled using the between-subjects predictor age group and the within-subjects predictors learning condition (IL, OL) and trial number (1:8).

Mixed effects model formula:

We treated the between-subjects predictor age group as a fixed effects factor, whereas all within-subjects’ predictors of interest were treated as fixed and random effects at the individual subject level. We additionally included intelligence as predictor to control for age differences in intelligence. The categorical predictors were contrast-coded, the continuous predictor trial was mean-centered. Regression weights (beta values), z-values and corresponding p-values are reported (see Supplementary Table 1 ).

The learning rates and inverse temperature of the best fitting models per conditions across age were significantly non-normally distributed ( p ’s < 0.05). Thus, we used robust mixed effects model to test for age and condition effects on learning rates and inverse temperature, respectively as implemented in the robustlmm package (see Supplementary Tables 2 and 3 ). Note that due to convergence problems we use a simpler fixed and random effect structure in these models (see Eq. ( 8 )).

Mixed effects model formula (example):

Finally, we assessed whether behavior per condition (i.e., accuracy) related to condition-specific prediction-error activation. We used multiple linear regression models predicting PE activation by proportion of optimal choice (accuracy), age, and their interaction, controlling for learning in the other condition, and intelligence. The categorical predictors were contrast-coded, the continuous predictor trial was mean-centered. Regression weights (beta values), z-values and corresponding p-values are reported (see supplementary files and Supplementary Table 9 ). P -values in brain-behavioral analyses were considered significant when evaluated against an FDR-corrected threshold that included all effects across the multiple ROIs (8) that were examined.

Multiple linear models’ formula (example):

MRI data acquisition

MRI data were acquired with a standard whole-head coil using a 3-T Philips Achieva scanner. T2*-weighted echoplanar images (EPIs) were obtained during three functional runs, in which the first two volumes were discarded to allow for equilibration of T1 saturation effects. Volumes covered the whole brain (38 slices; 2.75 mm slice thickness; ascending acquisition) and were acquired every 2200 ms (TE = 30 ms). A high resolution T1-weighted anatomical scan was included at the end of the imaging protocol (140 slices; TR = 9.76 ms; TE = 4.59 ms; flip angle = 8°; FOV = 224 × 177.33 × 168 mm; in-plane resolution = 0.875 × 0.875 mm; slice thickness = 2 mm). Visual stimuli were projected onto a screen that was visible for participants via a mirror attached to the head coil. Before the experiment, children were trained with a mock-scanning procedure. All participants were reminded during the session not to move during scanning, and head motion was restricted by using foam padding.

fMRI preprocessing and model specification

Data preprocessing and analysis were conducted using SPM8 (Welcome Department of Cognitive Neurology, London). Images were corrected for differences in timing of slice acquisition, followed by rigid body motion correction. The T1 structural image was co-registered to the functional images and segmented according to gray matter, white matter, and cerebrospinal fluid. Functional images were then spatially normalized using the normalization parameters obtained from the segmentation procedure. The normalization algorithm used a 12-parameter affine transformation together with a nonlinear transformation involving cosine basis functions. During normalization the data was re-sampled to 3-mm cubic voxels. Templates were based on the MNI305 stereotaxic space 71 . Functional volumes were smoothed with a 6-mm full-width at half maximum isotropic Gaussian kernel. Statistical analyses were performed on individual subjects’ data using the General Linear Model (GLM).

To investigate the neural responses to own and other’s outcomes and PE’s, we modeled in separate regressors the onset of the choice stimuli with the reaction time as the duration in the observational and the action phase for both conditions. Choice value, derived from the reinforcement learning model, was included as a parametric modulator of the choice regressor in the observational (OL), and of the choice regressor in the action phase (OL, IL). The onset of the outcome was modeled with a stick function. Separate outcome regressors were created for own and other’s outcomes in the OL condition, and for own and no-outcomes in the IL condition. In addition, three outcome regressors (own and other’s outcomes in the OL, and own outcomes in the IL condition) included a parametric modulation of trial-wise PE’s derived from the reinforcement learning model. Trials in which participants did not respond on time and censored motion trials were modeled separately as regressors of no interest. Finally, 6 head-motion parameters were included as nuisance regressors.

Our main analyses include the comparison between own outcomes in the IL (action phase), and other’s outcomes in the OL condition (observational phase, see Fig. 1 ). For completeness we include whole-brain maps of the non-modulated feedback event in Supplementary Fig. 8 . As a control analysis to investigate whether substantial differences would arise when considering one’s own outcomes in the IL and OL conditions, we compared PE-related activation for self-outcomes in both conditions (further details can be found in the supplementary files and Supplementary Fig. 9 ). It is important to note that PE activation patterns largely overlapped, and no significant differences were observed. We chose to use data from the IL condition exclusively to maintain a similar number of trials entered into the analysis. Also, the action phase in the IL condition is less affected by the information from the observational phase.

Unless stated otherwise, whole-brain results comparing learning conditions were considered significant if they exceeded an FWE cluster-corrected threshold of p  < 0.05, with an initial threshold of p  < 0.001. Age-related differences were tested with an FDR cluster-corrected threshold of p  < 0.05, with an initial threshold of p  < 0.001. We used the MarsBaR toolbox 72 for SPM8 to extract beta-values from cluster of activation observed in our contrasts of interest, which were used in subsequent correlations with performance and parameter estimates.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

The data that support the findings of this study can be found in the Leiden Repository ( https://doi.org/10.34894/W4WMPZ ).

Code availability

All relevant R codes can be found in the Leiden Repository ( https://doi.org/10.34894/W4WMPZ ).

Albert, D., Chein, J. & Steinberg, L. The Teenage Brain: Peer Influences on Adolescent Decision Making. Curr. Dir. Psychol. Sci. 22 , 114–120 (2013).

Article   PubMed   PubMed Central   Google Scholar  

Blakemore, S.-J. The social brain in adolescence. Nat. Rev. Neurosci. 9 , 267–277 (2008).

Article   CAS   PubMed   Google Scholar  

Chein, J., Albert, D., O’Brien, L., Uckert, K. & Steinberg, L. Peers increase adolescent risk taking by enhancing activity in the brain’s reward circuitry. Dev. Sci. 14 , F1–F10 (2011).

Steinberg, L. et al. Age differences in sensation seeking and impulsivity as indexed by behavior and self-report: Evidence for a dual systems model. Dev. Psychol. 44 , 1764–1778 (2008).

Article   PubMed   Google Scholar  

Olsson, A., Knapska, E. & Lindström, B. The neural and computational systems of social learning. Nat. Rev. Neurosci. 21 , 197–212 (2020).

Towner, A. V., Kock, A. A., Stopforth, C., Hurwitz, D. & Elwen, S. H. Direct observation of killer whales preying on white sharks and evidence of a flight response. Ecology 104 , e3875 (2023).

Topel, S. et al. Expecting the unexpected: a review of learning under uncertainty across development. Cogn. Affect. Behav. Neurosci. 23 , 718–738 (2023).

Nussenbaum, K. & Hartley, C. A. Reinforcement learning across development: What insights can we draw from a decade of research? Dev. Cogn. Neurosci. 40 , 100733 (2019).

Rodriguez Buritica, J. M., Eppinger, B., Schuck, N. W., Heekeren, H. R. & Li, S.-C. Electrophysiological correlates of observational learning in children. Dev. Sci. 19 , 699–709 (2016).

Rodriguez Buritica, J. M., Heekeren, H. R., Li, S.-C. & Eppinger, B. Developmental differences in the neural dynamics of observational learning. Neuropsychologia 119 , 12–23 (2018).

Burke, C. J., Tobler, P. N., Baddeley, M. & Schultz, W. Neural mechanisms of observational learning. Proc. Natl Acad. Sci. 107 , 14431–14436 (2010).

Article   CAS   PubMed   PubMed Central   ADS   Google Scholar  

Hill, M. R., Boorman, E. D. & Fried, I. Observational learning computations in neurons of the human anterior cingulate cortex. Nat. Commun. 7 , 12722 (2016).

Aquino, T. G. et al. Value-Related Neuronal Responses in the Human Amygdala during Observational Learning. J. Neurosci. 40 , 4761–4772 (2020).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Charpentier, C. J., Iigaya, K. & O’Doherty, J. P. A Neuro-computational Account of Arbitration between Choice Imitation and Goal Emulation during Human Observational Learning. Neuron 106 , 687–699.e7 (2020).

Hackel, L. M. & Amodio, D. M. Computational neuroscience approaches to social cognition. Curr. Opin. Psychol. 24 , 92–97 (2018).

Lockwood, P. L. & Klein-Flügge, M. C. Computational modelling of social cognition and behaviour—a reinforcement learning primer. Soc. Cogn. Affect. Neurosci . nsaa040, https://doi.org/10.1093/scan/nsaa040 (2020).

Zhang, L. & Gläscher, J. A brain network supporting social influences in human decision-making. Sci. Adv. 6 , eabb4159 (2020).

Article   PubMed   PubMed Central   ADS   Google Scholar  

Sutton, R. S. & Barto, A. G. Reinforcement learning: an introduction (MIT Press, 1998).

Schultz, W. Dopamine reward prediction-error signalling: a two-component response. Nat. Rev. Neurosci. 17 , 183–195 (2016).

Schultz, W. & Dickinson, A. Neuronal Coding of Prediction Errors. Annu. Rev. Neurosci. 23 , 473–500 (2000).

Diederen, K. M. J., Spencer, T., Vestergaard, M. D., Fletcher, P. C. & Schultz, W. Adaptive Prediction Error Coding in the Human Midbrain and Striatum Facilitates Behavioral Adaptation and Learning Efficiency. Neuron 90 , 1127–1138 (2016).

Schultz, W. Reward prediction error. Curr. Biol. 27 , R369–R371 (2017).

Joiner, J., Piva, M., Turrin, C. & Chang, S. W. Social learning through prediction error in the brain. NPJ Sci. Learn. 2 , 1–9 (2017).

Article   Google Scholar  

Apps, M. A. J. & Sallet, J. Social Learning in the Medial Prefrontal Cortex. Trends Cogn. Sci. 21 , 151–152 (2017).

Sul, J. H., Kim, H., Huh, N., Lee, D. & Jung, M. W. Distinct Roles of Rodent Orbitofrontal and Medial Prefrontal Cortex in Decision Making. Neuron 66 , 449–460 (2010).

Lockwood, P. L., Apps, M. A. J., Valton, V., Viding, E. & Roiser, J. P. Neurocomputational mechanisms of prosocial learning and links to empathy. Proc. Natl Acad. Sci. 113 , 9763–9768 (2016).

Westhoff, B., Blankenstein, N. E., Schreuders, E., Crone, E. A. & van Duijvenvoorde, A. C. K. Increased Ventromedial Prefrontal Cortex Activity in Adolescence Benefits Prosocial Reinforcement Learning. Dev. Cogn. Neurosci. 52 , 101018 (2021).

Collette, S., Pauli, W. M., Bossaerts, P. & O’Doherty, J. Neural computations underlying inverse reinforcement learning in the human brain. eLife 6 , e29718 (2017).

Yoshida, K., Saito, N., Iriki, A. & Isoda, M. Social error monitoring in macaque frontal cortex. Nat. Neurosci. 15 , 1307–1312 (2012).

Apps, M. A. J., Rushworth, M. F. S. & Chang, S. W. C. The Anterior Cingulate Gyrus and Social Cognition: Tracking the Motivation of Others. Neuron 90 , 692–707 (2016).

Behrens, T. E. J., Hunt, L. T. & Rushworth, M. F. S. The Computation of Social Behavior. Science 324 , 1160–1164 (2009).

Article   CAS   PubMed   ADS   Google Scholar  

Hampton, A. N., Bossaerts, P. & O’Doherty, J. P. Neural correlates of mentalizing-related computations during strategic interactions in humans. Proc. Natl Acad. Sci. 105 , 6741–6746 (2008).

Achterberg, M. et al. The neural and behavioral correlates of social evaluation in childhood. Dev. Cogn. Neurosci. 24 , 107–117 (2017).

O’Doherty, J. P. Reward representations and reward-related learning in the human brain: insights from neuroimaging. Curr. Opin. Neurobiol. 14 , 769–776 (2004).

Burnside, R., Fischer, A. G. & Ullsperger, M. The feedback‐related negativity indexes prediction error in active but not observational learning. Psychophysiology , e13389, https://doi.org/10.1111/psyp.13389 (2019).

van Duijvenvoorde, A. C. K., Zanolie, K., Rombouts, S. A. R. B., Raijmakers, M. E. J. & Crone, E. A. Evaluating the Negative or Valuing the Positive? Neural Mechanisms Supporting Feedback-Based Learning across Development. J. Neurosci. 28 , 9495–9503 (2008).

Eppinger, B., Mock, B. & Kray, J. Developmental differences in learning and error processing: Evidence from ERPs. Psychophysiology 46 , 1043–1053 (2009).

Crone, E. A., Jennings, J. R. & Van der Molen, M. W. Developmental Change in Feedback Processing as Reflected by Phasic Heart Rate Changes. Dev. Psychol. 40 , 1228–1238 (2004).

Hämmerer, D., Li, S.-C., Müller, V. & Lindenberger, U. Life Span Differences in Electrophysiological Correlates of Monitoring Gains and Losses during Probabilistic Reinforcement Learning. J. Cogn. Neurosci. 23 , 579–592 (2011).

van den Bos, W., Cohen, M. X., Kahnt, T. & Crone, E. A. Striatum–Medial Prefrontal Cortex Connectivity Predicts Developmental Changes in Reinforcement Learning. Cereb. Cortex 22 , 1247–1255 (2012).

Decker, J. H., Lourenco, F. S., Doll, B. B. & Hartley, C. A. Experiential reward learning outweighs instruction prior to adulthood. Cogn. Affect. Behav. Neurosci. 15 , 310–320 (2015).

Thompson-Schill, S. L., Ramscar, M. & Chrysikou, E. G. Cognition Without Control: When a Little Frontal Lobe Goes a Long Way. Curr. Dir. Psychol. Sci. 18 , 259–263 (2009).

Gopnik, A. Childhood as a solution to explore–exploit tensions. Philos. Trans. R. Soc. B Biol. Sci. 375 , 20190502 (2020).

Lockwood, P. L. et al. Neural mechanisms for learning self and other ownership. Nat. Commun. 9 , 4747 (2018).

Decety, J. & Lamm, C. The Role of the Right Temporoparietal Junction in Social Interaction: How Low-Level Computational Processes Contribute to Meta-Cognition. Neuroscientist 13 , 580–593 (2007).

Frith, C. D. & Frith, U. The Neural Basis of Mentalizing. Neuron 50 , 531–534 (2006).

Lee, D. & Seo, H. Neural Basis of Strategic Decision Making. Trends Neurosci. 39 , 40–48 (2016).

Lockwood, P. L., Apps, M. A., Roiser, J. P. & Viding, E. Encoding of Vicarious Reward Prediction in Anterior Cingulate Cortex and Relationship with Trait Empathy. J. Neurosci. 35 , 13720–13727 (2015).

Dunne, S. & O’Doherty, J. P. Insights from the application of computational neuroimaging to social neuroscience. Curr. Opin. Neurobiol. 23 , 387–392 (2013).

Fliessbach, K. et al. Social Comparison Affects Reward-Related Brain Activity in the Human Ventral Striatum. Science 318 , 1305–1308 (2007).

Bellebaum, C., Kobza, S., Thiele, S. & Daum, I. It Was Not M. Y. Fault: Event-Related Brain Potentials in Active and Observational Learning from Feedback. Cereb. Cortex 20 , 2874–2883 (2010).

Davidow, J. Y., Foerde, K., Galván, A. & Shohamy, D. An Upside to Reward Sensitivity: The Hippocampus Supports Enhanced Reinforcement Learning in Adolescence. Neuron 92 , 93–99 (2016).

Cohen, J. R. et al. A unique adolescent response to reward prediction errors. Nat. Neurosci. 13 , 669–671 (2010).

Galvan, A. et al. Earlier Development of the Accumbens Relative to Orbitofrontal Cortex Might Underlie Risk-Taking Behavior in Adolescents. J. Neurosci. 26 , 6885–6892 (2006).

Gould, R. L., Brown, R. G., Owen, A. M., ffytche, D. H. & Howard, R. J. FMRI BOLD response to increasing task difficulty during successful paired associates learning. NeuroImage 20 , 1006–1019 (2003).

Lewandowska, M., Piatkowska-Janko, E., Bogorodzki, P., Wolak, T. & Szelag, E. Changes in fMRI BOLD response to increasing and decreasing task difficulty during auditory perception of temporal order. Neurobiol. Learn. Mem. 94 , 382–391 (2010).

Sunaert, S., Van Hecke, P., Marchal, G. & Orban, G. A. Attention to Speed of Motion, Speed Discrimination, and Task Difficulty: An fMRI Study. NeuroImage 11 , 612–623 (2000).

Wechsler, D. WISC-3: Wechsler intelligence scale for children: manual (Psychological Corporation, 1991).

Brainard, D. H. The Psychophysics Toolbox. Spat. Vis. 10 , 433–436 (1997).

Windell, O. Vector Snowflake Application. Retrieved September 7, 2012, from http://www.evilmadscientist.com/2008/vector-snowflake-application/ (2008).

Hampton, A. N. & O’Doherty, J. P. Decoding the neural substrates of reward-related decision making with functional MRI. Proc. Natl Acad. Sci. 104 , 1377–1382 (2007).

Gläscher, J. P. & O’Doherty, J. P. Model‐based approaches to neuroimaging: combining reinforcement learning theory with fMRI data. WIREs Cogn. Sci. 1 , 501–510 (2010).

Kahnt, T. et al. Dorsal Striatal–midbrain Connectivity in Humans Predicts How Reinforcements Are Used to Guide Decisions. J. Cogn. Neurosci. 21 , 1332–1345 (2009).

Apps, M. A. J., Lesage, E. & Ramnani, N. Vicarious Reinforcement Learning Signals When Instructing Others. J. Neurosci. 35 , 2904–2913 (2015).

Palminteri, S., Kilford, E. J., Coricelli, G. & Blakemore, S.-J. The Computational Development of Reinforcement Learning during Adolescence. PLOS Comput. Biol. 12 , e1004953 (2016).

Farrell, S. & Lewandowsky, S. Computational modeling of cognition and behavior , (Cambridge University Press, 2018).

Gläscher, J., Daw, N., Dayan, P. & O’Doherty, J. P. States versus Rewards: Dissociable Neural Prediction Error Signals Underlying Model-Based and Model-Free Reinforcement Learning. Neuron 66 , 585–595 (2010).

O’Doherty, J. P., Hampton, A. & Kim, H. Model-Based fMRI and Its Application to Reward Learning and Decision Making. Ann. N. Y. Acad. Sci. 1104 , 35–53 (2007).

Article   PubMed   ADS   Google Scholar  

Wilson, R. C. & Niv, Y. Is Model Fitting Necessary for Model-Based fMRI? PLOS Comput. Biol. 11 , e1004237 (2015).

Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting Linear Mixed-Effects Models Using lme4. J. Stat. Softw. 67 , 1–48 (2015).

Cocosco, C. A., Kollokian, V., Kwan, R. K.-S., Pike, G. B. & Evans, A. C. Brainweb: Online interface to a 3D MRI simulated brain database (Citeseer, 1997).

Brett, M., Anton, J., Valabregue, R. & Poline, J. Region of interest analysis using an SPM toolbox. NeuroImage 16 , 497 (2002).

Google Scholar  

Download references

Acknowledgements

We thank Chris Burke and Geert-Jan Will for valuable input; Iris Koele, Laura Steinmann, and Sibel Altikulaç for help with data collection and project management; Stella Berboth for help with manuscript preparation and Rasmus Bruckner for help with the computational modeling. This work is supported by the Social Resilience and Security program (Leiden University) and by the Netherlands Organization for Scientific Research (NWO) (Open Research Area 464-15-176) awarded to A.C.K.v.D.; by a PhD DAAD Stipend and CIC (Center for International Cooperation FU Berlin) awarded to JRB and the German Research Foundation (DFG) (SFB 940/2 B7) awarded to B.E.

Open Access funding enabled and organized by Projekt DEAL.

Author information

Julia M. Rodriguez Buritica

Present address: Department of Education and Psychology, Freie Universität Berlin, Berlin, Germany

Authors and Affiliations

Department of Psychology, University of Greifswald, Greifswald, Germany

Julia M. Rodriguez Buritica, Ben Eppinger & Hauke R. Heekeren

Berlin School of Mind and Brain & Department of Psychology, Humboldt University of Berlin, Berlin, Germany

Department of Education and Psychology, Freie Universität Berlin, Berlin, Germany

Ben Eppinger

Department of Psychology, Concordia University, Montreal, Canada

Department of Psychology, Technische Universität Dresden, Dresden, Germany

Executive University Board, Universität Hamburg, Hamburg, Germany

Hauke R. Heekeren

Department of Psychology, Education and Child Studies, Erasmus University Rotterdam, Rotterdam, Netherlands

Eveline A. Crone

Institute of Psychology, Leiden University, Leiden, The Netherlands

Eveline A. Crone & Anna C. K. van Duijvenvoorde

Leiden Institute for Brain and Cognition, Leiden, The Netherlands

You can also search for this author in PubMed   Google Scholar

Contributions

J.M.R.B., B.E. and A.C.K.vD. designed the research. J.M.R.B., and A.C.K. vD. performed the experiments and J.M.R.B., and A.C.K.vD. analyzed data. J.M.R.B., B.E., A.C.K.v.D., H.R.H. and E.A.C. interpreted findings and wrote the manuscript.

Corresponding authors

Correspondence to Julia M. Rodriguez Buritica or Anna C. K. van Duijvenvoorde .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary files, reporting summary, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Rodriguez Buritica, J.M., Eppinger, B., Heekeren, H.R. et al. Observational reinforcement learning in children and young adults. npj Sci. Learn. 9 , 18 (2024). https://doi.org/10.1038/s41539-024-00227-9

Download citation

Received : 15 June 2022

Accepted : 21 February 2024

Published : 13 March 2024

DOI : https://doi.org/10.1038/s41539-024-00227-9

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

6.4 Observational Learning (Modeling)

Learning objectives.

By the end of this section, you will be able to:

• Define observational learning
• Discuss the steps in the modeling process
• Explain the prosocial and antisocial effects of observational learning

Previous sections of this chapter focused on classical and operant conditioning, which are forms of associative learning. In observational learning , we learn by watching others and then imitating, or modeling, what they do or say. For instance, have you ever gone to YouTube to find a video showing you how to do something? The individuals performing the imitated behavior are called models . Research suggests that this imitative learning involves a specific type of neuron, called a mirror neuron (Hickock, 2010; Rizzolatti, Fadiga, Fogassi, & Gallese, 2002; Rizzolatti, Fogassi, & Gallese, 2006).

Humans and other animals are capable of observational learning. For example, in a study of social learning in chimpanzees, researchers gave juice boxes with straws to two groups of captive chimpanzees. The first group dipped the straw into the juice box, and then sucked on the small amount of juice at the end of the straw. The second group sucked through the straw directly, getting much more juice. When the first group, the “dippers,” observed the second group, “the suckers,” what do you think happened? All of the “dippers” in the first group switched to sucking through the straws directly. By simply observing the other chimps and modeling their behavior, they learned that this was a more efficient method of getting juice (Yamamoto, Humle, and Tanaka, 2013).

Imitation is sometimes called the highest form of flattery. But consider Claire’s experience with observational learning. Claire’s nine-year-old son, Jay, was getting into trouble at school and was defiant at home. Claire feared that Jay would end up like her brothers, two of whom were in prison. One day, after yet another bad day at school and another negative note from the teacher, Claire, at her wit’s end, beat her son with a belt to get him to behave. Later that night, as she put her children to bed, Claire witnessed her four-year-old daughter, Anna, take a belt to her teddy bear and whip it. Claire was horrified, realizing that Anna was imitating her mother. It was then that Claire knew she wanted to discipline her children in a different manner.

Link to Learning

Are chimps smarter than children? Watch this video showing chimps and children performing tasks and contemplate who performed the task better. How about quicker?

Like Tolman, whose experiments with rats suggested a cognitive component to learning, psychologist Albert Bandura’s ideas about learning were different from those of strict behaviorists. Bandura and other researchers proposed a brand of behaviorism called social learning theory, which took cognitive processes into account. According to Bandura , pure behaviorism could not explain why learning can take place in the absence of external reinforcement. He felt that internal mental states must also have a role in learning and that observational learning involves much more than imitation. In imitation, a person simply copies what the model does. Observational learning is much more complex. According to Lefrançois (2012) there are several ways that observational learning can occur:

• You learn a new response. After watching your coworker get chewed out by your boss for coming in late, you start leaving home 10 minutes earlier so that you won’t be late.
• You choose whether or not to imitate the model depending on what you saw happen to the model. Remember Julian and his father? When learning to surf, Julian might watch how his father pops up successfully on his surfboard and then attempt to do the same thing. On the other hand, Julian might learn not to touch a hot stove after watching his father get burned on a stove.
• You learn a general rule that you can apply to other situations.

Bandura identified three kinds of models: live, verbal, and symbolic. A live model demonstrates a behavior in person, as when Ben stood up on his surfboard so that Julian could see how he did it. A verbal instructional model does not perform the behavior, but instead explains or describes the behavior, as when a soccer coach tells his young players to kick the ball with the side of the foot, not with the toe. A symbolic model can be fictional characters or real people who demonstrate behaviors in books, movies, television shows, video games, or Internet sources ( Figure 6.17 ).

Latent learning and modeling are used all the time in the world of marketing and advertising. This Ford commercial starring Derek Jeter played for months across the New York, New Jersey, and Connecticut areas. Jeter was an award-winning baseball player for the New York Yankees. The commercial aired in a part of the country where Jeter is an incredibly well-known athlete. He is wealthy, and considered very loyal and good looking. What message are the advertisers sending by having him featured in the ad? How effective do you think it is?

Steps in the Modeling Process

Of course, we don’t learn a behavior simply by observing a model. Bandura described specific steps in the process of modeling that must be followed if learning is to be successful: attention, retention, reproduction, and motivation. First, you must be focused on what the model is doing—you have to pay attention. Next, you must be able to retain, or remember, what you observed; this is retention. Then, you must be able to perform the behavior that you observed and committed to memory; this is reproduction. Finally, you must have motivation. You need to want to copy the behavior, and whether or not you are motivated depends on what happened to the model. If you saw that the model was reinforced for their behavior, you will be more motivated to copy them. This is known as vicarious reinforcement . On the other hand, if you observed the model being punished, you would be less motivated to copy them. This is called vicarious punishment . For example, imagine that four-year-old Allison watched her older sister Kaitlyn playing in their mother’s makeup, and then saw Kaitlyn get a time out when their mother came in. After their mother left the room, Allison was tempted to play in the make-up, but she did not want to get a time-out from her mother. What do you think she did? Once you actually demonstrate the new behavior, the reinforcement you receive plays a part in whether or not you will repeat the behavior.

Bandura researched modeling behavior, particularly children’s modeling of adults’ aggressive and violent behaviors (Bandura, Ross, & Ross, 1961). He conducted an experiment with a five-foot inflatable doll that he called a Bobo doll. In the experiment, children’s aggressive behavior was influenced by whether the teacher was punished for her behavior. In one scenario, a teacher acted aggressively with the doll, hitting, throwing, and even punching the doll, while a child watched. There were two types of responses by the children to the teacher’s behavior. When the teacher was punished for her bad behavior, the children decreased their tendency to act as she had. When the teacher was praised or ignored (and not punished for her behavior), the children imitated what she did, and even what she said. They punched, kicked, and yelled at the doll.

Watch this video clip about the famous Bobo doll experiment to see a portion of the experiment and an interview with Albert Bandura.

What are the implications of this study? Bandura concluded that we watch and learn, and that this learning can have both prosocial and antisocial effects. Prosocial (positive) models can be used to encourage socially acceptable behavior. Parents in particular should take note of this finding. If you want your children to read, then read to them. Let them see you reading. Keep books in your home. Talk about your favorite books. If you want your children to be healthy, then let them see you eat right and exercise, and spend time engaging in physical fitness activities together. The same holds true for qualities like kindness, courtesy, and honesty. The main idea is that children observe and learn from their parents, even their parents’ morals, so be consistent and toss out the old adage “Do as I say, not as I do,” because children tend to copy what you do instead of what you say. Besides parents, many public figures, such as Martin Luther King, Jr. and Mahatma Gandhi, are viewed as prosocial models who are able to inspire global social change. Can you think of someone who has been a prosocial model in your life?

The antisocial effects of observational learning are also worth mentioning. As you saw from the example of Claire at the beginning of this section, her daughter viewed Claire’s aggressive behavior and copied it. Research suggests that this may help to explain why victims of abuse often grow up to be abusers themselves (Murrell, Christoff, & Henning, 2007). In fact, about 30% of child abuse victims become abusive parents (U.S. Department of Health & Human Services, 2013). We tend to do what we know. Children who grow up witnessing their parents deal with anger and frustration through violent and aggressive acts often learn to behave in that manner themselves.

Some studies suggest that violent television shows, movies, and video games may also have antisocial effects ( Figure 6.18 ) although further research needs to be done to understand the correlational and causational aspects of media violence and behavior. Some studies have found a link between viewing violence and aggression seen in children (Anderson & Gentile, 2008; Kirsch, 2010; Miller, Grabell, Thomas, Bermann, & Graham-Bermann, 2012). These findings may not be surprising, given that a child graduating from high school has been exposed to around 200,000 violent acts including murder, robbery, torture, bombings, beatings, and rape through various forms of media (Huston et al., 1992). Not only might viewing media violence affect aggressive behavior by teaching people to act that way in real life situations, but it has also been suggested that repeated exposure to violent acts also desensitizes people to it. Psychologists are working to understand this dynamic.

View this video about the connection between violent video games and violent behavior to learn more.

What Do You Think?

Violent media and aggression.

Does watching violent media or playing violent video games cause aggression? Albert Bandura's early studies suggested television violence increased aggression in children, and more recent studies support these findings. For example, research by Craig Anderson and colleagues (Anderson, Bushman, Donnerstein, Hummer, & Warburton, 2015; Anderson et al., 2010; Bushman et al., 2016) found extensive evidence to suggest a causal link between hours of exposure to violent media and aggressive thoughts and behaviors. However, studies by Christopher Ferguson and others suggests that while there may be a link between violent media exposure and aggression, research to date has not accounted for other risk factors for aggression including mental health and family life (Ferguson, 2011; Gentile, 2016). What do you think?

As an Amazon Associate we earn from qualifying purchases.

This book may not be used in the training of large language models or otherwise be ingested into large language models or generative AI offerings without OpenStax's permission.

Want to cite, share, or modify this book? This book uses the Creative Commons Attribution License and you must attribute OpenStax.

Access for free at https://openstax.org/books/psychology-2e/pages/1-introduction

• Authors: Rose M. Spielman, William J. Jenkins, Marilyn D. Lovett
• Publisher/website: OpenStax
• Book title: Psychology 2e
• Publication date: Apr 22, 2020
• Location: Houston, Texas
• Book URL: https://openstax.org/books/psychology-2e/pages/1-introduction
• Section URL: https://openstax.org/books/psychology-2e/pages/6-4-observational-learning-modeling

© Jan 6, 2024 OpenStax. Textbook content produced by OpenStax is licensed under a Creative Commons Attribution License . The OpenStax name, OpenStax logo, OpenStax book covers, OpenStax CNX name, and OpenStax CNX logo are not subject to the Creative Commons license and may not be reproduced without the prior and express written consent of Rice University.

• Search Menu
• Sign in through your institution
• Advance articles
• Author Guidelines
• Submission Site
• About Social Work Research
• National Association of Social Workers
• Editorial Board
• Advertising and Corporate Services
• Journals Career Network
• Self-Archiving Policy
• Dispatch Dates
• Journals on Oxford Academic
• Books on Oxford Academic

Article Contents

• < Previous

Determining the Effectiveness of Behavior Skills Training and Observational Learning on Classroom Behaviors: A Case Study

• Article contents
• Figures & tables
• Supplementary Data

Thea Ervin, Alyssa N Wilson, Brandy R Maynard, Tracy Bramblett, Determining the Effectiveness of Behavior Skills Training and Observational Learning on Classroom Behaviors: A Case Study, Social Work Research , Volume 42, Issue 2, June 2018, Pages 106–117, https://doi.org/10.1093/swr/svy005

• Permissions Icon Permissions

Approximately three to four students in an average classroom engage in disruptive behaviors that interfere with normal academic and social development. School social work interventions to prevent and reduce challenging behaviors in the classroom can be used to improve behavior and academic success; however, there is a lack of research on classroom-based interventions social workers can deliver. The current study used a single-subject multiple-baseline across-classrooms design to examine the effects of behavior skills training (BST) paired with observational learning of students’ engagement in and responses to disruptive behavior in the classroom setting. Six students (ages eight through 18) with emotional and behavioral disorders were randomly selected as models ( n = 2) or observers ( n = 4). During training, each model was trained to ignore, walk away, or engage in a calming strategy when peers engaged in disruption, while observers watched. Using a concurrent multiple-baseline across-classrooms design, student engagement in disruptive behavior and response to peers’ disruptive behavior were observed before and after BST across classrooms. All students demonstrated an increase in appropriately responding to disruptive behavior following BST, and instances of disruptive behavior decreased.

Email alerts

Citing articles via.

• Recommend to your Library

Affiliations

• Online ISSN 1545-6838
• Print ISSN 1070-5309
• Copyright © 2024 National Association of Social Workers
• About Oxford Academic
• Publish journals with us
• University press partners
• What we publish
• New features  
• Open access
• Institutional account management
• Rights and permissions
• Get help with access
• Accessibility
• Advertising
• Media enquiries
• Oxford University Press
• Oxford Languages
• University of Oxford

Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide

• Copyright © 2024 Oxford University Press
• Cookie settings
• Cookie policy
• Privacy policy
• Legal notice

This Feature Is Available To Subscribers Only

Sign In or Create an Account

This PDF is available to Subscribers Only

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

Want to create or adapt books like this? Learn more about how Pressbooks supports open publishing practices.

Observational Learning

Not all forms of learning are accounted for entirely by classical and operant conditioning. Imagine a child walking up to a group of children playing a game on the playground. The game looks fun, but it is new and unfamiliar. Rather than joining the game immediately, the child opts to sit back and watch the other children play a round or two. Observing the others, the child takes note of the ways in which they behave while playing the game. By watching the behavior of the other kids, the child can figure out the rules of the game and even some strategies for doing well at the game. This is called observational learning .

Children observing a social model (an experienced chess player) to learn the rules and strategies of the game of chess. [Image: David R. Tribble, https://goo.gl/nWsgxI, CC BY-SA 3.0, https://goo.gl/ uhHola]

Observational learning is a component of Albert Bandura’s Social Learning Theory ( Bandura, 1977 ), which posits that individuals can learn novel responses via observation of key others’ behaviors. Observational learning does not necessarily require reinforcement, but instead hinges on the presence of others, referred to as social models . Social models are typically of higher status or authority compared to the observer, examples of which include parents, teachers, and police officers. In the example above, the children who already know how to play the game could be thought of as being authorities—and are therefore social models—even though they are the same age as the observer. By observing how the social models behave, an individual is able to learn how to act in a certain situation. Other examples of observational learning might include a child learning to place her napkin in her lap by watching her parents at the dinner table, or a customer learning where to find the ketchup and mustard after observing other customers at a hot dog stand.

Bandura theorizes that the observational learning process consists of four parts. The first is attention —as, quite simply, one must pay attention to what s/he is observing in order to learn. The second part is retention : to learn one must be able to retain the behavior s/he is observing in memory.The third part of observational learning, initiation , acknowledges that the learner must be able to execute (or initiate) the learned behavior. Lastly, the observer must possess the motivation to engage in observational learning. In our vignette, the child must want to learn how to play the game in order to properly engage in observational learning.

Researchers have conducted countless experiments designed to explore observational learning, the most famous of which is Albert Bandura’s “Bobo doll experiment.”

In this experiment ( Bandura, Ross & Ross 1961 ), Bandura had children individually observe an adult social model interact with a clown doll (“Bobo”). For one group of children, the adult interacted aggressively with Bobo: punching it, kicking it, throwing it, and even hitting it in the face with a toy mallet. Another group of children watched the adult interact with other toys, displaying no aggression toward Bobo. In both instances the adult left and the children were allowed to interact with Bobo on their own. Bandura found that children exposed to the aggressive social model were significantly more likely to behave aggressively toward Bobo, hitting and kicking him, compared to those exposed to the non-aggressive model. The researchers concluded that the children in the aggressive group used their observations of the adult social model’s behavior to determine that aggressive behavior toward Bobo was acceptable.

While reinforcement was not required to elicit the children’s behavior in Bandura’s first experiment, it is important to acknowledge that consequences do play a role within observational learning. A future adaptation of this study ( Bandura, Ross, & Ross, 1963 ) demonstrated that children in the aggression group showed less aggressive behavior if they witnessed the adult model receive punishment for aggressing against Bobo. Bandura referred to this process as vicarious reinforcement , as the children did not experience the reinforcement or punishment directly, yet were still influenced by observing it.

Bobo [Image: © Sémhur / Wikimedia Commons / CC-BY-SA-3.0 (or Free Art License), https://goo.gl/uhHola]

Observational Learning Copyright © by Philip Smith is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License , except where otherwise noted.

Share This Book

Bandura’s Bobo Doll Experiment on Social Learning

Saul Mcleod, PhD

Editor-in-Chief for Simply Psychology

BSc (Hons) Psychology, MRes, PhD, University of Manchester

Saul Mcleod, PhD., is a qualified psychology teacher with over 18 years of experience in further and higher education. He has been published in peer-reviewed journals, including the Journal of Clinical Psychology.

Learn about our Editorial Process

Olivia Guy-Evans, MSc

Associate Editor for Simply Psychology

BSc (Hons) Psychology, MSc Psychology of Education

Olivia Guy-Evans is a writer and associate editor for Simply Psychology. She has previously worked in healthcare and educational sectors.

During the 1960s, Albert Bandura conducted a series of experiments on observational learning , collectively known as the Bobo doll experiments. Two of the experiments are described below:

Bandura (1961) conducted a controlled experiment study to investigate if social behaviors (i.e., aggression) can be acquired by observation and imitation.

Bandura, Ross, and Ross (1961) tested 36 boys and 36 girls from the Stanford University Nursery School aged between 3 to 6 years old.

The researchers pre-tested the children for how aggressive they were by observing the children in the nursery and judged their aggressive behavior on four 5-point rating scales.

It was then possible to match the children in each group so that they had similar levels of aggression in their everyday behavior. The experiment is, therefore, an example of a matched pairs design .

To test the inter-rater reliability of the observers, 51 of the children were rated by two observers independently, and their ratings were compared. These ratings showed a very high-reliability correlation (r = 0.89), which suggested that the observers had a good agreement about the behavior of the children.

A lab experiment was used, in which the independent variable (the type of model) was manipulated in three conditions:

• Aggressive model is shown to 24 children
• Non-aggressive model is shown to 24 children
• No model is shown (control condition) – 24 children

Stage 1: Modeling

In the experimental conditions, children were individually shown into a room containing toys and played with some potato prints and pictures in a corner for 10 minutes while either:

• 24 children (12 boys and 12 girls) watched a male or female model behaving aggressively towards a toy called a “Bobo doll”. The adults attacked the Bobo doll in a distinctive manner – they used a hammer in some cases, and in others threw the doll in the air and shouted “Pow, Boom.”
• Another 24 children (12 boys and 12 girls) were exposed to a non-aggressive model who played in a quiet and subdued manner for 10 minutes (playing with a tinker toy set and ignoring the bobo-doll).
• The final 24 children (12 boys and 12 girls) were used as a control group and not exposed to any model at all.

Stage 2: Aggression Arousal

All the children (including the control group) were subjected to “mild aggression arousal.” Each child was (separately) taken to a room with relatively attractive toys.

As soon as the child started to play with the toys, the experimenter told the child that these were the experimenter’s very best toys and she had decided to reserve them for the other children.

Stage 3: Test for Delayed Imitation

• The next room contained some aggressive toys and some non-aggressive toys. The non-aggressive toys included a tea set, crayons, three bears and plastic farm animals. The aggressive toys included a mallet and peg board, dart guns, and a 3 foot Bobo doll.
• The child was in the room for 20 minutes, and their behavior was observed and rated though a one-way mirror. Observations were made at 5-second intervals, therefore, giving 240 response units for each child.
• Other behaviors that didn’t imitate that of the model were also recorded e.g., punching the Bobo doll on the nose.
• Children who observed the aggressive model made far more imitative aggressive responses than those who were in the non-aggressive or control groups.
• There was more partial and non-imitative aggression among those children who had observed aggressive behavior, although the difference for non-imitative aggression was small.
• The girls in the aggressive model condition also showed more physically aggressive responses if the model was male, but more verbally aggressive responses if the model was female. However, the exception to this general pattern was the observation of how often they punched Bobo, and in this case the effects of gender were reversed.
• Boys were more likely to imitate same-sex models than girls. The evidence for girls imitating same-sex models is not strong.
• Boys imitated more physically aggressive acts than girls. There was little difference in verbal aggression between boys and girls.

Bobo doll experiment demonstrated that children are able to learn social behavior such as aggression through the process of observation learning, through watching the behavior of another person. The findings support Bandura’s (1977) Social Learning Theory .

This study has important implications for the effects of media violence on children.

There are three main advantages of the experimental method .

• Experiments are the only means by which cause and effect can be established. Thus, it could be demonstrated that the model did have an effect on the child’s subsequent behavior because all variables other than the independent variable are controlled.
• It allows for precise control of variables. Many variables were controlled, such as the gender of the model, the time the children observed the model, the behavior of the model, and so on.
• Experiments can be replicated. Standardized procedures and instructions were used, allowing for replicability. In fact, the study has been replicated with slight changes, such as using video, and similar results were found (Bandura, 1963).

Limitations of the procedure include:

• Many psychologists are very critical of laboratory studies of imitation – in particular, because they tend to have low ecological validity. The situation involves the child and an adult model, which is a very limited social situation and there is no interaction between the child and the model at any point; certainly the child has no chance to influence the model in any way.
• Also, the model and the child are strangers. This, of course, is quite unlike “normal” modeling, which often takes place within the family.
• Cumberbatch (1990) found that children who had not played with a Bobo Doll before were five times as likely to imitate the aggressive behavior than those who were familiar with it; he claims that the novelty value of the doll makes it more likely that children will imitate the behavior.
• A further criticism of the study is that the demonstrations are measured almost immediately. With such snapshot studies, we cannot discover if such a single exposure can have long-term effects.
• It is possible to argue that the bobo doll experiment was unethical. For example, there is the problem of whether or not the children suffered any long-term consequences as a result of the study. Although it is unlikely, we can never be certain.

Vicarious Reinforcement Bobo Doll Study

An observer’s behavior can also be affected by the positive or negative consequences of a model’s behavior.

So we not only watch what people do, but we watch what happens when they do things. This is known as vicarious reinforcement. We are more likely to imitate behavior that is rewarded and refrain from behavior that is punished.

Bandura (1965) used a similar experimental set up to the one outlined above to test vicarious reinforcement. The experiment had different consequences for the model’s aggression to the three groups of children.

One group saw the model’s aggression being rewarded (being given sweets and a drink for a “championship performance,” another group saw the model being punished for the aggression (scolded), and the third group saw no specific consequences (control condition).

When allowed to enter the playroom, children in the reward and control conditions imitated more aggressive actions of the model than did the children in the punishment condition.

The children in the model punished group had learned the aggression by observational learning, but did not imitate it because they expected negative consequences.

Reinforcement gained by watching another person is known as vicarious reinforcement.

Bandura, A. (1965). Influence of models” reinforcement contingencies on the acquisition of imitative responses . Journal of personality and social psychology, 1(6) , 589.

Bandura, A., Ross, D. & Ross, S.A. (1961). Transmission of aggression through imitation of aggressive models .  Journal of Abnormal and Social Psychology , 63, 575-82.

Bandura, A., Ross, D., & Ross, S. A. (1963). Imitation of film-mediated aggressive models . The Journal of Abnormal and Social Psychology , 66(1), 3.

Bandura, A. (1977). Social Learning Theory . Englewood Cliffs, NJ: Prentice Hall.

Further Information

• Bandura’s Social Learning Theory
• Bobo Doll Study Summary
• BBC Radio 4 Programme: The Bobo Doll
• Bobo Doll Summary PowerPoint

Related Articles

Famous Experiments , Social Science

Solomon Asch Conformity Line Experiment Study

Learning Theories

Aversion Therapy & Examples of Aversive Conditioning

Learning Theories , Psychology , Social Science

Albert Bandura’s Social Learning Theory

Learning Theories , Psychology

Behaviorism In Psychology

Bloom’s Taxonomy of Learning

Child Psychology , Learning Theories

Jerome Bruner’s Theory Of Learning And Cognitive Development

36 Observational Learning (Modeling)

[latexpage]

Learning Objectives

By the end of this section, you will be able to:

• Define observational learning
• Discuss the steps in the modeling process
• Explain the prosocial and antisocial effects of observational learning

Previous sections of this chapter focused on classical and operant conditioning, which are forms of associative learning. In observational learning , we learn by watching others and then imitating, or modeling, what they do or say. The individuals performing the imitated behavior are called models . Research suggests that this imitative learning involves a specific type of neuron, called a mirror neuron (Hickock, 2010; Rizzolatti, Fadiga, Fogassi, & Gallese, 2002; Rizzolatti, Fogassi, & Gallese, 2006).

Humans and other animals are capable of observational learning. As you will see, the phrase “monkey see, monkey do” really is accurate ( [link] ). The same could be said about other animals. For example, in a study of social learning in chimpanzees, researchers gave juice boxes with straws to two groups of captive chimpanzees. The first group dipped the straw into the juice box, and then sucked on the small amount of juice at the end of the straw. The second group sucked through the straw directly, getting much more juice. When the first group, the “dippers,” observed the second group, “the suckers,” what do you think happened? All of the “dippers” in the first group switched to sucking through the straws directly. By simply observing the other chimps and modeling their behavior, they learned that this was a more efficient method of getting juice (Yamamoto, Humle, and Tanaka, 2013).

Imitation is much more obvious in humans, but is imitation really the sincerest form of flattery? Consider Claire’s experience with observational learning. Claire’s nine-year-old son, Jay, was getting into trouble at school and was defiant at home. Claire feared that Jay would end up like her brothers, two of whom were in prison. One day, after yet another bad day at school and another negative note from the teacher, Claire, at her wit’s end, beat her son with a belt to get him to behave. Later that night, as she put her children to bed, Claire witnessed her four-year-old daughter, Anna, take a belt to her teddy bear and whip it. Claire was horrified, realizing that Anna was imitating her mother. It was then that Claire knew she wanted to discipline her children in a different manner.

Like Tolman, whose experiments with rats suggested a cognitive component to learning, psychologist Albert Bandura’s ideas about learning were different from those of strict behaviorists. Bandura and other researchers proposed a brand of behaviorism called social learning theory, which took cognitive processes into account. According to Bandura , pure behaviorism could not explain why learning can take place in the absence of external reinforcement. He felt that internal mental states must also have a role in learning and that observational learning involves much more than imitation. In imitation, a person simply copies what the model does. Observational learning is much more complex. According to Lefrançois (2012) there are several ways that observational learning can occur:

Bandura identified three kinds of models: live, verbal, and symbolic. A live model demonstrates a behavior in person, as when Ben stood up on his surfboard so that Julian could see how he did it. A verbal instructional model does not perform the behavior, but instead explains or describes the behavior, as when a soccer coach tells his young players to kick the ball with the side of the foot, not with the toe. A symbolic model can be fictional characters or real people who demonstrate behaviors in books, movies, television shows, video games, or Internet sources ( [link] ).

Latent learning and modeling are used all the time in the world of marketing and advertising. This commercial played for months across the New York, New Jersey, and Connecticut areas, Derek Jeter, an award-winning baseball player for the New York Yankees, is advertising a Ford. The commercial aired in a part of the country where Jeter is an incredibly well-known athlete. He is wealthy, and considered very loyal and good looking. What message are the advertisers sending by having him featured in the ad? How effective do you think it is?

STEPS IN THE MODELING PROCESS

Of course, we don’t learn a behavior simply by observing a model. Bandura described specific steps in the process of modeling that must be followed if learning is to be successful: attention, retention, reproduction, and motivation. First, you must be focused on what the model is doing—you have to pay attention. Next, you must be able to retain, or remember, what you observed; this is retention. Then, you must be able to perform the behavior that you observed and committed to memory; this is reproduction. Finally, you must have motivation. You need to want to copy the behavior, and whether or not you are motivated depends on what happened to the model. If you saw that the model was reinforced for her behavior, you will be more motivated to copy her. This is known as vicarious reinforcement . On the other hand, if you observed the model being punished, you would be less motivated to copy her. This is called vicarious punishment . For example, imagine that four-year-old Allison watched her older sister Kaitlyn playing in their mother’s makeup, and then saw Kaitlyn get a time out when their mother came in. After their mother left the room, Allison was tempted to play in the make-up, but she did not want to get a time-out from her mother. What do you think she did? Once you actually demonstrate the new behavior, the reinforcement you receive plays a part in whether or not you will repeat the behavior.

Bandura researched modeling behavior, particularly children’s modeling of adults’ aggressive and violent behaviors (Bandura, Ross, & Ross, 1961). He conducted an experiment with a five-foot inflatable doll that he called a Bobo doll. In the experiment, children’s aggressive behavior was influenced by whether the teacher was punished for her behavior. In one scenario, a teacher acted aggressively with the doll, hitting, throwing, and even punching the doll, while a child watched. There were two types of responses by the children to the teacher’s behavior. When the teacher was punished for her bad behavior, the children decreased their tendency to act as she had. When the teacher was praised or ignored (and not punished for her behavior), the children imitated what she did, and even what she said. They punched, kicked, and yelled at the doll.

Watch this video clip to see a portion of the famous Bobo doll experiment, including an interview with Albert Bandura.

What are the implications of this study? Bandura concluded that we watch and learn, and that this learning can have both prosocial and antisocial effects. Prosocial (positive) models can be used to encourage socially acceptable behavior. Parents in particular should take note of this finding. If you want your children to read, then read to them. Let them see you reading. Keep books in your home. Talk about your favorite books. If you want your children to be healthy, then let them see you eat right and exercise, and spend time engaging in physical fitness activities together. The same holds true for qualities like kindness, courtesy, and honesty. The main idea is that children observe and learn from their parents, even their parents’ morals, so be consistent and toss out the old adage “Do as I say, not as I do,” because children tend to copy what you do instead of what you say. Besides parents, many public figures, such as Martin Luther King, Jr. and Mahatma Gandhi, are viewed as prosocial models who are able to inspire global social change. Can you think of someone who has been a prosocial model in your life?

The antisocial effects of observational learning are also worth mentioning. As you saw from the example of Claire at the beginning of this section, her daughter viewed Claire’s aggressive behavior and copied it. Research suggests that this may help to explain why abused children often grow up to be abusers themselves (Murrell, Christoff, & Henning, 2007). In fact, about 30% of abused children become abusive parents (U.S. Department of Health & Human Services, 2013). We tend to do what we know. Abused children, who grow up witnessing their parents deal with anger and frustration through violent and aggressive acts, often learn to behave in that manner themselves. Sadly, it’s a vicious cycle that’s difficult to break.

Some studies suggest that violent television shows, movies, and video games may also have antisocial effects ( [link] ) although further research needs to be done to understand the correlational and causational aspects of media violence and behavior. Some studies have found a link between viewing violence and aggression seen in children (Anderson & Gentile, 2008; Kirsch, 2010; Miller, Grabell, Thomas, Bermann, & Graham-Bermann, 2012). These findings may not be surprising, given that a child graduating from high school has been exposed to around 200,000 violent acts including murder, robbery, torture, bombings, beatings, and rape through various forms of media (Huston et al., 1992). Not only might viewing media violence affect aggressive behavior by teaching people to act that way in real life situations, but it has also been suggested that repeated exposure to violent acts also desensitizes people to it. Psychologists are working to understand this dynamic.

View this video to hear Brad Bushman, a psychologist who has published extensively on human aggression and violence, discuss his research.

According to Bandura, learning can occur by watching others and then modeling what they do or say. This is known as observational learning. There are specific steps in the process of modeling that must be followed if learning is to be successful. These steps include attention, retention, reproduction, and motivation. Through modeling, Bandura has shown that children learn many things both good and bad simply by watching their parents, siblings, and others.

Review Questions

The person who performs a behavior that serves as an example is called a ________.

In Bandura’s Bobo doll study, when the children who watched the aggressive model were placed in a room with the doll and other toys, they ________.

• ignored the doll
• played nicely with the doll
• played with tinker toys
• kicked and threw the doll

Which is the correct order of steps in the modeling process?

• attention, retention, reproduction, motivation
• motivation, attention, reproduction, retention
• attention, motivation, retention, reproduction
• motivation, attention, retention, reproduction

Who proposed observational learning?

• Ivan Pavlov
• John Watson
• Albert Bandura
• B. F. Skinner

Critical Thinking Questions

What is the effect of prosocial modeling and antisocial modeling?

Prosocial modeling can prompt others to engage in helpful and healthy behaviors, while antisocial modeling can prompt others to engage in violent, aggressive, and unhealthy behaviors.

Cara is 17 years old. Cara’s mother and father both drink alcohol every night. They tell Cara that drinking is bad and she shouldn’t do it. Cara goes to a party where beer is being served. What do you think Cara will do? Why?

Cara is more likely to drink at the party because she has observed her parents drinking regularly. Children tend to follow what a parent does rather than what they say.

Personal Application Question

What is something you have learned how to do after watching someone else?

Share This Book

• Increase Font Size

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Aust Prescr
• v.41(3); 2018 Jun

Observational studies and their utility for practice

Julia fm gilmartin-thomas.

2 Research Department of Practice and Policy, University College London, School of Pharmacy, London

1 Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, Melbourne

Ingrid Hopper

Randomised controlled clinical trials are the best source of evidence for assessing the efficacy of drugs. Observational studies provide critical descriptive data and information on long-term efficacy and safety that clinical trials cannot provide, at generally much less expense.

Observational studies include case reports and case series, ecological studies, cross-sectional studies, case-control studies and cohort studies. New and ongoing developments in data and analytical technology, such as data linkage and propensity score matching, offer a promising future for observational studies. However, no study design or statistical method can account for confounders and bias in the way that randomised controlled trials can.

Clinical registries are gaining importance as a method to monitor and improve the quality of care in Australia. Although registries are a form of cohort study, clinical trials can be incorporated into them to exploit the routine follow-up of patients to capture relevant outcomes.

Introduction

Observational studies involve the study of participants without any forced change to their circumstances, that is, without any intervention. 1 Although the participants’ behaviour may change under observation, the intent of observational studies is to investigate the ‘natural’ state of risk factors, diseases or outcomes. For drug therapy, a group of people taking the drug can be compared to people not taking the drug.

The main types of observational studies used in health research, their purpose and main strengths and limitations are shown in the Table . 2 - 8

Their purpose may be descriptive, analytical or both.

• Descriptive studies are primarily designed to describe the characteristics of a studied population.
• Analytical studies seek to address cause-and-effect questions.

Case reports and case series

Case reports and case series are strictly speaking not studies. However, they serve a useful role in describing new or notable events in detail. These events often warrant further formal investigation. Examples include reports of unexpected benefits or adverse events, such as a case report describing the use of high-dose quetiapine in treatment-resistant schizophrenia after intolerance to clozapine developed 9 and a case report of a medication error involving lookalike packaging. 10

Ecological studies

Ecological studies are based on analysis of aggregated data at group levels (for example populations), and do not involve data on individuals. These data can be analysed descriptively, but not definitively for causation. Typical examples include studies that examine patterns of drug use over time. One example is the comparison of the use of non-steroidal anti-inflammatory drugs and COX-2 inhibitors in Australia and Canada. 11 Sometimes ecological studies describe associations between drugs and outcomes, such as changes in the rates of upper gastrointestinal haemorrhage after the introduction of COX-2 inhibitors. 12 However, because individual-level data are not presented, causality is at best only implied in ecological studies. The 'ecological fallacy' refers to the error of assuming that associations observed in ecological studies are causal when they are not.

Cross-sectional studies

Cross-sectional studies collect data at a single point in time for each single individual, but the actual data collection may take place over a period of time or on more than one occasion. There is no longitudinal follow-up of individuals. Cross-sectional studies represent the archetypal descriptive study. 1 Typically, they provide a profile of a population of interest, which may be broad, like the Australian Health Survey undertaken intermittently by the Australian Bureau of Statistics, 13 or focused on specific populations, such as older Australians. 14

Case-control studies

Case-control studies focus on determining risk factors for an outcome of interest (such as a disease or a drug’s adverse effect) that has already occurred. 5

• those who already have the outcome (cases)
• those who do not have the outcome (controls), who are often matched to the cases to make them similar and reduce bias.

Second, data on previous exposure to selected risk factors are collected and compared to see if these risk factors are more (or less) common among cases versus controls. Case-control studies are useful for studying the risk factors of rare outcomes, as there is no need to wait for these to occur. Multiple risk factors can be studied, but each case-control study can involve only one outcome. 5 One example explored the relationship between the use of antiplatelet and anticoagulant drugs (risk factor) and the risk of hospitalisation for bleeding (outcome) in older people with a history of stroke. 15 Another case-control study explored the risk factors for the development of flucloxacillin-associated jaundice (outcome). 16

Cohort studies

Cohort studies compare outcomes between or among subgroups of participants defined on the basis of whether or not they are exposed to a particular risk or protective factor (defined as an exposure). They provide information on how these exposures are associated with changes in the risk of particular downstream outcomes. Compared to case-control studies, cohort studies take individuals with exposures and look for outcomes, rather than taking those with outcomes and looking for exposures. Cohort studies are longitudinal, that is they involve follow-up of a cohort of participants over time. This follow-up can be prospective or retrospective. Retrospective cohort studies are those for which follow-up has already occurred. They are typically used to estimate the incidence of outcomes of interest, including the adverse effects of drugs.

Cohort studies provide a higher level of evidence of causality than case-control studies because temporality (the explicit time relationship between exposures and outcomes) is preserved. They also have the advantage of not being limited to a single outcome of interest. Their main disadvantage, compared to case-control studies, has been that longitudinal data are more expensive and time-consuming to collect. However, with the availability of electronic data, it has become easier to collect longitudinal data.

One prospective cohort study explored the relationship between the continuous use of antipsychotic drugs (exposure) and mortality (outcome) and hospitalisation (outcome) in older people. 17 In another older cohort, a retrospective study was used to explore the relationship between long-term treatment adherence (exposure) and hospital readmission (outcome). 18

Observational studies versus randomised controlled trials

Compared to randomised controlled trials, observational studies are relatively quick, inexpensive and easy to undertake. Observational studies can be much larger than randomised controlled trials so they can explore a rare outcome. They can be undertaken when a randomised controlled trial would be unethical. However, observational studies cannot control for bias and confounding to the extent that clinical trials can. Randomisation in clinical trials remains the best way to control for confounding by ensuring that potential confounders (such as age, sex and comorbidities) are evenly matched between the groups being compared. In observational studies, adjustment for potential confounders can be undertaken, but only for a limited number of confounders, and only those that are known. Randomisation in clinical trials also minimises selection bias, while blinding (masking) controls for information bias. Hence, for questions regarding drug efficacy, randomised controlled trials provide the most robust evidence.

New and upcoming developments

New methods of analysis and advances in technology are changing the way observational studies are performed.

Clinical registries

Clinical registries are essentially cohort studies, and are gaining importance as a method to monitor and improve the quality of care. 19 These registries systematically collect a uniform longitudinal dataset to evaluate specific outcomes for a population that is identified by a specific disease, condition or exposure. This allows for the identification of variations in clinical practice 20 and benchmarking across practitioners or institutions. These data can then be used to develop initiatives to improve evidence-based care and patient outcomes. 21

An example of a clinical registry in Australia is the Australian Rheumatology Association Database, 22 which collects data on the biologic disease-modifying antirheumatic drugs used for inflammatory arthritis. Clinical data from treating specialists are combined with patient-reported quality of life data and linked to national databases such as Medicare and the National Death Index. This registry has provided insight into the safety and efficacy of drugs and their effect on quality of life. It was used by the Pharmaceutical Benefits Advisory Committee to assess cost-effectiveness of these drugs. 23

Another example is the Haemostasis Registry. It was used to determine the thromboembolic adverse effects of off-label use of recombinant factor VII. 24

Clinical registries can also be used to undertake clinical trials which are nested within the registry architecture. Patients within a registry are randomised to interventions and comparators of interest. Their outcome data are then collected as part of the routine operation of the registry. The key advantages are convenience, reduced costs and greater representativeness of registry populations as opposed to those of traditional clinical trials.

One of the first registry-based trials was nested within the SWEDEHEART registry. 25 This prospectively examined manual aspiration of thrombus at the time of percutaneous coronary intervention in over 7000 patients. 26 The primary endpoint of all-cause mortality was ascertained through linkage to another Swedish registry. The cost of the trial was estimated to be US$400 000, which was a fraction of the many millions that a randomised controlled trial would have cost.

Propensity score matching

Even without randomising people within cohorts, methods have emerged in recent years that allow for less biased comparisons of two or more subgroups. Propensity score matching is a way to assemble two or more groups for comparison so that they appear like they had been randomised to an intervention or a comparator. 27 In short, the method involves logistic regression analyses to determine the likelihood (propensity) of each person within a cohort being on the intervention, and then matching people who were on the intervention to those who were not on the basis of propensity scores. Outcomes are then compared between the groups. Propensity score analysis of a large cohort of patients with relapsing remitting multiple sclerosis found that natalizumab was superior to interferon beta and glatiramer acetate in terms of improved outcomes. 28

Data technology

Increasing sophistication in techniques for data collection will lead to ongoing improvements in the capacity to undertake observational studies (and also clinical trials). Data linkage already offers a convenient way to capture outcomes, including retrospectively. However, ethical considerations must be taken into account, such as the possibility that informed consent might be required before linking data. Machine learning will soon allow for easy analyses of unstructured text (such as free text entries in an electronic prescription). 29 Patient-reported outcome measures are important and in future will be greatly facilitated by standardised, secure hardware and software platforms that allow for their capture, processing and analyses.

While clinical trials remain the best source of evidence regarding the efficacy of drugs, observational studies provide critical descriptive data. Observational studies can also provide information on long-term efficacy and safety that is usually lacking in clinical trials. New and ongoing developments in data and analytical technology offer a promising future for observational studies in pharmaceutical research.

Conflict of interest: Julia Gilmartin-Thomas is a Dementia research development fellow with the National Health and Medical Research Council (NHMRC) - Australian Research Council (ARC). Ingrid Hopper is supported by an NHMRC Early Career Fellowship.

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

Loading metrics

Open Access

Peer-reviewed

Research Article

Learning by Observation: Insights from Williams Syndrome

Contributed equally to this work with: Francesca Foti, Deny Menghini

* E-mail: [email protected]

Affiliations Department of Developmental and Social Psychology, University “Sapienza” of Rome, Rome, Italy, IRCCS Santa Lucia Foundation, Rome, Rome, Italy

Affiliation Child Neuropsychiatry Unit, Neuroscience Department, “Children’s Hospital Bambino Gesu’”, Rome, Rome, Italy

Affiliations School of Movement Sciences (DiSIST), University of Naples “Parthenope”, Naples, Italy, IRCCS Santa Lucia Foundation, Rome, Rome, Italy

Affiliation Department of Developmental and Social Psychology, University “Sapienza” of Rome, Rome, Italy

Affiliations Department of Psychology, University “Sapienza” of Rome, Rome, Italy, IRCCS Santa Lucia Foundation, Rome, Rome, Italy

• Francesca Foti, 
• Deny Menghini, 
• Laura Mandolesi, 
• Francesca Federico, 
• Stefano Vicari, 
• Laura Petrosini

• Published: January 10, 2013
• https://doi.org/10.1371/journal.pone.0053782
• Reader Comments

Observing another person performing a complex action accelerates the observer’s acquisition of the same action and limits the time-consuming process of learning by trial and error. Observational learning makes an interesting and potentially important topic in the developmental domain, especially when disorders are considered. The implications of studies aimed at clarifying whether and how this form of learning is spared by pathology are manifold. We focused on a specific population with learning and intellectual disabilities, the individuals with Williams syndrome. The performance of twenty-eight individuals with Williams syndrome was compared with that of mental age- and gender-matched thirty-two typically developing children on tasks of learning of a visuo-motor sequence by observation or by trial and error. Regardless of the learning modality, acquiring the correct sequence involved three main phases: a detection phase, in which participants discovered the correct sequence and learned how to perform the task; an exercise phase, in which they reproduced the sequence until performance was error-free; an automatization phase, in which by repeating the error-free sequence they became accurate and speedy. Participants with Williams syndrome beneficiated of observational training (in which they observed an actor detecting the visuo-motor sequence) in the detection phase, while they performed worse than typically developing children in the exercise and automatization phases. Thus, by exploiting competencies learned by observation, individuals with Williams syndrome detected the visuo-motor sequence, putting into action the appropriate procedural strategies. Conversely, their impaired performances in the exercise phases appeared linked to impaired spatial working memory, while their deficits in automatization phases to deficits in processes increasing efficiency and speed of the response. Overall, observational experience was advantageous for acquiring competencies, since it primed subjects’ interest in the actions to be performed and functioned as a catalyst for executed action.

Citation: Foti F, Menghini D, Mandolesi L, Federico F, Vicari S, Petrosini L (2013) Learning by Observation: Insights from Williams Syndrome. PLoS ONE 8(1): e53782. https://doi.org/10.1371/journal.pone.0053782

Editor: Tricia A. Thornton-Wells, Vanderbilt University, United States of America

Received: July 12, 2012; Accepted: December 4, 2012; Published: January 10, 2013

Copyright: © 2013 Foti et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This work was supported by the Fondation Jérôme Lejeune to L.M. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

In humans and other animals new competencies may be learned through active experience and through observation of others’ experience [1] , [2] . Observing another person performing a complex action accelerates the observer’s acquisition of the same action and limits the time-consuming process of learning by trial and error [3] – [5] . Indeed, observational learning does not just involve copying an action and requires that the observer transforms the observation into an action as similar as possible to the model in terms of the goal to be reached and the motor strategies to be applied [5] – [10] .

Observational learning is already present at birth [5] , [11] – [13] and it is crucial for developing complex abilities such as language, social responsiveness, use of instruments to get things done [9] , [14] . Thus, in children, learning new competencies by observing adults or peers is a central process in cognitive development [15] .

By using an innovative task based on learning to detect a visuo-motor sequence, we demonstrated that in the presence of dyslexia the ability to learn by observation a previously observed visuo-motor sequence was markedly impaired, while the ability to detect a correct sequence by trial and error was preserved [16] . In the present research we focused on a population with learning as well as intellectual disability (ID), the Williams syndrome (WS) whose well-known neuropsychological profile with specific points of strengths and weaknesses allowed singling out cognitive processes working as learning went by. WS individuals show severely impaired visuo-spatial processing, planning and implicit learning [17] – [22] , while they exhibit relatively preserved perception of the visual characteristics of objects and face recognition [23] . WS individuals have specific difficulty in maintaining visuo-spatial information in working memory and in performing long-term memory tasks [24] , [25] , consistently with a deficit of dorsal stream. Considering that the visuo-motor task to be learned by observation required to translate visual information into action, specific function of dorsal stream network [26] , [27] , WS individuals appear to be the ideal participants to investigate the cognitive processes involved in the observational learning. Performances of a group of WS individuals were compared with those of a mental age- and gender-matched group of typically developing (TD) children on a task requiring the learning of a visuo-motor sequence. The participants learned the sequence either by performing the task after observing an actor detect the sequence of correct items by trial and error (observational training) or by actually performing the task by trial and error ( Fig. 1 ).

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

Condition 1: Learning by Trial and Error followed by Observational Learning: participants detected a sequence by trial and error (TE1), then they observed an actor detecting a sequence different from the one they had previously detected (observational training) and, finally, they reproduced the observed sequence (OBS2). Condition 2: Observational Learning followed by Learning by Trial and Error: participants were submitted to an observational training, then they reproduced the observed sequence (OBS1) and, finally, detected by trial and error a different sequence they had never observed (TE2). The incorrect positions touched by the actor during the observational training are evidenced in grey. S: starting point; F: final point.

https://doi.org/10.1371/journal.pone.0053782.g001

Materials and Methods

Participants.

Twenty-eight WS participants and 32 TD children (used as controls) matching the WS individuals for mental age (MA) have been examined in the present study constituted by two experimental conditions: Learning by Trial and Error followed by Observational Learning (Condition 1); Observational Learning followed by Learning by Trial and Error (Condition 2) ( Table 1 ). Only WS individuals with mental age (MA) of at least 5 years were included in the present research because participants with inferior MA did not succeed in completing the task. No significant differences in chronological age (CA), MA and IQ ( P always >0.2) among participants performing Conditions 1 and 2 were found ( Table 2 ).

https://doi.org/10.1371/journal.pone.0053782.t001

https://doi.org/10.1371/journal.pone.0053782.t002

The clinical diagnosis of WS was confirmed by fluorescence in situ hybridization (FISH) genetic investigation, which showed the characteristic deletion on chromosome band 7q11.23. WS participants were part of a larger pool of individuals with learning disabilities attending the Children’s Hospital Bambino Gesù of Rome for clinical and rehabilitative follow-up. All of them lived at home with their families. The parents of all individuals who participated in the study provided written informed consent. This study was approved by the Ethic Committee of the Children’s Hospital Bambino Gesù of Rome and conducted according to the Helsinki declaration.

WS individuals were tested in a quiet room at the Children’s Hospital Bambino Gesù. TD children were individually tested in a quiet room at their schools.

Intelligence Evaluation and Neuropsychological Assessment

In the present study, the brief version of the Leiter International Performance Scale–Revised [28] was employed (four out of 10 subtests: Figure Ground, Form Completion, Sequential Order and Repeated Patterns). The brief IQ and the corresponding mental ages were computed. Visuo-motor integration [29] , visuo-spatial perception [30] and memory [31] were assessed ( Table 3 ).

https://doi.org/10.1371/journal.pone.0053782.t003

Experimental Procedure

Each participant was sat in front of a computer touch screen (distance 60 cm). In both Conditions, the experimenter acting as the actor (F.F.) was sat near the participant. A 8×8 black matrix appeared on the touch screen. The subject was asked to find a hidden sequence of “correct” squares prepared in advance by the experimenters. The sequence was composed of 10 adjacent spatial positions in the matrix, which formed a “snake-like” pattern ( Fig.1 ). To explain the task to each participant the experimenter used the same verbal instructions: “You have to find a route formed by ten squares. When you touch a correct square it will be turned grey and you will hear a sound; conversely, if you touch a wrong square, it will be turned red. In this case, you have to find a new grey square. You have to start the route each time you find a new correct square. After finding the whole route, you have to re-touch it three times without making lighted red squares”. The participants started touching a grey square, which was the first element of the sequence and was always lit up. In the search for the second correct square, the participants had to touch one of the four squares bordering the grey square by moving in the matrix vertically or horizontally, but never diagonally. Each touched square (correct or incorrect) was lit up for 500 ms and then lighted off again; thus, no trace of the performed sequence remained on the screen. In learning the sequence by trial and error, the participants tried to find the correct sequence immediately after the verbal instructions. Conversely, in the observational learning task after the verbal instructions the participants observed the experimenter while she detected a 10-item sequence by trial and error (observational training). The experimenter performed the task by always making the same errors in the same positions, so that all participants observed the same pattern of correct and incorrect touches. Two minutes after the end of the observational training the participants were required to actually reproduce the observed correct sequence.

The tasks involved three phases: the Detection Phase (DP) that ended once the participants found the tenth correct position; the Exercise Phase (EP) in which they had to repeat the 10-item sequence until their performance was error-free; the Automatization Phase (AP) that ended when the correct sequence was repeated three consecutive times without errors.

Error parameters: DP errors, calculated as the number of incorrect items touched in detecting the ten correct positions; EP repetitions, calculated as the number of replications needed to reach the error-free performance. Time parameters: AP times (in msec), calculated as the time spent carrying out each of the three repetitions of the sequence.

Analysis of Error

To assess the kind of error further parameters were taken into account considering the two phases DP and EP together: the number of sequence errors , as touching a “correct” square in “wrong” moment (e.g. touching E7 before than F7); side-by-side errors , as touching the squares bordering the “correct” sequence (e.g. E8); illogical errors , as touching any other square (e.g. B5); perseverations, as consecutively touching the same item or a fixed sequence of items. Furthermore, in the task of observational learning we calculated the number of imitative errors , as touching the same squares wrongly touched by the actor during the observational training (e.g. F4) ( Fig.1 ).

Condition 1: Learning by Trial and Error Followed by Observational Learning

Fourteen WS and 16 TD individuals ( Table 1 ) firstly detected a sequence by Trial and Error (TE1) and, after ten minutes from task end, they were submitted to the observational training. After two minutes, participants were required to actually reproduce the observed sequence (OBS2). There was no fixed time limit for executing the task.

A pilot study was conducted to verify if the two sequences arranged to be used as “TE” and “OBS” sequences did not differ as to degree of difficulty. Six TD children [3 M] of mental age 6.10±0.3 detected the two different sequences by trial and error; presentation order was randomized among participants. Errors made in detecting each sequence were calculated by one-way ANOVA with repeated measures. The analysis failed to reveal any significant difference between sequences ( F (1,5)  = 0.63, P  = 0.46), confirming sequences of the same difficulty.

Condition 2: Learning by Observation Followed by Learning by Trial and Error

Fourteen WS and 16 TD individuals ( Table 1 ) first observed the experimenter detect a sequence (OBS1) and then actually reproduce it. After ten minutes from task end, they detected a different sequence by trial and error (TE2). Thus, the difference of the two conditions was that participants reproduced a sequence learned by observation after (Condition 1) or before (Condition 2) the detection of a different sequence by trial and error.

To evaluate mental representative mapping abilities, at the end of the reproduction of the sequence participants were required to draw the arrangement of the sequence on a 8×8 matrix sketched on a paper sheet. Thus, any participant drew the arrangement of two sequences, one learned by observation and the other one by trial and error. Mapping abilities were evaluated by tabulating the variable “errors” into three categories: “no error”, “one error” and “more than one error”.

Attentional Task

The sustained attentional abilities of all participants were tested. Participants sat in front of a computer monitor and were required to put their left index fingers on the A key of the keyboard and to put their right index fingers on the L key. The visual stimulus was a grey circle presented on monitor center for a duration varying from 1400 (short) to 2600 (long) msec in steps of 200 msec in a randomized order. Participants were submitted to a brief training in which they were instructed to judge 20 stimuli as short or long and to press the A or L keys, respectively. In the testing phase the participants had to judge the duration of 70 stimuli (10 stimuli of each of the 7 durations) and to press the A or L keys as quickly as possible after the stimulus appeared. The computer program recorded reaction times (with 1-ms resolution) and accuracy of the response. The responses were then analyzed by clustering them in blocks of ten (regardless of stimulus duration) (i.e. 1–10, 11–20, 21–30….61–70).

Statistical Analyses

The data were first tested for normality (Shapiro-Wilk normality test) and homoscedasticity (Levene test) and then compared by using two-way, three-way or four-way analyses of variance (ANOVAs). The two-way ANOVAs were performed by applying the mixed model for independent variable (group) and repeated measures (error, square or block). Three-way ANOVAs (group×condition×task) were performed on most parameters, while the four-way ANOVA on the three AP times was performed by applying the mixed model for independent variables (group, condition and task) and repeated measures (time). These analyses were followed by post-hoc multiple comparisons using Newman–Keuls test. In evaluating mapping abilities the error categories were analyzed by Chi-Square.

Because the 28 WS participants were differently aged (N = 9 age range: 8;9–14;1; N = 10 age range: 14;9–19;9; N = 9 age range: 22;9–35;3), we verified the sample homogeneity by comparing the performances of three differently aged WS sub-groups on three main parameters of the learning tasks they performed (DP errors; EP repetitions and AP times) by using MANOVAs. These analyses revealed no significant difference among WS sub-groups’ performances. Namely, in the tasks of learning by trial and error (TE1–TE2), the MANOVA revealed a not significant sub-group effect (F (2,25)  = 0.12, P  = 0.88) and a significant parameter effect (F (2,50)  = 154.54, P <0.0001). The interaction was not significant (F (4,50)  = 0.13, P  = 0.96). In the tasks of observational learning (OBS1–OBS2) the MANOVA also revealed a not significant sub-group effect (F (2,25)  = 0.47, P  = 0.62) and a significant parameter effect (F (2,50) =  85.46, P <0.0001). The interaction was not significant (F (4,50)  = 0.47, P  = 0.75). Thus, we pooled together the 28 differently aged WS individuals.

All statistical analyses were performed by using Statistica 8.0 for Windows and the significance level was established at P< 0.05.

Learning Tasks

WS participants performed a number of DP errors not significantly different from TD children after the observational trainings (OBS1–OBS2) and were significantly impaired in detecting the sequence by trial and error in TE1 compared with any other intra- or inter-group condition ( Fig. 2A ), as revealed by post-hoc comparisons (always P <0.001) on the second-order interaction (F (1,56)  = 8.37, P =  0.0054) of the three-way ANOVA (group×condition×task).

DP: Detection Phase; EP: Exercise Phase; AP: Automatization Phase. Data are expressed as mean ± SEM. The asterisks indicate the significance level of post hoc comparisons between groups (*** P <0.001).

https://doi.org/10.1371/journal.pone.0053782.g002

In EP, when individuals repeated the sequence until their performance was error-free, WS participants needed a significantly higher number of repetitions in comparison to TD children regardless of condition (1 or 2) and trial (OBS or TE), as revealed by the group effect (F (1,56)  = 9.58, P  = 0.0030) of the three-way ANOVA ( Fig. 2B ). The analysis of the three AP times revealed that although all participants exhibited significantly reduced times as the task went by (F (2,112)  = 27.62, P <0.00001), WS individuals were significantly slower than TD children (F (1,56)  = 10.37, P  = 0.0021), revealing a difficulty in automatizing the sequence ( Fig. 2C ).

In TE1, although WS and TD participants did not differ in the number of illogical errors, WS individuals exhibited values of sequence, side-by-side and perseverative errors higher than TD children, as revealed by post-hoc comparisons made on the interaction (F (3,84)  = 3.14, P  = 0.029) of the two-way ANOVA (group×kind of error) ( Fig. 3 ). The highest number of sequence errors of WS individuals was found in E7 and F7 squares when a change of strategy was required (i.e. after an error re-starting the sequence from the first item rather than continuing along on the “snake”) ( Fig. 1 ), as revealed by post-hoc comparisons made on the interaction (F (9,252)  = 1.96, P  = 0.044) of the two-way ANOVA (group×square) ( Fig. 4 ). As for side-by-side errors, the high number of errors of WS individuals was due to their significantly more frequent touching of a wrong square when a change of direction was required (squares: D7, F6, E1) ( Fig. 1 ), as revealed by post-hoc comparisons made on the interaction (F (27,756)  = 2.42, P <0.0001) of the two-way ANOVA (group×square) ( Fig. 4 ).

Data are expressed as mean ± SEM. The asterisks indicate the significance level of post hoc comparisons between groups (* P <0.05; ** P <0.005; *** P <0.001).

https://doi.org/10.1371/journal.pone.0053782.g003

On the right, the chromatic scale indicates the sum of incorrectly touched items (brown and blue denote maximal and minimal values, respectively). S: starting point; F: final point.

https://doi.org/10.1371/journal.pone.0053782.g004

The analysis of error in the remaining tasks (OBS2, OBS1 and TE2) revealed no significant difference between groups, even if significant difference among kind of errors was found (always P <0.00001) ( Fig. 3 ).

Mapping Abilities

Mental representative mapping abilities of the participants were evaluated by having them draw the arrangement of sequences they had just performed. No significant difference among categories of errors and between groups was found in any sequence ( P at least >0.4).

Two-way ANOVAs (group×block) on reaction times or response accuracy of the WS and TD groups revealed no attentional decay in both groups, as indicated by not significant difference in the reaction times in the seven blocks (F (6,348)  = 1.55, P  = 0.15). A similar result was obtained when response accuracy was analyzed (F (6,348)  = 1.80, P  = 0.10). Notably, a significant difference was found between WS and TD groups on reaction times (F (1,58)  = 13.52, P  = 0.00051), given WS participants pressed the keys at the appearance of the stimulus more quickly than TD children ( Fig. 5 ).

Data are expressed as mean ± SEM.

https://doi.org/10.1371/journal.pone.0053782.g005

Our study adopted a matched-group design to determine whether the learning performance of WS individuals was above or below that expected given their general level of intellectual functioning indexed as MA. However, although this design is one the most commonly employed measures of matching in ID research, we are aware that it has limitations in respect to ID-matched control group design that takes into account the cognitive profile of the specific pathology. Nevertheless, even the ID-matched control group cannot be taken as a guarantee of normative group, due cognitive profiles among different etiological groups with ID exhibit different peaks and troughs [32] . In an attempt to overcome the difficulties in matching individuals of different groups on any one particular measure it has been proposed the use of regression techniques that take the factors related to task performance into account [32] . However, this measure requires specific statistical properties of the data (as homogeneity of regression slopes or sample size), hardly available in studies on population affected by rare genetic conditions as WS.

The present study documented as WS participants significantly beneficiated of observational training as TD MA-matched children. This was true specifically in the DPs of learning tasks, while as for EPs and APs, in all tasks regardless of presentation order (1 or 2) or learning modalities (OBS or TE), WS participants performed significantly worse than TD children. The powerfully positive effect of observational training was present not only in reproducing the previously observed sequences (OBS1 and OBS2) but also affected the subsequent detection of a sequence by trial and error (TE2). However, the practice effect, inevitably present in any second task, potentially could affect performances.

Since WS individuals exhibit difficulties in maintaining visuo-spatial information in working memory and in performing spatial long-term memory tasks ( Table 3 ) [20] , [24] , [25] , their heavily impaired performances in all EPs appear linked to spatial working memory deficits and difficulties in bringing together the short sequences detected during DP, in maintaining them in working memory to recall the whole sequence trace and in monitoring the correct execution of the sequence. These findings indicate that the observational training exerts beneficial effects mainly on the acquisition of strategies to be applied.

In both Conditions, WS participants displayed AP times longer than TD children, even if progressively diminishing as the task went by. This finding was not a consequence of the fine motor deficits usually reported in WS individuals. In fact, consistently with Vicari et al. (2007) [22] , the reaction times exhibited by WS group in the Attentional Task were even shorter than those of TD group. Thus, the longer WS times were related to deficits in automatization processes increasing efficiency and speed of the response to reach highest levels of performance [33] . Automatizing skills is mainly linked to the functions of sub-cortical structures, as the cerebellum and basal ganglia and to their bidirectional interconnections with cortical structures [34] , [35] , [36] , [37] . The cause of automatization and procedural deficits of WS individuals could be their remarkable hypoplasia of the basal ganglia [38] and the disproportionate enlargement of the cerebellum [39] , [40] , [41] , [42] , [43] . Indeed, in WS individuals skill-learning abilities are impaired, as revealed by their performance in Tower of London test [21] , Serial Reaction Time task [22] and Radial or Multiple Reward Mazes [44] , [45] , [46] .

By analyzing the kind of errors, some remarks can be made. First of all, both groups made a very low number of illogical errors, thus suggesting all participants similarly managed task fundamentals ( Fig. 3 and 4 ). As for imitative errors, no difference between groups was found, thus suggesting participants did imitate but did not hyperimitate [47] . Conversely, WS individuals made more sequence and side-by-side errors than TD children in TE1, particularly when a change of direction was required. Errors in stopping the more easy “keep-straight” response and performing the more demanding “turn-left” response resulted by the WS difficulty in suppressing a previously correct but then inappropriate response. Correctly responding requires executive control processes based on frontal lobe function, as response inhibition, cognitive flexibility and attentional shifting [48] , [49] , [50] , [51] . WS individuals are impaired in spatial planning, working memory, cognitive flexibility and inhibiting well-learned responses become inappropriate to the situation [52] . Indeed, the executive function deficits that impaired WS performance dramatically reduced after the observational training, once more indicating the teaching power of the observation. WS participants made perseverative errors that could result from difficulties in withholding the inappropriate repetition of a response despite knowing that it was not the correct one. This is an important component of top-down executive control. Notably, perseverations may be symptom not only of prefrontal dysfunction but also of cerebellar and basal ganglia damage provoking “frontal-like” cognitive deficits [34] , [53] , [54] , [55] , [56] , [57] .

The prevalently frontal processes require a modulation in more posterior brain systems, via the attention networks. Basic aspects of attentional processing are selective spatial attention that allows maintaining the focus of processing between spatial locations, and the attentional processing that allows a kind of “selection for action” [58] , [59] , [60] . Namely, the action of reaching the right square required attentional modulation to plan, select and initiate the appropriate behavior, to direct it toward the selected goal, and to inhibit actions inappropriate for the current goal. Because many brain structures that are part of the attention networks are included within the dorsal stream network [27] , it is not surprising that WS participants performed more errors when behavioral inhibition and attentive shifting were required but no help from observing the actor was provided [61] . The “dorsal-stream vulnerability” in WS [62] is manifested not only in the spatial and visual processing occurring within the occipital and parietal areas but also in the processing of spatial information by frontal control systems, as reported in an fMRI study [63] .

At the end of testing, mental representative mapping abilities were evaluated by drawing the arrangement of squares just discovered. WS and TD participants were similarly able to represent the shape of the “snake”. This finding is consistent with the observation that WS individuals exhibit no difficulty in mentally visualizing objects without spatially manipulating them [25] , [64] and supports the indication that the present learning protocols encompassed requests of visual imagery.

If observative experience functions as a catalyst for executed action, it can be advanced that observing a sequence prior to experiencing it primes subjects’ interest in the actions to be performed to detect rules and sequence. In fact, the present results indicate that the observation of action has a strong impact on action memory. The influence of action perception on action production requires cross-modal information be coordinated. Action and perception share the distal reference and are coded in a common representational medium [65] , so that perceiving an action activates the corresponding motor representation within the observer automatically and without conscious effort [66] , [67] , [68] .

The close interplay between observation and execution of actions found in the present study is supported by studies providing evidence of a striking overlap in the brain systems recruited for one’s own action, observation of others’ action and imitation of action [69] . In particular, when imitation is aimed at learning novel actions, the activation of the “core circuit for imitation” [70] involving the inferior frontal gyrus, the inferior parietal lobule and the superior temporal sulcus seems to be integrated with activation of the dorso-lateral and ventro-medial prefrontal cortex, for selection of motor acts and error prediction [71] , of the premotor areas, for motor preparation [70] , [72] , [73] as well as of cerebellar areas, whether or not it is accompanied by actual motor acts [74] , [75] , [76] .

The existence of direct feed-forward connections from perceptual to motor processes allows observation sculpting motor abilities by exploiting the functional overlapping between perception and action systems. It has been suggested that observation of actions engages motor-related processes similar to those of actual execution, promoting the development of an efference copy of the descending motor commands, which in combination with a forward model provides a prediction of sensory consequences [77] , [78] , [79] , [80] . Thus, action observation, efficiently translated into the matching motor representation, powerfully activates the feed-forward predictive processes, so that learning does occur. Notably, even in WS individuals the beneficial effect of observation was evident although linked only to the DP. Action observation seems to result in an amelioration of frontal functions, as motor strategy planning, decision-making processes or response inhibition needed to guide planned sequential actions. Thus, observational training allows the acquisition of the strategies to be applied to identify and learn the visuo-spatial sequence. Notably, when the observation did not play any role (as in the DP of TE1), frontal deficits markedly affected WS performance. However, it has to be underlined that as far as the observational training was beneficial, in WS individuals it did not succeed in smoothing out the deficits in processing visuo-spatial information mainly linked to their repeatedly described dorsal stream vulnerability [62] .

Acknowledgments

We would like to thank the individuals with Williams syndrome and their parents for making this study possible. Moreover, we wish to thank Dr. P. Pani for his kind and competent help in analyzing data.

Author Contributions

Conceived and designed the experiments: F. Foti DM LM F. Frederico SV LP. Performed the experiments: F. Foti. Analyzed the data: F. Foti DM LP. Contributed reagents/materials/analysis tools: DM LM SV LP. Wrote the paper: F. Foti DM LP.

• View Article
• Google Scholar
• 12. Nadel J, Butterworth G (1998) Imitation in infancy. Cambridge: Cambridge University Press.
• 13. Nadel J (2002) The imitative mind. In Meltzoff A Prinz W editors. Cambridge: Cambridge University Press. 42–62.
• 28. Roid GH, Miller LJ (2002) Leiter – R, Leiter International Performance Scale–Revised. Giunti O.S. Organizzazioni Speciali, Firenze.
• 29. Beery KE (2000) VMI: Developmental test of visual-motor integration. Italian Adaptation. Organizzazioni Speciali, Firenze.
• 30. Hammill D, Pearson N, Voress J (1994) TPV. Test di percezione visiva e integrazione visuo-motoria. Erickson, Trento.
• 31. Vicari S (2007) In O. S. Giunti (Ed.), PROMEA: Prove di memoria e apprendimento. Firenze: Organizzazioni Speciali.
• 60. Hoffstaedter F, Grefkes C, Zilles K, Eickhoff SB (2012) The “What” and “When” of Self-Initiated Movements. Cereb Cortex In press .

• Study Protocol
• Open access
• Published: 28 May 2024

MEM&SO protocol: understanding the determinants of social learning in neurodegenerative diseases

• Pauline Saliou   ORCID: orcid.org/0009-0001-1408-5581 1 ,
• Julien Chavant   ORCID: orcid.org/0000-0002-0343-925X 1 ,
• Serge Belliard 2 ,
• Catherine Merck 2 ,
• Vincent de La Sayette 1 ,
• David Wallon 3 ,
• Olivier Martinaud 1 ,
• Francis Eustache 1 &
• Mickaël Laisney   ORCID: orcid.org/0000-0003-3571-9333 1  

BMC Psychology volume  12 , Article number:  307 ( 2024 ) Cite this article

Metrics details

People with neurodegenerative diseases may have difficulty learning new information, owing to their cognitive impairments. Teaching them techniques for learning in social contexts could alleviate this difficulty. The present study will examine the performances of patients with Alzheimer’s disease and patients with the semantic variant of primary progressive aphasia on a memory test administered in three social contexts. The protocol will make it possible to identify determinants of social interactions, social abilities, cognition, and personality that can explain the potentially beneficial effect of social context on learning in these patients.

Thirty dyads (patient with primary memory impairment who meets criteria for Alzheimer’s disease paired with caregiver), 16 dyads (patient meeting criteria for semantic variant of primary progressive aphasia paired with caregiver), and 46 dyads (healthy controls with no cognitive complaints) will be recruited. A nonverbal memory test (social memory task) will be administered to each dyad in three different social contexts (presence-only, observation, collaboration). Patients and healthy controls will also undergo a neuropsychological assessment to measure social (interactions and abilities), cognitive and personality aspects. Patients will be compared with controls on differential social scores calculated between the presence-only and collaboration contexts, and between the presence-only and observation contexts. A multiple comparative case study will be conducted to identify social, cognitive and personality variables that potentially explain the differential scores in the collaboration and observation contexts.

For the first time, memory will be assessed in patients with Alzheimer’s disease and patients with the semantic variant of primary progressive aphasia in three different contexts (presence-only, observation, collaboration). The multiple comparative case study will make it possible to identify the determinants of memory performance in the social context, in order to create the most beneficial learning context for individual patients, according to their profile.

Trial registration

This study was approved by the Ile de France XI institutional review board (2022-A00198-35), and registered on ClinicalTrials.gov (no. NCT05800028), on April 27, 2023.

Peer Review reports

Neurodegenerative diseases place patients in situations of dependence in which they have to interact with people who are often new to them and adapt to new living environments. These life changes involve learning and memorizing new information, which is particularly difficult for people with cognitive disorders, such as memory impairment in Alzheimer’s disease (AD) and language deficits in the semantic variant of primary progressive aphasia (svPPA). The amnesic form of AD, characterized by an initial and predominant memory disorder, is the commonest and most typical form of the disease [ 1 ]. Severe deficits in learning new information are a core feature of AD, although deficits in storage and retrieval cannot be excluded [ 2 , 3 ]. A progressive loss of semantic knowledge is observed in svPPA [ 4 ]. The latter is characterized by an initial and predominant impairment of language, with poor verbal comprehension and profound word-finding difficulties that interfere with daily activities and include the impairment of confrontation naming and single-word comprehension. Current and proposed classifications suggest that svPPA and semantic behavioural variant frontotemporal dementia [ 5 ] are both early forms of semantic dementia. The latter is characterized by a gradual loss of semantic knowledge evidenced by difficulty finding words and identification deficits for objects and/or persons, and impaired word comprehension [ 6 , 7 ].

Different learning techniques have been developed to try to compensate for the learning and memory deficits of patients with AD, or improve the naming and use of objects for patients with svPPA, in order to maintain their autonomy. Errorless learning acts on the encoding phase, the aim being to avoid learning incorrect information [ 8 , 9 , 10 , 11 ]. The vanishing cues method facilitates the retrieval of information by providing cues that gradually disappear [ 12 , 13 ]. The purpose of space retrieval is to improve long-term retention through the repetition of learning with increasingly long intervals between the encoding of information and its retrieval [ 14 , 15 ]. In patients with svPPA, the semantic features of concepts are enriched [ 16 , 17 ], and object use is relearned [ 18 ].

All these techniques involve the creation of materials for each learning event and require numerous repetitions, but they do not always have observable long-term effects, and generalization to other types of material is not always effective or tested [ 19 ]. For patients with PPA, the most beneficial strategy is functional communication, where communication strategies that are already used by the patient are practised with a communication partner [ 20 ]. However, this and other similar techniques need to be performed with others, and cannot be practised alone.

Numerous social psychology studies among healthy individuals have shown that the presence of others can have, either a positive effect on cognitive performance (i.e., social facilitation ) or a negative effect, (i.e., social inhibition ). In memory tasks, both social facilitation and inhibition depend, on memory consolidation [ 21 ]: if learning is not consolidated, the presence of others during encoding has a negative impact on learning speed. Although there have been relatively few studies of memory tasks, results suggest that the social effect differs according to recall interval. While short-term recall (2 min) is impaired when individuals are observed during encoding [ 22 ], their performance improves as the time between encoding and recall increases (i.e., 15 min [ 23 ], and 45 min [ 22 ]). By contrast, when individuals are observed during a neuropsychological assessment, their cognitive performance decreases (i.e., social inhibition), with attention and memory tasks (immediate and delayed recall) being most negatively affected [ 24 ]. The nature of the social effect (i.e., facilitation vs. inhibition) may also depend on the real or perceived difficulty of the task [ 25 ], as well as on the individual’s assessment of the other person [ 26 ] or of the context in general. In other words, depending on the individual’s personality traits, assessment of the context, and feelings towards the task, learning in the presence of others may generate pressure through apprehension of the task and the fact of having to tackle its difficulties alone.

This pressure may be alleviated and cognitive load reduced in a context where individuals can observe someone else carrying out the task, insofar as they no longer have to tackle it themselves. Observational learning (i.e., receiving information, then using it) builds habits and improves observers’ skills [ 27 ]. However, although observational learning is advantageous because it avoids the cost of trial-and-error learning, individuals need to be selective, only using the relevant information that comes from learning from others [ 28 , 29 ]. Observational learning of a motor sequence seems to be efficient in patients with AD [ 30 ].

Collaborative learning, through partners’ co-construction of ideas within what can be regarded as a particular memory group [ 31 ], is more effective than learning on one’s own [ 32 ]. Collaborative learning is defined as interaction between peers with the aim of jointly performing a task, and meets three criteria: communication, reciprocity, and common goal [ 33 ]. Collaboration is said to be beneficial for learning [ 34 ], and contributes to participants’ social, emotional and psychological wellbeing (e.g., collaborative learning at university [ 35 ]). However, collaboration may have some disadvantages, such as the possible dilution of motivation, decreased productivity if there are uneven contributions from members, and possible encoding of errors produced by collaborators and played back in subsequent recall [ 32 ]. Collaborative learning outperforms learning alone, both among older people undergoing the normal age-related deterioration in memory [ 36 ] and among patients with AD [ 37 , 38 ]. The social dimension, including common idea building and conversational skills, therefore seems to constitute a favourable learning environment. Using a collaborative trial in which patients self-generated labels with a familiar partner, Duff and colleagues [ 37 ] showed that patients with AD can perform comparably to healthy individuals. Beyond their ability to build a common representation with their partner, patients with AD are capable of knowing that other people do not share this representation [ 39 ]. However, the mechanisms behind this effect are still poorly understood, given that these patients have impaired theory of mind (ToM; i.e., ability to attribute mental states to others) [ 40 ], an essential skill for acting appropriately in social exchanges. The presence of a social context may help to compensate for patients’ apparent difficulty constructing representations of others’ mental states [ 41 ]. Several neurological diseases are characterized by ToM disorders [ 42 ]. In semantic dementia, for instance, both cognitive and affective forms of ToM are affected [ 43 ], mainly as a result of right anterior temporal atrophy in the early stage of disease [ 5 ] or after bilateralization. Atrophy of the right anterior temporal lobe is correlated with impaired emotion recognition and person identification [ 5 , 44 , 45 , 46 ]. Some patients may exhibit egocentric behaviour [ 47 ], and respond with their own personal preferences when judging the preferences of others [ 42 ]. These social behavioural disorders can be attributed to the ToM impairment reported in these patients [ 43 ], as well as to the impairment of semantic knowledge about social norms [ 42 ].

The main objective of the present study will be to assess whether social context (collaboration or observation) improves the memory performances of patients with a neurodegenerative disease who exhibit social cognition, memory or language disorders. A secondary objective will be to identify social (interactions and abilities), cognitive and personality variables that could explain differences in patients’ performances between collaborative and observation contexts.

To this end, a nonverbal memory test (social memory task) will for the first time be administered in three different social contexts (presence-only, observation, collaboration), to patients with AD, patients with svPPA, and healthy older people with no cognitive disorders. Participants will also undergo a neuropsychological assessment probing social (interactions and abilities), cognitive and personality aspects.

A multiple comparative case study will be conducted [ 48 , 49 , 50 ] to identify variables (social interactions and social abilities, cognitive skills, and personality traits) that could explain better performances in a collaboration or observation context than in a presence-only context.

Participants

We will recruit 184 participants in order to create 92 dyads: 30 dyads in which a person who meets the criteria for AD with primary memory impairment [ 1 ] is paired with a caregiver, 16 dyads in which a person who meets the criteria for svPPA [ 4 ] is paired with a caregiver, and 46 pairs of healthy controls with no cognitive complaints (HC).

The dyad partners must have socialized together for at least 2 h per week for at least 5 years. Patients and HC must be aged 50–85 years. Patients’ caregivers and HC must have no cognitive complaints and a Montreal Cognitive Assessment (MoCA) [ 51 ] score > 25.

Exclusion criteria for all participants will be (1) concurrent participation in a therapeutic drug trial, (2) prior neurological disorders (stroke, epilepsy, head injury with loss of consciousness lasting more than 1 h), (3) chronic alcoholism or drug addiction, (4) a clinically severe major psychiatric disorder within the previous 10 years, and (5) use of psychotropic medication.

The size of the patient groups was determined according to the feasibility of including this type of patient, based on studies carried out for more than 15 years in our research unit [ 43 , 52 , 53 ]. The size of the HC group was determined in relation to the size of the patient groups. To ensure that HC are not cognitively impaired, a quick cognitive test will be performed at the beginning of the psychological interview. Individuals who perform below the normal range will be excluded from the study and replaced.

Social memory task

The social memory task takes the form of a game, and has been specially designed to be feasible in different social contexts and adapted to patients with cognitive disorders. The game consists in constructing 12 pairs of pictures by selecting a picture in a draw pile (always visible) and finding the same picture in a 12-box grid (hidden face; see Fig.  1 ). Participants (here, either patients or HC) are instructed to memorize the position of that match in the grid, as this will be tested later. A recall phase takes place 20 min after this learning phase. The game will be repeated in each of the three social contexts (presence-only, observation, collaboration; see “Procedure for social memory task” subsection below), with three different sets of pictures.

Screenshot showing learning phase of social memory task, displayed on touchscreen of computer

This test will be displayed on the touchscreen of a computer with a detachable keyboard. The 12.3-inch touchscreen will lie flat on the table for the duration of the test. The material used for this task was taken from Snodgrass and Vanderwart’s set of object pictures [ 54 ]. These images were improved with colour and texture [ 55 ], and ranked according to familiarity and image agreement. We selected the first 72 pictures, and randomly assigned them to one of the three contexts (i.e., 24 pictures per context). There were therefore 12 target pictures and 12 distractor pictures for each context. This task comes in two phases: picture location learning, and location recall.

Learning phase

Participants will have to match each of the 12 target pictures in the pile (lefthand side of screen) with an identical one hidden in the grid. In order to encourage them to explore the whole grid and avoid making the test too simple, for each trial, four boxes will be blocked: the box identified in the previous trial, and three others at random. When a pair is found, the relevant picture on the lefthand side of the screen will be marked with a green tick, and the participant will no longer be able to click on it. The learning phase will end when all the pairs have been constructed. To make the social memory task appear more like a game, participants will have to match each pair before the cursor on the righthand side of the screen is lowered. To avoid setting participants up for failure, the cursor will only be lowered after a pair is found, and not after each trial.

Recall phase

The recall phase will take place after an interference task (completion of a questionnaire), and will consist of picture recognition and grid location recall. For the picture recognition, all 24 pictures (i.e., 12 targets and 12 distractors) will be displayed one by one in random order. For each picture, participants will have to say whether or not they saw the picture in the learning phase. For the location recall, the 12 target pictures will be displayed one by one, and participants will have to place each one in the box it occupied in the learning phase.

Procedure for social memory task

The learning phase will take place in three social contexts: (1) presence-only context, where the participant performs the task alone, with the dyadic partner present in the same room but behind a screen; (2) observation context, where the participant observes the dyadic partner performing the task; and (3) collaboration context, where the participant and the dyadic partner perform the task together. Each dyad will perform the task in all three contexts. The order of the three contexts will be randomly counterbalanced across the dyads.

During the learning phase, regardless of the context (presence-only, observation, or collaboration), the participant, dyadic partner, and experimenter will all be present in the room (see Fig.  2 for room layout). The tables of the two members of the dyad will be placed side by side facing a wall, and the experimenter’s desk will be perpendicular to these two tables, more than 1 m away. In the presence-only context, the touchscreen will be placed between the participant’s table and the dyadic partner’s table. The dyadic partner will complete a questionnaire during this learning phase. In the observation and collaboration contexts, the dyadic members’ tables will be fixed together, and the touchscreen will be placed in the middle.

Layout of experiment room in ( A ) presence-only condition, ( B ) observation condition, ( C ) collaboration condition, and ( D ) recall phase

During in the recall phase, only the participant and the experimenter will be present in the room. The participant will be seated at the same table as the experimenter.

Outcome of social memory task

For the learning phase, and for each social context, several scores will be collected: the number of trials performed before all the pairs of pictures are found, the number of clicks on each picture on the lefthand side of the screen, and on each picture in the grid. We will examine whether these scores are related to delayed recall performance, to establish whether a picture is remembered better and more accurately located if it has been seen several times.

For the picture recognition in the recall phase, we will record the numbers of correct responses (i.e., saying “yes” when the picture has previously been seen, and saying “no” when the picture has not been seen before) for each social context.

Finally, for the location recall, we will calculate an overall score for each social context. If a picture is correctly located in the grid, 1 point will be awarded, and if it is located in an adjacent box, 0.5 point will be awarded. If not, 0 point will be awarded.

We will assess whether social context influences the picture recognition and location recall scores, and use these scores to identify factors that could explain potential differences in recognition performance between social learning contexts.

To clarify the psychological mechanisms associated with performance in different social contexts, a debriefing phase will take place at the end of each learning session. Using Likert-type scales, participants will rate their confidence in the quality of the learning achieved, their pleasure and satisfaction during the task, their anxiety during the learning process, the effort put into learning, the attention paid to the task, and perceived task difficulty.

Neuropsychological and clinical data

All the questionnaires and tests that will be used in this study are listed below. The scores they yield will allow us to identify determinants (social interactions, social abilities, cognitive skills, and personality traits) that could explain the potentially beneficial impact of the different social contexts on participants’ learning. These will then be used to define explanatory variables for differences in memory performance depending on social learning context.

Social interactions and social cognition

Three questionnaires will be used to identify the patients’ social interaction skills, in order to find out which social variables could explain differences in performance between the collaboration or observation context and the presence-only context. These questionnaires assess the quality of people’s social relationships [ 56 ], changes in socio-emotional [ 57 ] and exo-/egocentric behaviour since the onset of the disease, and social vulnerability [ 58 ].

Social cognition skills will be assessed with an empathy questionnaire [ 59 , 60 ], and three tests developed in our research unit: a test of knowledge of social rules [ 61 ], a preference judgment test adapted from [ 62 ], and a social learning information test.

Memory abilities

This protocol will include several tests used in clinical neuropsychology to determine memory abilities. More specifically, we will assess visual episodic memory [ 63 ], working memory [ 64 ], and semantic memory [ 65 , 66 ].

Personality traits

Personality traits will be assessed with the Temperament and Character Inventory developed by Cloninger [ 67 ]. This inventory is based on two fundamental components: temperament, which qualitatively describes the psychobiological dimensions of the individual’s personality; and character, which describes the individual’s levels of adaptation and maturity. The two subscales that are most relevant for the purposes of the study concern the character component. They are Self-Directedness , which corresponds to the level of individual maturity, and Cooperativeness , which reflects social maturity.

Additional assessment

Global cognitive impairment will be assessed using the MoCA [ 51 ]. Finally, executive functions in everyday life will be assessed with a questionnaire [ 68 ]. An MRI scan may be carried out to document patients’ brain lesions.

Demographic data

The relationship with the dyadic partner will be documented: nature of relationship (relative, friend, other), feeling of closeness, and number of years they have known each other. Some demographic information will also be collected: age, sex, highest degree obtained, living alone or not.

MEM&SO study procedure

MEM&SO is a comparative cross-sectional study. Participants will be recruited from several centres (Caen, Rennes, Rouen, and memory clinics in Normandy). HC will be recruited in and around Caen, and the examinations of all participants will be performed in Caen. Registered on ClinicalTrials.gov (no. NCT05800028).

After the inclusion phase, the study will take place over 3 half-days, corresponding to the three study visits (see Fig.  3 ). Each visit will last 90 min, with a 10-minute break in the middle.

Schedule for assessment visits. *The order of the three contexts will be randomly counterbalanced across the dyads

Data analysis plan

Each participant’s performance in the presence-only context will serve as a reference score, and a differential score will be calculated by subtracting this reference score from the participant’s memory performance in the collaboration context. The z score of this differential score will then be calculated, based on the mean and standard deviation for the HC group. This will show whether each patient’s differential score differs significantly from that of the HC group. The same analyses will then be carried out for performance in the observation context.

As this study is designed to identify determinants (social interactions and abilities, cognitive skills, and personality traits) that could explain the differential scores for the collaboration and observation contexts, we will carry out multiple comparative case analyses [ 48 , 49 , 50 ], based on the differential scores for the collaboration context of at least three patients whose memory performance is substantially better in the collaboration condition, compared with the presence only condition, and for three patients whose performance is only minimally better. These six patients will be compared on their social skills (social interactions and abilities), personality (especially the cooperation and determination dimensions), and memory performance in classic neuropsychological tests. Following a logic of replication, other patients will be added to the analyses, to see whether a regular pattern emerges. The same analyses will be conducted on differential scores for the observation context.

Given that individuals with neurodegenerative diseases may need to learn new information despite their cognitive disorders, establishing whether social interaction favours this learning could lead to the development of new cognitive rehabilitation techniques. The purpose of the present study will therefore be to find out whether a particular social context can improve the memory performance of patients with neurodegenerative diseases, and shed light on cognitive, social and personality variables that could subtend differences in memory performance between social learning contexts (i.e., collaboration or observation vs. presence-only).

As this is a multiple comparative case study, cognitive, social and personality profiles will be established for each target population, in order to identify the most beneficial learning context for each patient. Depending on the patient’s profile, strategies can then be devised to compensate for cognitive deficits.

The MEM&SO protocol will therefore have both theoretical and practical implications in the field of social cognition, bringing social psychology concepts to neuropsychology.

Data availability

Not applicable.

Abbreviations

Memory and social interactions in neurodegenerative disease

• Alzheimer disease

Semantic variant of primary progressive aphasia

Theory of mind

Healthy control

Montreal cognitive assessment

McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR, Kawas CH, et al. The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 2011;7:263–9.

Article   PubMed   Google Scholar  

Ergis A-M, Eusop-Roussel E. Les troubles précoces de la mémoire épisodique dans la maladie d’Alzheimer. Rev Neurol (Paris). 2008;164:S96–101.

Eustache F, Faure S, Desgranges B. Manuel De Neuropsychologie – 5e éd. Dunod; 2018.

Gorno-Tempini ML, Hillis AE, Weintraub S, Kertesz A, Mendez M, Cappa SF, et al. Classification of primary progressive aphasia and its variants. Neurology. 2011;76:1006–14.

Article   PubMed   PubMed Central   Google Scholar  

Younes K, Borghesani V, Montembeault M, Spina S, Mandelli ML, Welch AE, et al. Right temporal degeneration and socioemotional semantics: semantic behavioural variant frontotemporal dementia. Brain. 2022;145:4080–96.

Belliard S, Bon L, LeMoal S, Jonin PY, Vercelletto M, LeBail B. Semantic dementia. Psychol Neuropsychiatr Vieil. 2007;5:127–38.

PubMed   Google Scholar  

Moreaud O, Belliard S, Snowden J, Auriacombe S, Basaglia-Pappas S, Bernard F, et al. Démence sémantique: réflexions d’un Groupe De travail pour des critères de diagnostic en français et la constitution d’une cohorte de patients. Rev Neurol (Paris). 2008;164:343–53.

Clare L, Jones RSP. Errorless learning in the rehabilitation of memory impairment: a critical review. Neuropsychol Rev. 2008;18:1–23.

Middleton EL, Schwartz MF. Errorless learning in cognitive rehabilitation: a critical review. Neuropsychol Rehabil. 2012;22:138–68.

Jokel R, Anderson ND. Quest for the best: effects of errorless and active encoding on word re-learning in semantic dementia. Neuropsychol Rehabil. 2012;22:187–214.

Jokel R, Rochon E, Anderson ND. Errorless learning of computer-generated words in a patient with semantic dementia. Neuropsychol Rehabil. 2010;20:16–41.

Glisky EL. Acquisition and transfer of word processing skill by an amnesic patient. Neuropsychol Rehabil. 1995;5:299–318.

Article   Google Scholar  

Kessels RPC, Haan EHF. Implicit learning in memory rehabilitation: a meta-analysis on errorless learning and vanishing cues methods. J Clin Exp Neuropsychol. 2003;25:805–14.

Schacter DL, Rich SA, Stampp MS. Remediation of memory disorders: experimental evaluation of the Spaced-Retrieval technique. J Clin Exp Neuropsychol. 1985;7:79–96.

Erkes J, Raffard S, Meulemans T. Utilisation de la technique de récupération espacée dans la prise en charge des patients atteints de maladie d’Alzheimer. Revue critique et applications cliniques. Psychol Neuropsychiatr Vieil. 2009;7:275–86.

Bier N, Macoir J, Gagnon L, van der Linden M, Louveaux S, Desrosiers J. Known, lost, and recovered: efficacy of formal-semantic therapy and spaced retrieval method in a case of semantic dementia. Aphasiology. 2009;23:210–35.

Suárez-González A, Savage SA, Caine D. Successful short-term re-learning and generalisation of concepts in semantic dementia. Neuropsychol Rehabil. 2018;28:1095–109.

Bozeat S, Patterson K, Hodges J. Relearning object use in semantic dementia. Neuropsychol Rehabil. 2004;14:351–63.

Grandmaison E, Simard M. A critical review of memory stimulation programs in Alzheimer’s disease. J Neuropsychiatry Clin Neurosci. 2003;15:130–44.

Volkmer A, Spector A, Meitanis V, Warren JD, Beeke S. Effects of functional communication interventions for people with primary progressive aphasia and their caregivers: a systematic review. Aging Ment Health. 2020;24:1381–93.

Pessin J, Husband RW. Effects of social stimulation on human maze learning. J Abnorm Soc Psychol. 1933;28:148–54.

Deffenbacher KA, Platt GJ, Williams MA. Differential recall as a function of socially induced arousal and retention interval. J Exp Psychol. 1974;103:809.

Geen RG. Effects of being observed on short- and long-term recall. J Exp Psychol. 1973;100:395–8.

Eastvold AD, Belanger HG, Vanderploeg RD. Does a third party observer affect neuropsychological test performance? It depends. Clin Neuropsychol. 2012;26:520–41.

Zajonc RB. Social facilitation. Science. 1965;149:269–74.

Cottrell NB, Wack DL, Sekerak GJ, Rittle RH. Social facilitation of dominant responses by the presence of an audience and the mere presence of others. J Pers Soc Psychol. 1968;9:245–50.

Danchin É, Giraldeau L-A, Valone TJ, Wagner RH. Public information: from nosy neighbors to cultural evolution. Science. 2004;305:487–91.

Grüter C, Leadbeater E, Ratnieks FLW. Social learning: the importance of copying others. Curr Biol. 2010;20:R683–5.

Rendell L, Boyd R, Cownden D, Enquist M, Eriksson K, Feldman MW, et al. Why copy others? Insights from the social learning strategies tournament. Science. 2010;328:208–13.

van Tilborg IADA, Kessels RPC, Hulstijn W. Learning by observation and guidance in patients with Alzheimer’s dementia. NeuroRehabilitation. 2011;29:295–304.

Harris CB, Barnier AJ, Sutton J, Keil PG. Couples as socially distributed cognitive systems: remembering in everyday social and material contexts. Mem Stud. 2014;7:285–97.

Andrews JJ, Rapp DN. Benefits, costs, and challenges of collaboration for learning and memory. Transl Issues Psychol Sci. 2015;1:182–91.

Chalmeau R, Gallo A. La coopération Chez les primates. Année Psychol. 1995;95:119–30.

Barron B. Achieving coordination in collaborative problem-solving groups. J Learn Sci. 2000;9:403–36.

Gaudet AD, Ramer LM, Nakonechny J, Cragg JJ, Ramer MS. Small-group learning in an upper-level university biology class enhances academic performance and student attitudes toward group work. PLoS ONE. 2010;5:e15821.

Derksen BJ, Duff MC, Weldon K, Zhang J, Zamba KD, Tranel D, et al. Older adults catch up to younger adults on a learning and memory task that involves collaborative social interaction. Memory. 2015;23:612–24.

Duff MC, Gallegos DR, Cohen NJ, Tranel D. Learning in Alzheimer’s disease is facilitated by social interaction. J Comp Neurol. 2013;521:4356–69.

Paek EJ, Yoon SO. Partner-specific communication deficits in individuals with Alzheimer’s disease. Am J Speech Lang Pathol. 2021;30:376–90.

Yoon SO, Duff MC, Brown-Schmidt S. Learning and using knowledge about what other people do and don’t know despite amnesia. Cortex J Devoted Study Nerv Syst Behav. 2017;94:164–75.

Laisney M, Bon L, Guiziou C, Daluzeau N, Eustache F, Desgranges B. Cognitive and affective theory of mind in mild to moderate Alzheimer’s disease. J Neuropsychol. 2013;7:107–20.

Duclos H, Bejanin A, Eustache F, Desgranges B, Laisney M. Role of context in affective theory of mind in Alzheimer’s disease. Neuropsychologia. 2018;119:363–72.

Duclos H, Desgranges B, Eustache F, Laisney M. Impairment of social cognition in neurological diseases. Rev Neurol (Paris). 2018;174:190–8.

Duval C, Bejanin A, Piolino P, Laisney M, de La Sayette V, Belliard S, et al. Theory of mind impairments in patients with semantic dementia. Brain. 2012;135:228–41.

Kumfor F, Landin-Romero R, Devenney E, Hutchings R, Grasso R, Hodges JR, et al. On the right side? A longitudinal study of left- versus right-lateralized semantic dementia. Brain. 2016;139:986–98.

Thompson SA, Patterson K, Hodges JR. Left/right asymmetry of atrophy in semantic dementia: behavioral–cognitive implications. Neurology. 2003;61:1196–203.

Ulugut Erkoyun H, Groot C, Heilbron R, Nelissen A, van Rossum J, Jutten R, et al. A clinical-radiological framework of the right temporal variant of frontotemporal dementia. Brain. 2020;143:2831–43.

Belliard S, Jonin PY, Merck C. Actualités sur la démence sémantique. Rev Neuropsychol. 2010;2:31–7.

Google Scholar  

Eisenhardt KM. Building theories from case study research. Acad Manage Rev. 1989;14:532–50.

Eisenhardt KM. Better stories and better constructs: the case for rigor and comparative logic. Acad Manage Rev. 1991;16:620–7.

Eisenhardt KM, Graebner ME. Theory building from fases: opportunities and challenges. Acad Manage J. 2007;50:25–32.

Nasreddine ZS, Phillips NA, Bédirian V, Charbonneau S, Whitehead V, Collin I, et al. The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc. 2005;53:695–9.

Bejanin A, Chételat G, Laisney M, Pélerin A, Landeau B, Merck C, et al. Distinct neural substrates of affective and cognitive theory of mind impairment in semantic dementia. Soc Neurosci. 2017;12:287–302.

Bertoux M, Duclos H, Caillaud M, Segobin S, Merck C, de La Sayette V, et al. When affect overlaps with concept: emotion recognition in semantic variant of primary progressive aphasia. Brain. 2020;143:3850–64.

Snodgrass JG, Vanderwart M. A standardized set of 260 pictures: norms for name agreement, image agreement, familiarity, and visual complexity. J Exp Psychol [Hum Learn]. 1980;6:174–215.

Rossion B, Pourtois G. Revisiting Snodgrass and Vanderwart’s object Pictorial Set: the role of surface detail in basic-level object recognition. Perception. 2004;33:217–36.

Thomacos N. A measure to assess attachment and social support: the interview schedule for Social Interaction – questionnaire version (ISSI-Q). Stress Anxiety - Theory Pract Meas. 2020;:129–44.

Beni C, Rochat L, Malysse N, Delecroix H, Arnould A, Azouvi P, et al. L’échelle Des changements de comportements socio-émotionnels de genève (ECCSEG): validation auprès de patients victimes d’un traumatisme crânio-cérébral. [The Geneva Scale of Socio-emotional Behavior Change (ECCSEG): validation with patients suffering from cranio-cerebral trauma]. Can J Behav Sci Rev Can Sci Comport. 2017;49:7–17.

Pinsker DM, Stone V, Pachana N, Greenspan S. Social Vulnerability Scale for older adults: Validation study*. Clin Psychol. 2006;10:109–19.

Gilet A-L, Mella N, Studer J, Grühn D, Labouvie-Vief G. Assessing dispositional empathy in adults: a French validation of the interpersonal reactivity index (IRI). Can J Behav Sci Rev Can Sci Comport. 2013;45:42–8.

Davis M. A multidimensional approach to individual differences in empathy. JSAS Cat Sel Doc Psychol. 1980;10.

Duclos H, de La Sayette V, Bonnet A-L, Viard A, Eustache F, Desgranges B, et al. Social cognition in the frontal variant of Alzheimer’s disease: a case study. J Alzheimers Dis. 2017;55:459–63.

Baron-Cohen S, Campbell R, Karmiloff-Smith A, Grant J, Walker J. Are children with autism blind to the mentalistic significance of the eyes? Br J Dev Psychol. 1995;13:379–98.

Violon A, Wijns C. Test De La Ruche. Test de perception et d’apprentissage progressif en mémoire visuelle. L’application Tech Mod; 1984.

Wechsler D. WAIS-IV: échelle d’intelligence de wechsler pour adultes. Éditions du Centre de psychologie appliquée; 2011.

Godefroy O. le GREFEX. Fonctions exécutives et pathologies neurologiques et psychiatriques: évaluation en pratique clinique. Groupe de Boeck. 2008.

Merck C, Charnallet A, Auriacombe S, Belliard S, Hahn-Barma V, Kremin H, et al. La Batterie d’évaluation des connaissances sémantiques du GRECO (BECS-GRECO): validation et données normatives. Rev Neuropsychol. 2011;3:235–55.

Cloninger CR, Svrakic DM, Przybeck TR. A psychobiological model of temperament and character. Arch Gen Psychiatry. 1993;50:975–90.

Roy A, Besnard J, Fournet N, Lancelot C, Le Gall D. Brief-A inventaire d’Evaluation Comportementale Des Fonctions Exécutives—Version Adulte—Adaptation Française. Hogrèfe Fr Ed Paris; 2015.

Download references

Acknowledgements

We would like to thank Donna Pinsker for providing us with the Social Vulnerability Scale, and Nikos Thomacos for the Interview Schedule for Social Interaction. We also thank Alice Pélerin and Alice Poissonnier, who will be responsible for recruiting the patients and carrying out the experiments.

This work is supported by a Fondation France Alzheimer et Maladies Apparentées grant and by Normandy Regional Council (PhD funding for Pauline Saliou).

Author information

Authors and affiliations.

Inserm, U1077, EPHE, UNICAEN, Normandie Université, PSL Université Paris, CHU de Caen, GIP Cyceron, Neuropsychologie et Imagerie de la Mémoire Humaine (NIMH), Caen, 14000, France

Pauline Saliou, Julien Chavant, Vincent de La Sayette, Olivier Martinaud, Francis Eustache & Mickaël Laisney

Département de Neurologie, CHU Pontchaillou, Rennes, France

Serge Belliard & Catherine Merck

Univ Rouen Normandie, Normandie Univ, Inserm U1245 and CHU Rouen, Department of Neurology and CNRMAJ, Rouen, F- 76000, France

David Wallon

You can also search for this author in PubMed   Google Scholar

Contributions

OM is the principal investigator. VdLS, SB, and DW are associate investigators. ML is the scientific investigator, and designed the study, obtained funding, and supervised the drafting of the manuscript and the applications for ethical approval. ML, FE and PS reflected on the theoretical aspects of the project in collaboration with OM, VdLS, SB, CM and DW, who contributed their clinical experience. ML and PS designed the experimental tasks. JC thought about the structure of the hardware and then developed the computerized tasks. PS drafted the manuscript, and the applications for ethical approval. PS will implement this procedure with patients and healthy controls. All authors contributed to the reviewing of the manuscript and approved the final version.

Corresponding author

Correspondence to Mickaël Laisney .

Ethics declarations

Ethics approval and consent to participate.

The present study has been approved by an institutional review board (CPP Ile de France XI; no. ID-RCB: 2022-A00198-35). It will be conducted in accordance with the Declaration of Helsinki and good clinical practice. Participants must each give their free and informed consent in writing, prior to inclusion. The data we collect will be anonymized, and analyses will not include any information that could allow the individuals who participate in the research to be identified. The results will be disseminated via peer-review journals, conferences, and clinical meetings. The social memory task will be made available online for the benefit of the scientific community.

Consent for publication

Trial status.

The recruitment phase began in June 2023. To date, only 5 dyads (10 participants) have been included in the study (April 2024).

Patient and public involvement

This study is based on the fact that with advancing age, people sometimes have to learn new information, especially if they have neurodegenerative diseases. This learning may concern new people (home carers, staff of a residential facility), and new living environments. The purpose of the study is therefore to ask whether a particular social context enables people with a neurodegenerative disease to learn better. In order to set up the protocol to answer this question, funding was obtained from Fondation France Alzheimer et Maladies Apparentées, which brings together the families of patients with AD or related diseases. The scientific committee that assesses research projects is made up of doctors and researchers in the field of AD who are appointed by the board of directors. The study was presented at the World Alzheimer’s Day, in order to make contact with patients and their families, and initiate the recruitment process. Recruitment will also be possible through the various patient charities in the region. As for the study procedure, each patient will be paired with a longstanding caregiver. Regarding the new experimental tasks developed by our research unit, members of the public participated in pilot studies and gave their feedback (duration, difficulty, comprehension) so that our team could improve them. The long-term aim of the study is to produce guidelines for families and health professionals describing the most beneficial social learning context for individual patients according to their cognitive and social disorders and their personality. The aim is to draw up these guidelines with input from the patients’ families and health professionals.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Saliou, P., Chavant, J., Belliard, S. et al. MEM&SO protocol: understanding the determinants of social learning in neurodegenerative diseases. BMC Psychol 12 , 307 (2024). https://doi.org/10.1186/s40359-024-01791-w

Download citation

Received : 12 April 2024

Accepted : 15 May 2024

Published : 28 May 2024

DOI : https://doi.org/10.1186/s40359-024-01791-w

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Aphasia, primary progressive
• Social learning
• Personality
• Social interaction.

BMC Psychology

ISSN: 2050-7283

• General enquiries: [email protected]

Observational Case Studies

• First Online: 01 January 2014

Cite this chapter

• Roel J. Wieringa 2  

1 Citations

An observational case study is a study of a real-world case without performing an intervention. Measurement may influence the measured phenomena, but as in all forms of research, the researcher tries to restrict this to a minimum.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

• Available as EPUB and PDF
• Read on any device
• Instant download
• Own it forever
• Compact, lightweight edition
• Dispatched in 3 to 5 business days
• Free shipping worldwide - see info
• Durable hardcover edition

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

D. Damian, J. Chisan, An empirical study of the complex relationships between requirements engineering processes and other processes that lead to payoffs in productivity, quality and risk management. IEEE Trans. Softw. Eng. 32 (7), 433–453 (2006)

Article   Google Scholar  

M. Denscombe, The Good Research Guide For Small-Scale Social Research Projects , 4th edn. (Open University Press, Maidenhead, 2010)

Google Scholar  

K.M. Eisenhardt, Building theories from case study research. Acad. Manag. Rev. 14 (4), 532–550 (1989)

B. Flyvberg, Five misunderstandings about case-study research. Qual. Inq. 12 (2), 219–245 (2006)

R.L. Glass, Pilot studies: What, why, and how. J. Syst. Softw. 36 , 85–97 (1997)

M.M. Kennedy, Generalizing from single case studies. Eval. Q. 3 (4), 661–678 (1979)

B. Kitchenham, L. Pickard, S.L. Pfleeger, Case studies for method and tool evaluation. IEEE Softw. 12 (4), 52–62 (1995)

C. Robson, Real World Research , 2nd edn. (Blackwell, Oxford, 2002)

P. Runeson, M. Höst, A. Rainer, B. Regnell, Case Study Research in Software Engineering: Guidelines and Examples (Wiley, Hoboken, 2012)

Book   Google Scholar  

J.M. Verner, J. Sampson, V. Tosic, N.A.A. Bakar, B.A. Kitchenham, Guidelines for industrially-based multiple case studies in software engineering, in Research Challenges in Information Science, 2009. RCIS 2009. Third International Conference on , 2009, pp. 313–324

L. Warne, D. Hart, The impact of organizational politics on information systems project failure-a case study, in Proceedings of the Twenty-Ninth Hawaii International Conference on System Sciences , vol. 4, 1996, pp. 191–201

R.J. Wieringa, Towards a unified checklist for empirical research in software engineering: first proposal, in 16th International Conference on Evaluation and Assessment in Software Engineering (EASE 2012) , ed. by T. Baldaresse, M. Genero, E. Mendes, M. Piattini (IET, Ciudad Real, 2012), pp. 161–165

R.J. Wieringa, A unified checklist for observational and experimental research in software engineering (version 1). Technical Report TR-CTIT-12-07, Centre for Telematics and Information Technology University of Twente (2012)

R.K. Yin, Case Study research: Design and Methods (Sage, Thousand Oaks, 1984)

R.K. Yin, Case Study research: Design and Methods , 3rd edn. (Sage, Thousand Oaks, 2003)

Download references

Author information

Authors and affiliations.

University of Twente, Enschede, The Netherlands

Roel J. Wieringa

You can also search for this author in PubMed   Google Scholar

Rights and permissions

Reprints and permissions

Copyright information

© 2014 Springer-Verlag Berlin Heidelberg

About this chapter

Wieringa, R.J. (2014). Observational Case Studies. In: Design Science Methodology for Information Systems and Software Engineering. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-662-43839-8_17

Download citation

DOI : https://doi.org/10.1007/978-3-662-43839-8_17

Published : 20 August 2014

Publisher Name : Springer, Berlin, Heidelberg

Print ISBN : 978-3-662-43838-1

Online ISBN : 978-3-662-43839-8

eBook Packages : Computer Science Computer Science (R0)

Share this chapter

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Publish with us

Policies and ethics

• Find a journal
• Track your research

Academia.edu no longer supports Internet Explorer.

To browse Academia.edu and the wider internet faster and more securely, please take a few seconds to  upgrade your browser .

Enter the email address you signed up with and we'll email you a reset link.

• We're Hiring!
• Help Center

Case Study based on Albert Banduras theory

Related Papers

International Journal of Advanced Research (IJAR)

Aman Tiwari

This research is set out to carefully analyze the psyche of the criminals and to identify commonality in their behaviour irrespective of the demographic aspect. The research aims at analyzing criminal law theories and their practical usage in the modern scenario. Crime is defined as an act of deviance from socially accepted norms translated as criminal code. Countries all around the world have a definite set of criminal code compromising principles of morality and ethics as per their unique culture and society. But in practicality, these principles have failed to culminate the desired result of prevention of crime upon implementation by the traditional methods of enforcement. Deviance relates to the subjectivity of society. Considering morality as a subjective aspect would propagate deviance among individuals formed out of different circumstances than the majority. Hence, crime is often committed by the minority upon the majority in a society. The ultimate aim of the research is to identify the principle of causality in relation to crime and eventually portraying an effective approach for the prevention of crime.

Hande Sahin

— Crime is a complex phenomenon and therefore has many definitions. At the same time it is a major problem for the entire world. The crime rates and also juvenile crime has increased over the world. Therefore especially social sciences focus on subject of crime. There are also so many theories about crime. Each of them focuses on a different aspect of the crime. This study focuses on the social aspect of crime. The subject of this study is the relationship between crime and education. The study aims to reveal the crime can be learned. Learning theory is a functional to discuss all these aims. The main assumption of the study is that crime can be learned with process of socialization like other social phenomenon.Failed and inadequate socialization are key phenomenon in this process. People learn criminal behaviour through interaction with intimate person. Or they can commit crime because they don't know correct norms of society because of limited education. The method of the study is that nine master and three doctoral theses, four extensive field researches and government statistics about juvenile delinquency in Turkey have been evaluated related to the subject of the study. Their data has been collected. Some results have been achieved: The education level of juvenile offenders and their families are too low. The number of offenders in families and relatives of these children is quite.

Julius A Adinoyi

Nadia Rahaman

Aggressive Behavior

Anthony Mawson

Karen moore

Diana Shlyapnikova

Through many centuries, criminal behavior was mistakenly explained mostly with biological defects of the individual. It was believed, that abnormal actions, negatively affected society, were preferably conducted by those, who suffered from serious physical deviations in development. However, as the science was progressing, scholars have begun searching other possible causes of the formation of antisocial behavior and psychology began to actively explore this area. Nowadays, it is known, that incorrect parenting style, negative influence of environment, and formation of improper role models may significantly result inner conflicts and thus, led to the abnormal behavior. In this essay, we would further examine, how these aspects can influence the behavior of individuals and lead to criminal conduct, illustrating it with examples from biographies of famous criminals to show how concepts can be applied in reality. First of all, it is necessary to draw a distinction between biological and psychological factors, influencing the behavior of individual. Under biological factors we mean physical anomalies, such as neurotransmitter dysfunction, bradygenesis, and other problems of neural development, genetically predisposed or caused by trauma or injury. Under psychological factors we mean mental disorders, and deviation in education and mental development of the individual. However, despite the fact that the differences seem obvious, both areas are interconnected between deeply. As we know, the connection of nature and nature is rooted into every individual. As the course textbook states " everything psychological is simultaneously biological " (Meyers, 1999, Chapter 2, p.47). Thus, various theories, designed to explain the emerging and development of deviant behavior, combine both approaches. For example, theory of personality and crime by Hans J. Eysenck, British psychologist, states that each individual has innate hereditary predisposition towards asocial behavior, which discloses in certain circumstances. Thus, " criminal behavior is the result of an interaction between certain environmental conditions and features of the nervous system " (Bartol & Bartol, 2005). However, Eysenck does not say, that deviant behavior inborn, but may be caused by the compound of heredity and environment. It is not itself, or criminality that is innate; it is certain peculiarities of the central and autonomic nervous system that react with the environment, with upbringing, and many other environmental factors to increase the probability that a given person would act in a certain antisocial manner (Eysenck & Gudjonsson, 1989)

Journal of emerging technologies and innovative research

trimurthi rao

IJPUBLICATION

Srinivas Katherasala

Juvenile delinquency is an enormous problem in India by which most of the youth ruin their lives. On account of adolescent wrongdoing and relate issues youth, their families and the whole society endure numerous fallouts. Besides the fact that the issue influences the survivors of the wrongdoing; it additionally influences the adolescent reprobates family, The most significant result of violations perpetrated by adolescents conveys due to financial and mental issues which consider their relatives and the general public. The primary goal of this paper is to concentrate on the frequency of adolescent misconduct regarding mental viewpoints. The adolescent who carry out serious wrongdoings challenge their future to fight apparent maltreatments that have been executed against them. "Current cultures appear to see youngsters in a fairly irresolute way. The last many years of the 20th century gave us two or three telling models from a few nations. From one perspective, the kids and adolescents are seen as requiring care and security. Assuming that they become delinquent, instructive measures are believed to be the fitting in the event that not vital response to the issues they are causing or may experience the ill effects of. This is the overall disposition among everybody in most of nations on the planet, insofar as the offenses committed by youths don't rise steeply in numbers and stay frivolous or moderate in quality.

Cumhuriyet Medical Journal

RELATED PAPERS

Sandor Markon

Talita de Souza Dias

Nouf Alotaiby

TESOL Journal

Natasha Heny

Parent-Child Relations: A Guide to Raising Children (Revised Edition)

Dr. Hisham Altalib د. هشام الطالب

Giuliano Guzzone

Uniedusul Editora eBooks

Salete Zanella

Ilaria Brambilla

Netherlands Journal of Geosciences - Geologie en Mijnbouw

Anne Schulp

Veterinary Parasitology: Regional Studies and Reports

Marcelo Molento

Sveikatos mokslai

Nora Šiupšinskienė

Journal of Clinical Medicine

Chiara Busso

Revista Chilena De Literatura

Aquaculture International

Florence Cornette

Paolo De Stefani

Duglas choc

International Review for Spatial Planning and Sustainable Development

Bian Lanchun

Nicoletta Marcelli

Pediatric Anesthesia

Narasimhan Jagannathan

RELATED TOPICS

•   We're Hiring!
•   Help Center
• Find new research papers in:
• Health Sciences
• Earth Sciences
• Cognitive Science
• Mathematics
• Computer Science
• Academia ©2024