Skip to main content

Emotional Learning in Social Interaction

Final Report Summary - ELSI (Emotional Learning in Social Interaction)

The project Emotional Learning in Social Interaction (ELSI) has added in several important ways to our knowledge about how learning about emotional information is affected by social information, and how such learning affects decision-making. First, it has provided new information about the behavioral, neural and computational mechanisms enabling social learning about threatening events (people and objects), for example how threat information is transmitted between individuals trough observation. We have shown that this learning is stronger than previously believed. Indeed, we have shown that this indirect or “vicarious” learning can affect our behavior and decisions even when it contradicts our own personal experiences. In separate studies, we have established that fear learning from another individual is facilitated through (1) enhanced empathy and physiological synchronization with the individual’s expressions of pain and/or anxiety; (2) higher perceived dominance and level of intentionality of the individual; and (3) shared social (racial, ethnic and cultural) group identity. We have shown this using measures of both verbal and non-verbal behavior, as well as physiology; bodily arousal and brain activity. In addition, ELSI has provided evidence of that vicarious threat learning is dependent on specific systems of endogenous chemical components, such as opioids that regulate both self-experienced and vicarious pain, as well as the updating of threat learning through both of these routes.
In terms of the underlying brain mechanisms, ELSI has shown that vicarious threat learning partially draws on the same mechanisms as direct (Pavlovian) threat learning, including basic regions coding for basic emotional qualities and learning, such as the amygdala, anterior cingulate and the anterior insula cortices. In addition, it is distinguished by the interaction of these mechanisms with brain processes known from social and cognitive neuroscience, involving the medial wall of the prefrontal cortex, and more posterior parts of the brain known to contribute to the attribution of mental states. Our results have also provided details about the computational mechanisms underlying this form of learning, in other words, revealing not only where learning occurs, but also how it is instantiated in the brain.
Our studies on patients suffering from autism spectrum disorders (ASD), characterized by the impaired ability to attribute mental states to others, show both similarities and differences in the ability to learn through observation. In brief, our studies suggest that different types of ‘strategies’ to express successful social learning might be used in ASD patients and typically developed individuals.
ELSI has provided the first evidence of that safety learning through observation of a ‘safe’ other individual outperforms direct, personal, safety learning (i.e. traditional extinction). Of particular note is our finding that social safety learning blocks the return of fear, hinting towards a previously unknown mechanism of ‘unlearning’, highlighting possible routes of treatment for anxiety disorders. Interestingly, similar to the acquisition of fear via observation, social safety learning is dependent on social group belonging. For example, observational safety learning is only successful when the ‘safe’ individual belongs to the learner’s social in-group.
Additional studies have proposed a novel psychological mechanism that can help to explain how observational fear learning is transmitted across generations of learners (modeled in the lab by the transmission between multiple naive participants, and simulated in a virtual environment of interacting agents), providing new insights into the possible evolutionary role of this basic kind of social learning. Addressing the objective to study learning to fear or avoid others, ELSI has so far provided new evidence that the bias to stronger associate individuals from another, as compared to the own, social group with an aversive outcome, is generalized to actual, interactive, behaviors. Our research has also offered the first description of the brain processes underlying such discriminating behaviors. Finally, ELSI has generated several of novel experimental paradigms that are now used to study the mechanisms underlying social interaction in dynamic situations, and how learning unfolds over time during active turn-taking. These and several others novel paradigms are currently developed further and will continue to enable our understanding of emotional learning in social interaction.