Social Concepts Simplify Complex Reinforcement Learning.

Humans often generalize rewarding experiences across abstract social roles. Theories of reward learning suggest that people generalize through model-based learning, but such learning is cognitively costly. Why do people seem to generalize across social roles with ease? Humans are social experts who easily recognize social roles that reflect familiar semantic concepts (e.g., "helper" or "teacher"). People may associate these roles with model-free reward (e.g., learning that helpers are rewarding), allowing them to generalize easily (e.g., interacting with novel individuals identified as helpers). In four online experiments with U.S. adults (N = 577), we found evidence that social concepts ease complex learning (people generalize more and at faster speed) and that people attach reward directly to abstract roles (they generalize even when roles are unrelated to task structure). These results demonstrate how familiar concepts allow complex behavior to emerge from simple strategies, highlighting social interaction as a prototype for studying cognitive ease in the face of environmental complexity.

Propagation of Economic Inequality Through Reciprocity and Reputation.

Reciprocity and reputation are powerful tools for encouraging cooperation on a broad scale. Here, we highlight a potential side effect of these social phenomena: exacerbating economic inequality. In two novel economic games, we manipulated the amount of money with which participants were endowed and then gave them the opportunity to share resources with others. We found that people reciprocated more toward higher-wealth givers, compared with lower-wealth givers, even when those givers were equally generous. Wealthier givers also achieved better reputations than less wealthy ones and therefore received more investments in a social marketplace. These discrepancies were well described by a formal model of reinforcement learning: Individuals who weighted monetary outcomes, rather than generosity, when learning about interlocutors also most strongly helped wealthier individuals. This work demonstrates that reciprocity and reputation-although globally increasing prosociality-can widen wealth gaps and provides a precise account of how inequality grows through social processes.

On the neural implausibility of the modular mind: Evidence for distributed construction dissolves boundaries between perception, cognition, and emotion.

Firestone & Scholl (F&S) rely on three problematic assumptions about the mind (modularity, reflexiveness, and context-insensitivity) to argue cognition does not fundamentally influence perception. We highlight evidence indicating that perception, cognition, and emotion are constructed through overlapping, distributed brain networks characterized by top-down activity and context-sensitivity. This evidence undermines F&S's ability to generalize from case studies to the nature of perception.

Simplifying social learning.

Social learning is complex, but people often seem to navigate social environments with ease. This ability creates a puzzle for traditional accounts of reinforcement learning (RL) that assume people negotiate a tradeoff between easy-but-simple behavior (model-free learning) and complex-but-difficult behavior (e.g., model-based learning). We offer a theoretical framework for resolving this puzzle: although social environments are complex, people have social expertise that helps them behave flexibly with low cognitive cost. Specifically, by using familiar concepts instead of focusing on novel details, people can turn hard learning problems into simpler ones. This ability highlights social learning as a prototype for studying cognitive simplicity in the face of environmental complexity and identifies a role for conceptual knowledge in everyday reward learning.

Two modes of social impressions and their effects on choice.

Our memories of other people shape how we interact with them. Yet, even when we forget exactly what others said or did, we often remember impressions that capture a general gist of their behavior-whether they were forthright, friendly, or funny. Drawing on fuzzy trace theory, we propose two modes of social impression formation: impressions formed based on ordinal gist ("more competent," "less competent") or categorical gist ("competent," "incompetent"). In turn, we propose that people gravitate toward the simplest representation available and that different modes of memory have distinct consequences for social decisions. Specifically, ordinal impressions lead people to make decisions based on an individual's standing relative to others, whereas categorical impressions lead people to make decisions based on discrete classifications that interpret behavior. In four experiments, participants learned about two groups of individuals who differed in their competence (Studies 1a, 2, and 3) or generosity (Study 1b). When participants encoded impressions as ordinal rankings, they preferred to hire or help a relatively good target from a low-performing group over a relatively bad target from a high-performing group, even though both targets behaved identically and accuracy was incentivized. However, when participants could use categorical boundaries to interpret behavior, this preference was eliminated. In a final experiment, changing the category participants used to encode others' generosity changed their impressions, even when accounting for memory for verbatim details. This work links social impressions to theories of mental representation in memory and judgment, highlighting how distinct representations support divergent patterns of social decision-making. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Instrumental learning of social affiliation through outcome and intention.

To build social ties, humans need to find others who want to interact with them. How do people learn, over time, to interact with partners who want to affiliate with them? Theories of social cognition suggest that people try to infer whether others value them, but theories of instrumental learning suggest that rewarding outcomes reinforce choices. In three studies, we provide evidence that both social acceptance outcomes and cues to a partner's acceptance intentions reinforce social partner choices. Even when outcomes were experimentally dissociated from a partner's intentions, outcomes influenced how people felt, which partners people chose, and how well people believed they were liked by partners. Finally, people acted kindlier both to partners who demonstrated acceptance intentions and to partners who provided acceptance outcomes. These findings support an integrative instrumental learning model of social affiliation, wherein social cognition and rewarding outcomes jointly shape affect, partner choice, and prosocial behavior. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Context-dependent learning in social interaction: Trait impressions support flexible social choices.

[Correction Notice: An Erratum for this article was reported in Vol 123(4) of Journal of Personality and Social Psychology (see record 2023-02979-001). In the error, the Study 2 heading Computational Mode of Learning should instead appear as Computational Model of Learning. All versions of this article have been corrected.] How do humans learn, through social interaction, whom to depend on in different situations? We compared the extent to which inferred trait attributes-as opposed to learned reward associations previously examined as part of feedback-based learning-could adaptively inform cross-context social decision-making. In four experiments, participants completed a novel task in which they chose to "hire" other players to solve math and verbal questions for money. These players varied in their trait-level competence across these contexts and, independently, in the monetary rewards they offered to participants across contexts. Results revealed that participants chose partners primarily based on context-specific traits, as opposed to either global trait impressions or material rewards. When making choices in novel contexts-including determining who to choose for social and emotional support-participants generalized trait knowledge from past contexts that required similar traits. Reward-based learning, by contrast, demonstrated significantly weaker context-sensitivity and generalization. These findings suggest that people form context-dependent trait impressions from interactive feedback and use this knowledge to make flexible social decisions. These results support a novel theoretical account of how interaction-based social learning can support context-specific impression formation and adaptive decision-making. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Shifting prosocial intuitions: neurocognitive evidence for a value-based account of group-based cooperation.

Cooperation is necessary for solving numerous social issues, including climate change, effective governance and economic stability. Value-based decision models contend that prosocial tendencies and social context shape people's preferences for cooperative or selfish behavior. Using functional neuroimaging and computational modeling, we tested these predictions by comparing activity in brain regions previously linked to valuation and executive function during decision-making-the ventromedial prefrontal cortex (vmPFC) and dorsolateral prefrontal cortex (dlPFC), respectively. Participants played Public Goods Games with students from fictitious universities, where social norms were selfish or cooperative. Prosocial participants showed greater vmPFC activity when cooperating and dlPFC-vmPFC connectivity when acting selfishly, whereas selfish participants displayed the opposite pattern. Norm-sensitive participants showed greater dlPFC-vmPFC connectivity when defying group norms. Modeling expectations of cooperation was associated with activity near the right temporoparietal junction. Consistent with value-based models, this suggests that prosocial tendencies and contextual norms flexibly determine whether people prefer cooperation or defection.

Model-Based and Model-Free Social Cognition: Investigating the Role of Habit in Social Attitude Formation and Choice.

Do habits play a role in our social impressions? To investigate the contribution of habits to the formation of social attitudes, we examined the roles of model-free and model-based reinforcement learning in social interactions - computations linked in past work to habit and planning, respectively. Participants in this study learned about novel individuals in a sequential reinforcement learning paradigm, choosing financial advisors who led them to high- or low-paying stocks. Results indicated that participants relied on both model-based and model-free learning, such that each type of learning was expressed in both advisor choices and post-task self-reported liking of advisors. Specifically, participants preferred advisors who could provide large future rewards as well as advisors who had provided them with large rewards in the past. Although participants relied more heavily on model-based learning overall, they varied in their use of model-based and model-free learning strategies, and this individual difference influenced the way in which learning related to self-reported attitudes: among participants who relied more on model-free learning, model-free social learning related more to post-task attitudes. We discuss implications for attitudes, trait impressions, and social behavior, as well as the role of habits in a memory systems model of social cognition.

Computational neuroscience approaches to social cognition.

How do we form impressions of people and groups and use these representations to guide our actions? From its inception, social neuroscience has sought to illuminate such complex forms of social cognition, and recently these efforts have been invigorated by the use of computational modeling. Computational modeling provides a framework for delineating specific processes underlying social cognition and relating them to neural activity and behavior. We provide a primer on the computational modeling approach and describe how it has been used to elucidate psychological and neural mechanisms of impression formation, social learning, moral decision making, and intergroup bias.

Frontal cortical effects on feedback processing and reinforcement learning: Relation of EEG asymmetry with the feedback-related negativity and behavior.

Reinforcement learning refers to the acquisition of approach or avoidance action tendencies through repeated reward/nonreward feedback. Although much research on reinforcement learning has focused on the striatum, the prefrontal cortex likely modulates this process. Given prior research demonstrating a consistent pattern of lateralized frontal cortical activity in affective responses and approach/avoidance tendencies in the EEG literature, we aimed to elucidate the role of frontal EEG asymmetry in reinforcement learning. Thirty-two participants completed a probabilistic selection task in which they learned to select some targets and avoid others though correct/incorrect feedback. EEG indices of frontal cortical asymmetry were computed from alpha power recorded at baseline and during task completion. We also examined the feedback-related negativity ERP component to assess feedback processing associated with activity in the dorsal anterior cingulate cortex. Results revealed that greater right-lateralized frontal cortical activity during learning was associated with better avoidance learning, but neither left- nor right-sided asymmetry reliably related to approach learning. Results also suggested that left frontal activity may relate to reinforcement feedback processing, as indicated by the feedback-related negativity (FRN). These findings offer preliminary evidence regarding the role of frontal cortical activity in reinforcement learning while integrating classic and contemporary research on lateralized frontal cortical functions.

The Influence of Emotional State on Learning From Reward and Punishment in Borderline Personality Disorder.

Despite preliminary evidence that individuals with borderline personality disorder (BPD) demonstrate deficits in learning from corrective feedback, no studies have examined the influence of emotional state on these learning deficits in BPD. This laboratory study examined the influence of negative emotions on learning among participants with BPD (n = 17), compared with clinical (past-year mood/anxiety disorder; n = 20) and healthy (n = 23) controls. Participants completed a reinforcement learning task before and after a negative emotion induction. The learning task involved presenting pairs of stimuli with probabilistic feedback in the training phase, and subsequently assessing accuracy for choosing previously rewarded stimuli or avoiding previously punished stimuli. ANOVAs and ANCOVAs revealed no significant between-group differences in overall learning accuracy. However, there was an effect of group in the ANCOVA for postemotion induction high-conflict punishment learning accuracy, with the BPD group showing greater decrements in learning accuracy than controls following the negative emotion induction.

Social identity shapes social valuation: evidence from prosocial behavior and vicarious reward.

People frequently engage in more prosocial behavior toward members of their own groups, as compared to other groups. Such group-based prosociality may reflect either strategic considerations concerning one's own future outcomes or intrinsic value placed on the outcomes of in-group members. In a functional magnetic resonance imaging experiment, we examined vicarious reward responses to witnessing the monetary gains of in-group and out-group members, as well as prosocial behavior towards both types of individuals. We found that individuals' investment in their group-a motivational component of social identification-tracked the intensity of their responses in ventral striatum to in-group (vs out-group) members' rewards, as well as their tendency towards group-based prosociality. Individuals with strong motivational investment in their group preferred rewards for an in-group member, whereas individuals with low investment preferred rewards for an out-group member. These findings suggest that the motivational importance of social identity-beyond mere similarity to group members-influences vicarious reward and prosocial behavior. More broadly, these findings support a theoretical framework in which salient social identities can influence neural representations of subjective value, and suggest that social preferences can best be understood by examining the identity contexts in which they unfold.

Instrumental learning of traits versus rewards: dissociable neural correlates and effects on choice.

Humans learn about people and objects through positive and negative experiences, yet they can also look beyond the immediate reward of an interaction to encode trait-level attributes. We found that perceivers encoded both reward and trait-level information through feedback in an instrumental learning task, but relied more heavily on trait representations in cross-context decisions. Both learning types implicated ventral striatum, but trait learning also recruited a network associated with social impression formation.