Adopted utility calculus: Origins of a concept of social affiliation

Abstract

To successfully navigate their social world, humans need to understand and map enduring relationships between people: we need a concept of social affiliation. Here I propose that our initial concept of social affiliation, available to human infants, is based on the extent to which one individual consistently takes on the goals and needs of another. This proposal grounds affiliation in a commonsense psychology that treats individuals as rational actors, as formalized in the naive utility calculus model. A concept of affiliation based on interpersonal utility adoption can account for findings from studies of infants’ reasoning about imitation and similarity, helpful and fair individuals, “ritual” behaviors, and social groups, without the need for additional innate mechanisms such as a coalitional psychology, moral sense, or general social preference for similar others. This concept of affiliation also offers a new view on the prosocial nature of imitation, placing the value of aligning with someone else’s behavior in the adoption of their goals rather than in mere similarity. I propose further tests of this concept of affiliation, and also discuss how it is likely to be relevant to reasoning and learning about social relationships across the lifespan.

A hierarchical Bayesian model of adaptive teaching

“…For example, the learner may want to maintain their reputation, or the teacher may want to communicate that they care about a learner's well-being. Currently, our utility functions have two terms -informativity and communication cost -but it is also straightforward to account for these other goals, by adding more terms to the teacher and learner utility functions (Yoon, Tessler, Goodman, & Frank, 2020;Powell, 2022;Hung, Thomas, Radkani, Tenenbaum, & Saxe, 2022;.…”

Toddlers prefer agents who help those facing harder tasks

“…First, despite the early-emerging expectation that agents act efficiently to minimize the costs of their own actions (5,6,8,9) and an early-emerging preference for agents who act efficiently (11), the toddlers in our study preferred an agent who chose to perform a more costly action for the benefit of someone else. This seeming discrepancy can be explained by extending recent proposals that apply the Naïve Utility Calculus framework to social interactions (28,29). Extending the idea that helping NEED-BASED EVALUATIONS 14 involves an agent "adopting" the utility of another agent (29), we speculate that toddlers understand the actions presented in the current paper by considering the utilities of both potential beneficiaries: the agent who needed help more and the agent who needed help less (i.e., both agents would benefit from being helped but one would benefit more; Fig.…”

“…Within the framework of the Naïve Utility Calculus, prosocial actions like helping and teaching can be formalized as one agent adopting or taking on another agent's utility function (28,29): Specifically, one agent (a "helper" or a "teacher") benefits another agent (a "beneficiary") by engaging in an action that increases the beneficiary's rewards and reduces the beneficiary's costs. Yet, this formalization does not readily explain how we might evaluate the relative helpfulness of actions that vary in their consequences for those involved.…”

Infants and toddlers leverage their understanding of action goals to evaluate agents who help others

“…Infants and toddlers could have formed preferences in the present experiments without representing the helper's action plan, by focusing on the outcomes of each helper's action for the protagonist and favoring the helper that produced a positive outcome. A growing body of research, however, supports the possibility that infants and toddlers understand that a protagonist's goal is embedded within the goal of a helper (see Hamlin et al, 2013;Powell, 2021;Ullman et al, 2010;Woo & Spelke, 2020;Woo et al, 2017). If infants and toddlers in the present experiments indeed represented the helpers' actions as guided by hierarchically structured social goals, then these findings stand in contrast to all the findings with which this paper began, showing that young infants struggle to reason about hierarchically organized nonsocial goals when viewing a solitary agent who engages in a means-end action.…”