Reasoning about social choices and social relationships
2014; Wiley; Volume: 36; Issue: 36 Linguagem: Inglês
ISSN
1551-6709
Autores Tópico(s)Psychology of Social Influence
ResumoReasoning about social choices and social relationships Alan Jern Charles Kemp jern@rose-hulman.edu Department of Humanities and Social Sciences Rose-Hulman Institute of Technology ckemp@cmu.edu Department of Psychology Carnegie Mellon University Abstract Kemp, 2011). In social settings, however, the chooser’s utility may depend on the utility experienced by others. One way to capture this dependence is to suppose that the chooser’s util- ity function is a weighted combination of the utilities directly experienced by all affected individuals (Wyer, 1969; McClin- tock, 1972; Griesinger & Livingston, Jr., 1973). We propose that people represent different social relationships as differ- ent utility weighting functions. We show that combining this proposal with the inverse reasoning approach can account for a wide array of social inferences, including inferences about whether a pair of people are more likely to be friends, ene- mies, or strangers. Our proposal is conceptually related to previous computa- tional approaches that have been used to explain how people infer social goals like “helping” (Baker, Goodman, & Tenen- baum, 2008; Ullman et al., 2009). These approaches have focused on how people reason about sequences of actions that extend through time and space. By contrast, we explore one of the simplest possible settings that supports inferences about social choices and how such choices are affected by social relationships. The next two sections introduce our formal approach in more detail. We then evaluate our approach in an experiment in which participants made several kinds of inferences about social choices. We study inferences about social choices—choices that affect people besides the chooser. Social choices depend on the rela- tionships between the people involved: for example, whether they are friends, strangers, or enemies. We propose that these different social relationships correspond to different ways in which the chooser weights another person’s utility relative to her own. We describe a probabilistic model of social reasoning that incorporates this notion of weighted utility, and evaluate it in an experiment in which participants made inferences about others’ social choices. The results support our probabilistic model and expose some of the assumptions that people tend to make when reasoning about social choices. Keywords: social cognition; social reasoning; folk social psy- chology; probabilistic models People frequently engage in folk psychological reasoning. We explain and predict other people’s behavior and draw in- ferences about other people’s thoughts and feelings. This reasoning sometimes depends on knowledge of social rela- tionships. For example, if you know that Alice and Bob are friends, you might predict that Alice would be willing to make a financial sacrifice to help Bob, perhaps by offering him a loan. If you know that Alice and Bob are enemies, you might predict that Alice would not be willing to make a finan- cial sacrifice to benefit Bob, but she might be willing to make a financial sacrifice to harm Bob, perhaps by turning down a mutually beneficial business opportunity. Alice’s choices above are instances of social choices— choices that affect people besides the chooser. Her choices each result in a cost to herself, but also a benefit or cost to Bob. These examples illustrate how knowledge of the rela- tionship between two people can inform expectations about the social choices that they will make. Conversely, observing a social choice may allow us to infer something about the re- lationship between the people involved. Despite the fact that people commonly reason about social choices and social re- lationships, there are few formal proposals about how people perform this sort of reasoning (Haslam, 1994). We suggest in this paper that inferences about social choices and relation- ships can be viewed as a kind of probabilistic reasoning. Previous research has explored how people reason about other people’s non-social choices, like choosing which shirt to buy. Standard choice models can be used to predict the choices that follow from a given set of preferences, and “in- verting” these models provides a way to reason backward and infer the preferences that likely motivated an observed choice. Several studies have shown that this inverse reasoning ap- proach accounts well for experiments that focus on reasoning about non-social choices (Lucas et al., 2014; Bergen, Evans, & Tenenbaum, 2010; Jern & Kemp, 2011; Jern, Lucas, & A social choice model We propose that people reason about social choices by invert- ing a simple model of how utilities give rise to social choices. Consistent with previous approaches (Train, 2009), we as- sume that utilities are additive and that people tend to choose options with greater utilities. With social choices, the notion of utility can be confusing because utility is not necessarily identical to a direct reward. For example, if Alice’s choice can either benefit or harm Bob, Alice’s utility may depend on the effect her choice has on Bob in addition to any benefit her choice provides for herself. To alleviate this confusion, we will refer to the utilities assigned to rewards or payouts as direct utilities and will use the term “utility” to refer to a chooser’s total utility. How much utility a chooser assigns to different options in a social choice depends on how the chooser weights the direct utilities of everyone affected by the choice. Here we assume that there is only one other person affected by the choice, but our approach can be straightforwardly extended to include any number of people. We will henceforth refer to the chooser as Alice and the person affected by the choice as Bob. Let w A be the weight that Alice assigns to her own direct utility and
Referência(s)