How do people translate their experiences into abstract attribute preferences?

In many literatures, scholars study summarized attribute preferences: overall evaluative summaries of an attribute (e.g., a person's liking for the attribute “attractive” in a mate). But we know little about how people form these ideas about their likes and dislikes in the first place, in part because of a dearth of paradigms that enable researchers to experimentally change people's attribute preferences. Drawing on theory and methods in covariation detection and social cognition, we developed a paradigm that examines how people infer summarized preferences for novel attributes from functional attribute preferences: the extent to which the attribute predicts an individual's evaluations across multiple targets (e.g., a person's tendency to positively evaluate mates who are more vs. less attractive). In three studies, participants encountered manipulated information about their own functional preference for a novel attribute in a set of targets. They then inferred a summarized preference for the attribute. Summarized preferences corresponded strongly to the functional preference manipulation when targets varied on only one attribute. But additional complexity (in the form of a second novel attribute) caused summarized and functional preferences to diverge, and biases emerged: Participants reported stronger summarized preferences for the attribute when the population of targets possessed more of the attribute on average (regardless of functional preference strength). We also documented some support for a standard-of-comparison mechanism to explain this inferential bias. These studies elucidate factors that may warp the translation process from people's experienced evaluative responses in the world to their overall, summary judgments about their attribute preferences