This action provides funds to the Air Force Office of Scientific Research for NSF's share of this jointly funded research project, the principal investigator of which is Dr. Daniel Kahneman of the Psychology Department of the University of California-Berkeley. The research will study (1) the development of expectations about objects that are observed on repeated occasions, and (2) the rules that govern the norms and standards to which people compare objects and events, for example in judging that a duck is large, or that an athlete moves awkwardly. The research will employ a new experimental paradigm, the mini-event paradigm. Observers facing a computer or TV screen will be exposed to a series of mini-events, in which one or two small objects are seen in motion for 5-10 seconds. The same "individual" objects will reappear periodically in a series of episodes. The behavior of individual objects will not be strictly repetitive, but each one will show some consistency, allowing observers to form distinct impressions of each individual object. Observers will rate each event on a scale of surprise, from "highly surprising" to "completely unsurprising". Following a series of mini-events, observers will be shown test events, and will compare each test event to a designated standard, for example "evaluate how fast each object moves, relative to the speed with which objects of the same color moved in the first series of events". The surprise measure will provide a direct measure of the development of expectations, and, in properly designed series of mini-events, answers to questions such as the following: (i) How quickly do expectations develop? (ii) What are the long-term effects on expectations of a single instance of a behavior that is highly unusual for a particular object? Will a recurrence of that behavior after a long interval be judged relatively unsurprising? (iii) When one object behaves in a new way, how does this affect the norms and expectations for the behavior of other objects? The comparative judgments that will be obtained at the end of the session require the observer to form a representation of a collection of instances specified by the experimenter. These judgments are similar to those commonly made in everyday life, in which people are asked to think about arbitrary categories, for example, to answer a question about the friendliness of neighbors on the other side of the street, or about the size of their dogs. The experiments will explore the rules that determine which instances come to mind when we try to construct such categories. The experiments will also provide another measure of spontaneous categorization. The expectation is that observers will be fastest and most accurate when they are asked to relate a test event to its "natural" category, the one to which it would have been spontaneously assigned.
