9514331 MADDOX The long-term objective of this research is to better understand the perceptual and cognitive processes involved when a person identifies or categorizes objects. This research will investigate the complex interplay between perceptual and decisional processes within the theoretical framework of General Recognition theory which is a multivariate generalization of Signal Detection Theory. This research will investigate several important aspects of human behavior. First, this research will investigate the optimality of human categorization performance when category baserates and category payoffs are unequal. This is an important issue, because baserate and payoff differences are ubiquitous in nature; for example, disease categories often differ in their frequency and in the costs and benefits associated with various diagnoses. Since the goal of the physician is to diagnose a patient accurately, efficient use of baserate and payoff information is extremely important. This research will not only allow one to determine the optimality of human performance, but will assist in identifying the locus of any potential suboptimalities. Second, this research will examine the time it takes an individual to arrive at an identification or categorization decision. The focus will be on identifying the environmental factors that influence response time, such as whether the components of the object can be attended selectively, or whether the identification or categorization problem is complex, etc. A major goal will be to test several extant theories of response time, including the response time predictions from General Recognition Theory. This work complements the baserate and payoff analyses because not only does the physician want to make an accurate diagnosis, but in many cases the physician is under severe time pressure. Third, this research will investigate the relationship between the identification and categorization of objects. The focus will be on understanding how perceptual and decisional processes differ across tasks. The results of this research will allow us to better understand human identification and categorization behavior, and ultimately influence these behaviors in a productive and useful manner. ***
