Comparison and Categorization in Visual Arrays

This research examines how comparison of objects with common and distinctive features underlies categorization, an essential component of human cognition. Classification probabilities have been found to vary as a function of similarity, though it is unclear if the differences are due to a failure to notice lower levels of match, a post sampling evaluation of object similarity, or a combination of both processes. A binomial category sorting task in which participants viewed a 4 x 4 array of 16 occluded objects visible one at a time and provided labels for each object at a family-species level was utilized in the current research. Our primary objective was to learn what kinds of set-level structure provide the most natural or intuitive basis for visual categorization, and to gain insight into the strategies used when engaging in free categorization of novel stimuli. Labeling data and the pattern of visits to other objects while labels are considered for a given target object in each array was analyzed. Objects at a superordinate level, based on overall alignability or structural correspondence, tend to be grouped together significantly more than objects of differing structural configuration, and received more withincategory comparisons than between-category comparisons during labeling. Repeated subordinate level instances with one to four (out of four) matching features were compared and received the same species labels significantly more than other discretely varying alignable objects sharing no identical features. Explanations for scanning patterns at different levels of discrete feature match and future lines of research will be discussed.