Abstract
Visual search and classification require several cognitive processes to work together, such as visual memory, decision-making, and hypothesis-testing. How are cognitive resources allocated dynamically to these processes during natural tasks? In the current study, a natural search and classification task was devised in which cognitive complexity was varied and both manual and oculomotor search performance were used to assess memory load. Subjects had to find 3 objects in a field of 9 hidden objects that belonged to a given category. Five categories, defined over some subset of the objects' four features, were varied systematically in their complexity. In the manual search task, subjects clicked on object locations with a mouse to reveal the object in that location. In the eye movement task, an object was visible only when its location was fixated. In both experiments, objects were only visible one at a time and revisits were permitted. Stimuli were chosen randomly such that the performance of an ideal searcher (no memory loss) would be the same across category complexity. Search performance reflected memory and cognitive load: the frequency of revisits to previously viewed objects (double-checking) increased as complexity of the category increased in both the manual search (R2;=.873) and the eye movement tasks (R2;=.892). However, subjects in the eye movement task, who searched with saccades requiring less motor effort than moving a mouse, produced an average of 52% more revisits overall (p=.03), reflecting a preference for more frequent double-checking over finishing the task more quickly. These results indicate that cognitive complexity significantly affects memory load as reflected in search behavior, which suggests a unitary pool of resources that must be strategically managed during natural tasks involving several cognitive processes.
Supported by: NSF DGE0549115 and NIH EY15522