Visual pattern complexity determination for enhanced usability in cognitive testing

Abstract

Computerized cognitive tests often entail tasks related to visual stimuli. An efficient complexity measure for these tasks can enhance their cognitive evaluation accuracy, specifically for elderly and cognitively impaired subjects. This paper details the design, implementation, and testing of a visual pattern complexity determination algorithm. The patterns used for the study are sixteen-bit binary patterns taken from computerized cognitive assessments. Three complexity levels were defined based on the visual perception of human subjects: easy, medium, and hard. The algorithm was tested on three hundred patterns and the results were compared to the parallel complexities perceived by human judges. Correlations of 72%, 74%, and 61% between human perception and the algorithm’s predictions were obtained for the easy, medium, and hard patterns, respectively. The algorithm has potential to become an accurate measure of visual pattern complexity in computerized assessment, and could improve the usability of these tests for psychometric and cognitive evaluations.