Expectations influence visual search

Abstract

Satisfaction of search errors, also called subsequent search misses, are a costly visual search problem, particularly in radiology. To date, research on causes and interventions for satisfaction of search errors has focused on properties of the stimuli and the mechanics of the search process. I present evidence to support a new theoretical understanding of some of the underlying cognition that drives search behavior and that can predict visual search errors. An eye-tracked experiment that manipulated participant expectations of target characteristics and number of targets demonstrated that participant expectations, generated based on environmental cues and long-term memory, influence search behavior. Through exemplar training, participants learned to associate cues with target sets that varied in color of target and number of targets. Participants were instructed to utilize these learned relationships to facilitate their visual search. Analysis of response time, fixation data, and miss errors indicated that expectation was a significant predictor of search behavior, with lower expectations for secondary targets being associated with shorter response times, more miss errors, and fewer fixations to unexpected colors. In a first step towards understanding the cognitive mechanisms behind visual search misses for secondary targets, a cognitive process model was developed. This model integrated hypothesis-guided search with visual search to predict participant behavior. The model was tested against the empirical data and successfully captured the high-level results of the experiment. Future iterations of the model will seek to better fit the more subtle complexities of the empirical results.