No exploitation of temporal predictive context during visual search

Abstract

The human visual system can rapidly extract regularities from our visual environment, generating predictive context. It has been shown that spatial predictive context can be used during visual search. We set out to see whether observers can additionally exploit temporal predictive context, using an extended version of a contextual cueing paradigm. Though we replicated the contextual cueing effect, repeating search scenes in a structured order versus a random order yielded no additional behavioural benefit. This was true both for participants who were sensitive to spatial predictive context, and for those who were not. We argue that spatial predictive context during visual search is more readily learned and subsequently exploited than temporal predictive context, potentially rendering the latter redundant. In conclusion, unlike spatial context, temporal context is not automatically extracted and used during visual search.