Examining differential effects of target and context repetition in visual search: Insights from a big data approach

Much remains unknown about how visual statistical learning—incidental learning of stimulus and display regularities—influences future performance. Visual search paradigms may be useful for investigating visual statistical learning given that recent stimulus repetitions facilitate search performance and the effects decay over time. Specifically, past studies on the “Previous Trial Effect"—same/different content in preceding trials influences performance— have found priming boosts consecutive trial performance and decays over ~5 trials (Maljkovic & Nakayama, 1994). Additionally, contextual cueing—implicit learning of repeated contextual information—shows performance improves following the repetition of context, regardless of its task relevance (Chun & Jiang, 1998). As the extent and potential overlap of such priming effects remain unclear, the current study aimed to determine how target repetition was affected, and possibly strengthened, by general search context (e.g., background characteristics). With a massive dataset (~3.8B trials, ~15.5M participants) of visual search data collected from the Airport Scanner mobile application (Kedlin Co.), it is possible to look at subtle and large changes in target and context repetitions, and at varying delay intervals. For example, performance was evaluated for a specific trial as a factor of whether a preceding trial contained (1) the same target and context, (2) the same target but different context, (3) a different target but same context, or (4) a different target and context. Moreover, these conditions were assessed for multiple preceding trials to measure decay and other time-focus effects. Search was faster in all match conditions compared to baseline performance, with larger benefits for the matching target than context alone. However, response times were fastest following recent trials where the combination of matching target and context occurred, suggesting a possible benefit for stimulus and display associations. These and other results provide insights from visual search performance on how statistical regularities influence cognitive performance.