Context-based guidance versus context suppression in contextual learning: Role of un-/certainty in the target-context relations in visual search.

Standard investigations of contextual facilitation typically use invariant distractor arrangements predicting a fixed target location. In the real world, however, invariant spatial contexts are not always predictive. We examined how facilitation is influenced by uncertainty in target location prediction: comparing conditions where old contexts were 100% versus minimally (3%) predictive (Experiment 1), 80% predictive (20% nonpredictive) versus 20% predictive (Experiment 2), or a trial-wise mixed condition where 80% predicted a fixed location and 20% a random location (Experiment 3). New-context displays with matching target-location probabilities served as baselines. The results revealed both fully predictive and minimally predictive old contexts to expedite the search, but facilitation was larger for the former (Experiment 1). This held even when the display types were randomly intermixed at an 80:20 cross-trial uncertainty ratio (Experiment 3). However, when old displays predicted the target location in 80% of trials (Experiment 2), facilitation dropped to the level of minimally predictive displays. This indicates only fully predictive old displays support acquiring contextual cues that guide attention. The facilitation seen with 80% predictive contexts likely involves a less efficient process: singling out the target by context suppression. These findings can be incorporated into a neural-network model of context effects: When distractor representations are suppressed, the formation of facilitative links between distractor representations and the target location on the priority map becomes unlikely. (PsycInfo Database Record (c) 2025 APA, all rights reserved).