Abstract
In visual search, attention can be guided on the basis of statistical regularities in display composition learned through experience – a form of statistical learning (SL) in the attention domain. Specifically, higher priority is conferred to display locations that more often contain the target, relative to locations with rare targets. Likewise, lower priority is conferred to locations that more often contain a salient distractor, relative to locations with rare distractors. In a recent study, we found cross-talk between these two forms of SL, with target SL influencing distractor interference, and vice versa, which is compatible with the notion that both forms of SL modify priority maps of space. Here we tested this idea further by asking how the system would be affected when target and distractor SL are tipped one against the other. Specifically, we had distinct display locations where the probability of both target and distractor were high or, respectively, low. Based on prior evidence, high target frequency should increase priority at the corresponding location, but high distractor frequency at the same location should reduce priority. Similarly, low target frequency should decrease priority at the corresponding location, but low distractor frequency at the same location should increase priority. The results showed clear item-specific modulation of performance at the two critical locations: target selection was enhanced at the high vs. low target-probability location; however, at the same time, the distractor cost was greater at the low vs. high-distractor probability location. Furthermore, the two effects were anti-correlated, indicating a different susceptibility to the two forms of SL across participants. These findings indicate that at least to some extent target and distractor SL can co-occur at the same location, even when they dictate opposite changes in priority. This indicates some independence of the underlying neural mechanisms, likely reflecting a feature-based component.
Meeting abstract presented at VSS 2018