6533b85bfe1ef96bd12bbe93

RESEARCH PRODUCT

Reliable Correlational Cuing While Controlling for Most-Recent-Pairing Effects.

subject

description

Irrelevant aspects of the environment or irrelevant attributes of task-relevant stimuli can have important and reliable effects on behavior. When the specific values of an irrelevant stimulus attribute are correlated with different responses, a correlational-cuing effect is observed: faster and more accurate responses when the correlation is positive than when it is negative. Previous work has shown that this effect is not due to simple differences in how often the specific stimuli are being presented, and most explanations of the effect have stressed the clear parallels with classical associative learning. Very recently, however, evidence in favor of an alternative model has been presented. This model focuses on transient linkages or bindings between the irrelevant stimulus attribute and one of the possible responses. This model is quite distinct from that based on classical learning as it only depends on most-recent-pairing of stimulus and response, instead of all previous experiences. In the recent work, no evidence of correlational cuing was found when most-recent-pairing was taken into account, but we argue that the experimental conditions employed were not optimized for associative learning. A new experiment that was designed to favor associative learning was conducted and produced reliable evidence of correlational cuing even when most-recent-pairing was controlled for. We conclude that the correlation-cuing effect is due to both associative learning and transient binding, with the relative contributions of each mechanism depending on specific experimental conditions.