Purchase this article with an account.
Sunghyun Kim, Melissa R. Beck; Influences of Prediction Errors in Establishment of Attentional Control Settings during Incidental Associative Learning. Journal of Vision 2019;19(10):140c. doi: https://doi.org/10.1167/19.10.140c.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Prediction errors, differences between predicted and actual stimuli, play a critical role in associative learning: when events momentarily cease to accurately predict their consequence, associability may increase or decrease. However, previous studies on whether incidental associative learning affects establishment of attentional control settings (e.g., contextual cueing) have a biased implicit assumption: when contexts no longer reliably predict characteristics of search targets, the contexts are no longer used for search. In an attentional capture paradigm, we explored how prediction errors influence attentional control settings during incidental associative learning. In Experiment 1, while searching for target letters, the shape of task-irrelevant placeholders (contexts) accurately predicted task-irrelevant color of targets in the learning and test sessions. Critically, two mismatch trials, where the context-color associations were violated (prediction errors), were inserted between the sessions. Results showed that the associations were used for search only after the mismatch trials, suggesting that prediction errors triggered the effect. However, it was unclear whether the effect was driven by the learning or test session because the same associations were used in the two sessions. Experiment 2, where the associations were reversed between the sessions, showed no effect after the mismatch trials, possibly because prediction errors allowed the associations both after and before the prediction errors to be active and cancel each other out. This possibility was confirmed in Experiment 3, where the color-context associations were present in the learning session but absent in the test session. An effect was observed only in the test session, suggesting that the effect came from learning during the learning session. This study suggests that prediction errors not only facilitate learning of associations after the prediction errors, but also restore associations learned before the prediction errors for establishment of attentional control settings.
This PDF is available to Subscribers Only