Abstract
An illusory conjunction is a feature binding error wherein a person incorrectly combines features from two different objects. Behavioral investigations have implicated spatial attention in successful feature binding, suggesting that failures of spatial attention lead to illusory conjunctions. We examined this phenomenon using electroencephalography (EEG) to identify neural markers of binding successes and failures. We hypothesized that EEG markers of covert spatial attention, the N1 and N2pc, would be greater when features were correctly bound compared with illusory conjunctions. We briefly presented (150 ms) two characters, an L or 7 and a 0, flanked by $s on either side. The characters, in one of three colors (red, green, or blue), varied in position. Stimuli were presented in the right or left visual field with distractors ($$$$) presented at the opposite location, and were bilaterally pre- and post-masked with distractors. After 1000 ms, participants reported the target identity (L or 7) and the color of that target. Illusory conjunctions occurred when participants reported, for example, a red L, when presented with a red 0 and a blue L. In 10 participants with adequate trials for EEG analysis, participants reported the correct target shape (L or 7) and color in an average of ~61% of trials, and illusory conjunctions in and average of ~20% of trials. We randomly down-sampled correct trials to equate conditions for each participant. We analyzed ERPs over posterior-lateral sites for correct responses and illusory conjunctions, for contralateral and ipsilateral target presentation. Though a marginal N2pc was present (contralateral more negative than ipsilateral, p = .069), there was no interaction between binding accuracy and laterality, contrary to our hypothesis,. However, correct trials demonstrated marginally greater mean amplitudes than illusory conjunctions for N1 and P3 (p = .054 and p =.068, respectively). These differences suggest physiological correlates required for successful feature integration.
Meeting abstract presented at VSS 2015