Abstract
Objects in the real world do not appear in a void. Rather, they are typically found within a broader context, having relationships with the environment. Numerous studies have shown that these relationships affect subjects' performance, so that incongruent objects (i.e., objects that appear in unexpected scenes) are typically identified less accurately and slower than congruent objects. Yet an ongoing debate involves the source of this effect, and its prerequisites: does it stem from a genuine difficulty to identify incongruent objects, or does it reflect a later, post-perceptual process? And does it depend on explicit congruency detection or on conscious perception? Here, we present results from three EEG experiments; the first focuses on the N300 component, held to reflect object identification processes. By presenting congruent/incongruent objects that are either intact or scrambled, we show that the congruency-evoked N300 is different from identifiability–evoked N300, yet we still find a later divergence of waveforms evoked by intact incongruent objects from scrambled ones (as compared with the divergence of congruent intact and scrambled objects), providing first direct evidence for the difficulty to identify incongruent objects. We then focus on the scene-N400 component, held to index integration attempts between the object and the scene, and show that it is modulated both by explicit vs. implicit detection of scene incongruency (Exp. 2), and by conscious vs. unconscious processing of the scene (Exp. 3).
Meeting abstract presented at VSS 2018