Abstract
Humans are extraordinarily quick at processing natural scenes. Furthermore, good exemplars of natural scene categories are not only categorized more easily but are also more readily detected than bad exemplars. However, it remains unclear when and how this good exemplar advantage arises. To address this question, we measured event-related potentials (ERPs) while participants viewed (and made a delayed judgment about) good and bad exemplars of six scene categories: beaches, city streets, forests, highways, mountains and offices. Good and bad exemplars first evoked differential ERPs 250 to 350ms after onset, with bad exemplars producing greater frontal negativity than good exemplars. This effect is consistent with the N3 complex, previously associated with global structure in an image. The results thus indicate that structural processing is easier for good than for bad exemplars. Good exemplars then elicited a larger late positive complex (LPC) response, likely reflecting more confident judgments for these items. Furthermore, the separation between natural vs. man-made ERP waveforms was both earlier and larger in amplitude for good exemplars as compared to bad exemplars. Overall, the results indicate that the good exemplar advantage may not only extend to eased cognitive processing, but also to perceptual processing, having its roots in higher order visual processing and possible feedback to lower level processing areas
Meeting abstract presented at VSS 2015