Purchase this article with an account.
Jason Haberman, Timothy F. Brady, George A. Alvarez; Independent ensemble processing mechanisms for high-level and low-level perceptual features. Journal of Vision 2014;14(10):1322. doi: 10.1167/14.10.1322.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The ability to extract a summary representation for a group of related objects (ensemble perception) operates across a host of visual domains: People can readily perceive the average emotion of crowds of faces, the average size of dots, and the average orientation of Gabors. Do these ensemble representations rely on a common underlying mechanism, or are there separate ensemble processing mechanisms for different stimulus domains? Here, we address this question using an individual differences approach. In a series of experiments, we assessed performance on pairs of ensemble dimensions, including high-level dimensions (facial expression, facial identity, animal shape) and low-level dimensions (Gabor orientation, triangle orientation, color). For example, some participants saw sets of faces and sets of oriented Gabors, while other participants saw sets of Gabors and sets of oriented triangles. Each set (e.g., 4 faces) was displayed for one second, after which participants adjusted a test item to match the average feature (e.g., average expression) of the preceding set. We measured the error in responses for each feature dimension, and then correlated performance on the two features across subjects (N=100 per experiment). The results revealed a striking disconnect between high-level and low-level ensemble domains. For example, performance in representing average facial identity did not predict performance in representing average orientation (r=0.20; no higher than the correlation with an unrelated non-ensemble task). In contrast, performance within high-level domains was strongly correlated (e.g., average identity vs. average animal shape, r=0.72), as was performance within low-level domains (e.g., average triangle orientation vs. average color, r=0.66). Overall, these experiments reveal that the cognitive architecture of ensemble perception reflects a strong divide between high-level and low-level feature domains. This division could reflect perceptual noise that is correlated within but not across levels, or the existence of separate high-level and low-level ensemble mechanisms.
Meeting abstract presented at VSS 2014
This PDF is available to Subscribers Only