Purchase this article with an account.
Assaf Harel; P2: A novel ERP marker of global scene perception. Journal of Vision 2020;20(11):908. https://doi.org/10.1167/jov.20.11.908.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Humans are extremely adept at recognizing complex visual scenes, an ability supported by a network of dedicated scene-selective cortical areas. In spite of the growing knowledge of these areas, much less is known about the temporal dynamics underlying scene recognition. Since scenes vary on multiple dimensions, ranging from low-level image-statistics to high-level semantics and action affordances, the key question is which types of information get utilized at which timepoint. Our goal, therefore, in the current work is to identify the earliest electrophysiological markers of scene perception, and establish the scene properties they are most sensitive to. We report a series of ERP experiments demonstrating that the first signature of scene-selectivity (greater activity to scenes compared with faces and objects) occurs at posterior-lateral sites, at the P2 time window, 220ms post-stimulus onset. We demonstrate that functionally, the P2 indexes the processing of high-level global scene information: First, P2 is the only visually-evoked component to be modulated by scene inversion, as would be expected if global information is extracted during this period. Second, P2 amplitude is sensitive to global scene properties (GSPs), such as spatial expanse (closed/open) and naturalness (manmade/natural), and these effects are evident across a variety of stimulus presentation conditions and scene types. Moreover, P2 response to GSPs is largely unperturbed by manipulations of local texture information (in contrast to earlier visually-evoked components). Fourth, P2 is the first component to carry significant information about the potential for navigation in a scene (navigability affordances). Lastly, P2 responses to GSPs are hardly modulated by observers' recognition goals, suggesting rapid and mandatory processing of global scene information. Together, our findings suggest that global scene information is processed robustly and automatically around 220 milliseconds and that P2 can be used as an index of global scene perception
This PDF is available to Subscribers Only