Behavioral Relevance Impacts Utilization of Diagnostic Information for Scene Categorization at Multiple Time Windows: Electrophysiological Evidence

Document Type


Publication Date



Recent work aimed at uncovering the temporal dynamics of scene processing (Harel et al., 2016) has found that global scene properties modulate the amplitude of early ERP components. Key among these is the P2, peaking 220ms post-stimulus onset, which is sensitive to both the naturalness of the scene and its spatial boundary. It is still not clear, however, how much these ERP signatures are impacted by observer-based goals, such as task context, or whether they reflect automatic, stimulus-driven processing of scene information. To address this question, we investigated how task context impacts the electrophysiological responses to global scene properties. Participants viewed 96 scene images while their ERPs were recorded. The scenes spanned two dimensions: naturalness (manmade/natural) and spatial boundary (open/close). In half of the trials the participants determined whether the scene was manmade or natural (naturalness task), and in the other half whether it was open or closed (spatial boundary task). Thus, one scene dimension was task-relevant in half of the trials but task-irrelevant in the others. We found that the P2 amplitude could distinguish between open and closed scenes and between manmade and natural scenes, independently of whether the participants were attending to the relevant dimension or not. However, P2 amplitude was higher in the spatial boundary task than in the naturalness task across all scene dimensions. This could not be attributed to task-difficulty, as response times in both tasks were equivalent. A significant effect of task context on scene properties was found earlier, at the P1 level. P1 amplitude varied as a function of spatial boundary, but only in the spatial boundary task. When participants viewed the same images but attended their naturalness, this effect dissipated. These findings highlight the temporal dynamics of task-specific and stimulus-driven utilization of diagnostic scene information underlying human scene categorization.


This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.


Presented at the Vision Sciences Society Seventeenth Annual Meeting, St. Pete Beach, FL, May 19-24, 2017