Abstract
The object content of most real-world scenes is structured: It is characterized both by the types of individual objects (e.g., a living room typically consists of a sofa, a table, a lamp and a carpet) and by their locations (e.g., a carpet typically lies on the floor, whereas a lamp is hanging from the ceiling). The repeated co-occurrence of specific object types and locations prompts the hypothesis that object processing should be most efficient when the objects appear in their typical locations (e.g., a lamp in the upper visual field), relative to atypical locations (e.g., a lamp in the lower visual field). Here, we present behavioral and neural data that provide converging support for this hypothesis. In a continuous flash suppression paradigm, observers were faster in detecting an object under conditions of inter-ocular suppression when the object was shown in its typical location. This benefit in behavioral performance suggests a processing enhancement for typically positioned objects, allowing them to more efficiently compete in inter-ocular suppression. Using multivariate classification of both fMRI and EEG data, we spatiotemporally characterize this processing enhancement on a neural level: Representations in object-selective regions of the lateral occipital cortex (fMRI) and 140 ms after stimulus onset (EEG) allow for increased decoding of object information if the object is shown in its typical, relative to an atypical, location. Together, these data suggest that objects presented in typical locations gain a processing advantage during early, presumably bottom-up, stages of object processing. The early sensitivity for positional regularities may be highly beneficial for object detection in naturalistic environments where multiple, but regularly positioned, objects need to be processed.
Meeting abstract presented at VSS 2018