Abstract
Object identification is quite rapid and efficient but as the stimulus load increases, bottlenecks in visual performance become evident. These bottlenecks occur when object presentation rates exceed 8 – 10 items per second and are likely to reflect inherent processing limitations of the human visual system. Here, we investigated whether this bottleneck is a result of insufficient central attentional resources and therefore generalizable across all stimulus types, or due to the limited neural capacity of category-specific visual areas. We measured fMRI activity while participants viewed sequences of faces, houses, or alternating faces and houses. Items were presented serially at rates of 2.3, 4.7, 9.4, 18.8, or 37.5 items per second which allowed us to manipulate the difficulty of discerning individual items. Brain areas concerned with processing faces and houses, the Fusiform Face Area (FFA) and Parahippocampal Place Area (PPA) respectively, showed an inverted ‘U’ shape function when presented with their preferred stimuli (i.e., faces only or houses only). The greatest neural activity occurred at the middle presentation rates, and declined at the higher presentation rates. However, when presented with sequences of alternating faces and houses, the peak in fMRI activity for the FFA and PPA occurred at a higher presentation rate than for sequences containing only one stimulus type. This shift in peak activity suggests that the processing limit of each visual area depended on the presentation rate of the preferred stimulus, and not just any stimulus. Neural activity in early visual areas did not show differences between faces, houses, or alternating faces and houses, suggesting that this type of processing limitation occurs beyond striate cortex. The results support the notion that bottlenecks in object perception result from processing limits of category-specific brain areas rather than a general limit in attentional resources.