Abstract
When specific stimuli are repeated, neural activity to these stimuli in the ventral stream is reduced. This is referred to as repetition-suppression (RS). It is unknown whether RS as measured with fMRI reflects reduced firing of the neural populations responding to the repeated stimulus proportional to the initial response (referred to as “scaling”), or whether RS reflects a reorganization of the neural representation in which less neurons respond to repeated stimuli and the tuning of neurons selective to the stimulus becomes narrower (referred to as “sharpening”). Critically, scaling predicts that RS is largest to the preferred stimulus, while sharpening predicts smallest RS to the preferred stimulus. Therefore we examined the relation between RS and category selectivity in the human: (i) object-selective cortex (the lateral occipital complex, LOC), (ii) face-selective regions in the fusiform gyrus (Fus), the inferior occipital gyrus (IOG), and the posterior superior temporal sulcus (pSTS), (iii) body part-selective regions in the Fus and the IOG, and (iv) house-selective regions in the parahippocampal gyrus (PHG). Ten subjects performed a categorization task while viewing repeated and nonrepeated images of faces, limbs, flowers, cars, guitars, and houses in a high-resolution fMRI scan (1.5 × 1.5 × 1.5 mm). We measured the effects of RS and category selectivity within these regions of interest (ROIs) that were defined from separate localizer scans. Our results indicate that significant RS occurred for all categories in the LOC as well as within the category-selective regions. That is, within all ROIs, RS occurred to both preferred and nonpreferred categories and RS was largest for the preferred category. Furthermore, the level of RS was linearly correlated with the initial response - larger responses were coupled with larger RS. Overall, our data supports the scaling model for RS in the ventral stream.
Support: Whitehall Foundation NEI 5 R21 EY016199 to KGS.