fMR-adaptation reveals a distributed representation of inanimate objects and places in human visual cortex

Abstract

The way information about objects is represented in visual cortex remains controversial. It is unclear, for example, whether information is processed in modules, specialized for different categories of objects or whether information is represented in a distributed fashion across a large network of overlapping visual areas. In this study, we used fMR-adaptation to investigate the extent to which ‘specialized’ regions of visual cortex are involved in representing information about inanimate objects and places. We found adaptation in the object-selective lateral occipital complex (LOC) following repeated presentations of the same inanimate object. However, we also found fMR-adaptation to inanimate objects in fusiform face area (FFA) and the parahippocampal place area (PPA). Furthermore, this adaptation was not affected by changes in the size of the stimulus. In the second part of the experiment, we found adaptation to repeated images of places in the place-selective PPA, which was both size- and viewpoint-invariant. fMR-adaptation to repeated images of places was also observed in the LOC, but not in the FFA. These results suggest that the representation of inanimate objects and places is not restricted to those regions showing maximal responses to these particular categories of objects, but is distributed across human visual cortex and can include ‘face-selective’ regions such as the FFA.

Introduction

Visual areas involved in object recognition form a ventral processing stream that projects toward the temporal lobe (Ungerleider and Mishkin, 1982, Milner and Goodale, 1995). Lesions to this region of the brain often result in difficulties in recognizing, identifying, and naming different categories of objects (Farah, 1990). The concept that discrete areas of the human temporal lobe are specialized for different categories of objects is supported by a number of physiological studies. For example, a region in the fusiform gyrus has been shown to be more responsive to faces than to other complex objects (Allison et al., 1994, Kanwisher et al., 1997; see however, Gauthier et al., 1999, Gauthier et al., 2000). Similar category-specific visual responses have been found for inanimate objects (Malach et al., 1995), buildings and scenes (Epstein and Kanwisher, 1998), human body parts (Downing et al., 2001), and letter strings (Allison et al., 1994). These results are consistent with single-neuron recordings in humans that have also revealed category-specific responses for faces, natural scenes, houses, and animals (Fried et al., 1997).

Selectivity of neural response need not, however, imply that the perception of different categories of objects is only coded by a particular neuronal population. This is because the neural response to any category of object is not restricted to the area that responds maximally to that particular category; many brain regions show significant responses to many different categories of objects (Ishai et al., 1999, Andrews and Schluppeck, 2004, Andrews and Ewbank, 2004). Thus, the functional significance of neural responses to ‘non-preferred’ stimuli is unclear (Cohen and Tong, 2001, Andrews, 2005). An alternative model of object perception proposes that information about different object categories is represented by a widely distributed population response in which both strong and weak responses play a central role in recognition (Haxby et al., 2001). The implication is that specialized regions of visual cortex, such as the fusiform face area (FFA), could also be contributing to the perception of object categories such as inanimate objects and places.

However, it remains unclear whether non-preferred responses play an important role in perception or just reflect a non-specific activation of the visual system that does not lead to recognition (Spiridon and Kanwisher, 2002, Andrews and Schluppeck, 2004). To address this issue, we previously used fMR-adaptation (the reduction in fMRI activity that follows the repeated presentation of identical images; Grill-Spector and Malach, 2001) to ask how different face- and object-selective regions of visual cortex contribute to specific aspects of face perception (Andrews and Ewbank, 2004). We found that activity in the FFA was reduced following repeated presentations of the same face. However, despite the fact that object- and place-selective regions of visual cortex responded to photographs of faces, we failed to find adaptation to face images. Although this finding challenges the view that faces are coded by a distributed representation across all regions of the ventral visual pathway, it is not clear whether other categories of objects are represented in a similar way.

Here, we used fMR-adaptation to determine how inanimate objects and places are represented in visual cortex. Imaging studies have revealed a region in the lateral occipital lobe (LOC) that responds more strongly to whole objects than to scrambled images or textured patches (Malach et al., 1995, Grill-Spector and Malach, 2001, Moore and Engel, 2001). Whereas a region in the medial temporal lobe, known as the parahippocampal place area (PPA), has been shown to respond more strongly to scenes depicting places and buildings than to other kinds of visual stimuli (Epstein and Kanwisher, 1998). Previous studies have reported fMR-adaptation to inanimate objects in the LOC (Grill-Spector et al., 1998, Kourtzi and Kanwisher, 2001) and to places in the PPA (Avidan et al., 2002, Epstein et al., 2003). While these results are consistent with a modular view of cortical processing, adaptation to inanimate objects and places has also been reported in brain regions that are not selective for these object categories (Avidan et al., 2002). However, it could be argued that the reduced activity in non-selective regions of visual cortex might not reveal high-level object recognition, but may simply reflect adaptation to low-level features of the stimulus. To address these issues, we have determined whether a reduction in response to repeated presentations of the same image is specific to particular regions of the brain and also whether this adaptation is evident when low-level attributes (such as the size or the viewpoint) of the images are changed.