Encoding the identity and location of objects in human LOC

Abstract

We are able to recognize objects independent of their location in the visual field. At the same time, we also keep track of the location of objects to orient ourselves and to interact with the environment. The lateral occipital complex (LOC) has been suggested as the prime cortical region for representation of object identity. However, the extent to which LOC also represents object location has remained debated. In this study we used high-resolution fMRI in combination with multivoxel pattern classification to investigate the cortical encoding of three object exemplars from four different categories presented in two different locations. This approach allowed us to study location-tolerant object information and object-tolerant location information in LOC, both at the level of categories and exemplars. We found evidence for both location-tolerant object information and object-tolerant location information in LOC at the level of categories and exemplars. Our results further highlight the mixing of identity and location information in the ventral visual pathway.

Introduction

When we see an object in one location of the visual scene, and at some other occasion in a different location, we can typically identify it without effort. This seemingly simple recognition involves two important computational problems: (1) we recognize the identity of the object independent of its location, and (2), simultaneously, in order to be able to interact with an object, we need information about its location. The lateral occipital complex (LOC) has been suggested to be the prime locus of object identity representation (Malach et al., 1995, Grill-Spector, 2003). LOC is a functionally defined set of regions in lateral occipital and inferior temporal cortex responding stronger to pictures of intact objects than to scrambled pictures (Malach et al., 1995). However, the way in which LOC represents objects (Op de Beeck et al., 2008b) and the extent of its location tolerance remain debated (Kravitz et al., 2008). In particular, three issues remain to be addressed.

Location tolerance is a necessary requirement for a representation to underlie object recognition: only representations that are tolerant to changing viewing conditions can reliably signal the presence of an object (Riesenhuber and Poggio, 2002, DiCarlo and Cox, 2007). To which extent object representations in LOC are location-tolerant remains debated. Prior fMRI studies using BOLD activation and fMRI adaptation differed in their estimate of the degree of tolerance with which LOC responds to objects at different locations in the visual field (Grill-Spector et al., 1998, Grill-Spector et al., 1999, Niemeier et al., 2005, Hemond et al., 2007, MacEvoy and Epstein, 2007). Similarly, studies using multivoxel pattern classification differed in their estimate: While some studies found evidence for location-tolerant object representations (Schwarzlose et al., 2008, Sayres and Grill-Spector, 2008, Williams et al., 2008, Carlson et al., in press), others did not (Kravitz et al., 2010).

In typical everyday situations we recognize and automatically categorize objects based on visual evidence at the level of exemplars within a category (Rosch et al., 1976, Mervis and Rosch, 1981). However, most fMRI studies investigated object representation at the level of categories by averaging over brain responses for single exemplars within a category (Haxby et al., 2001, Grill-Spector et al., 1998, O'Toole et al., 2005, Spiridon and Kanwisher, 2002). Only recently, fMRI and multivoxel pattern classification have been combined to explore object representation below the level of category (Op de Beeck et al., 2010.; Kriegeskorte et al., 2007, Kriegeskorte et al., 2008, Eger et al., 2008a, Eger et al., 2008b, Kravitz et al., 2010). Thus, in this study we investigated whether location-tolerance in LOC holds for the representation of exemplars as well as for the representation of categories.

Besides recognizing and categorizing objects, the visual system must also keep track of the location of objects to allow us to orient ourselves and to interact with the environment. An influential view of the visual system (Ungerleider and Mishkin, 1982) claims that location information is not contained in the ventral stream. However, later interpretations of the dorsal and ventral stream distinction claimed that there is location information in the ventral stream, though in a different format than in the dorsal stream (Milner and Goodale, 2006, Milner and Goodale, 2008). Also, recent studies showed the encoding of location information in LOC (Schwarzlose et al., 2008, Sayres and Grill-Spector, 2008, Carlson et al., in press, Kravitz et al., 2010). Furthermore, there is retinotopic information at least in parts of LOC (Brewer et al., 2005, Larsson and Heeger, 2006, Sayres and Grill-Spector, 2008), and LOC exhibits an eccentricity bias (Levy et al., 2001, Malach et al., 2002, Hasson et al., 2002). In consequence, the degree of location information in LOC remains under investigation. We therefore investigated whether LOC contains information about the location of objects independent of object identity.

In this study we used high-resolution fMRI and multivoxel pattern classification (Mika et al., 2001, Haxby et al., 2001, Spiridon and Kanwisher, 2002, Cox and Savoy, 2003, Carlson et al., 2003, Kamitani and Tong, 2005, Kamitani and Tong, 2006, Haynes and Rees, 2005, Haynes and Rees, 2006, Norman et al., 2006, Haynes et al., 2007) to estimate two crucial types of information encoded in LOC. We assessed location-tolerant object information and object-tolerant location information using visual stimuli at the level of exemplars within a category.