Selectivity for large nonmanipulable objects in scene-selective visual cortex does not require visual experience

Highlights

• We explore the PPA responses to nonmanipulable objects without visual experience.

• Large nonmanipulable object selectivity in PPA is independent of visual experience.

• “Scene regions” show selectivity for nonmanipulable objects in blind and sighted.

• Functional connectivity associated with PPA does not require visual experience.

Abstract

The principles that determine the organization of object representations in ventral temporal cortex (VTC) remain elusive. Here, we focus on the parahippocampal place area (PPA), a region in medial VTC that has been shown to respond selectively to pictures of scenes. Recent studies further observed that this region also shows a preference for large nonmanipulable objects relative to other objects, which might reflect the suitability of large objects for navigation. The mechanisms underlying this selectivity remain poorly understood. We examined the extent to which PPA selectivity requires visual experience. Fourteen congenitally blind and matched sighted participants were tested on an auditory size judgment experiment involving large nonmanipulable objects, small objects (tools), and animals. Sighted participants additionally participated in a picture-viewing experiment. Replicating previous work, we found that the PPA responded selectively to large nonmanipulable objects, relative to tools and animals, in the sighted group viewing pictures. Importantly, this selectivity was also observed in the auditory experiment in both sighted and congenitally blind groups. In both groups, selectivity for large nonmanipulable objects was additionally observed in the retrosplenial complex (RSC) and the transverse occipital sulcus (TOS), regions previously implicated in scene perception and navigation. Finally, in both groups the PPA showed resting-state functional connectivity with TOS and RSC. These results provide new evidence that large object selectivity in PPA, and the intrinsic connectivity between PPA and other navigation-relevant regions, do not require visual experience. More generally, they show that the organization of object representations in VTC can develop, at least partly, without visual experience.

Introduction

The functional organization of object representations in the human visual cortex, especially the ventral temporal cortex (VTC), has been the focus of much recent research. Functional neuroimaging studies have provided evidence that different object domains evoke distinct responses in VTC. For example, specific regions of VTC respond selectively to particular object categories, such as faces, bodies, words, or places (Bracci et al., 2010, Chao et al., 1999, Cohen and Dehaene, 2004, Downing et al., 2001, Downing et al., 2006, Epstein and Kanwisher, 1998, Kanwisher, 2010, Kanwisher et al., 1997, Peelen and Downing, 2005).

A particularly strong type of categorical selectivity is observed with scene stimuli. Compared to pictures of faces, common objects or scrambled pictures, pictures of scenes or places elicit stronger activation in a region in the parahippocampal gyrus (the parahippocampal place area, PPA), along with two additional regions in the retrosplenial complex (RSC) and the transverse occipital sulcus (TOS) (e.g., Epstein and Kanwisher, 1998, Goh et al., 2004). These findings motivated hypotheses about the function of PPA, including that it processes peripheral visual information, certain geometrical features about openness or closeness, or spatial properties (e.g., Kravitz et al., 2011, Levy et al., 2001, Park et al., 2011, Ross and Oliva, 2010). Interestingly, a series of recent studies showed that PPA activity is also modulated by the type of objects, preferring objects that are part of a scene (e.g., buildings, Maguire et al., 2001), large (Konkle and Oliva, 2012, Mullally and Maguire, 2011, Troiani et al., in press), with strong contextual associations (e.g., Bar and Aminoff, 2003), or those that more easily evoke a sense of space or place (Mullally and Maguire, 2011).

The degree to which visual object properties underlie the observed object categorical effects in PPA remains debated. One type of proposal is that the preference for scenes and some types of objects is driven by its sensitivity to certain visual aspects that are shared between scenes and these objects, e.g., peripheral visual information being more important (e.g., Konkle and Oliva, 2012). Alternatively, it might be because these regions are at least partly engaged in the more abstract interpretation of the stimulus, and the selectivity reflects how strongly the objects imply a scene/place and information useful for spatial navigation (e.g., Troiano et al., in press). In the present study we tested congenitally blind participants to investigate whether knowledge of visual object properties is required for object category selectivity in the PPA.

While the potentially relevant object properties driving PPA, such as size, are predominantly obtained through the visual modality in normal circumstances, they can nonetheless be obtained through other modalities. Save et al. (1998) reported that early blind rats exhibited place cell firing activities highly similar to sighted rats. In humans, Wolbers et al. (2011) reported that both blind and sighted participants showed stronger PPA activation when they explored Legos as a scene layout compared to when they explored the stimuli as objects, suggesting that PPA's engagement in scene processing can develop without visual experience.

Other aspects of object categorical distributions in VTC have also been shown to be independent of visual experience. Pietrini et al. (2004) reported category-related fMRI response patterns for faces and manmade objects in VTC in congenitally blind participants during tactile object recognition. Reich et al. (2011) found that congenitally blind individuals show selective fMRI responses to Braille word stimuli in a region in left VTC that closely corresponds to the “visual word form area” in sighted individuals. Mahon et al. (2009) observed a preference for inanimate over animate objects in the medial portion of bilateral VTC in both sighted and three congenitally blind participants performing an auditory size judgment task.

In the present study, we investigated whether the selectivity for large nonmanipulable objects in PPA requires previous visual experience. We compared PPA responses to objects that are large and typically nonmanipulable/nonportable, relative to small manipulable objects (tools) and animals in sighted and congenitally blind participants. Both participant groups listened to the names of these objects, and sighted participants additionally viewed pictures. If the selectivity to large nonmanipulable objects in PPA was driven primarily by certain visual properties specifically associated with these objects, we expect that congenitally blind participants would show different (i.e., weaker, null, or disordered) patterns in comparison to the sighted participants. If, however, such selectivity in PPA originated from non-visual processes such as multi-modal spatial navigation, we expect that congenitally blind participants would develop PPA selectivity patterns similar to sighted participants. Furthermore, to better understand the role of visual experience in shaping PPA's functional profile, we explored the intrinsic functional connectivity pattern of PPA using resting-state fMRI in both participant groups, aiming to examine whether the spontaneous functional network associated with PPA is affected by visual experience.