The Occipital Place Area Is Causally and Selectively Involved in Scene Perception

Daniel D. Dilks, Joshua B. Julian, Alexander Paunov, Nancy KanwisherJournal of Neuroscience 2013 · DOI

Abstract

Functional magnetic resonance imaging has revealed a set of regions selectively engaged in visual scene processing: the parahippocampal place area (PPA), the retrosplenial complex (RSC), and a region around the transverse occipital sulcus (previously known as "TOS"), here renamed the "occipital place area" (OPA). Are these regions not only preferentially activated by, but also causally involved in scene perception? Although past neuropsychological data imply a causal role in scene processing for PPA and RSC, no such evidence exists for OPA. Thus, to test the causal role of OPA in human adults, we delivered transcranial magnetic stimulation (TMS) to the right OPA (rOPA) or the nearby face-selective right occipital face area (rOFA) while participants performed fine-grained perceptual discrimination tasks on scenes or faces. TMS over rOPA impaired discrimination of scenes but not faces, while TMS over rOFA impaired discrimination of faces but not scenes. In a second experiment, we delivered TMS to rOPA, or the object-selective right lateral occipital complex (rLOC), while participants performed categorization tasks involving scenes and objects. TMS over rOPA impaired categorization accuracy of scenes but not objects, while TMS over rLOC impaired categorization accuracy of objects but not scenes. These findings provide the first evidence that OPA is causally involved in scene processing, and further show that this causal role is selective for scene perception. Our findings illuminate the functional architecture of the scene perception system, and also argue against the "distributed coding" view in which each category-selective region participates in the representation of all objects.

Introduction

While functional MRI (fMRI) has identified several brain regions that respond preferentially to visual scenes—specifically the PPA, RSC, and the area formerly known as the transverse occipital sulcus (TOS)—correlation does not equal causation. While clinical lesion studies have established the necessity of the PPA and RSC for navigation and recognition, the functional role of the TOS remained unclear. This study renames this region the Occipital Place Area (OPA) and utilizes "virtual lesions" to determine if it is a necessary, specialized module for scene perception or a non-essential participant in a distributed network.

Methods

The researchers employed a double dissociation design across two experiments using Transcranial Magnetic Stimulation (TMS) to temporarily disrupt neural activity in targeted regions.

  1. Localization: Individual fMRI scans were used to precisely locate the rOPA, the face-selective right occipital face area (rOFA), and the object-selective right lateral occipital complex (rLOC) for each participant.
Localization of target regions OPA (A), OFA (B), and LOC (C) in the brain using fMRI.
  1. Experiment 1 (Scenes vs. Faces): Participants performed a fine-grained discrimination task using morphed images of computer-generated houses (scenes) and male faces. TMS was delivered to either the rOPA or the rOFA.
  2. Experiment 2 (Scenes vs. Objects): Participants performed a categorization task (e.g., distinguishing between beaches and kitchens or cameras and chairs) using images from the SUN Database. TMS was delivered to either the rOPA or the rLOC.

Results

The results demonstrated a clear double dissociation in both experiments. In Experiment 1, TMS to the rOPA significantly increased the discrimination threshold for scenes—meaning participants required more visual information to perform the task—but had no effect on face discrimination. Conversely, TMS to the rOFA impaired face discrimination but left scene perception intact.

Scene discrimination thresholds under different TMS conditions; higher thresholds indicate impaired performance.

In Experiment 2, TMS to the rOPA reduced categorization accuracy specifically for scenes, while TMS to the rLOC impaired object categorization without affecting scene processing. These findings indicate that the OPA is not just active during scene viewing but is functionally necessary for it.

Discussion

This study provides the first causal evidence that the OPA is selectively involved in scene processing. The specificity of the impairment—where disrupting the OPA affects scenes but not faces or objects—strongly supports a modular view of brain organization rather than a distributed coding model.

Because the OPA is located more posteriorly than the PPA or RSC, it likely represents the first stage in a hierarchical scene-processing network. Future research is needed to determine the specific visual features (such as 3D geometry or spatial layout) that the OPA extracts and how it communicates with other regions in the scene-processing system.

Summary updated

Abstract

Functional magnetic resonance imaging has revealed a set of regions selectively engaged in visual scene processing: the parahippocampal place area (PPA), the retrosplenial complex (RSC), and a region around the transverse occipital sulcus (previously known as "TOS"), here renamed the "occipital place area" (OPA). Are these regions not only preferentially activated by, but also causally involved in scene perception? Although past neuropsychological data imply a causal role in scene processing for PPA and RSC, no such evidence exists for OPA. Thus, to test the causal role of OPA in human adults, we delivered transcranial magnetic stimulation (TMS) to the right OPA (rOPA) or the nearby face-selective right occipital face area (rOFA) while participants performed fine-grained perceptual discrimination tasks on scenes or faces. TMS over rOPA impaired discrimination of scenes but not faces, while TMS over rOFA impaired discrimination of faces but not scenes. In a second experiment, we delivered TMS to rOPA, or the object-selective right lateral occipital complex (rLOC), while participants performed categorization tasks involving scenes and objects. TMS over rOPA impaired categorization accuracy of scenes but not objects, while TMS over rLOC impaired categorization accuracy of objects but not scenes. These findings provide the first evidence that OPA is causally involved in scene processing, and further show that this causal role is selective for scene perception. Our findings illuminate the functional architecture of the scene perception system, and also argue against the "distributed coding" view in which each category-selective region participates in the representation of all objects.
BibTeX
@article{dilks2013occipital,
  author = {Daniel D. Dilks and Joshua B. Julian and Alexander Paunov and Nancy Kanwisher},
  title = {{The Occipital Place Area Is Causally and Selectively Involved in Scene Perception}},
  journal = {Journal of Neuroscience},
  year = {2013},
  doi = {10.1523/jneurosci.4081-12.2013},
}
Figure