Theta-burst TMS of lateral occipital cortex reduces BOLD responses across category-selective areas in ventral temporal cortex

Abstract

Human visual cortex contains three scene-selective regions in the lateral, medial and ventral cortex, termed the occipital place area (OPA), medial place area (MPA) and parahippocampal place area (PPA). Using functional magnetic resonance imaging (fMRI), all three regions respond more strongly when viewing visual scenes compared with isolated objects or faces. To determine how these regions are functionally and causally connected, we applied transcranial magnetic stimulation to OPA and measured fMRI responses before and after stimulation, using a theta-burst paradigm (TBS). To test for stimulus category-selectivity, we presented a range of visual categories (scenes, buildings, objects, faces). To test for specificity of any effects to TBS of OPA we employed two control conditions: Sham, with no TBS stimulation, and an active TBS-control with TBS to a proximal face-selective cortical region (occipital face area, or OFA). We predicted that TBS to OPA (but not OFA) would lead to decreased responses to scenes and buildings (but not other categories) in other scene-selective cortical regions. Across both ROI and whole-volume analyses, we observed decreased responses to scenes in PPA as a result of TBS. However, these effects were neither category specific, with decreased responses to all stimulus categories, nor limited to scene-selective regions, with decreases also observed in face-selective fusiform face area (FFA). Furthermore, similar effects were observed with TBS to OFA, thus effects were not specific to the stimulation site in the lateral occipital cortex. Whilst these data are suggestive of a causal, but non-specific relationship between lateral occipital and ventral temporal cortex, we discuss several factors that could have underpinned this result, such as the differences between TBS and online TMS, the role of anatomical distance between stimulated regions and how TMS effects are operationalised. Furthermore, our findings highlight the importance of active control conditions in brain stimulation experiments to accurately assess functional and causal connectivity between specific brain regions.

Trial registration: ClinicalTrials.gov NCT00001360 NCT01617408.

Copyright © 2021. Published by Elsevier Inc.

Figures

Fig. 1.

Example stimuli, task schematic, experimental design and TBS target sites. A, Example stimuli from eight categories: Four scene-categories and four non-scene categories. B, Task schematic. Stimuli were presented in blocks. Participants fixated centrally and were required to push a button every time a 2-back repeat occurred. C, Schematic of experimental design. Each session began with a T1-weighted scan, followed by four Pre-TBS runs of the task, each lasting ~7 mins. Following the end of Pre Run4, participants were removed from the scanner and TBS was performed. Four Post-TBS runs were then acquired followed by a second T1-weighted image. The first Post-TBS volumes were acquired ~3 mins following the cessation of TBS. D, Left: Individual TMS stimulation sites (red = OPA, blue = OFA). Individual participant (n = 16) stimulation sites are overlaid onto a partially inflated lateral view of the right hemisphere (gyri = light gray, sulci = dark gray). Right: Group ROIs overlaid onto a medial view of the right hemisphere.

Fig. 2.

TMS effects in OPA and PPA. A, Bars represent the group-average response in OPA for each of the stimulus category blocks presented in the experiment (F = faces, On = natural objects, Om = man-made objects, B = buildings, Smo = Manmade open scenes, Smc = manmade closed scenes, Sno = natural open scenes, Snc = closed natural scenes) during Pre (gray) and Post (white) sessions measured during TMS to OPA (left), TMS to OFA (middle) and Sham (right). Individual participant data are overlaid and linked in each case. B, Bars represent the response difference between the Pre and Post session for TMS to OPA (red), TMS to OFA (blue) and Sham (gray). Error bars represent S.E.M. across subjects. C-D, same as A-B, but for PPA.

Fig. 3.

TMS effects in OFA and FFA. A, Bars represent the group-average response in OFA for each of the stimulus category blocks presented in the experiment (F = faces, On = natural objects, Om = man-made objects, B = buildings, Smo = Manmade open scenes, Smc = manmade closed scenes, Sno = natural open scenes, Snc = closed natural scenes) during Pre (gray) and Post (white) sessions measured during TMS to OPA (left), TMS to OFA (middle) and Sham (right). Individual participant data are overlaid and linked in each case. B, Bars represent the response difference between the Pre and Post session for TMS to OPA (red), TMS to OFA (blue) and Sham (gray). Error bars represent S.E.M. across subjects. C-D, same as A-B, but for FFA. Note that for the Sham condition in FFA, the apparent increase in the Post relative to the Pre session is largely driven by one outlier subject. Statistical results were qualitatively the same when excluding this subject from analysis.

Fig. 4.

Linear mixed effects results in the stimulated right hemisphere (RH). A, Locations of significant Session by TMS Condition interaction effects (p = 0.0000024, q = 0.000042) are overlaid onto lateral (top) and medial (bottom) views of a partially inflated surface reconstruction of the right hemisphere (sulci=dark gray, gyri=light gray). Interaction effects are evident overlapping with (or in close proximity to) group-based definitions of OPA, OFA (left), PPA, MPA, FFA (right) as well as early visual cortex (V1). B, The effect of Session (Pre, Post) following OPA stimulation is overlaid onto the same views as A and thresholded on the interaction shown in A. Cold colors represent a decrease in response following TBS of OPA, with hot colors representing an increase. On the whole, TBS of OPA caused a reduction in response. Local decreases are evident inside OPA, OFA, anterior PPA and FFA. C, Same as B but for OFA stimulation. Here, positive responses are present on the lateral surface and within OPA and OFA specifically. On the ventral surface, responses are negative and numerically larger than those following OPA stimulation. D, Same as B but for the Sham condition. Responses are largely positive throughout the brain. E, Bars represent the mean t-value for the effect of Session during OPA (red), OFA (blue) and Sham (gray) conditions. Considering the entire hemisphere (left plot), both stimulation conditions resulted in reduced responses, whereas a positive response followed the Sham condition. Additional plots represent the mean difference in activation across the nodes showing a significant interaction effect within OPA, OFA, PPA and FFA. See Supplementary Figure S5 for a version of this analysis that included random slopes for each participant in addition to random intercepts, yielding qualitatively similar results.