A previously undescribed scene-selective site is the key to encoding ego-motion in naturalistic environments

  1. Bryan Kennedy
  2. Sarala N Malladi
  3. Roger BH Tootell
  4. Shahin Nasr  Is a corresponding author
  1. Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, United States
  2. Department of Radiology, Harvard Medical School, United States
13 figures and 1 additional file

Figures

Distribution of scene-selective areas within the human visual cortex.

Panel (A) shows the group-averaged (n = 14) response to ‘scenes > faces’ contrast (Experiment 1). Areas parahippocampal place area/temporal place area (PPA/TPA), restrosplenial cortex/medial place area (RSC/MPA), and transverse occipital sulcus/occipital place area (TOS/OPA) are localized within the temporal, medial and posterior-lateral brain surfaces, respectively. To show consistency with our previous reports (Nasr et al., 2011), data from individual subjects was largely smoothed (FWHM = 5 mm) and the group-averaged maps were generated based on fixed- rather than random-effects (see also Figure 3). The resultant map was thresholded at p<10–25 and overlaid on the common brain template (fsaverage). Panel (B) shows the activity map in one randomly selected subject (see also Figure 2), evoked in response to the same stimulus contrast as in panel (A) Here, the activity map was only minimally smoothed (FWHM = 2 mm). Consequently, multiple smaller scene-selective sites could be detected across the cortex, including posterior intraparietal gyrus scene-selective site (PIGS) (black arrow), located within the posterior intraparietal gyrus. Traditionally, these smaller activity patches are treated as noise in measurements and discarded. For ease in comparing the two panels, the individual’s data was also overlaid on the fsaverage.

Activity evoked by ‘scene > face’ contrast in seven individual subjects, other than the one shown in Figure 1.

Panel (A) shows the significance of evoked activity in the left hemisphere (LH) of one individual subject. The inset shows the enlarged activity map within the intraparietal region. The three scene-selective areas, along with area posterior intraparietal gyrus scene-selective site (PIGS), are indicated in the map with arrows. The location of the parieto-occipital sulcus (POS), the intraparietal sulcus (IPS), and the calcarine sulcus (CS) is also indicated in the inset. Panel (B) shows the result from six other individuals. In this panel, the first two columns show the activity within the LH, while the next two columns show the activity within the right hemisphere (RH) of the same subjects. In all subjects, PIGS (black arrow) is detectable bilaterally within the posterior portion of the intraparietal gyrus, near (but outside) the POS. All activity maps were overlaid on the fsaverage to highlight the consistency in PIGS location across the subjects.

Posterior intraparietal gyrus scene-selective site (PIGS) was detected in group-averaged activity maps across two non-overlapping populations.

Panel (A) shows the group-averaged activity, evoked within the intraparietal region of 14 subjects who participated in Experiment 1. Panel (B) shows the group-averaged activity, evoked within the intraparietal region of 31 subjects who participated in Experiment 4. Importantly, PIGS was evident in both groups bilaterally in the corresponding location (black arrows). Thus, despite its small size, this area was detectable even in the group-averaged activity maps based on large populations. Notably, in both panels, maps were generated based on random-effects, after correction for multiple comparisons. In both maps, the location of restrosplenial cortex/medial place area (RSC/MPA) and transverse occipital sulcus/occipital place area (TOS/OPA) are respectively indicated with white and green arrows.

Posterior intraparietal gyrus scene-selective site (PIGS) was detected consistently across sessions.

Panel (A) shows the stimuli used for localizing PIGS during 7T scans. Stimuli including indoor, manmade outdoor, and natural outdoor scenes and faces other than those used in Experiment 1. Panels (B) and (C) show the significance (p<10–2) of activity evoked by ‘scene > face’ contrast in the 3T scans (Experiment 1), overlaid on subjects own reconstructed brain. Panel (D) shows the significance (p<0.05) of activity evoked by ‘scene > face’ contrast during 7T scans (Experiment 2). Despite the difference in scanners (3T vs. 7T) and stimuli, the location of PIGS remained mostly unchanged. Panel (E) shows the location of PIGS, measured in 3T (black dashed lines) and 7T (green dashed lines) relative to the location of area V6 (white arrow), localized functionally based on the response to ‘optic-flow > random motion’ (Experiment 3 a). In all subjects, the center of scene- and optic-flow-selective responses was adjacent, but not overlapping.

Area posterior intraparietal gyrus scene-selective site (PIGS) is located outside the parieto-occipital sulci (POS) and adjacent to the functionally localized area V6.

Panels (A) and (B) show the probabilistic localization of areas PIGS and V6, respectively (see ‘Methods’). Panel (C) shows the probabilistic localization of areas restrosplenial cortex/medial place area (RSC/MPA) and transverse occipital sulcus/occipital place area (TOS/OPA). All probability maps are thresholded at 20–50% (red-to-yellow) and overlaid on the fsaverage. Panel (D) shows the relative location of these sites. Consistent with the results from the individual maps (Figure 4E), PIGS and V6 were located adjacent to each other, such that V6 was located within the POS and PIGS was located outside the POS (within the intraparietal gyrus) with minimal overlap between the two regions.

Localization of posterior intraparietal gyrus scene-selective site (PIGS) and transverse occipital sulcus/occipital place area (TOS/OPA) relative to the retinotopic visual areas in the right hemisphere of two subjects.

The right and left columns show respectively the polar angle and scene > face response mapping, collected in a 7T scanner on two different days. In both subjects, PIGS was located close to areas IPS3-4. Area TOS/OPA overlapped with areas V3A/B and IPS0 (V7). The borders of visual areas (defined based on the polar angle mapping) are indicated by dashed black lines. Notably, for both subjects, maps were overlaid on their own reconstructed flattened cortex. No activity smoothing was applied to the collected data (i.e., FWHM = 0; see ‘Methods’). Similar results were also found in the opposite hemispheres (not shown here). On the right column, the scale bars indicate 1 cm.

Probabilistically generated labels can be used to measure posterior intraparietal gyrus scene-selective site (PIGS) response in different experimental conditions.

Panel (A) shows the activity evoked by scenes and faces, across PIGS, V6, restrosplenial cortex/medial place area (RSC/MPA), and transverse occipital sulcus/occipital place area (TOS/OPA), all of them localized based on probabilistically generated labels based on a different group of subjects. Panel (B) shows the level of scene-selective activity, measured as ‘scene – face’, within these regions. Despite the small size of PIGS, the probabilistic label could detect the scene-selective activity within this area and the level of this activity was significantly higher than the adjacent area V6. In all panels, each dot represents the activity measured in one subject.

Posterior intraparietal gyrus scene-selective site (PIGS) could also be detected based on the ‘scene > object’ contrast.

Panels (A) and (D) show the stimuli used in Experiments 5a and 5b respectively. Panels (B) and (E) show the activity maps evoked by ‘scene > object’ contrast in two different individuals who participated in Experiments 5a and 5b. Panels (C) and (F) show the activity maps evoked by a different set of scenes and faces (used in Experiments 1 and 4) in the same individuals. The location of PIGS, as indicated by the blacks, remained unchanged between the two maps.

The application of probabilistically generated labels to measure the posterior intraparietal gyrus scene-selective site (PIGS) response to ‘scene vs. object’ stimuli.

Panels (A) and (C) show the activity evoked by scenes and objects in Experiments 5a and 5b, respectively. Panels (B) and (D) show the level of scene-selective activity within the regions of interest. As in Experiment 4, the probabilistic label detected the scene-selective activity within PIGS and the level of this activity was significantly higher than the adjacent area V6. Other details are similar to Figure 7.

Example of stimuli used in Experiment 6.

Coherently changing scenes implied ego-motion as if the observer was jogging through a trail. Incoherently changing scenes consisted of the same scene images as the coherently changing scenes but presented in a pseudo-random order. Face stimuli consisted of a mosaic of faces. These stimuli were different than those used in the previous experiments.

Scene-selective response to coherently vs. incoherently changing scenes within the intraparietal region (Experiment 6).

Panels (A) and (B) show respectively the group-averaged activity evoked by coherently and incoherently changing scenes relative to faces. Panel (C) shows the group-averaged response evoked by the ‘coherently > incoherently changing scenes’ contrast. Among scene-selective areas, only posterior intraparietal gyrus scene-selective site (PIGS) showed significant sensitivity to the observer ego-motion. Besides PIGS, this contrast also evoked activity within area MT (cyan arrows), also within more dorsal portions of the parietal cortex. Panel (D) shows the location of scene-selective areas in the same group of subjects based on an independent set of scene and face stimuli (Experiment 1). In all panels, the location of PIGS outside the parieto-occipital sulci (POS) defined based on panel (D) is indicated by black dashed lines. All maps were generated based on random-effects, after correction for multiple comparisons.

The scene-selective activity evoked within posterior intraparietal gyrus scene-selective site (PIGS) is influenced by the ego-motion.

Panel (A) shows the scene-selective activity evoked by the coherently (red) and incoherently changing scenes (blue), measured relative to the response to the faces, across areas PIGS, V6, restrosplenial cortex/medial place area (RSC/MPA), transverse occipital sulcus/occipital place area (TOS/OPA), and parahippocampal place area/temporal place area (PPA/TPA). Panel (B) shows the level of difference between the response evoked by ‘coherently – incoherently changing scenes’ across the regions of interest. While all regions showed a significantly stronger response to scenes compared to faces, PIGS showed the strongest impact of ego-motion on the scene-selective response. Other details are similar to Figure 7.

The group-averaged activity map evoked by the ‘biological > translational motion’ contrast.

Despite the low threshold used to generate these maps, we did not detect any significant activity evoked by the ‘biological > translational motion’ contrast within the posterior intraparietal gyrus scene-selective site (PIGS) and/or the other scene-selective areas. Rather, this contrast evoked a significant activity mainly within the inferior temporal sulcus (ITS), medial temporal sulcus (MTS), and superior temporal sulcus (STS). In both hemispheres, the location of PIGS, detected based on an independent set of scene and face stimuli (as in Experiment 1), is indicated by black dashed lines. The location of other scene-selective areas is indicated by white dashed lines.

Additional files

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Bryan Kennedy
  2. Sarala N Malladi
  3. Roger BH Tootell
  4. Shahin Nasr
(2024)
A previously undescribed scene-selective site is the key to encoding ego-motion in naturalistic environments
eLife 13:RP91601.
https://doi.org/10.7554/eLife.91601.3