Human cortical areas involved in perception of motion-in-depth

2563 

Abstract Submission 

Sean Lee¹, Sylvia van Stijn², Bo-yong Park³

¹Max Planck Institute for Empirical Aesthetics, Frankfurt, Germany, ²Ernst Strüngmann Institute for Neuroscience in Cooperation with Max Planck Society, Frankfurt, Germany, ³Inha University, Incheon, Incheon

Sean Lee  
Max Planck Institute for Empirical Aesthetics
Frankfurt, Germany

Sylvia van Stijn  
Ernst Strüngmann Institute for Neuroscience in Cooperation with Max Planck Society
Frankfurt, Germany

Bo-yong Park  
Inha University
Incheon, Incheon

The middle temporal area (MT/V5) in the primate brain is thought to play a major role in processing of motion. For human motion-in-depth perception, the MT complex (hMT+) and its vicinities have been suggested to be involved (Likova and Tyler, 2007; Rokers et al., 2009). The human MT+ is a cluster of half a dozen subregions including MT. To answer which of these and possibly what other cortical areas are involved in the perception of motion-in-depth we measured the BOLD signal while the participant viewed the 3-D motion stimuli in a series of fMRI experiments.

To maintain overall appearance of the stimuli constant across conditions we employed a dynamic random-dot stereogram (RDS) to elicit perception of a frontoparallel plane moving in 3D. New sets of random-dots were plotted on eleven disparity-defined layers every 50ms. One of the layers contained white dots and the rest contained black dots. In each frame, the white dots were assigned to a different layer. When successive assignment occurred in a neighboring layer toward one direction, observers perceived the white plane smoothly traversing in depth (motion-in-depth, MID); when the switching occurred in arbitrary layers in succession, observers saw the white plane but perceived no coherent motion (shuffled-depth-plane, SDP. In the control condition the white dots were evenly distributed to all layers, in which the white dots formed no surface but appeared as amorphic cloud of white dots in depth (RDRD).
Six adults were scanned in a 3T MRI scanner with an 8-channel RF head coil. The 1-mm isotropic whole-brain anatomical MRI was taken followed by the fMRIs (EPI: TR 2000 ms; TE 30 ms; FA 62, FOV 192 x 192 mm, 32 slices). The participant viewed the stimuli through the MR-compatible goggles and fixated a small red corss to perform a contrast discrimination task every 2 - 9 seconds. There were three contrast pairs in blocked design (12 seconds per block, repeated 8 times): MID-SDP, SDP-RDRD and MID-RDRD. Each contrast was ran 4 times (4mts each).
EPI data processing: BrainVoyager QX software including field-distortion correction, slice-scan-time correction, motion correction, high-pass temporal filtering, and we used general linier model convolved with hemodynamic response function and tested in a univariate way. For multiple runs/subjects, GLM multi-study with a fixed effect analysis and z-transformation. The anatomical scan was processed with FreeSurfer and Werkbench to reconstruct the cortical surface and pacellated (Glasser et al., 2017). The brain areas labled according to Glasser et al.

By alternating MID and RDRD conditions in a blocked design, we identified two concurrently activated areas: one in the dorso-occipital area (V3A/B) and the other in the tempero-occipital area (MST) anterior to MT. To determine which of the two areas is more directly involved in the perception of motion in depth, we compared the response pattern (BOLD percent signal change) of the two contrasts, MID-SDP and SDP-RDRD, in blocked designs. In the MID-SDP contrast, MST showed more prominent response compared to that of V3A/V3AB. In the SDP-RDRD contrast, V3A/V3B showed more prominent response than that of MST (Figure 1).

Our RDS stimuli were devised to control a single feature at a time. Between the MID and SDP conditions, disturbing only the time-series of the occurrence of the white-dot surface destroyed the perception of smooth motion in depth. Note that all other features, such as the number of random dots, overall contrast, and overall disparity distribution, remained identical for all stimulus conditions. Therefore, with a single-eye-only view, all three stimulus conditions appeared identical.
Because seeing the surface in our RDS is a prerequisite for perceiving motion in depth, we conclude that V3A/B is a preceding area that is involved in segmenting/representing the surface in depth in the scene. On the other hand, MST showed a more direct signal change toward motion in depth.

Attention: Visual ²

Perception: Visual ¹

Cognition

Cortex

ELECTROPHYSIOLOGY

FUNCTIONAL MRI

Machine Learning

MRI

MRI PHYSICS

Neuron

Perception

STRUCTURAL MRI