Cross-decoding perceived and intended visual motion in hMT+

Poster Presentation 53.431: Tuesday, May 19, 2026, 8:30 am – 12:30 pm, Pavilion
Session: Motion: Mechanisms, models

Wentao Si1 (), Eunhye Choe1, Nathan H. Heller2, Patrick Cavanagh3, Viola S. Störmer1, Peter U. Tse1; 1Dartmouth College, 2Johns Hopkins School of Medicine, 3Glendon College, York University

The apparent motion quartet is a bistable stimulus in which two dot pairs alternate in succession: one appearing in the upper-left and lower-right positions, and the other in the opposite corners of an invisible rectangle. This produces alternating percepts of horizontal and vertical motion. Prior psychophysical work has shown that volitional intention – choosing to perceive vertical or horizontal motion – modulates the perceptual bistability of this ambiguous stimulus (Kohler et al., 2008; Sun et al., 2017). Here we investigated the neural basis of intending to see a specific motion direction, using fMRI with a cue-delay, “single-shot alternation” paradigm. Methods: We cued observers to perceive either vertical or horizontal motion and measured neural activity using 3T fMRI during the variable delay period (6s-10s) prior to an instantaneous dot-pair switch. In separate scans, we also recorded neural responses while observers passively viewed apparent or physical motion along the illusory trajectories. Results: Using MVPA and cross-decoding techniques, we found that apparent motion elicited activation patterns in hMT+ similar to those elicited by physical motion, broadly replicating a previous 7T high resolution study (Schneider et al., 2019). In addition to replicating this finding, we also found robust cross-decoding between perceiving physical motion and intending to perceive motion during the delay period in which neither physical nor apparent motion was present. Finally, we also observed reliable (though weaker) cross-decoding performance between apparent motion percepts and intended motion within hMT+. This suggests that similar representations in hMT+ exist while observers intend to see vertical versus horizontal motion and when they actually perceive corresponding motion paths.

Acknowledgements: NSF Award ID 2241975