Abstract
Experimental stimuli are often constrained to what can be presented on a flat monitor. While approximating retinal input in this way has laid the foundation for understanding mechanisms of visual processing, the majority of visual information encountered in the real world is 3D. Although cortical area MT has been one of the most thoroughly studied areas of the primate brain, recent evidence has shown that MT neurons are selective for not only 2D retinal motions, but also exhibit selectivity for 3D motion directions (Czuba et al., 2014; Sanada & DeAngelis, 2014). We’ve shown that a model of MT that incorporates the projective geometry of binocular vision is predictive of human perceptual errors in 3D motion estimation (Bonnen et al., 2018). Importantly, perceptual errors and neural response predictions are strongly influenced by viewing distance(VD). We therefore measured the responses of MT neurons in awake macaque to binocular 3D moving dot stimuli rendered with full geometric cues using a motorized projection display that allowed us to precisely and dynamically control physical viewing distance in a range of 30–120 cm. Tuning for 3D direction was widely evident in neurons and similar at different VDs (regardless of disparity tuning). Many neurons with 3D direction tuning changed in overall response level across viewing distances. Moreover, we observed orderly transitions between interdigitated regions of 3D and 2D selectivity across linear array recordings tangential to the cortical surface. Interestingly, regions of 3D selectivity were not necessarily co-localized with classic disparity selectivity. Robust selectivity for 3D motion & space in MT extends beyond simple interaction of known selectivities, and may reflect a transition of information from retinal input space to environmental frames of reference. This finding reinforces the importance of stimuli that more fully encompass both the geometry of retinal projection & statistical regularities of the natural environment.
Acknowledgement: NIH R01-EY027023, NIH R01-EY020592, NIH T32 EY21462