Research Topics
Functional Mapping of Face-Selective Regions of the Marmoset's Cortex -using fMRI
It is an open question whether marmoset, a New World monkey, possess a similar extrastriate functional organization specialized for face processing, as shown in macaques and humans.To address this question, we measured BOLD fMRI responses in the brain of awake, behaving marmosets in a 7T scanner. We report that at least five face-selective patches mark the occipitotemporal pathway of the marmoset, with the most anterior patches showing the strongest preference for faces over other stimuli. The face patches located either within or ventral to superior temporal sulcus (STS), closely resembling that observed in macaques and humans.
Visual Response Properties to Complex Stimuli of Ventral Stream Visual areas - using ECoG (electrocorticography)
Medial Axis Shape Coding in Macaque Inferotemporal Cortex
We implanted a pair of 32-channel ECoG arrays across a large swath of the occipitotemporal cortex ranging from V1 to TE in two marmosets. The high-gamma band powers of local field potentials (LFP) showed strong stimulus categories selectivity. Four cluster of electrode sites demonstrated face selectivity. We observed an apparent division between dorsal and ventral face patches in the response pattern. Specifically, the high-gamma responses to faces in the dorsal areas PD and MD were transient. By contrast, PV and MV showed more sustained responses with a longer delay.
The basic, still unanswered question about visual object representation is this: what specific information is encoded by neural signals? I designed an algorithm to generate parametric-defined 3D random shapes with modifiable surface and skeletal features. Using an adaptive shape-sampling algorithm, we probed the single unit response properties of macaque’s inferotemporal cortex (IT). Our metric shape analyses revealed a coding continuum, along which most neurons represent a configuration of both medial axis and surface components. The neurons also showed broad tolerance but not total invariance to 3D rotation of the shapes, suggesting that they were tuned to view-specific 3D models. In summary, this work demonstrated that IT embodies a rich basis set of single unit activities for simultaneously representing skeletal and external shape features. This would be especially useful for representing biological shapes, which are often characterized by both complex, articulated skeletal structure and specific surface features.
Ongoing Projects
Cross Species (Marmoset vs. Macaque) Comparison of Cortical Visual Responses to Naturalistic Movie Viewing
We investigated neural responses during free viewing of naturalistic movies using fMRI in both marmoset and macaque (collaborated with Dr. Russ). Whole brain fMRI was carried out during the viewing of the same movies in the two species. In the main analysis, we created low-level (e.g. luminance, contrast, motion) and high-level (e.g. faces, bodies, scenes) feature models extracted from the evolving content in each of the movies. We then generated functional maps, based on the similarity between hemodynamic time courses and feature models. This analysis revealed similar robust principles of cortical processing based on both ECoG and fMRI results. First, most of the visual responsive voxels shared a common mode, which was modulated by the contrast of the movie. Second, motion signals drove large regions of the visual cortex, particularly along superior temporal sulcus(STS), including area MT and FST, whereas foveal V1-V4 were less sensitive to stimuli motion. Currently, we are comparing the responses temporal profiles within individual marmoset's and macaque's face patches and hope to establish functional homology between them.
Probe Functional Interdependencies among Face Patches - using electrophysiology & optogenetics
Because of the marmoset’s small and lissencephalic cortex, I can use a single chronic cranial chamber to cover most of the ventral stream, and study simultaneously multiple face patches on the lateral cortical surface. The chamber will be combined with an artificial dura design, through which I can deploy diverse experimental methods, including: single unit recording with penetrating electrodes, viral or tracer injections with glass pipettes, optical imaging both at the single-cell or mesoscopic levels, and optogenetic manipulation. This unique experimental setup allows me to probe causal relationships between the different face patches and the individual contributions of each face patch to visual
perception by combining simultaneous multi-site recordings with optogenetic silencing of different areas.