RT @EricMTrautmann: I'm thrilled to announce that our work to develop the Neuropixels-NHP 1.0 neural recording probe is now out in print in….

Our paper on developing Neuropixels 1.0 NHP for deep recordings in the primate brain is now out in @NatureNeuro.

RT @WadiaVarun: 1/10 Very excited to announce that my thesis project is now a preprint! We present the first detailed study of mental image….

RT @doristsao: This (Tuesday) afternoon at #sfn2024 ! Come check out @HesseJanis talk about his work on conscious perception in MCP Room S1….

Thrilled to share the discovery that face cells rapidly code-switch, within 20 ms, from a tuning optimized for face detection to a completely different tuning optimized for face discrimination. This switch is gated by successful detection of a face and doesn’t happen for objects.

The discovery of segmentation patches makes future studies on how the brain segments the visual scene more tractable. E.g., in the future, it would be interesting to see whether inhibition of these patches leads to behavioral deficits related to scene segmentation. (8/8)

We could also decode segmentation from cells in segmentation patches (86%, chance:25%), significantly better than outside the patches. Interestingly, color could be decoded less well from segmentation patches compared to outside regions. (7/8)

Remarkably, within fMRI-identified patches we found clusters of cells that signaled border-ownership (which is an important segmentation signal) invariant to cue! (6/8)

We next inserted electrodes into these regions to record from single cells. Importantly, previous studies had found that single cells from random locations may signal segmentation for a single cue but were inconsistent for other cues (. (5/8).

We compared fMRI activation for stimuli containing figures versus only background and found a set of patches that were activated consistently regardless of cue (texture, luminance, motion or disparity). These patches were mostly non-overlapping with color-activated regions. (4/8)

A rich literature has shown modularity in retinotopic cortex for specific low-level features such as color or orientation. Here, through combining fMRI and electrophysiology, we found modules encoding the output of a specific *computation*, visual scene segmentation. (3/8).

Segmentation is an important step in visual processing as it allows your brain to distinguish which regions of an image belong to different objects. Are there specialized regions in the brain that compute the segmentation of a visual scene? (2/8).

Excited to share that we found modules in visual cortex for segmenting a visual scene into objects. Find more details in our paper just published in PNAS ( or the summary thread below: (1/8).

Nice commentary by Steve Fleming.

Thank you, Sarah, that's a perfect one-sentence summary of our current work.

RT @sarahanmy: Some slides from .@doristsao of one of the most fascinating talks ever at #ASSC26 (Ned Block's words). In my very meager 1 s….

RT @PsySciBerkeley: Alex and Allyson Grey (@alexgreycosm and @allysongreycosm) are responsible for the most influential and visionary piece….

RT @doristsao: Happy this is now published in elife:

Grateful to have been a part of this collaboration. The NHP Neuropixels 1.0 probes described in this preprint might well present a little revolution for primate neuroscience, as they allow recordings from over a thousand neurons deep in the primate brain.

RT @doristsao: Interested in helping run a lab to understand visual perception in primates? Tsao lab @UCBerkeley is looking for a new lab m….