Saloni Sharma
@amrahs-inolas.bsky.social
Former PhD student @KULeuven. Imposter syndrome-ing postdoc at Livingstone Lab @harvardmed. Simultaneously experiencing child-like joy and crippling nihilism.
Pinned
Saloni Sharma
@amrahs-inolas.bsky.social
· Nov 23
Face cells encode object parts more than facial configuration of illusory faces
Nature Communications - Macaque face cells respond to objects humans perceive as illusory faces, yet the specific features that drive these responses remain unclear. Here, the authors show face...
rdcu.be
Our new work is out in @NatureComms! rdcu.be/d0ea2. Face cells are pareidolia-selective, but this selectivity 1) is not explained by human ratings, 2) did not require a face-like configuration, but was 3) driven by local features 4) explained by features present in non-pareidolia nonface objects.
Curious how the face-processing system recovers when early face experience is missing? Come see me present new data on recovery and limits of plasticity at #SfN2025 @sfn.org on Tues, Nov 18 at 10:45 am (SDCC Rm 33)!
www.abstractsonline.com/pp8/#!/21171...
www.abstractsonline.com/pp8/#!/21171...
Program Planner
www.abstractsonline.com
November 13, 2025 at 7:07 PM
Curious how the face-processing system recovers when early face experience is missing? Come see me present new data on recovery and limits of plasticity at #SfN2025 @sfn.org on Tues, Nov 18 at 10:45 am (SDCC Rm 33)!
www.abstractsonline.com/pp8/#!/21171...
www.abstractsonline.com/pp8/#!/21171...
Reposted by Saloni Sharma
These findings suggest that the predictive power of LMss for human visual cortex responses is not due to the evolution or learning of language per se, but rather reflects the statistical structure of the visual world as captured by natural language
www.biorxiv.org/content/10.1...
6/6
www.biorxiv.org/content/10.1...
6/6
Monkey See, Model Knew: Large Language Models Accurately Predict Visual Brain Responses in Humans and Non-Human Primates
Recent progress in multimodal AI and language-aligned visual representation learning has rekindled debates about the role of language in shaping the human visual system. In particular, the emergent ab...
www.biorxiv.org
March 14, 2025 at 4:21 PM
These findings suggest that the predictive power of LMss for human visual cortex responses is not due to the evolution or learning of language per se, but rather reflects the statistical structure of the visual world as captured by natural language
www.biorxiv.org/content/10.1...
6/6
www.biorxiv.org/content/10.1...
6/6
Reposted by Saloni Sharma
New preprint “Monkey See, Model Knew: LLMs accurately predict visual responses in humans AND NHPs”
Led by Colin Conwell with @emaliemcmahon.bsky.social Akshay Jagadeesh, Kasper Vinken @amrahs-inolas.bsky.social @jacob-prince.bsky.social George Alvarez @taliakonkle.bsky.social & Marge Livingstone 1/n
Led by Colin Conwell with @emaliemcmahon.bsky.social Akshay Jagadeesh, Kasper Vinken @amrahs-inolas.bsky.social @jacob-prince.bsky.social George Alvarez @taliakonkle.bsky.social & Marge Livingstone 1/n
March 14, 2025 at 4:14 PM
New preprint “Monkey See, Model Knew: LLMs accurately predict visual responses in humans AND NHPs”
Led by Colin Conwell with @emaliemcmahon.bsky.social Akshay Jagadeesh, Kasper Vinken @amrahs-inolas.bsky.social @jacob-prince.bsky.social George Alvarez @taliakonkle.bsky.social & Marge Livingstone 1/n
Led by Colin Conwell with @emaliemcmahon.bsky.social Akshay Jagadeesh, Kasper Vinken @amrahs-inolas.bsky.social @jacob-prince.bsky.social George Alvarez @taliakonkle.bsky.social & Marge Livingstone 1/n
Our new work is out in @NatureComms! rdcu.be/d0ea2. Face cells are pareidolia-selective, but this selectivity 1) is not explained by human ratings, 2) did not require a face-like configuration, but was 3) driven by local features 4) explained by features present in non-pareidolia nonface objects.
Face cells encode object parts more than facial configuration of illusory faces
Nature Communications - Macaque face cells respond to objects humans perceive as illusory faces, yet the specific features that drive these responses remain unclear. Here, the authors show face...
rdcu.be
November 23, 2024 at 10:46 PM
Our new work is out in @NatureComms! rdcu.be/d0ea2. Face cells are pareidolia-selective, but this selectivity 1) is not explained by human ratings, 2) did not require a face-like configuration, but was 3) driven by local features 4) explained by features present in non-pareidolia nonface objects.