Happy to share my second paper with Peter Dayan on our decision-theoretic approach to perceptual multistability (see the tweeprint of the first paper here x.com/neuroprincip...). Paper: A decision-theoretic model of multistability biorxiv.org/content/10.1... - ThreadSky

neuroprinciplist.bsky.social • 75 days ago

Happy to share my second paper with Peter Dayan on our decision-theoretic approach to perceptual multistability (see the tweeprint of the first paper here https://x.com/neuroprinciples/status/1564271812573372417).

Paper:
A decision-theoretic model of multistability https://biorxiv.org/content/10.1101/2024.12.06.627286v1

Comments

cimoore444.bsky.social•70 days ago

Hey, sorry we didn’t figure out connecting awhile back, would be great to have a zoom discussion of your stuff. Perhaps in January?

neuroprinciplist.bsky.social•69 days ago

I thought you are busy, and I didn't bother you with more emails. Anyhow, having a discussion over Zoom in January would be lovely. I will get in touch after the holidays to set it up. Very much look forward to it!

cimoore444.bsky.social•69 days ago

Wonderful, and you were right, been a hectic time even by my often rushing around standard! And thanks for being understanding, looking fwd to Jan

hadivafaii.bsky.social•74 days ago

You framed this as Helmholtzian inference versus RL, but there is a line of work that casts RL as inference (see review by Sergey Levine: https://arxiv.org/abs/1805.00909).

So my question is: given this connection, can we interpret your results under the unifying framework of RL as inference?

neuroprinciplist.bsky.social•69 days ago

Thanks very much Hadi, the paper seems very interesting (and wasn't aware of it). Let me check it and come back it again (but I guess, yes).

neuroprinciplist.bsky.social•75 days ago

1/11 We can take an internal reinforcement learning perspective to perception and conceive perceptual multistability as value-based decision-making (if you like, internal foraging) instead of passive inference and perceptual selections are internal actions like other sorts of actions

neuroprinciplist.bsky.social•75 days ago

2/11 Perceptual multistability is the perceptual phenomena of famous illusions like duck-rabbit (image below), binocular rivalry, Necker cube …

neuroprinciplist.bsky.social•75 days ago

3/11 Mulistability has been studied for more than 2 centuries! Huh! Centuries? Yes (image below) and attracted many prominent theoreticians (e.g.,
Bill Bialek) and experimentalists (e.g., @profgeraintrees.bsky.social) over the last three decades.

neuroprinciplist.bsky.social•75 days ago

4/11 So far, the dominant computational view on perceptual multistability is based on the Helmholtzian account of perception. There are several great works on that, e.g., studies by Alex Pouget, @biyuhe.bsky.social, @morenobote.bsky.social, @gershbrain.bsky.social @hohwy.bsky.social

neuroprinciplist.bsky.social•75 days ago

5/11 But Helmholtzian view cannot fully explain, e.g., why our perceptions are modulated by reward, the aesthetic and emotional content of images, social information, and task (e.g., see many great works of Sterzer & also our previous paper https://cell.com/neuron/fulltext/S0896-6273(22)00669-9).

neuroprinciplist.bsky.social•75 days ago

6/11 Based on a large body of neural and behavioral data, we formulate perceptual multistability as a *decision process*

kayson.bsky.social•75 days ago

Existence of a second paper with the same authors on the same framework strongly suggests a third paper?

Comments

Posting Rules

Reply