Hey #PredictiveProcessing people, do you have any tips on who is currently studying how and when the precision of the prediction error is calculated? My focus is perception and learning but I'm happy to hear about papers from other related fields.

Thanks so much for any pointers!!

Hi everyone 👋​! Another #introduction as I just moved instance.

I am an Associate Prof. at Harvard Medical School, and am interested in figuring out how the brain works 🧠​. In particular, how do its neural dynamics implement the probabilistic computations required for efficient behavior? We ask this question from the theoretical perspective 💻​, using tools from machine learning and physics, and collaborate with experimentalists 🐭​🙈​👤​ to test the arising theories.

In the past I have worked a lot on decision-making - both perceptual and value-based - normative models for the speed-accuracy trade-off, and how this relates to decision confidence and attention. Currently, we are looking into probabilistic reasoning about more structured objects, like hierarchies, and at efficient navigation in light of uncertain sensory information.

Happy to be here!

#neuroscience
#neuroAI

After many years in the making and sitting as a preprint, our opinion paper on why McGurk stimuli may not be the best for #audiovisual #speech processing is now out. In brief:

• For in-person speech, the auditory and visual speech information are always congruent (it's the talker's mouth producing the speech!), contrary to McGurk stimuli;
• Real speech is usually words and sentences, McGurk stimuli are often phonemes;
• Many McGurk-based paradigms require metalinguistic decisions ("what sound did you hear?") not present in everyday conversation.

And perhaps most damningly: Performance on McGurk tasks does not seem to correlate with perception of non-McGurk audiovisual stimuli.

Van Engen KJ, Dey A, Sommers MS, Peelle JE. Audiovisual speech perception: Moving beyond McGurk. The Journal of the Acoustical Society of America 152:3216–3225.

https://asa.scitation.org/doi/10.1121/10.0015262

With Musk’s call to prosecute Fauci, I’m thinking it’s finally time to step back from Twitter.

https://twitter.com/elonmusk/status/1601894132573605888

I’m also enjoying Mastodon more.

There are still powerful communities there doing good (URMs; advocacy) and I don’t fault anyone for staying there for that. But I don’t think I can justify my own contributions there, ethically.

How are all of you thinking about this topic?