The problem with "prediction error encoding" is the part about "encoding".

But there is a ton of evidence for prediction error or novelty modulating sensory responses, e.g.
while people watch movies the entire brain seems to be modulated by surprise:
www.nature.com/articles/s41...

We developed new state-of-the-art AI to detect breast cancer in MRI. This foundation model was trained on 30,000 MRI exams.

Paper: doi.org/10.1148/ryai...
Code: github.com/lkshrsch/Bre...

No Kings in Kingston, NY #NoKingsDay

Evidence in support of the "Brain from the inside out":

Response to movies is dominated by intrinsic brain dynamics. The dynamic itself is barely alternatered by the task of watching movies.

Human intracranial recordings from 5000 electrodes, now out in eLife.

elifesciences.org/reviewed-pre...

The intrinsic dynamic of the brain dominates its response to external stimulation. ... Think of waves on a quiet lake in response to pebbles you throw in.

elifesciences.org/reviewed-pre...

NYC had enough!

Finally a sensible discussion of oscillations in the brain!

www.sciencedirect.com/science/arti...

I love how their figures match our view of feed-forward and recurrent processing in the visual stream.

journals.plos.org/ploscompbiol...

Nice
@martinavinck.bsky.social @canalesjohnson.bsky.social

Large crowd in New York City at Stand Up for Science.

In @elife.bsky.social: Intrinsic dynamic shapes responses to external stimulation in the human brain doi.org/10.7554/eLif...

All new ideas come from young people in academia. In AI, creativity drops after they leave for industry. 99% of new medication in the last 15 years were supported by the NIH. Prosperity and health are on the line. It's common sense. Don't be silent. Speak your mind. March 7 NYC.

A bit of causal inference for everyone: Granger analysis when there are internal dynamics and external inputs. This happens everywhere. New method with examples from neuroscience, physiology, sociology and econometrics.

Code: Matlab, python, R

doi.org/10.1371/jour...

Remember, the human brain only uses ~20 Watt (estimate below based on ATP consumption).

Compute with less power is an engineering challenge worth solving in the coming decade.

OpenAI won't figure it out. Their VC bro only throws money at the same uninspired approach.

doi.org/10.1038/jcbf...

Backprop is the workhorse of AI, but unrealistic in biology, and way too inefficient in recurrent networks (BPTT).

One can also propagate sensitivity forward in time. This can be done efficiently and exactly for a large class of recurrent networks.

💥 Problem solved! 😁

arxiv.org/abs/2411.16695

As usual, Bernie speaks with absolute clarity on the disastrous performance of the Democratic Party ...

City University of New York, Morning of October 7, 2024.

Blinks change what we see. I think we need to include blinks when analyzing visual processing, on par with saccades.

www.pnas.org/doi/10.1073/...

Talking face video from speech. A little creepy

www.microsoft.com/en-us/resear...