Claude Code is incredible, but if you push it far you will see it very quickly gets stuck in loops and I become its therapist.

Just wait until Schmidhuber discovers this thread.

Congrats!!

Don’t know if these things count since it’s debatable if they share anything in common with a human mind

I’m not convinced

@andpru.bsky.social @neurobeats.bsky.social @mkashefi.bsky.social @sujayane.bsky.social

Huge congratulations to first author Sujaya Neupane for leading this effort, and to our collaborators at Western University – Jessica Grahn, Andrew Pruszynski, Mehrdad Kashefi, and Rhonda Kersten, for their ideas, expertise, and support.

Strikingly, this encoding persisted in the GPi even after rewards were withdrawn, suggesting that the basal ganglia could act as an interface where extrinsic rewards are transformed into 'intrinsic value', effectively unmasking latent audiomotor connections.

However, once tones were paired with reward, a profound reorganization occurred: robust encoding emerged across the motor network, particularly in PMd and globus pallidus interna (GPi).

We found that motor areas do not spontaneously respond to auditory patterns during passive listening, but care a lot about trial structure and reward.

Using high-density Neuropixels probes, we recorded from neural populations across the motor and auditory systems of a rhesus macaque – first during passive listening to see how these circuits respond to auditory patterns.