@bonsai-lab.bsky.social @bennybakes.bsky.social

Lots of detail and examples that should help philosophers thinking about representation in cog sci. Plus, we end with a provocation re. those debates, concerning the way concepts like representation are typically studied (p 23).

We emphasize the way brains use representations, but also — and most importantly — how scientists use *the concept* of representation to direct and organize their research.

This is finally out! Details below, but basically: we gathered a group of philosophers and cognitive/computer scientists to explain what’s up with neural representation. #philsci #philosophy #philmind

I’m not great with the 300-character format, but I’m just finishing a review of this body of work in philosophy of cog sci — happy to share if interested.

Refs.

On representation: Frances Egan (mitpress.mit.edu/978026255160...) & myself (link.springer.com/article/10.1..., also this interdisciplinary collaboration: arxiv.org/pdf/2604.13829)

On computation: Egan (philarchive.org/rec/EGAMAC-2) & myself (onlinelibrary.wiley.com/doi/full/10....).

This can be true even if there’s no such thing as representation: no target for a theory of representation. That doesn’t stop the concept from doing a lot of work, and it doesn’t make our use of the concept misguided. Similar things can likely be said about many (not all) scientific terms.

But words do more than pick out properties. Describing brain activity as a representation may just be a way to avail yourself of the resources associated with that concept: modeling strategies, explanatory frameworks, ways of directing attention and framing questions, etc.

Really interesting question! I think this counts as 4: some terms don’t seem to pick out categories/targets at all. E.g., when we describe some brain activity as a representation, we *might* be saying it has a property — representation — which can then be theorized (on either option 1 or 2).

I volunteer to be the guy

Lisa Tessman has a couple books on this (her answer is yes, because of some real-world cases like the ones in the other comment: you should do A, you should do B, but doing one precludes doing the other).

Basically, there’s a clear logic by which things like Swampman are informative about the real world/real kinds. So Millikan and Dennett and the rest might be right: these creations aren’t good *examples* of scientific kinds. But they're a well-grounded *methodology* for inquiring into those kinds.

Awkward news for us empirically-minded philosophers of science: it turns out that wildly unrealistic thought experiments are good, actually. If you want an excuse to stop scrolling for a bit, I just had a paper published on Swampman & co. — link below!

#philsci #philosophy #philmind

AI and the Pedagogical Relationship As AI is being developed to serve educational roles, we should think about the human and social aspects of education, and whether AI can replicate them.

Some thoughts on AI Ed tools and what they do to the pedagogical relationship #edusky #philtech

(1) Swampmen and crazy thought experiments. I resuscitate Swampman and give a logic for thought experiments like him, show how they can be informative even when they’re ridiculously unrealistic.

Squat > deadlift? Huge red flag

... are you supposed to use it as a noun?