David Barack's avatar

David Barack

@dlbarack.bsky.social

497 followers 372 following 149 posts

Philosopher and neuroscientist | Studying reasoning and foraging | Conceptual and theoretical foundations of cognition | #T1D

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

It doesn't always happen but I had a wonderful R1 on a paper that just came back as R&R. Great comments, could not 'unreservedly' recommend pub, but complimented the paper and offered helpful suggestions. +1, would do again

0 replies 0 reposts 3 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

This looks cool

0 replies 1 reposts 3 likes

Reposted by David Barack

The Memory Palace's avatar The Memory Palace @philmemopalace.bsky.social
[ View ]

We are happy to announce the launch of The Memory Palace! The Memory Palace is a space for sharing scholarly work in and around the philosophy of memory. Subscribe at thememorypalacephil.substack.com to receive an email each Tuesday as a new post debuts. (1/5)

1 replies 8 reposts 14 likes

Reposted by David Barack

Steve Wise's avatar Steve Wise @granularprefrontal.bsky.social
[ View ]

My new book, Cortical Evolution in Primates, is now shipping from both Amazon and Oxford Univ. Press USA.

0 replies 9 reposts 13 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

Neither do I! I just think they may be explanatorily epiphenomenal.

1 replies 0 reposts 1 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

Literal garden path sentence

0 replies 0 reposts 2 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

And Cole did something like this in the early 2010s, iirc, but with neuroimaging

0 replies 0 reposts 0 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

I think Buschman has been trying to get there—see the Tafazoli preprints. But they use a very small number of tasks iirc

1 replies 0 reposts 1 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

But what are these “functional gradients”? Computing the value of a location in the VF to direct gaze doesn’t seem to admit a gradient. I agree that some functions do; untangling object representations might for instance. So how to come up with a principled understanding of a func gradient?

2 replies 0 reposts 0 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

100%. Something like what Driscoll et al 2022 did for RNNs.

1 replies 0 reposts 1 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

Multifunctionality and variable function assignment are conceptually orthogonal.

0 replies 0 reposts 0 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

Sure. But there is room for context-sensitive, multifunctional invariant functions. Context might determine the function a part performs, with different functions performed in different contexts, but the part still has all those functions b/c it is competent to perform them in the right context.

2 replies 0 reposts 0 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

Ah, that’s diachronic mixed selectivity (over time). Mixed selectivity can also be synchronic (at a time): for ex, a cell is mixed selective at a time if it codes for a mixture of task conditions. And that kind of coding can be stable across time!

0 replies 0 reposts 1 likes

Reposted by David Barack

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

100% agreed. I think we often just haven't nailed down what the functions are! This relates to my earlier comments about using more complex tasks; some (most?) of the tasks in e.g. primate neuro don't seem to me to be complex enough to get what those regions are likely doing

1 replies 0 reposts 2 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

Of course data and evidence are key. But so are clear definitions of the functions we are trying to localize! Is it that we have failed to localize PFC functions or that we have failed to correctly specify them? The latter is a conceptual and theoretical issue that also involves formalization.

1 replies 0 reposts 1 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

This might just be that the functions are often ill-specified!

1 replies 0 reposts 1 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

The functions that fail to localize are typically folk psychological--beliefs, desires, etc. Things aren't that much better for control or attention. But once we move to more computational constructs like salience, cognitive maps, coordinate transformations, etc., localization is more successful.

1 replies 0 reposts 1 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

While I agree that dynamics and electrical influences trascend anatomy, it does not follow that you can't localize function! What is meant by function is key here--what typically disintegrates is localization of folk psych function. The headier, more scientific senses of function are better off

1 replies 0 reposts 1 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

One of my absolute favorite books !

0 replies 0 reposts 0 likes

Reposted by David Barack

Lauren McKenzie [free 🇵🇸]'s avatar Lauren McKenzie [free 🇵🇸] @themckenziest.gay
[ View ]

Holy shit did you see the riot police response to the peaceful student protests at Pomona College yet??

19 replies 130 reposts 288 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

Absolutely. I've taught some myself! I think we need those complex tasks to get at how items are manipulated in WM. I've always wanted to do something like an 'item construction task' precisely to see how items might be manipulated in WM, which I believe might occur within, not just across, areas.

1 replies 0 reposts 0 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

And the values of options, of course. The question is, what about more complex cognitive activity? The sorts of transformations that are purported to occur in WM go well-beyond most macaque study designs

1 replies 0 reposts 1 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

Like so much in biology, it's probably both in diff populations and in the same. For within the same population, can be modeled as RNN; trajectories through the (sometimes low-dim) state space are the transformations. That's my take on it.

1 replies 0 reposts 1 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

And what does that even look like?! For example, if the maintenance is some pattern, then the transformation of the information is a change from one pattern to another. How do those transformations of patterns of activity occur in the same population?

1 replies 0 reposts 0 likes

Reposted by David Barack

Olivia Guest · Ολίβια Γκεστ's avatar Olivia Guest · Ολίβια Γκεστ @olivia.science
[ View ]

I'm looking back at this paper written 2019 & published 2021 with @andreaeyleen.bsky.social: How Computational Modeling Can Force Theory Building in Psychological Science (pdf: oliviaguest.com/doc/guest_ma...) We presented a positive view and kind, if I do say so myself, compatibilist stance. 1/

1 replies 7 reposts 21 likes

Reposted by David Barack

Reposted by David Barack

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

Another manuscript is away!

0 replies 0 reposts 1 likes

Reposted by David Barack

Ted Underwood 🦋's avatar Ted Underwood 🦋 @tedunderwood.me
[ View ]

An accessible introduction to the state-space models that excel at long contexts and are starting to compete with (or complement) Transformers. #MLSky 🤖

1 replies 5 reposts 17 likes

Reposted by David Barack

hakwan lau's avatar hakwan lau @hakwan.bsky.social
[ View ]

i plan to move to Korea later this year, & will soon hire at all levels (students, postdocs, staff scientists, junior PIs) -

docs.google.com/document/d/1...

my lab in Japan will run till at least 2025. my job has been nothing but the dream job. but hopefully the above explains it.

#neuroscience

5 replies 33 reposts 44 likes

Reposted by David Barack

Reposted by David Barack

Elihu M. Gerson's avatar Elihu M. Gerson @emgerson.bsky.social
[ View ]

Study of interdisc. research at UC Davis has open 2-year post-docs, 1 in network & bibliometric analysis, 2 in sociol. of work &/or phil. of bio., studying www.biologicalpurpose.org. Familiarity with evol. biol. strongly preferred. Apply by April 30, 2024. Info at:
griesemer.net/call-for-pos...

0 replies 14 reposts 13 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

5.     This is the culmination of several years of collaborative, concerted effort and intellectual labor. I think it is a game changer and I’m honored to be part of this team. Congrats to all! (fin)

0 replies 0 reposts 1 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

4.     These criteria are defined in information-theoretic terms and are related to common analysis methods, including correlation analysis, encoding and decoding models, representational similarity analysis, and statistical testing.

1 replies 0 reposts 1 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

3.     For a neural response to be evidence of representation, it must be sensitive to the feature, it must be specific to the feature, it must be invariant to other features, and it must be functional—used downstream in the brain.

1 replies 0 reposts 3 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

2.     We detail four criteria for when a neural response represents some feature in the world and illustrate with several examples from the literature, including representations of orientation, numerosity, faces, and spatial location.

1 replies 0 reposts 4 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

1.     Excited to see this out in the world! We (Stephan Pohl, Edgar Walker, myself, Jenn Lee, Rachel Denison, @neddo.bsky.social , Florent Meyniel and Wei Ji Ma) present an account of evidence for representations in the brain. arxiv.org/abs/2403.14046

1 replies 14 reposts 19 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

This is amazing--goldfish do not show observing responses (a la Bromberg-Martin and many, many others)--!

0 replies 0 reposts 1 likes

Reposted by David Barack

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

Just now catching up: another one down...

0 replies 0 reposts 0 likes

David Barack's avatar David Barack @dlbarack.bsky.social
[ View ]

Thank you! Helpful just to have some idea

0 replies 0 reposts 1 likes

Reposted by David Barack

Andrew MacAskill's avatar Andrew MacAskill @macaskillaf.bsky.social
[ View ]

!! come join us as a group leader !! we have 2 lectureships just opened for applications in our department at UCL! Join one of the best, the most exciting and most friendly neuroscience departments!

2 replies 11 reposts 7 likes