haley keglovits
@haleykegl
Followers
193
Following
2K
Media
23
Statuses
320
no longer using this platform, sorry if I ignore you // phd student @brownclps studying cog control @badrelab // previously @cal @ccnlab // go bears // she/her
Joined December 2013
Apoorva's comparison is right, and he really carried this ring through the finish line! đź’Ť I feel so lucky to have this project be my first major work in grad school and learned so much from him.
This is the hardest project I've worked on: extensive methods development, painstaking piloting, writing two grants, intensive data collection, and a LOT of thinking. It needed a huge dose of patience as we carried its burden over many years. A bit like Frodo carrying the ring.
0
0
3
RT @apaxon: Collectively, studying representations of two different task structures in the same subjects revealed generalizable principles….
0
1
0
RT @apaxon: The flat task showed local high-dim structure and orthogonality across clusters that were unrelated to the structure of the tas….
biorxiv.org
0
1
0
RT @apaxon: However, there were clues in the data suggesting lPFC may have started with a task-agnostic, high-dim representation with learn….
0
1
0
RT @apaxon: Therefore, at least in highly trained subjects, lPFC learned task-tailored representations that recapitulated the structure of….
0
2
0
RT @apaxon: On the other hand, in the flat task, a global axis encoded the response-relevant, XOR categories abstractly. Category-specific….
0
2
0
RT @apaxon: In the hierarchy task, the global axis abstractly encoded higher-level context, while low-dimensional, context-specific local g….
0
2
0
RT @apaxon: Using a series of decoding analyses, we comprehensively worked out the detailed local structure within each cluster in both tas….
0
1
0
RT @apaxon: Nevertheless, lPFC representational geometry for each task was highly tailored to its structure. In each task, clustering cre….
0
1
0
RT @apaxon: Across both tasks, inputs were encoded on manifolds of intermediate dimensionality, with at least some non-linear mixing of inp….
0
1
0
RT @apaxon: With decoding analyses, across both task structures, we found lPFC coding diverse task-relevant information. On the other hand,….
0
1
0
RT @apaxon: As we have previously shown, lPFC representations are hard to study with fMRI, with poor pattern reliability and small effects.….
0
1
0
RT @apaxon: One task used a categorization rule for mapping inputs to outputs. The other used a flat, XOR structure. We yoked & counterbala….
0
1
0
RT @apaxon: To really test these two accounts apart, one needs to characterize lPFC representations in 2 very different tasks in the same s….
0
1
0
RT @apaxon: Another account is that lPFC flexibility is a consequence of representation learning. lPFC just learns specialized, task-tailor….
0
1
0
RT @apaxon: One account, popularized by Mattia Rigotti & Stefano Fusi, is that lPFC non-linearly mixes inputs, projecting them on a high-di….
0
1
0
RT @apaxon: Dozens of studies show that lPFC neurons are highly flexible, coding whatever task is being performed. How lPFC accommodates di….
0
1
0
so excited to share this!! thread coming. .
New preprint! Joint work w/ @haleykegl & @BadreLab. lPFC flexibly codes tasks of diff structure. How? We test 2 prevalant ideas 1) it uses a high dim, expressive geometry, agnostic to structure 2) it learns tailored geometries for each structure. tldr - Its 1, with an asterisk👇.
0
0
0
RT @browndoglab: 🔍Recent publication highlight: our grad student, @madelinepelgrim, and PI, Dr. Daphna Buchsbaum, recently published in Pro….
0
19
0
specifically ability to work with .py files (not notebooks) and to run a kernel for line by line development. in line visuals etc are a plus but I'm more concerned about my python workflow for code development, especially with big data.
0
0
0