@adredish@neuromatch.social avatar

adredish

@adredish@neuromatch.social

Scientist studying learning, memory, and decision-making. Poet and Playwright.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

ct_bergstrom, (edited ) to random
@ct_bergstrom@fediscience.org avatar

My colleague Kevin Gross and I have a new preprint up on the arXiv.

Just for fun, rather than a simple text explainer, a thread with some slides for a talk I'm giving at https://www.icssi.org/ tomorrow.

Here's the paper itself: Rationalizing risk aversion in science. https://arxiv.org/abs/2306.13816

adredish,
@adredish@neuromatch.social avatar

@ct_bergstrom

What's the data that scientists actually don't take risks? What's the data that "risky science" is "better science" and moves fields forward faster?

I know that these are both well-worn myths, but I don't know of any good data supporting these hypotheses.

Moreover, the "risk aversion" of funding agencies is not consistent at all with my personal experience on grant review committees or my own experience of grant funding/rejection. Nor is the "risk aversion" of scientific projects. Nor, for that matter, is the fact that most of the discoveries where I do know the origin stories are due to serendipity and careful observation, not "risk seeking".

adredish,
@adredish@neuromatch.social avatar

@elduvelle @neuralreckoning @ct_bergstrom

People get all worked up over negative results, but a real negative result is actually really hard to show. The problem is that a negative result can be negative for lots of reasons, most of them boring.

For example, maybe the DREADD didn't affect rat behavior because there was no DREADD in the virus. (That happened to us once.) Or maybe the human subject pool was tasked wrong because they didn't understand the instructions. (That happened to us once.)

To get a real negative result, you have to have positive controls to show that all of the techniques are doing what you think they are and that the negative result is not a consequence of a trivial outcome.

Yes, you need controls for positive results as well, but it's easier to determine what those controls are, and reviewers tend to demand those controls. People who try and fail to publish negative results almost never have the right controls for those negative results (which are not the same at all as the controls you need for the positive results). A well-structure negative result experiment should be very publishable.

adredish, to random
@adredish@neuromatch.social avatar

Question for neurophysiologists. Anyone who uses the Intan recording system. How do you sync up other signals (like video recordings) with Intan's neurophys? We have a hacked together system that works but I'd really like to clean it up. What do others do?

@elduvelle @hugospiers @katejjeffery

adredish,
@adredish@neuromatch.social avatar

@chrisXrodgers @elduvelle @hugospiers @katejeffrey @BorisBarbour

Video is by in-house coded Matlab (only works on older Matlab systems since the newer versions added too much overhead and can't keep up). We send a pulse and Matlab-time code every second via arduino to two of the Intan channels, but it requires complex post-recording analysis to sync them up. Was hoping for some way to send the Intan timecode out or to save the video linked... somehow. (One advantage of the integrated Neuralynx systems.)

Theoretically, OpenEphys can communicate via ZeroMQ, but I never got that to work.

We do have the triggers (feeders, stim events, opto signals) sent into intan digital inputs. But video, even XY locations, are too large.

adredish,
@adredish@neuromatch.social avatar

@elduvelle @chrisXrodgers @BorisBarbour Yep, we have 30 Hz reliably using Matlab (only works in earlier than 2015b). It works fine.

I've heard good things about bonsai, although I've also heard it's pretty heavy (requiring a lot of additional packages), but maybe not. I don't really want to rebuild our whole video system unless I have to.

adredish,
@adredish@neuromatch.social avatar

@elduvelle @Andrewpapale @chrisXrodgers @BorisBarbour

You could probably time-stamp VTEs online, but only after the behavior completes. It would be hard to differentiate VTE vs not in the middle of the VTE. But it wouldn't be too hard to recognize that a VTE had just occurred.

adredish, to random
@adredish@neuromatch.social avatar

Are there better preprint servers than #bioRxiv? Every time I try to post something to #bioRxiv, I have to fight with them that a paper doing new analyses on real data is a "real paper" just because we also do a good job of placing our results in the literature.

Moreover, #bioRxiv has an explicit policy that new theoretical insights are "not suitable for posting" (meaning they don't think theory is a real contribution to the literature), which is bad for science. What are the better other options?

It's almost (almost) as bad as fighting with editorial desk-rejections at a real journal.

#neuroscience #preprints #sciencepublishing #theory #experiment

adredish,
@adredish@neuromatch.social avatar

@schoppik

Not yet! That's the point. It will be up as soon as we're done fighting with the editors (sorry, "moderators") at bioRxiv or as soon as we submit it to a different journal (sorry, "preprint server").

The two problems that seem to arise when dealing with bioRxiv (aside from the flat out "theory doesn't count" bs - which is why the super-important "What is theory in neuroscience" paper ended up at arXiv [JNsci: https://www.jneurosci.org/content/43/7/1074; Preprint: https://arxiv.org/abs/2003.13825]) is that they don't like large review introductions even if there is good data in it and that they don't like re-analyses of old data, even if those re-analyses are novel and important. So the paper, which lays out a thorough discussion of different types of decision-making systems and then tests those hypotheses by new analyses of old data requires a fight to get past the editors (sorry, "moderators").

adredish,
@adredish@neuromatch.social avatar

@elduvelle I think it's more that we did a thorough review of the issues before testing the hypotheses. But I don't know. We'll see what they say. I've appealed to the editors.

(I thought part of the point of preprints was to get away from this gatekeeping.)

adredish,
@adredish@neuromatch.social avatar

@Learning

I would love to know what you do to get your mathematical and computational work into bioRxiv, because we have struggled, presumably because we are selling it wrong.

charanranganath, to random

For those who haven't heard Mariam Aly of Columbia Psychology learned that she would not be promoted to Associate Professor w/o tenure, so she is being denied tenure w/o an organized external review process. See her tweet:
Today I learned that
@Columbia
will not be promoting me or putting me up for tenure. I have multiple active grants (including an R01 and a CAREER award) plenty of publications, a teaching award, and a history of DEI advocacy. Open to all opportunities in and out of academia.

adredish,
@adredish@neuromatch.social avatar

@charanranganath @mariam

This sounds like an opportunity. Mariam (who has done some really great work BTW!) should consider other programs who are willing to support her in her career journey.

BTW, apropos of nothing, Minnesota Department of Neuroscience (https://med.umn.edu/neuroscience) is hiring...

Assistant Professor (Tenure Track): https://hr.myu.umn.edu/jobs/ext/353650

Associate or Full Professor (Tenured): https://hr.myu.umn.edu/jobs/ext/353649

elduvelle, to random
@elduvelle@neuromatch.social avatar

Come on - does no one care about SFN? 👀

From: @elduvelle
https://neuromatch.social/@elduvelle/110468701179097408

adredish,
@adredish@neuromatch.social avatar

@elduvelle

SFN is my favorite meeting. The key is to see SFN as a deconstructed meeting, where everyone gets to create their own experience.

Because every neuroscience question is at SFN, we can all find our own paths through it. I can spend a morning talking addiction, go to lunch with a striatal colleague, and then spend an afternoon talking hippocampus. And then spend the evening talking neuroeconomics. Someone else might spend the morning in drosophila head direction systems and the afternoon talking rat navigation with me. Everyone finds their own path through SFN.

The problem I see people having with SFN is that they treat it like a normal conference. It's not. It's more like a chance to swim in a great science ocean and to see lots of friends from many different communities. If you go into SFN trying to see specific things, you'll likely fail and hate it. But if you go with the zen attitude of "I'm going to see some great science for a few days", then it's a wonderful meeting.

adredish,
@adredish@neuromatch.social avatar

@nadege @elduvelle SFN takes a few years to accommodate to the ocean. The first year is overwhelming, but once you've been to a few, it gets comfortable and one learns to find the spaces that are the best way for you and can be a wonderful meeting.

elduvelle, to random
@elduvelle@neuromatch.social avatar

neuroscientists:
Do you think that hippocampal can truly represent a future planning trajectory?
Or that all replay trajectories are actually related to consolidation / generalization / other memory-oriented mechanisms?

Discussion & questions welcome!
1/2
Poll in Post 2 ⤵️​

adredish,
@adredish@neuromatch.social avatar

@WorldImagining @tyrell_turing @elduvelle

To be clear, the data is pretty strong that (awake) hippocampal "replay" is more likely some sort of generative exploration of the possibilities in the world than actual planning of future paths. (This is why you get short-cuts, novel paths, backwards sequences, etc.)

Sleep replay, particularly early slow-wave-sleep replay, does seem to be more veridical and may well be some sort of reactivation for consolidation.

REM replay seems to be more different.

Observations of planning seem to be happening during theta sequences, not replay.

adredish,
@adredish@neuromatch.social avatar

@WorldImagining @elduvelle @tyrell_turing @tdverstynen

One of the complexities is whether SWRs can occur in the middle of theta. Personally, I have never seen anything that I would call an SWR in theta. But Csicsvari has argued that SWRs also occur in the middle of theta. (Jozsef has said as much at his posters in SFN meetings in the past. I don't know what his current take is.) Loren Frank has argued that SWRs and theta cycles are two edges of a continuum. (Loren has said this in person - I don't know if he's said it in print and I don't know what his current belief is.)

My current belief is that these are very different processes, that SWRs do not occur during theta, and that people often mislabel high-gamma (140-180 Hz) signals that do occur in theta cycles as SWRs because they stretch their SWR bandwidths too low.

Note: belief = best story so far (belief in a Bayesian sense. :)

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • JUstTest
  • GTA5RPClips
  • DreamBathrooms
  • cubers
  • mdbf
  • everett
  • magazineikmin
  • Durango
  • Youngstown
  • rosin
  • slotface
  • modclub
  • kavyap
  • ethstaker
  • provamag3
  • ngwrru68w68
  • thenastyranch
  • cisconetworking
  • khanakhh
  • osvaldo12
  • InstantRegret
  • normalnudes
  • Leos
  • tacticalgear
  • tester
  • megavids
  • anitta
  • lostlight
  • All magazines
    •