Anirudh Wodeyar @awodeyar@qoto.org

In a world where a bit of math can pass as human on the internet, we are obliged to be more unpredictable, more chaotic.

Be the language model failure mode you want to see in the taco.

You are here.

(March 2023 update)

Let's do #JoinMastodonDay on 9th of March!

Before: spread the word! Get to know Mastodon, choose a server, export contacts from 'other platforms', ...

On the 9th: Mastodon users write & boost tutorials and tips, share #Introduction posts, advertise servers, send invite links, ... and the Newcomers Come!

After: keep helping newcomers and sharing intro posts.

What say you? 😃​ Please boost if you're in!

an idea of @noellemitchell@mstdn.socialinspired by @alexwild 's #LeaveTwitterDay *

For #InternationalWomensDay, the entire collection of the Cambridge Elements series on women in the history of philosophy is available to download for free: https://www.cambridge.org/core/publications/elements/cambridge-elements-series/women-in-the-history-of-philosophy

#philosophy #HistoryOfPhilosophy #feminism #herstory @philosophy

Hello world! Elon Musk picking a twitter fight with Icelandic philanthropist and man of the year 2022 Haraldur Thorleifsson ended up being the final push that we needed, so now we are here on Mastodon. We don't know how any of this works, but follow us for posts on vision science, cognitive psychology, cognitive science, cognitive neuroscience, and stupid memes.

A short argument for why the big publishers cannot be part of a publishing reform effort

Science is stuck in a vicious cycle it is hard to escape from. The decision to publish a scientific paper is made based on an evaluation of its likely importance and technical correctness. Scientists are evaluated based on these publication decisions, and resources (jobs, grants and promotions) are distributed accordingly.

The current system distorts scientific priorities. Science is incredibly competitive, resources are allocated on a short term basis, and the primary metric used to evaluate scientists is their publication record. As a consequence, there is an unavoidable pressure to select problems and design studies that can lead to results that are likely to be favourably evaluated and published in the short term. This is in opposition to the long term scientific value, a fact that appears to be widely acknowledged by working scientists (https://www.vox.com/2016/7/14/12016710/science-challeges-research-funding-peer-review-process).

The current system is a vicious cycle and stable equilibrium. In principle, we could choose to evaluate scientists and their work in a better way. However, no individual or small group can do this alone. If an institution chooses to hire scientists who do work that they believe will be of enduring scientific value despite being unlikely to win short term grant funding, they will take a huge financial hit. Public research is under such severe resource constraints that this is simply not feasible for most institutions even if they wished to do so. Similarly, a public funding body that makes decisions based on long term scientific value and not short term publishability is likely to be able to count fewer high profile papers in their output, and compared to other funding bodies will appear to be underperforming when they are reviewed at the government level. Individual scientists have even less flexibility than these institutions.

Journal prestige cements this problem. It is the widespread availability of an easily calculated metric based on journal prestige that makes this cycle so hard to break. If there were no such metric, different groups could try different approaches and the effect would not be so obvious in the short term. The availability of the metric forces all institutions to follow the same strategy, and makes it hard to deviate from this strategy.

The majority of big publishers commercial value rests on their journal prestige. If there were no funding implications to publishing in one journal rather than another, scientists would be free to choose based on price or features. There are widely available solutions with better features at virtually no cost. Consequently, the entire business model of these publishers would collapse without the journal prestige signal.

Big publishers therefore cannot be part of the needed reforms. The success of these reforms would untie the evaluation of the quality of scientific work from the journal it is published in, and this would destroy the business model of these publishers. They will therefore do everything in their power to resist such reform.

Divorcing from the big publishers will not be enough. Journal prestige is the cement of the current negative stable equilibrium, but eliminating that will not guarantee a globally better system. We need systems for publishing and evaluating science that is diverse and under the control of researchers. This is what we intend to do with Neuromatch Open Publishing (https://nmop.io/).

Two excellent journalism discussions of the minefield that is ChatGPT. I really like thinking of it in terms of the classic ELIZA condition - that it is a mirror reflecting our own selves back at us. That the minefield is not that students will use it to write essays but that we will lose ourselves in the mirror.

(The disturbing part of ChatGPT professing its love and trying to get the NYT journalist to leave his wife is not that ChatGPT mimics emotion, but that a sufficiently large portion of long conversations on the internet are catfishing that it started mirroring that catfishing back.)

https://www.vox.com/future-perfect/23617185/ai-chatbots-eliza-chatgpt-bing-sydney-artificial-intelligence-history

https://theconvivialsociety.substack.com/p/the-prompt-box-is-a-minefield-ai

So... what are the differences between #ThetaSequences and #Replay (in the rodent #Hippocampus ?

Here is an illustration from real data!

Fig 1 shows ~70 #PlaceCells firing while a rat runs on a maze to reach a reward. See how the activity is different between running and pausing?
Fig 2 shows possible theta sequences.
Fig 3 shows possible replay.

Spot the differences & check alt text for more info!

PS: (very simple) #Matlab code to plot this is available here !

1/2

Cardiogenic control of affective behavioural state https://www.nature.com/articles/s41586-023-05748-8 ... (this looks wild, interesting, and important, #Deisseroth)

#Hippocampus neuroscientists:
Do you think that hippocampal #Replay 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 ⤵️​

#Oxford university and #Sackler money
Pleased to be interviewed by Antonia Cundy for this piece (£).
Empire of Pain by Patrick Radden Keefe is a compelling account of just why any institution that values ethics should have nothing to do with Sacklers or their money.
https://www.ft.com/content/6a42e764-46a1-4bdf-a76e-d548d78db8fc

This argues that - because of the "perils of a public good in private hands" - not just discussion should move from twitter to mastodon, scholarly institutions should now also create instances in the fediverse that make publicly available: papers, data & code.
https://www.nature.com/articles/d41586-023-00486-3

10 years after we created Registered Reports, the thing critics assured us would never (in a million years) happen has happened: @Nature is offering them.

The Registered Reports initiative just went up a gear and we are one step closer to eradicating publication bias and reporting bias from science.

Congratulations to all involved in achieving this milestone.

https://www.nature.com/articles/d41586-023-00506-2

Governments, banks, employers, and many other institutions make life-altering decisions about us by using machine learning to predict our future behavior. In a new paper, we challenge the legitimacy of this type of decision making, which we call “predictive optimization”. It is sold as accurate, fair, and efficient. But we show that it fails *on its own terms* and suffers from seven recurring flaws.

By @ang3linawang, @sayashk, Solon Barocas, and me.

https://predictive-optimization.cs.princeton.edu/

61 companies agreed to test out a 4-day workweek.

The results are in: Worker satisfaction went up, revenues remained steady — and nearly all of the companies plan to stick with it for now.
https://t.co/HtNzInmDzP

I will start this hippocampal 'history tour' with the contribution from Carolyn Harley. For those of you who did not know Carolyn (who passed away several moths ago) I will preface her thoughts by noting that she was an absolute gem of a scientist and person. Always cheerful, always helpful, and always curious. The past few years I was lucky enough to engage in weekly conversations with her (and recently including a few younger colleagues) to discuss the role of the locus coeruleus in modulating hippocmpal memory dynamics. Sadly our efforts though close to being publishable are now going to have to be finished up without her always thoughtful input.

With that as an introduction, here is what Carolyn had to say in response to the questions I posed:

1. What got you interested in the hippocampus? My supervisor, Dan Kimble, because of Dan’s lectures at the UO decades ago I switched my graduate studies plan from cross-cultural psychology to physiological psychology (as it was then called). His research interest was the hippocampus and he had studied with Bob Isaacson and spent postdoctoral time with Karl Pribram at Stanford. Because of his sojourn there I later had a chance to spend a summer in Chile with Theresa Pinto-Hamuy who had been with Dan at Stanford and who also deepened my interest in the biology of learning and memory more broadly. As a graduate student I spent a lot of time looking at rats with bilateral suction lesions of the hippocampus. We observed their maternal behaviour, sexual behaviour, orienting behaviour, dominance behaviour as well as testing them on various learning and memory tasks. I was struck by how relatively subtle the differences were. I tested the role of the hippocampus in attentional shifting for my PhD thesis but failed to support the hypothesis I was testing. Dan had predicted deficits in inhibiting attention to previously significant cues and I did not see such deficits. My most interesting results, failure to orient to a novel stimulus when competing stimuli were present, and fixated position responding when behavioural shifts were required in difficult learning tasks, could be explained, we then suggested, by subjects being prone to frustrative fixated behaviour patterns because they lacked inhibitory control over energizing hypothalamic systems. At a later time in my career as I turned to experiments involving the role of the locus coeruleus in modulating hippocampal input and in learning and memory I was struck by how similar the deficits after hippocampal lesions were to deficits after LC lesions. There is a link there somewhere. In any event by the end of my PhD thesis I thought the only way forward was to record from hippocampal neurons in behaving rats. I didn’t have the opportunity at that time for various personal reasons, but you will know from that observation what were the findings in the past 50 years that most excited me.

2. Aside from your own work, what findings about hippocampus (and related brain parts) in the past 50 years most excited you, and why? The findings showing that the hippocampus provides spatial, and likely, temporal structures to frame our episodic memories have most excited me. The idea of the transfer of individual life events to cortical encoding semantic records also had a revelatory quality for me.

3. Can you relate one personal story about interactions with colleagues that most exemplifies the world of hippocampal research? As a relatively young researcher in a relatively isolated university, the most amazing thing that happened to me, which I think exemplifies the world of hippocampal research, was when my husband wanted me to go to London for my sabbatical as he was interested in the poetry community there. By that time, the O’Keefe and Nadel book has appeared to revolutionize our thinking. With a friend, I had organized a senior seminar around that book and related findings. I had met John O’Keefe briefly at conferences but never had a real conversation with him. Of course that would have been my own dream sabbatical. So with spousal encouragement I phoned his UK lab (the Cerebral Functions Group as he had grandly named it) from Newfoundland to ask if I might spend a year there. He was so immediately encouraging (even seeming to know who I was) and welcoming that I was at lost for words. I spent a wonderful year in the UK and the very first experimental results he showed me were his first evidence that a hippocampal map provides a look at what the rat is thinking. He could predict the choices that would be made in a cue-less environment based on the map that had been called up earlier in the presence of cues (O’Keefe and Speakman Exp Brain Res. 1987;68(1):1-27). Andrew was still in the lab working on the data when I arrived. It was a glorious time and no experiment has ever made more of an impression on me than that did. Imagine looking at the brain and inferring accurately what another species is thinking and feeling that you understand why they do what they do subsequently. Listening to cells that year was all that I had hoped it would be. Despite John’s best efforts to design a poor lady’s Microdrive for me to take home, we only carried out a few experiments with stereotrode technology at my home University. (It did lead to an MSc thesis for John Huxter, who returned to John’s lab to work on phase precession.) But I think the poor lady’s Microdrive became the basis for the drives used by Kate Jeffries and Axona. Pat Wall and Clifford Woolf were also in the group at that time. They all shared as generously of their thoughts and their time as John and I also brought home new methods acquired from Cliff that led me to a better understanding of brain compartments and the suite of brain metabolic responses. Alan Ainsworth, John’s electrophysiological technician, was also wonderfully helpful. It was such a welcoming small group from top to bottom. I would run into Bernard Katz in the library and attended a rich assortment of local and visiting lectures, but it is the time in the informal (sneakers hung on the wall) environment of continual thought and discussion that has stayed with me most. It was so exciting. It is still my impression that the world of hippocampal researchers is a most welcoming one.

4. What would tell a young researcher interested in the hippocampus to focus on now?

I guess I see two rather different directions that would be exciting as foci for me if I were again a young researcher. One is to go in a network direction and look at what conversations are being held between the hippocampus and its subfields and other brain regions. The other is to ask more focused questions about the creation and support of hippocampal memories of varying durations. There are terrific new tools to investigate in both those directions or in any direction that captures your imagination. I began my studies of locus coeruleus/hippocampal interactions with the idea that I would come to include all the neuromodulatory hippocampal inputs and define their interactions as well as their roles in ‘isolation’ (if that ever happens). I am still finding new depths in that single input so must leave the larger conversation to others. It is an endlessly fascinating topic.