Blake Richards Profile picture
Researcher at @mcgillu combining AI and neuroscience. Also on Bluesky (@tyrellturing.bsky.social) and Mastodon: @tyrell_turing@fediscience.org.
Nov 2, 2022 6 tweets 4 min read
@EngertLab @kendmil Nice thread, Florian (and Ken)!

But, one thought that arises from the last thing you said, Florian, is that the strongest version of your hypothesis cannot be true, because there are some capabilities that we know *must* require experience in some animals. For example... @EngertLab @kendmil Consider technical skills, like computer programming in humans (or opening garbage cans in raccoons). There is no chance that evolution has genetically wired the brain to be able to do these things sans experience, since there would have been zero chance for selection on them.
Feb 8, 2022 16 tweets 4 min read
1/ A new paper on the “brain-computer metaphor debate” by myself and Tim Lillicrap is now out in Frontiers in Computer Science:

frontiersin.org/articles/10.33…

tl;dr: The debate is useless because it’s actually just a semantic disagreement. 2/ The perspective in this paper shows how my thoughts on the matter have evolved, in large part because of the conversations that I have had on Twitter. Though we all tire of these debates, it helped me to get clear on the matter at least! 😅
May 26, 2021 5 tweets 2 min read
Yup. Many ppl seem to think that North America's decision to give a majority of urban space over to cars was natural, because we had new cities and lots of space. That is not accurate. We had dense cities like Europe and we tore them down to make freeways and parking lots. I went to high school in a neighbourhood in downtown Toronto called Cork Town. When I was there, it was a ghost town of old freeway on ramps and warehouses. I was very surprised to learn later in life that Cork Town used to be a dense neighbourhood of Irish immigrants.
Feb 19, 2021 6 tweets 2 min read
1/n Well, I just finished my first ever week as a reviewer for @NSERC_CRSNG Discovery Grants. I'm really happy I participated and I have a few major take-aways I wanted to share with the community: 2/n Take-away 1: The process is very fair! For every grant *the grid is paramount*. It doesn't matter how many CNS papers you have, or which university you're at, or whether you're using fancy tech, etc. The Q is always, where is this on the grid?

nserc-crsng.gc.ca/_doc/Professor…
Jan 14, 2021 16 tweets 7 min read
1/ I'm very happy to give a little thread today on our paper accepted at ICLR 2021!

🎉🎉🎉

In this paper, we show how to build ANNs that respect Dale's law and which can still be trained well with gradient descent. I will expand in this thread...

openreview.net/forum?id=eU776… 2/ Dale's law states that neurons release the same neurotransmitter from all of their axonal terminals.

en.wikipedia.org/wiki/Dale%27s_…

Practically speaking, this implies that neurons are either all excitatory or inhibitory. It's not 100%, nothing is in biology, but it's roughly true.
Nov 11, 2020 9 tweets 5 min read
1/ We have a new lab website!

linclab.mila.quebec

There were a variety of reasons for the change, but the biggest ones were related to commitments we made as a lab during #ShutDownSTEM in June:



Specifically, we've made the following changes: 2/ We've created an online application form with clear instructions:

linclab.mila.quebec/opportunities

From now on, I will direct anyone who emails me to this form, which eliminates the "hidden curriculum" of how to write to a PI. Hopefully, this limits potential implicit bias from me.
Jun 30, 2020 13 tweets 3 min read
1/n) A small thread on argument structure... I've been thinking about this bc @TheFrontalLobe_ recently posted about Tim Van Gelder's famous paper, and I turned into a complete jerk in response (sorry again André). I was asking myself, why does that paper make me so irritable? 2/n) I realized why: it's the structure of the argument in the paper. I also realized that many other papers that get under my skin share the same structure. I need to learn to be less of a jerk on Twitter, yes, but I also want to highlight why this structure bothers me so.
Jun 10, 2020 20 tweets 6 min read
1/ For #ShutDownSTEM today, our lab put aside research and crafted some concrete ideas for what we can do to help reduce anti-black/indig. racism, and more broadly, increase diversity in STEM. Our focus was local, specific acts to the lab. I wanted to share what we came up with. 2/ (Item 1) we decided that we could alter the way in which members of the lab are selected, in order to reduce the potential influence of unconscious biases and barriers that could potentially keep BIPOC from getting into the lab.
Apr 5, 2020 34 tweets 13 min read
1/ Need a distraction from the pandemic? It's #tweeprint time!!!

I'm very excited to share here with you new work from myself, @NeuroNaud, @guerguiev, Alexandre Payeur, and @hisspikeness:

biorxiv.org/content/10.110…

We think our results are quite exciting, so let's go! 2/ Here, we are concerned with the credit assignment problem. How can feedback from higher-order areas inform plasticity in lower-order areas in order to ensure efficient and effective learning?
Feb 27, 2020 11 tweets 3 min read
1/ I want to very briefly address this article that came out today by @matthewcobb in the Guardian:

theguardian.com/science/2020/f… 2/ I'm tempted to ignore it, but I think that would actually be a shame, because in many ways, it's a good article. Yet, it is also a confused article, and I worry about it confusing both scientists and the public more broadly. So, I'll just quickly address the confusion.
Nov 29, 2019 7 tweets 9 min read
@GaryMarcus @r_chavarriaga @KordingLab @DeepMindAI You don't actually keep up with the neuroscience literature, do you? That has been evident in these conversations... Here, lemme give you a few examples: @GaryMarcus @r_chavarriaga @KordingLab @DeepMindAI 1) ANNs optimised on relevant tasks match the representations in human (and primate) cortical areas better than other models developed to date:

journals.plos.org/ploscompbiol/a…

sciencedirect.com/science/articl…

journals.plos.org/ploscompbiol/a…
Oct 28, 2019 17 tweets 7 min read
1/ SciTwitter: I'm very excited to share our new Perspective article out in Nature Neuroscience today!

nature.com/articles/s4159… 2/ In this piece, we argue that neuroscience would benefit from adopting a framework that parallels the approach to designing intelligent systems used in deep learning.
Oct 12, 2019 15 tweets 7 min read
1/ New tweeprint from my lab! This one is work done by the amazing @guerguiev, and was inspired by the work of @benlansdell and @KordingLab (who was also a collaborator in this project).

arxiv.org/abs/1910.01689 2/ Our focus in this paper is the question of weight alignment. If you try to work out how to efficiently estimate cost fn gradients in biological networks, you will find that it would be great if feedforward pathways and feedback pathways had symmetric synaptic weights.
Sep 14, 2019 19 tweets 6 min read
1/ #tweeprint time everybody! It's about neural coding (and I mean that literally). We asked the following Q: if info is encoded in the neocortex with both rate and synchrony of spikes, do different subtypes of neurons display differential sensitivity to these two info streams? 2/ In the example image above, a binary signal is encoded with either a rate or synchrony code. The rate code uses high rate = 1, low rate = 0. In contrast, in the synchrony code the cells have a constant rate-of-fire, but high synch = 1, low synch = 0.
Feb 6, 2019 25 tweets 5 min read
Thread for the interested (or masochistic) neurotwitter verse...

This is a great little essay by @RomainBrette because it truly does get at the heart (pun intended) of our disagreement:

IMO, Brette, and others with his perspective, take an approach to the concept of metaphor that is misleading, and which serves as a constant tool for obfuscation in these arguments.
Sep 21, 2018 11 tweets 3 min read
1/ A thread on the brain as a computer (tl; dr: 🧠=💻): 2/ I think many neuroscientists are confused by the terms "algorithm" and "computer". Very few seem to understand how Turing machines (TMs) play a part in defining those words. In fairness, it can be hard to understand if your background is in physiology and psychology.