Mac Shine
Computational Neurobiologist from Sydney, Australia. https://shine-lab.org. Banner image from https://www.gregadunn.com.
- [Not loaded yet]
- 1/X Excited to present this preprint on multi-tasking, with @david-g-clark.bsky.social and Ashok Litwin-Kumar! Timely too, as “low-D manifold” has been trending again. (If you read thru the end, we escape Flatland and return to the glorious high-D world we deserve.) www.biorxiv.org/content/10.6...
-
View full thread36/X Pooled across behavioral syllables, we should recover high dimensionality, with higher dimension the more types of behavioral syllables that are observed. But we should observe lower growth of dimension per unit recording time, compared to pooling across periods of no behavior.
- 37/X Overall, really enjoyed working on this with @david-g-clark.bsky.social and Ashok, and I’d love to chat about any part of it—the details behind the theory, the experimental implications, etc. Thanks for reading!
- 18/X We think of this chaotic state as the “spontaneous” state of the network, where no task is activated. A task activates when the noisy, linearized dynamics lose stability, so that the associated latent variables grow exponentially (before nonlinearly self-stabilizing) to dominate the network.
- 19/X This can be achieved through modulating the strength of the associated connectivity component. Because any one connectivity component is low rank, this can be biologically implemented via gain modulation of an external loop, eg through thalamus (as in Logiaco et al Cell Reports 2021).
- 10/X In this example, the two connectivities we superposed would have produced stable limit cycle dynamics and bistable dynamics, respectively, if each were the sole network connectivity. When combined (previous post), the limit cycle "wins" because it's marginally stronger in this case.
- 11/X Thus we identified a kind of “interference” between task-related connectivity components, and it’s fairly one-sided: at most one task’s associated latent dynamical system does its thing, while every other's is suppressed to the origin and does nothing at all.
- 8/X In particular, we ask what happens when we linearly superpose different connectivity matrices, each of which is constrained to be low rank (R) and would generate, on its own, some nonlinear dynamical system on some low-dimensional manifold.
- 9/X The answer is surprisingly simple: the “strongest” (in a sense we make precise) latent dynamical system wins and controls the dynamics of the whole network, while the weaker dynamics are suppressed to the origin. This happens for any initial condition.
- [Not loaded yet]
- I'll sleep in bedroom 4.
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- Interesting approach that aligns mouse and human brain-wide data using transcriptomics and structural connectivity. I'm curious if anyone here has tried using it yet? #neuroskyence www.biorxiv.org/content/10.1... transbrain.readthedocs.io/en/latest/
- [Not loaded yet]
- 𝗔 𝗡𝗘𝗨𝗥𝗢𝗘𝗖𝗢𝗟𝗢𝗚𝗜𝗖𝗔𝗟 𝗣𝗘𝗥𝗦𝗣𝗘𝗖𝗧𝗜𝗩𝗘 𝗢𝗡 𝗧𝗛𝗘 𝗣𝗥𝗘𝗙𝗥𝗢𝗡𝗧𝗔𝗟 𝗖𝗢𝗥𝗧𝗘𝗫 By Mars and Passingham "Understanding anthropoid foraging challenges may thus contribute to our understanding of human cognition" Going to the top of the reading list! doi.org/10.1016/j.ne... #neuroskyence
- So I get that a Neuroscientist Couldn’t Understand a Microprocessor, and TBH I’m ok with that. But could a neuroscientist understand a deep RNN? Because that seems like a more pressing issue. *assuming you think the brain operates through the parallel activity of many connected input/output units
- [Not loaded yet]
- Turns out @macshine.bsky.social scooped our title…. www.biorxiv.org/content/10.1...
- [Not loaded yet]
- [This post could not be retrieved]
- *New preprint from the lab* – “Granularity of thalamic head direction cells” doi.org/10.1101/2025... 🧠📈 🧪 1/11
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- @macshine.bsky.social I am (re)reading your (and collaborators) recent work from: Müller EJ, Munn BR, Shine JM (2025): The brain that controls itself. Current Opinion in Behavioral Sciences 63:101499. backwards. Do you see a connection between ....
-
View full threadThe short answer is yes, but we were quite limited in writing that paper to keep the text short and the references mostly to the last few years. I'm certainly not an expert on circular causality/constraint closure, but I do very much like the implications of the ideas
- The point would be (I think) that there must be (a) closure in the loops from thalamus and/or brain stem nuclei to cortex and back which is/instantiates (?) the control parameter (which is external in non-biological or non-brain complex systems). Could that be correct? @kathrynnave.bsky.social
- ... constraints through low-dimensional units (thalamus and/or brain stem) on the one hand and closure of constraints: Kate Nave (2025): A drive to survive; ISBN 978-0-262-55132-8 and Alicia Juarrero (2023): Context changes everything: how constraints create coherence; ISBN 978-0-262-37477-4?
- [Not loaded yet]
- [Not loaded yet]
- What changes and what stays the same as you scale from single neurons up to local populations of neurons up to whole brains? Michael and Mac @macshine.bsky.social on the systems approach to study brains across scales. braininspired.co/podcast/220/
- Excited to share my latest work with @jonathanamichaels.bsky.social @diedrichsenjorn.bsky.social & @andpru.bsky.social! We asked: How does the motor cortex account for arm posture when generating movement? Paper 👉 www.biorxiv.org/content/10.1... 1/10
-
View full thread3️⃣ A condition-independent shift dimension: a trajectory reflecting trial progression, unfolding similarly across all movements, regardless of direction or posture. 7/10
- Modeling: Using modular RNNs, we asked what’s required for this compositional structure to emerge. It turns out this solution is common whenever the effector is complex enough to demand posture-dependent control policies. 8/10
- Key finding: High-density recordings from M1 & PMd revealed a compositional neural geometry with 3 components. 1️⃣ A posture subspace: fixed points for each target, visited whenever the arm rested at that location before or after a reach. 5/10
- 2️⃣ Rotational dynamics: transitions linking posture-specific fixed points.These rotations were systematic—similar rotations produced similar reach directions—and their projection continuously updated posture. 6/10
- Why this problem exists: Most insights come from center-out tasks, where all movements start from one spot. Here, reach direction and final posture are always correlated—making it impossible to separate movement dynamics from posture encoding. 3/10
- Our approach: We trained monkeys to reach between all pairs of 5 targets. Each target was a start point on some trials and an end point on others. This design let us dissociate posture representations from movement dynamics. 4/10
- Adaptive learning is coordinated across behaviour, time and neurobiology (from synapses and dendrites, to astrocytes and the systems level). If you’ve ever wondered how noradrenaline helps shape these multiscale learning processes, you might like this: www.cell.com/trends/cogni...
- BREAKING: Minneapolis Mayor Jacob Frey JUST CALLED FOR a statewide and federal ban on assault weapons. Who’s with him?
- [Not loaded yet]
- In The Transmitter's reading list, we highlight the most anticipated upcoming neuroscience books and other notable 2025 releases. Featuring titles by @okaysteve.bsky.social, @dr-brein.bsky.social, @nicolecrust.bsky.social, Xiao-Jing Wang and many more. By @franciscorr25.bsky.social bit.ly/45MvK41
- [Not loaded yet]
- Theoretical neuroscience has room to grow — a Comment article by Ann Kennedy @antihebbiann.bsky.social #neuroscience #neuroskyence www.nature.com/articles/s41...
- [Not loaded yet]
- 🎊Ain’t no party like a thesis submission party🎊 Honored to have submitted my PhD as a recipient of the Paulette Isabel Jones Career Development Award from The University of Sydney! Thank you x1000000 to MVP supervisor @bendfulcher.bsky.social & co-supervisor @macshine.bsky.social 😊
- [Not loaded yet]
- On the left is a rabbit. On the right is an elephant. But guess what: They’re the *same image*, rotated 90°! In @currentbiology.bsky.social, @chazfirestone.bsky.social & I show how these images—known as “visual anagrams”—can help solve a longstanding problem in cognitive science. bit.ly/45BVnCZ
