Read on Twitter
Ever wondered how different systems in the brain work together to coordinate attention, cognition and awareness? If so, this is the thread for you...
Many of our best models for how the mammalian brain works are focussed on the cerebral cortex. When you look at a human brain, it's really hard to miss on the outer surface of the brain, and there's also ++ evidence from clinical neurology that lesions lead to specific symptoms
But if you track nervous systems back across phylogeny (e.g., Paul Cisek's brilliant work: http://doi.org/10.3758/s13414-019-01760-1), you start to realise that there is waaaaaay more to the brain than a cerebral cortex.
In fact, there are lots of highly conserved circuits all throughout the brain, although the connections between systems often seem like a tangled mess? Is there a way that we can look at the brain that betrays its function?
In a new perspective paper published in Progress in Neurobiology ( http://macshine.github.io/publications/2020_progneuro.pdf), I argue that a thalamo-centric view of the brain offers a great vantage point for understanding systems-level dynamic activity in the rest of the brain
Like the proverbial rug that ties the room together, a number of crucial circuits from different macrosystems of the brain intersect in the thalamus, and in a way that likely shapes ongoing activity within the rest of the brain
A really crucial factor that helps to make sense of the connections is to use an old scheme developed by the late Ted Jones, which splits thalamic nuclei into different categories according to their projections to the cortex and striatum. http://doi.org/10.1016/S0166-2236(00)01922-6
The basic idea is that some thalamic inputs project directly to granular layers of the cortex, whereas others project more diffusely to supragranular layers. For those interested, this is a modern take by @eli_j_muller: https://macshine.github.io/publications/2020_nimg.pdf
A key insight for me was that, within the ventral tier of the thalamus, these two populations receive inputs from the cerebellum and basal ganglia, respectively. In other words, the logic of their organisation can be tracked from subcortex to the cortex...
One of my favourites implications is that it links the basal ganglia with a non-linearity inherent within layer V pyramidal neurons -- the cells can transition into burst mode when feedforward signals coincide with supragranular activity. @mattlark has done amazing work on this
I argue that this connection allows the basal ganglia to add variability constrained by the current context into the dynamically evolving cortical state. @brembs has cogently argued that this is really all you need to see behaviour consistent with our notions of 'free will'
Two brilliant post-docs in my lab ( @eli_j_muller and @DrBMunn) have done some really beautiful modelling work on this idea too. Will be published shortly, but there's a preprint here for those that are interested: https://www.biorxiv.org/content/10.1101/2020.06.09.141416v1
Another cool implication is that the connections can help to explain how cerebellar activity can shape and constrain ongoing patterns in the ctx -- I argue that this circuit explains the distinction b/w deliberate vs. automatic behaviours. hat-tip: @n_ramnani @DrBalsters
With these ideas in mind, I needed a way to frame my thinking about the different connections and relationships. So I turned to dynamical systems, and thought about how the different connectivity motifs might impact the evolution of the brain state across an attractor landscape
For those interested, here is a really cool tutorial by @ncasenmare: https://ncase.me/attractors/ . And @WiringTheBrain has also written a lovely, clear intro the ideas on his excellent blog: http://www.wiringthebrain.com/
As an aside, I share @WiringTheBrain's desire to have even more cool attractor landscape explorables to play with. In case @DirkBrockmann would like an exciting new challenge...
The basic idea is quite simple -- the constellation of activity in the brain can be represented as a ball rolling around a hilly landscape, following only the topography and gravity.
The cool part is that there is a direct link between these notions and the kinds of equations we use to model neural activity at the mean-field level. Check out @DrBreaky's excellent review paper if you're interested in more details: http://doi.org/10.1038/nn.4497
To link it back to the brain: the idea is that different subcortical inputs to the thalamus change the topography of the attractor landscape, and thus change the manner in which the brain state will evolve over time.
Recruiting the basal ganglia will release the matrix thalami from inhibition, and hence promote constrained (local) variability in the cortex, and hence open up new options for the cortical state to evolve.
Recruiting the cerebellum will have a different effect -- likely facilitating parallel processing and relatively automaticity through its ability to encode robust, sequential patterns of activity within its modular circuitry
There's a bunch more in the paper, including what I think are really exciting implications for thinking about higher-level brain functions, including attention, working memory, dual-tasking and even conscious awareness.
And finally, if you're interested, I'll be giving a brief talk about the ideas at the @BrainMind_Usyd symposium next Friday at 2:30pm. The PDF is here if you'd like to read it: https://macshine.github.io/publications/2020_progneuro.pdf. I'd love to hear any and all thoughts/ideas/feedback. Thanks for listening...
