How about a thread on Richard Ivry’s comment. I see no problem. Each area has a unique pattern of inputs and outputs and so can perform a unique computation or transformation from inputs to outputs. But TASKS are not performed by areas but by networks. So what is the problem?
There is no problem but science is open to multiple views and not everyone sees the same.
Question: what is actually an area? That's a far from trivial question. In my view your camp has to offer a principled way to understand what an area is; we can't just assume they exist.
It can't be Brodmann right? (methods from over 100 years). Let's engage at the level of ideas, not just say "I see no problem" because that could be the default answer from all sides.
No?
It may not be Brodmann but it can be cytoarchitecure and myeloarchitecture as in the recentvmap from the Van Essen lab. Areas so defined also correspond to areas as defined by anatomical inputs and outputs and as Mars et al showed these also correspond closely with areas that are activated by tasks. I seriously don’t think that there is a problem here.
@dickretired
I collaborate with comparative neuroanatomists and they fight to death to define areas across species. Also I don't believe MRI is the basis to define much, such as in the Van Essen parcellation. fMRI activation even less useful.
So I see lots of challenges.
@PessoaBrain @dickretired Areas are perhaps useful when one cannot have actual synaptic circuits, but are ultimately not satisfactory. In the #Drosophila fly brain, early work mimicked the anatomical approach in mammals, defining areas and ascribing meaning and function to them; once the cellular #connectome was done, areas faded fast other than for referring to large chunks of brain. The #zebrafish brain connectome is also around the corner. Lots of change ahead.
@dickretired @PessoaBrain My approach to brain areas is a practical one: can one define a set of neurons whose synapses are exclusively within the area, a second set that projects into it (inputs) and a third set that projects out (outputs)? If so there’s grounds for an area, or module, because it can be abstracted to the transformation function that it implements. In practice extremely few fly brain regions are like this; mostly just first-order sensory neuropils.
@albertcardona @dickretired @PessoaBrain if "few regions are like that", why adopt this approach? Is it because it's the best we can do, or is it that defining brain areas is not useful (for your application) so it doesn't matter?
@christakou @dickretired @PessoaBrain To understand neural circuit function one needs synaptic-level circuits. The meso-circuits defined by inferred connectivity between areas on the basis of correlated activity or at best tractography, are, presumably, an intermediate level of resolution in mammals, particularly humans, because brain-wide synapse-level wiring diagrams are presently beyond reach.
#connectomics #neuroscience
I agree that we need synaptic level circuits to really understand how the brain works. But just as the resolution of FMRI is adequate for studying which areas are involved during a task, so the resolution of the connectome is adequate for studying how those areas are connected,
@albertcardona @PessoaBrain @dickretired Interesting insights! When adding high-density functional data to the mouse classical parcellations in cortex, brain regions also tend to blur away. If anatomical connectivity defines the roads that can be taken by neuronal activity, they are not always busy making it harder to delineate big uniform brain regions.
@pierre_le_merre @PessoaBrain @dickretired Thanks for your comment. I am having a hard time understanding the last sentence: did you mean “they are” rather than “they are not”?
@albertcardona @PessoaBrain @dickretired Sorry to be a bit unclear. What I meant is that anatomical connectivity will definitely constrain the flow of activity (as the possible roads to be taken) but these pathways can also be rather inactive, or dynamically active/inactive. So if one tries to define brain regions only by connectivity, a part of the information coming from the dynamical range of interactions might be missing.
@albertcardona @PessoaBrain
There is a difference between areas in drosophila and areas in the neocortex of a primate. But irrespective: there are two questions. 1 What does area A do in the network that area B does not. 2 What areas are involved in the network when the animal does task X.