A road map to the beauty of the brain 2 – how behavioural architecture emerges from cortical developmental principles

 Che bella cosa na jurnata ’e sole! ;-) 

After having some delicious meringues during the coffee break, we started off as Rodney reminded us of some important numbers re. cortical circuits. 

Lay design of canonical circuit to emergence of behaviour – a cortical sheet for computation

Rodney first laid out the basic design of the canonical cortical circuit – there are about 10-100K neurons and 3.2 X 10^8 synapses per mm^3 in the Human brain. Cortex is extremely (axon rich and) synapse-dense (where excitatory synapses are formed onto the spine head of excitatory pyramidal neurons, whereas inhibitory synapses are formed on the dendritic shaft).

As illustrated in the cartoon, feedforward thalamic drive (only 1%) excites spiny stellate cells in L4 that provides excitation to L2/3 pyramidal neurons (via first intra-cortical L4-L2/3 synapse). There are few interconnections between spiny stellate neurons (15%) compared to in L2/3 (30-40% recurrent connectivity). L2/3 connects to L5 that sends out projections to non-thalamic projecting sources (striatum, spinal cord) as well as L6. L6 send cortico-thalamic projections back to thalamus and connects also to L4. 

Rodney argued that cortex doesn’t generate function out of boring repetitive structural repeats – 80% of these (feedforward and feedback) inputs are local and 20% comes from other cortical areas, very little of these are feedforward thalamic inputs.  Higher-order thalamic structures and their interactions with primary cortical areas are only beginning to be appreciated (see Denis Jabaudon’s lecture; should we be thinking about distinct TC interactions between between lower-order (LO) and higher-order (HO) thalamic areas with cortex?).

Emergence of areas:

In the final stage of this lecture, Rodney explained why/how the two-dimensional sheet is the major source of their intelligent behaviour, especially for primates. I highly recommend reading his primer in Current Biology as it explains the concept in detail, which is beyond the scope of this short update (Douglas and Martin, 2012; Behavioral architecture of the cortical sheet - Volume 22, Issue 24, 1033–1038). One important question is: How do cortical areas with defined functions coordinate global function – Rodney suggests probably via Basal ganglia or thalamus! Interestingly, basal ganglia connectivity patterns are distinctly different from the cortex – here the projection neurons are ‘inhibitory’ and the local neurons are ‘excitatory’, opposite to what we see in the cortex! 

The central sulcus and inferior-central sulcus divide the 2D-sheet in the coronal plane keeping major spatial relationships of cortical areas. The temporal cortex is separated from the more medial cortex by the prominent lateral sulcus. In three dimensions, parts of the temporal and orbital cortex and the limbic components fold back under the cortical plate so that the divided septum and hippocampus in fact form continuous ‘medial’ structures. The more frontal areas in Rodney’s 2D map corresponds to cingulate and orbitofrontal areas that are associated with logical operation, evaluative and subjective experiences (egocentric), prediction and planning. The posterior areas of the central sulcus contribute to sensory processing and processing of ‘space’. The temporal lobe encodes objective structures such as places, objects, and faces (see attached fig. 3 from the 2012 paper).

 


















What does these all mean at the end? This essentially argues for an evolution of cortical structures through repetition of basic units where processing is more localised as high bandwidth long-range connections are physically improbable (Rodney gives example of macaque cortex) – this also argues against a long-standing view that the cortex is composed of a hierarchy of interconnected areas providing hierarchical processing steps, with increasing complexity. 

Rodney argues that the connections between these local cortical areas and their subcortical partners determines and sculpts the overall computational architecture of the brain. This argument is critical as in the next few days, we will discuss how these individual computational units (brain areas, neurons, and individual synapses) are constructed and deconstructed and how function arises from these distinct forms. 

Off to lunch now.. 🍽