A Perspective on Cortical Layering and Layer-Spanning Neuronal Elements


A new review of thinking about cortical layers (co-authored by Matthew Larkum) is out.

Read here.


Hi Sergey,
I read this paper and sent out a short email about it internally at Numenta. In case you or anyone else is interested I include my comments below. I read the paper quickly and my comments were written quickly, so please don’t view my comments any more than quick notes.

The paper is mostly to introduce a new fMRI technique and talk about how it might be useful for laminar distinctions in human neocortex. The technique, called “ultra high resolution fMRI” is basically fMRI with a stronger magnet. It allows voxels down to about 0.8mm3. When applied to humans it means they can measure relative energy consumption in upper, middle, and lower layers of the neocortex. That’s pretty cool, although it still has the other limitations of fMRI. They argue that most of the energy consumed by a neuron is at the synapse, so fMRI measures synaptic activity more than somatic activity.

The authors spend most of the paper talking about cellular layers. They point out that where the neuron body is (aka its layer) isn’t really important. What matters is where the input and output synapses are, and what they connect to. They also emphasize the different roles of apical and basal dendrites and the role of active dendrites. So a real theory of layers requires accounting for where the synapses are and what part of the neuron (apical vs. basal) the synapses are located. (Nothing new here for us.)

Therefore measuring spikes at the soma doesn’t say much, but with the new high res fMRI you can start to tease apart where synaptic activity is occurring. They mention that you might be able to detect activity (dendritic spikes) that doesn’t result in a somatic spike.

This technique might be useful for us.

Ps. They cite the 2009 George and Hawkins paper and not any of our newer papers. The 2009 George paper was in a list of papers that describe layers but don’t take account of different dendritic integration zones. Our newer papers of course are all about dendrites and our columns paper could have been cited. However, get this, this paper by Larkum was submitted on August 2, 2017, several months before our columns paper appeared. It has been in review for just shy of a year!