Sometimes it's good to read stuff that tells you you're wasting your time, to keep things in perspective and to keep you humble.

In December I blogged about the 2007 Douglas & Martin paper "Mapping the Matrix: The Ways of the Neocortex" in which the authors argued that the neocortical minicolumn is a poor functionally descriptive entity.

Well I just finished reading a similar paper that precedes Douglas & Martin by a couple of years: "The cortical column: a structure without a function" by Horton & Adams.

Their arguments are very similar to those of Douglas and Martin, though more detailed. They point out the imprecise usage of the terms "column", "minicolumn", and "macrocolumn". They're okay with the term "hypercolumn", coined by Hubel and Wiesel, which describes a region of cortex containing a full set of values for a given receptive field parameter. For example, there are orientation columns, small regions of your visual cortex that are sensitive to the orientation of lines. A given hypercolumn for orientation columns would contain those areas sensitive to all orientations from 0 to 180 degrees.

But the authors don't think a hypercolumn is neatly divisible into discrete columns or minicolumns because there seem to be overlapping areas of activation and no clear borders.

As I noted in my comments on the Douglas and Martin paper, virtually every biological concept is fuzzy in nature. "Gene", "species", and even "life" itself are all concepts that do not have clearly delineated boundaries. This doesn't keep them from being extremely useful descriptive entities.

But the authors think that researchers who buy into the column or minicolumn as a descriptive functional unit for the neocortex are clinging to a pleasant, outdated fantasy. They go so far as to compare cortical columns to spandrels, invoking Stephen J. Gould's metaphor for structures without functions that arise non-adaptively (this is basically comparing cortical columns to the belly button, the appendix, or nipples on men). They say:

It now seems doubtful that any single, transcendent principle endows the cerebral cortex with a modular structure. Each individual area is constructed differently, and each will need to be taken apart cell by cell, layer by layer, circuit by circuit and projection by projection to describe fully the architecture of the cortex.

That philosophical approach to understand the neocortex might render it almost impossible to understand. The human neocortex has roughly 20 billion neurons, and trillions of connections. It's going to take quite a long time to understand it if we creep through it a neuron at a time, without extracting out some common principles between cortical areas based on structural and functional regularities.

Another way to look at it is through comparative anatomy. The average chimpanzee brain is about 300-500g, while the human's is around 1300g. The largest differences are in the expansion of the neocortex. Yet the difference between our genomes is currently thought to be about 4%. How could evolution result in such a massive expansion of the cortical sheet in such a relatively small amount of time (~5-7 million years ago) unless the neocortex were based off a repeating template, such as the neocortical minicolumn?

The authors spend a fair amount of time talking about ontogenetic columns, vertical stacks of cells that radiate outward in the developing brain to form the neocortex. Rakic, who has done pioneering work in understanding how the neocortex develops, argues that changes in human DNA that increase the number of these precursor structures could explain the rapid expansion of the neocortex.

In other words, evolution found a highly modular structure that allowed for the scaling of the neocortex by massively increasing the number of modules. If we don't work from the hypothesis that there is this fundamental regularity, both structurally and functionally, from which to theorize about the neocortex, how do we explain the rapid expansion of the neocortex? How do we explain how a handful of genes encodes for billions of highly-differentiated and individualistic neurons and trillions of highly-specified connections?

At least Douglas and Martin tried to offer up some alternative organizing principles in their paper. Horton and Adams don't even bother with that. Suffice it to say I think they're wrong, but as I mentioned at the outset, it's always good to read critics of your assumptions, at the very least to keep things in perspective.

Horton, J.C., Adams, D.L. (2005). The cortical column: a structure without a function. Philosophical Transactions of the Royal Society B: Biological Sciences, 360(1456), 837-862. DOI: 10.1098/rstb.2005.1623