Sunday, June 8, 2008

Synapses Have Evolved a Lot

The manifest of proteins associated with neural synapses has continued to evolve and become more complex, according to this study.
"Our simple view that 'more nerves' is sufficient to explain 'more brain power' is simply not supported by our study," explained Professor Seth Grant, Head of the Genes to Cognition Programme at the Wellcome Trust Sanger Institute and leader of the project. "Although many studies have looked at the number of neurons, none has looked at the molecular composition of neuron connections. We found dramatic differences in the numbers of proteins in the neuron connections between different species".

"We studied around 600 proteins that are found in mammalian synapses and were surprised to find that only 50 percent of these are also found in invertebrate synapses, and about 25 percent are in single-cell animals, which obviously don't have a brain."
I'm trying to figue out this maps into TheRadicalModerate's Overly Broad Theory of Human Cognition, which contains (but is not limited to) the following tenets:
  • Cortex uses the same neural algorithms to learn and store patterns, pretty much everywhere.

  • The cortex processes patterns derived from somewhere between 50 and 100 sensory "inputs".

    Vaguely pertinent aside: Yes, we have a lot more than five senses. Sensory organs usually do more than one thing, and it's entirely reasonable to assume that each separate thing they do results in different neural inputs. On top of that, you get all sorts of funny chemical signals from the body that also act as sensory information.

    If you take all of that into account, an incomplete list of human senses might be: Visual edge detection, visual motion, color, binocular discrimination, pure tones, auditory transients, overt smells (many), pheromone "smells" (many undiscovered), sweet, sour, bitter, salty, umami, balance (from semicircular canals), balance (from toe position and pressure), pressure, temperature, hair follicle bending, skin friction, various kinds of pain, proprioception (i.e., the ability to sense body position), a variety of visceral senses (hunger, satiety, gastic distress, bowel sense, circulatory sense, and so on), and I'm not even going to attempt to go through all the hormonal signals, many of which act as both inputs and outputs. But I've managed to list 26 separate "senses" without trying very hard.

  • The neural processing for these inputs is widely varied and hard to deduce, but it's safe to say that all of these input systems have evolved to process signals so as to produce the most diverse set of patterns possible to the cortex.

  • Cortical regions receiving projections from input sources automatically learn to extract information from the input patterns, with no inherited magic required.

  • But projections from one cortical region to another are inherited, and determine the ultimate success of the mind that emerges from a particular brain.

  • I suspect that there is a much wider variety of neural "outputs" then one might suspect, as well, but I have put much thought into them yet.

Note that, if cortex is the same everywhere and that there's a nice, simple, genetic mechanism for producing a fairly varied set of projections, you've got a recipe for very rapid evolution of cognition. You can pretty much grow more and more cortex, subject to the ability of the body to feed and cool it, but that cortex doesn't do much until it's wired to some other chunk of cortex, or to an input or output region, in such a way that it doesn't drive the organism insane. But if the genetic mechanism to produce projections is fairly variable, you can still grow useful, non-insane cortex pretty quickly, at least in evolutionary terms. (See this post for random thoughts on what this might tell us about autism.)

It's possible that the variety of proteins expressed in synapses has much more to do with various forms of signal processing and motor control than it does with cortex, but that's no better than a guess.

Why is this finding important? If you buy this model and then want to produce computing machinery that might do similar things, you'd be tempted to use a fixed, fairly simple model for the behavior of cortical columns, then find nifty ways to wire up oddly processed signals to artifically designed "cortical regions", which in turn project to other regions, to see whether you can get something like intelligent behavior out of the system. If cortical processing actually turns out to be extremely complex and variable, that model starts to fall apart.

No comments: