Artificial Consciousness/Neural Correlates/Neural Groups/Marr's Codon Model
Marr's Codon Model
[edit | edit source]In the 1970s David Marr was studying at Trinity College in Cambridge, and he developed a method of using Probabilistic Mathematical Methods to attempt to understand how some areas of the brain worked. At the time, the reigning theory was that neural connections were mathematical functions, so in his Theory of Cerebral Cortex, for instance, he assumed that the basket cells in Layer 4 of the Cerebral Cortex were Division Functions, and that the connections would work out mathematically to have some meaning beyond his basic Probabilistic Methods. He was sorely disappointed, and died in 1980 before he had published more than a few models of brain functions.
One of the functions he had published was the Theory of Cerebral Cortex article that claimed to have defined the function of the cerebral cortex. Of course since then we have found that his work was not sufficient to capture the whole function of the cerebral cortex. But it was a solid starting point in that it attempted to explain at least the bottom four layers of the cerebral cortex.
Of course Marr was a man before his time, and made a number of unfortunate assumptions, not the least of which being that he could unravel the snarl of connections in the human brain and make sense of it.
One aspect of his research was that he theorized a CODON, a self-classifying Content addressable Memory element as the basis for the Cerebral Cortex. Of course that is not the only function that the cerebral cortex has, so his theory was proven to be inaccurate, but, I feel it is an honest characterization of the Heterogeneous Networks within the first four layers of the cortex, as long as they are isolated from the rest of the brain. Unfortunately for Marr, the Cerebral Cortex is not isolated like his mathematical formulae were, so other functions overlay the basic memory function and complicate proof of its suitability.
One of the main exceptions I have heard of to Marr's work, was experiments done using neural networks that failed to capture certain neural frequencies associated with the real cortex.
Today Marr's article is used as an example of a level of sophistication to reach for, and to cow overenthusiastic graduate researchers. It's main claim to fame however is the fact that it is so dry as to put readers to sleep. My own experience was that the one article put me to sleep 18 times before I read it through once. Some scientists believe that their research is not properly valued if they make it too easy to understand, and so they use literary tricks to increase rather than reduce the complexity of the writing. One way that Marr managed to increase the difficulty of his work was to mystify the reader as to his mathematical functions. cateogry: Neural Networks