As a log, this has not worked out so far. I have finished reading From Neuron To Brain by Stephen Kuffler, John Nicholls, and Robert Martin. A number of topics in the book would have made good blog entries. I even started one on Mauthner cells, which are large, complex, very interesting neurons found in fish, but did not complete it.
Except the early chapter on the visual cortex and the last chapter, "Genetic and Environmental Influences in the Mammalian Visual System," there is not much in the book that is directly helpful with issues of machine understanding.
And yet it does give an appreciation of the neural system, including its biochemistry. It would be a wonder of nature even if it were not capable, in human form, of smashing atoms and writing poetry. I had been working with simplistic ideas about how individual neurons work. That is actually fine for computer models. The fact that there are many modes of operation of neurons shows that evolution can make good use of both true redundancy and the fine tunings that come from slight variations.
I uncovered a small, common salamander today when pulling wood from the pile to bring up to the woodstove. There is no pond near the wood pile, so this creature had to wander some distance to get to this shelter. It is a good shelter too, complete with insects and other arthropods that make life easy for a salamander. I would tend to say that a salamander does not offer much in the way of understanding capabilities. But salamanders have been navigating the world and keeping alive, so even if we want to think of them as not capable of thought, still they have the necessary degree of intelligence to get them through their generations of life.
I am looking at the general issue of putting neurons together in patterns that could be said to be capable of at least the rudiments of understanding. There is nothing worth reporting on yet, so I'll probably go back to reporting on what Numenta is doing. The geniuses there are working on the problem full time, and are claiming some progress.
Wednesday, February 24, 2010
Subscribe to:
Posts (Atom)