>The first Hebbian cell-assembly will affect through the axon it's nearest
>neighbour, and the Cortex is such structured that they are 0.5 mm away from
>each other, and they are arranged in a roughly hexagonal pattern.
>There will of course be new ideas and new thoughts that will be initiated
>and die, but the "strongest" one will through what he call the Darwinian
>Machine develope and be presented as either as one part of a
>string-of-thought, or an action.
The competitive network mechanism requires, as far as I know, a convergence or
gating neuronal architecture to allow for a winner. This works well for small
to medium artificial network structures, but in functionally specialized cortex
where functions can involve the use (functional interconnection) of much of the
brain, it seems more untenable (a reference most recently mentioned on this
list to that effect would be something like: Mima et al., Transient
interhemispheric neuronal synchrony correlates with object recognition, J.
Neurosci. 21(11):3942-8, 2001).
Another issue seems to be with the cortical architecture being used. The
connectivity is a little more complex than the medium length, 0.5 mm axonal
range. The interconnectivity also displays short length (300 micron or smaller)
and long or wide arbor axons (greater than a millimeter), which form spanning
domains of overlapping influence with neighboring cortical modules (Check some
of the work by Patricia Goldman-Rakic).
Mike