Modern Gestalt Theory

x011 at Lehigh.EDU x011 at Lehigh.EDU
Tue Feb 14 10:26:49 EST 1995


In article <3hjvfe$jlo at portal.gmu.edu>, herwin at osf1.gmu.edu (HARRY R. ERWIN) wri
tes:
>I suspect that compact objects in the external world are mapped by a
>sequence of generalized Laplace transforms into a corresponding pattern of
>active primary (excitatory) neurons in the brain. This pattern or cloud of
>activation is recognized when it triggers a corresponding neural cellular
>array (NCA), linked by potentiated axon collaterals, that generates
>synchronized spike trains. To learn the pattern for an unknown object
>involves locus coeruleus reafference that increases the sensitivity of the
>activated primary neurons in the cloud to each other, followed by
>potentiation. Connections in the NCA that are not maintained by stimuli
>(sensory stimuli, sensory reafference, or dreaming) habituate
>exponentially and degrade over time and are forgotten. (They can also be
>erased by the median raphe nucleus using serotonin.) Hence we see only
>what we expect to see, but attending a novel pattern learns it quickly and
>allows us to see it. We can also initialize a cloud using sensory
>reafference.
>
>There's a lot of detail about how multiplicative normalization of the
>afferent stimuli maintains the signal to the primary neurons within a
>narrow band suitable for this mechanism to work. There's even evidence
>that this normalization process (which uses GABAergic neurons) is immature
>in small children, who must rely on their parents to provide suitable
>controls over the intensity of the stimuli that they are exposed to. The
>net result is that the activation levels of the primary neurons of the
>cortex appear to be maintained in a fairly narrow range suitable for the
>creation of these clouds of activation and NCAs.
>
>This theory can be labeled 'modern Gestalt' since it predicts external
>objects are represented in the brain as distributed objects. The memory
>pattern (NCA) corresponding to one of these distributed objects is also
>distributed, so constituting a pseudoholographic representation. The
>intersections of the NCAs for multiple distributed objects become
>interesting--they seem to correspond to abstractions. By the way, this
>model applies to the cognitive (or logical) system and not to the
>behavioral (or habitual) system (motor memory, mediated by acetycholine).
>I haven't much of a clue about the affective system.
>
>--
>Harry Erwin
>Internet: herwin at gmu.edu
>
Very interesting!  Physical support for your model may be abstracted from
Kenneth Showalter of West Virginia University (Science - February 10,
1995) work.  Kenneth's group report using chemical waves to find the
shortest distance in a chemical maze.  The wave propagations results in
a quasi-parallel approach with vector field information.  Almost any
point could be choosen at random and you would known automatically how
to get back to the starting point in the maze.
Ron Blue



More information about the Neur-sci mailing list