Newsgroups: sci.physics,comp.ai,sci.psychology,sci.bio,sci.skeptic,alt.consciousness
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!newsfeed.internetmci.com!panix!news.columbia.edu!news!ml.media.mit.edu!minsky
From: minsky@ml.media.mit.edu (Marvin Minsky)
Subject: Color perception (was quantum nonsense or something)
Message-ID: <1996Feb26.045129.3535@media.mit.edu>
Sender: news@media.mit.edu (USENET News System)
Organization: MIT Media Lab
References: <825037000snx@sv.span.com> <825289029snx@sv.span.com> <Pine.A32.3.91.960225183759.92928A-100000@black.weeg.uiowa.edu>
Date: Mon, 26 Feb 1996 04:51:29 GMT
Lines: 43
Xref: glinda.oz.cs.cmu.edu sci.physics:173403 comp.ai:37347 sci.skeptic:155071

> Brian J Flanagan <bflanagn@blue.weeg.uiowa.edu> writes:
>On Sun, 25 Feb 1996, Shez wrote:
>
>> One consequence of the overlap is that some individual wavelengths of
>> light at the blue/green end of the spectrum do not generate a colour
>> sensation if perceived in isolation, presumably because they excite all
>> cones to a similar degree. In such cases a second contrasting colour is
>> needed in the visual field to enable us to set our "white balance" and
>> see the original stimulus as a colour.
>
>And so, how is it that we "see the original stimulus as a colour"? 
>
>I see you are not completely stupid, but you have failed to grasp the 
>essential problem altogether. 

Indeed, the colour we "see" is related in a complex way to the
distribution of nearby colors.  There are several different technical
formulations for the rules of this, but they don't all agree on
different situations, so we have desctiprions emphasizing
distance-weight4ed distributions, triple-point balance-normalizing,
the classic tri-color vector system, and so forth.  In any case, the
"colors we see" depend on the image in complex ways.

As for why we see them as colors, this I suggest is because some of
your image describing brain centers notice that certain of their
inputs are correlated, as it happens (although the observer knows
nothing about this) the color information comes from the parvocellular
blob system of the lateral geniculate, which connects to the visual
cortical area 4a.  What you're doing when you identify a patch as
having a certain color is, presumably, classifying that input as in
the correlation class of activities in those brain areas -- plus the
various complex other associations such as green with edibiulity, blue
with unobstructed horizontal vision, red with dangers (such as
poisonous foods (mostly), fire, the mouths of predators, and what-not.

The important point is that the parts of your brain that make yyour
color judgements do not simply arise from seeing "the original
stimulus" at all, but from complex functions that are computed from
both local and non-local other inputs, from a variety of loci in the
brain.  Most of us share more or less the same structures, so we tend
to agree socially on those judgments.  In everyday life, your sort of
naive pop-psychology is quite adequate--but for scientists and
philosophers it is you who might seem the simplistic one.
