Bob Mottram wrote:
2008/5/30 Mike Tintner <[EMAIL PROTECTED]>:
Computer model reveals how brain represents meaning
[Modern human brain. Image source: Univ. of Wisconsin-Madison Brain
Collection.] Modern human brain.
Image source: Univ. of Wisconsin-Madison Brain Collection.
Being able to predict patterns of neural activity is definitely a step
forward, and it would also be interesting to know how similar or
different the representations are from one person to another, and
whether representations are different for people speaking different
languages.
Does anyone know what those 25 verbs are?
One other response to this study: quite often the results from this
kind of fMRI research are not as dramatic as they seem on the surface.
I have not read their paper (not freely available as far as I can see)
but I would not be surprised if they got their statistical result
because as little as one neuron in a million was consistent with the
pattern. While that is better than nothing, it could easily be a dead
end, in the sense that if you know that there is a marginal fluctuation
in favor of the pattern, it tells you almost nothing about how the words
are actually represented.
That's a little abstract, so here is a concrete illustration of what I
mean. Suppose that the words were represented by *moving* patterns of
activation (like solitons) navigating the brain in a quasi-random way.
Then, if you discovered that a word like "dumpling" caused localized
activity in Area 23163 of the brain, this might only mean that when the
"dumpling" pattern was wandering around interacting with the other
patterns, it passed through Area 23163 *slightly* more often than it
passed through other areas. This very slight localization would be
enough to cause the neuroscientists to declare that the "location" of
the dumpling word had been discovered, and I could believe that Mitchell
and Just might be able to predict this location on the basis of their
textual proximity analysis.
But under this interpretation, the conclusion would be meaningless.
This would not be the "location" of that word at all. Even the
slightest, most subtle bias would cause the researchers to publish a
brain scan image showing a colored patch for the "dumpling" area, but if
that is just a place that the wandering "dumpling" pattern visits more
often, what value would there be in the conclusion?
Richard Loosemore
-------------------------------------------
agi
Archives: http://www.listbox.com/member/archive/303/=now
RSS Feed: http://www.listbox.com/member/archive/rss/303/
Modify Your Subscription:
http://www.listbox.com/member/?member_id=8660244&id_secret=103754539-40ed26
Powered by Listbox: http://www.listbox.com
