simple wavelet-opponent process

Ron Blue rcb5 at msn.com
Fri Sep 21 12:38:39 EST 2001


Possible demonstration of an observable simple wavelet-opponent process is
suggested by examining the research presented in the following article:

Dehaene-Lambertz, G. & Dehaene, S. (1994, July 28). Speed and cerebral
correlates of syllable discrimination in infants. Nature v370, p292(4).

To simply illustrate in format using a typing system let (L) = a large
wavelet, (M) = a small wavelet, (S) = a small wavelet, M = a wavelet packet
of "ba", W = an inverse or opponent wavelet packet, -> = the direction of
the series, and MV = a novel wavelet packet "ga".

Using Ghislaine Dehaene-Lambertz and Stanislas Dehaene (1994) research on
speech sounds and changes in baby's brains which suggest a mirror image of
the stimulus wavelet, we would observe the following series in an average
evoked related potential relative to a stimulus series of  "ba", "ba", "ba",
"ga", "nothing"

->  (L)M  ->  (M)M  ->  (S)M -> (L)MV ->  (M)W

Notice that (M)M is the average of the series of the stimulus "ba".
Notice that (M)W is an opponent wavelet mirroring the stimulus "ba" when
"ba" was removed.

The idea is that an opponent wavelet filter for "ba" is being formed which
we see as habituation in the series.  The wavelet filter does not block the
stimulus "ga".  The lack of the stimulus "ba" causes a reciprocal inhibition
that is a mirror image imposed on a non-stimulus.

Ron Blue
http://turn.to/ai






---




More information about the Neur-sci mailing list