Re: R&M's "memory illusions" and functional verbal response classes

From: Sergio Navega (snavega_at_intelliwise.com)
Date: 07/28/04 

Date: Wed, 28 Jul 2004 10:55:58 -0300

"Eray Ozkural exa" <erayo@bilkent.edu.tr> escreveu na mensagem
news:fa69ae35.0407270155.59af2dd6@posting.google.com...
> "Sergio Navega" <snavega@intelliwise.com> wrote in message
> >
> > In a single paragraph you mentioned the idea of probabilistic
> > adjustments of activations, memory decay and interference
> > effects. Are these the repugnant conceptions of a mentalistic
> > scientist? Or are these abstract conceptions that may have
> > connections to the way neurons work? Just to illustrate,
> > using concepts very similar to the ones above researcher Pierre
> > Perruchet developed a model (called PARSER) of the segmentation
> > of words by 8 month old infants. Using this model as central
> > theme, he has published on the prestigious BBS (Behavioral
> > and Brain Sciences, one of the most important journals on
> > the matter, if you don't know about it you should!). A whole
> > theory of consciousness was founded because of this single
> > (and testable) starting point. I'm saying all this not to
> > show that I'm at Perruchet's side (I'm not), but
> > because I *disagree* of some characteristics of his
> > proposal. That's the real issue in science: if anyone
> > proposes something, it is worthless to say "you're using
> > the wrong methodologies, the wrong models, the wrong
> > assumptions, the wrong philosophy". One has to show how
> > and why that theory fails compared with another one.
> > In cognitive science (at least, the "healthy" segment of
> > it), one can do that: just make sure the theory is or
> > is not falsified by behavioral data.
>
> Yes, Glen has a remarkable talent.
>
> I found the model you mentioned interesting. What is this theory of
consciousness?
>

Perruchet and Vinter proposed in 1998 a computer model (Parser) that
was able to duplicate results from a 1997 paper by Saffran et al.
Basically, Saffran's paper was the result of an experiment with
adults listening to a sequence of syllables. Unknown to the subjects,
these syllables composed nonsense words (such as DUTABA, PATUBI, etc).
The words were assembled by combining syllables such as "DU",
"PA", "BA", "PU", etc. The statistical structure of the words
thus formed were interestingly difficult to be consciously spelled,
but nevertheless the subjects were able, in a subsequent tests,
to present much better than chance identification of the correct
words. This lead Saffran and others to conjecture that the
subjects abstracted some characteristics of the statistical
distribution. Perruchet and Vinter proposed a computer model that
was able to show the same behavior. Their model used notions such
as decay of memory traces, interference effects and clusters of
sensory fragments in a perceptual mechanism (something they
called "perceptual shaper"). From this theory, both Perruchet
and Vinter published in BBS a more bold theory (the SOC, Self-
Organizing Consciousness) which proposes a model somewhat different
than most of the current ideas about consciousness.

http://bbsonline.cup.cam.ac.uk/Preprints/Perruchet/

In their model there's no need of a "thick unconscious level", with
things in our awareness being directly the result of immediate
phenomenal experiences.

I'm not into the subject of discussing consciousness, but I find
Perruchet & Vinter's idea very interesting (mainly in relation
to Parser), along with several of the concepts they ellaborate.
I don't know of any other model of Saffran's 1997 results, except
for my own (I wrote a program that is capable of duplicating these
results, although using different notions than Perruchet's).

Sergio Navega.