Speech perception and PCT

[From Rick Marken (970306.1330 PST)]

Tracy Harms (970305) says:

I'm curious to see more examples of how experienced PCT psych-sorts
critique specific non-PCT theories.

I'm quite familiar with the theories of speech perception described
by Dennis Hughes in his "Position statement..." I don't really know how to
critique these theories from a PCT perspective. To the extent that they
actually are theories (working models that account for observations) then
they don't really conflict with PCT. From a PCT perspective they are
theories of the perceptual function, the mechanism that turns acoustical
signals into speech perceptions.

I think PCT is particularly compatible with a version of what Hughes calls
the "General Auditory Mechanisms" theory which says that speech perception
results from a mapping of acoustic signal into perceived sound. I think
speech perception is actually many perceptions happening simultaneously at
different levels of a perceptual hierarchy. We can hear speech as varying
intensities, sensations (timbres), configurations (phonemes, words),
transitions (inflection) etc. All of these perceptions are happening (and
some are being controlled) at the same time.

I believe that perceptual functions (neural networks) continuously convert
sensed aspects of the auditory stream into the perceptions of speech. I
also believe that these perceptual functions are hierarchically related;
higher order speech perceptions (like the perception of words, sentences,
etc) are derived (via higher order perceptual functions) from lower order
perceptions (timbres, changing timbres, etc).

The problem with all these theories of speech perception is not theoretical;
it's _factual_. To the extent that these theories are actually based on (and
account for) _behavioral_ data, it is data that was almost certainly
collected as though behavioral responses (judgements, ratings,
identifications) are an open loop function for acoustical input. If you have
been following the discussion about using people as "talking rulers" (in
speech perception they are used as "talking tape recorders") you can,
perhaps, see why this kind of behavioral data cannot be used to test these
models of speech perceeption.

What we need are new studies, along the lines of a masters thesis on vowel
perception done by a PCTer named Issac Kurtzer, that put the subject in
control of aspects of the acoustic signal that are presumed to be the basis
of some particular speech perception, and then test (by applying
disturbances) to determine whether or not this variable can be controlled
(and, hence, perceived).

Best

Rick

i.kurtzer (970308.0400)

What we need are new studies, along the lines of a masters thesis on vowel
perception done by a PCTer named Issac Kurtzer, that put the subject in
control of aspects of the acoustic signal that are presumed to be the
basis
of some particular speech perception, and then test (by applying
disturbances) to determine whether or not this variable can be controlled
(and, hence, perceived).

thankyou for the plug! if any else is interested come to the CSG
conference in July for a more formal elucidation of what speech
science theories have been explaining and how their methods have
systematically created artifacts; and you guessed it, it is these
artifacts that provide the foundation for which the theories are tested,
even the fancy
neural-like-like ones.

a side note on the control/perception-hence comment:
rick is absolutely right in standing firm to a positive and testable
criteria of perceptual status; not only does it dismiss outright the
status of percepts outside the domain of corroboration as outside the
domain of science, but refers to the best procedure by which to confirm or
disconfirm to what our percepts correspond to, the Test.

i.