Re: The Auditory Continuity Illusion/Temporal Induction

[Al Bregman <bregman@xxxxxxxxxxxxxxxxxxxx>]

Dear Dick,

Your reply to Fatima is much appreciated.  The question of "where" a
process takes place in the brain is very complex.  The auditory signal
starts at the object that makes the sound.  So a Gibsonian might say that
the external object participates in the process.  Then the sensory system
registers the sound.  If the stimulus features that you described as being
needed for induction to occur have to be registered by the peripheral
auditory system, then at least to that extent, the latter system is also
involved.  Then information is circulated to a number of brain areas.  It
is likely that without the participation of any of these sites, no
induction would be carried out.

A central question remains:  Is there an area of the CNS that is involved
in auditory induction but not in any other auditory process?  That is what
is required before you could argue that this specific area is the module
that takes care of auditory induction.  Using this strict criterion, we may
not have enough scientific information to say where *any* auditory
phenomenon is carried out, never mind auditory induction.  One might think
that we can obtain this information from studies involving brain-imaging .
However, the process of subtraction of a control condition from an
experimental condition in such studies may give misleading answers.

Let me illustrate this point with a fanciful analogy.  Suppose some Klingon
scientists are interested in exactly where an analysis of variance (ANOVA)
is carried out in a human-built computer (Klingons have enormous computers
that compute different functions in different places.)  They don't know
exactly how a human-built computer works, but they do know how to measure
things.   In particular, they are able to measure the activity level of
every part of the computer as it carries out a number of statistical
processes.   By the clever use of the subtraction technique, they find that
a certain area of the computer's memory (region A) is more active during an
ANOVA than is any other part of the computer. [Let me reveal that what they
have discovered is the high level program that specifies the ANOVA
computation; it is, however, written using other high-order processes, such
as "sum of squares", as primitives (these primitives being calculated by
helper programs located elsewhere in the computer).]

Our Klingon scientists conclude that the ANOVA is carried out in this
region A.  Are they right?  Well, yes and no.  The process is indeed
coordinated there, but is actually carried out all over the computer.
There are helper programs, such as sum of squares, that are located in
other memory regions.  There are also the pieces of hardware for shifting
numbers from one location to another,  testing the equality of two numbers,
adding, multiplying, inputting and outputting numbers, busses for
connecting these pieces of hardware, etc., which are located all over the
computer.  So it is more correct to say that a large part of the computer
calculates the ANOVA, not just region A.

It seems to me that this is a direct analogy to the use of the subtraction
method in fMRI studies.  Even assuming that its human practitioners are as
clever as the Klingon scientists in choosing all the necessary control
conditions, the subtraction method leads them to say that the process is
*carried out* in the region that shines forth after all subtractions.  The
method effectively discards the contributions of the rest of the brain to
the process.  Studies involving ablation or brain damage have the same
problem.  Finding out that the destruction of a particular region destroys
a certain capacity without affecting any others (a rare finding) still does
not tell us that the capacity is largely located in that region, only that
the region is essential for the process.

The conclusion is that imaging and destruction studies cannot in themselves
tell us where a process is taking place in the brain.  Since those are the
two most powerful techniques that we now have available, this conclusion is
discouraging.  What can we do about it?  Well, we can be aware that the
brain is an organ whose parts are highly interactive.  Maybe it is not
particularly productive to ask "where" a process is going on.  A better
question might be "how".  How can processes distributed all over the brain
collaborate in carrying out a process such as auditory induction.

Best holiday wishes to all,

Al

----------------------------------------------------
Albert S. Bregman,
Emeritus Professor
Psychology Dept., McGill University
1205 Docteur Penfield Avenue
Montreal, Quebec
Canada  H3A 1B1

Voice & Fax: +1 (514) 484-2592
----------------------------------------------------

----- Original Message ----- 
From: "Richard M. Warren" <rmwarren@xxxxxxx>
To: <AUDITORY@xxxxxxxxxxxxxxx>
Sent: Monday, December 05, 2005 12:37 PM
Subject: The Auditory Continuity Illusion/Temporal Induction


> Dear List,
>
> Questions concerning illusory auditory continuity (aka temporal
> induction) were raised by Fatima Husain.  Her first posting on October
> 25 dealt with a question concerning instructions employed for an
> experiment involving the illusory continuity of a pitch glide
> interrupted by noise, and this has already been answered by others.  I
> wish to respond to her posting of October 28 concerning the basis for
> illusory continuity.
>
> During the exchanges on the list, Al Bregman suggested that I might
> have something to contribute on the topic, and wrote “Dick, are you out
> there?”  Sorry for the tardy response, but I was tied up with several
> activities, and am just catching up with the postings.  I take issue
> with Fatima’s statement that “ I think that the continuity illusion is
> ‘cortical,’ as opposed to say being auditory (anywhere in the central
> auditory processing system) or peripheral.”  There have been a number
> of experiments dealing with the stimuli and peripheral auditory
> stimulation response that must be considered for any model of illusory
> continuity.  It appears that a “temporal induction” can induce a
> perceptual synthesis of signal fragments that have been masked or
> replaced by brief (up to a few hundred ms) bursts of a louder sound.
> This louder inducer must be a potential masker of the signal fragment,
> and studies have indicated that a portion of its auditory
> representation is reallocated for the generation of contextually
> appropriate segments of a fainter signal.
>
> Restoration of a portion of tonal frequency glides as employed by
> Fatima belongs to one of the three types of temporal induction; each of
> these has been studied in some detail:  Type 1 “homophonic induction”
> consists of the illusory continuity of the fainter of two alternating
> levels of the same sound (e.g., alternating levels of a 1,000 Hz tone
> that differ by at least 2 or 3 dB, alternating levels of white noise
> that differ by at least 0.5 dB); Type 2 “heterophonic induction”
> consists of the illusory continuity of the fainter of two alternating
> sounds that are qualitatively different (e.g., continuity of a 1,000 Hz
> tone alternating with a louder 1/3-octave band of noise centered on
> 1,000 Hz), alternating levels of a 1,000 Hz tone with a louder 970 Hz
> tone); Type 3 “contextual catenation” also consists of the restoration
> of a contextually appropriate fragment of a signal as do the other two
> types, but unlike the others, the signal is dynamic, and the restored
> fragment differs from the preceding and following portions of the
> signal (e.g., the restored fragment of a tone glide as studied by
> Fatima, the phonemic restoration of portions of speech, or the missing
> notes of familiar melodies played on a piano).
>
> In connection with the role played by the cortex in restoration, it
> should be kept in mind that different cortical loci and organizational
> rules are employed for the restoration of fragments of speech, familiar
> melodies, and pitch glides.  However, all three types of illusory
> continuity follow the same simple rule that the neural representation
> of the extraneous inducing sound include those units that would have
> responded to the absent segment.  There are also specific
> characteristics associated with the restoration of segments of
> particular sounds (e.g., the homophonic induction of tones, the
> phonemic restoration of speech) and these have provided new information
> about the processing of these signals when they are uninterrupted.
>
> More information on this topic is available in Chapter 6 “Perceptual
> Restoration of Missing Sounds,” in my book, Auditory Perception:  A New
> Analysis and Synthesis (Cambridge University Press, 1999), which is
> devoted to a summary and discussion of the literature on illusory
> continuity, including contributions from our lab.  I believe the book
> is now out-of-print, but I have just signed a contract for a new
> edition with the same publisher.  It should appear next year.
>
> Regards,
>
> Dick
>
> ---------------------------------------------
> Richard M. Warren
> University of Wisconsin-Milwaukee
> Department of Psychology
> PO Box 413
> Milwaukee, WI  53201-0413
>
> Phone:  (414) 229-5328
> Fax:      (414) 229-5219
> Email:  rmwarren@xxxxxxx
>