[bob]

On Jul 31, 6:29 pm, Mike wrote:
On Jul 29, 6:40 pm, bob wrote:
Please stop projecting your own misconceptions on me. All I've done is
to demand that YOU define the terms you use. After all, you can't
expect me to have a serious debate with someone who uses verbal
ambiguity to suit his "argument."

I started a discussion about the epistemology of audio testing. Maybe
you aren't interested in this kind of discussion.

What, you aren't required to define your terms in epistemology?

That's fine if we are
talking about, say, organic chemistry, but when we are talking about
the perception of music---sorry, nothing is precisely defined.

Exactly--you can't even define your own terms. So you're saying, "I
can't tell you what 'perception of music' means, but you have to prove
that DBTs don't inhibit it in any way." Can't you see how ridiculous
that is?

It is ridiculous. It's also a strawman.

Have you not demanded that I prove DBTs don't inhibit "the perception
of music"?

Have you not admitted that you can't define "the perception of music"?

So how is this a straw man? Seems to be exactly what you've said.

Anyone can explore the
perception of music. It especially helps to sing or play an
instrument. Many of us find a common language and feel that we are
communicating about the experience of music on a common level. Many of
us find that music has a meaning only in context, and that many DBT's
will distort the context. Therefore, I'm interested in finding some
way of constructing a DBT that doesn't distort the context.

Isn't that rather easy? Just listen to the music for as long as you
like. Listen to a whole symphony through amp A. Then listen again
through amp B. Decide which one you prefer. Repeat 20 times,
randomizing the order of A and B so you don't know which is which. See
if your preference is consistent. It'll take you a while, but you'll
get to enjoy the music in the meantime.

By the way, it won't work. You'll get negative results even when you
should get positive ones. But that's empiricism, not epistemology, so
maybe you don't care about it.

snip

In the gorilla experiment, subjects were given a restricted focus. A
DBT likewise encourages a restricted focus. There are many reasons. A
test based on snippets will restrict the focus to one short moment of
sound. A test based on quick switching will restrict the focus to the
moment of switching. Even a test based on longer listening segments
requires that the subject try to remember sound in a way that foreign
to listening for enjoyment.

Here, finally, is the nub of the issue. Your whole argument is just
the same old discredited "quick-switching" canard, dressed up in four-
bit pseudo-philosphical jargon. Guess we'll have to discredit it once
again.

Researchers do not use quick-switching tests because they are dumber
than you. Researchers use quick-switching because they are smarter
than you. They know, from experience or from prior research, that when
subjects can't switch instantaneously between A, B, and X, their
ability to hear small differences essentially disappears. Listening
long enough to appreciate the "musical context" confers no advantage
in detecing small differences. And giving up quick switching confers a
huge disadvantage.

You're trying to raise a question that's already been answered.

bob

[Steven Sullivan]

Mike wrote:
On Jul 29, 11:54 am, Steven Sullivan wrote:
Michael Mossey wrote:
On Jul 26, 6:59 pm, bob wrote:
On Jul 26, 7:43 pm, Mike wrote:

I'm not talking about masking which occurs at a fairly low level in
the perceptual mechanism, well before consciousness. Rather, the
problem is that you have many places to put your conscious attention,
choosing among those aspects of the signal which are available to
consciousness.

Indeed. So your argument, I take it, is that DBT results aren't
generalizable to other listening situations because, unlike in those
situations, in a DBT "you have many places to put your conscious
attention, choosing among those aspects of the signal which are
available to consciousness."

The trouble with this argument is that it contains a flawed premise.
I.e., that in other listening situations, you do NOT "have many places
to put your conscious attention, choosing among those aspects of the
signal which are available to consciousness."

In fact, you've already implicitly admitted that this is a false
premise, with your (narrowly correct, in this case) claim that we
experience music differently each time we hear it. The very reason we
experience it differently is that we concentrate on or note different
aspects of the music.

So, if we "concentrate on or note different aspects of the music" in
DBTs, and we "concentrate on or note different aspects of the music"
when we're aren't in DBTs, then how is a DBT different from other
forms of listening? Once again, you've bolstered my argument, not
yours.
A basic fact about listening to music is that context affects what you
can perceive. Musicians exploit this. MAGICians exploit this too in
the visual realm. Your INTENTION affects what you can perceive. Your
intention even affects how the lower neural circuits process raw
stimuli. For a discussion how your intention actually changes how you
process signals, see
http://www.nytimes.com/2005/11/22/science/22hypno.html?
ei=5070&en=968c0cb9ab848d69&ex=1185854400&adxnnl=1 &adxnnlx=1185702379-
xWx02NnMb9D0ytIAKCNlLw
and for a demonstration how you can completely miss important signals
if you are tied up looking for something else:
http://answers.google.com/answers/threadview?id=490946
In the face of evidence like this, the burden is on DBT advocates to
show that they _don't_ skew perception, if they wish us to believe
they are relevant to other listening contexts.

Actually following your logic, the burden is also on subjectivists
to show they their perception isn't skewed, as well.

Actually following your logic, the burden is also on subjectivists
to show they their perception isn't skewed, as well.

Sighted listening can be biased, I agree.

It's *always* a possibility. Don't try to soft-pedal what science
has recognized for decades.

To restate something important, there's no single concept "DBT." There
are many DBT protocols, many test conditions. It's not surprising that
some aspects of a signal can be perceived well under some conditions.

the unifying concept of DBT is simple: neither the proctor nor the
subject must know which 'treatment' has been applied.

Neither do I call for abandoning blind protocols. I call for
understanding better their relationship to some of the subtler
qualities of sound, such as musical feeling.

You went far, far beyond that call, into the realm of pure speculation.

YOu and your ilk keep going on and on about 'perception of music'.

Music is only one of the sounds and experiences provided by an audio
system, but it's a particularly interesting one. The way that context
affects perception of music, something exploited by musicians and
composers constantly, leads to a particularly difficult situation to
test.

speech is another particularly interesting experience, which we are
arguably most sensitive to. So, do your 'issues' apply there too?

--

___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Steven Sullivan]

Mike wrote:
On Jul 29, 6:39 pm, Keith Hughes wrote:

This infers, quite clearly, that the pattern processing (i.e. your
musical interpretation overlayed on the sensory input) will be very
difficult to overcome. Rather puts paid to the notion that the DBT is
*so* disruptive that the *music* can't come through.

But context affects musical interpretation, and many DBT protocols
highly distort the context. Basic facts about music suggest that
interesting and subtle qualities won't "come through".

Basic facts about human perception suggest that 'sighted' listening
is highly prone to bias.

See even *today's* (Monday's) New York Science Times front page article,
on unconscious 'priming'.

[Steven Sullivan]

Mike wrote:
On Jul 29, 6:40 pm, bob wrote:
On Jul 29, 11:03 am, Michael Mossey wrote:





A basic fact about listening to music is that context affects what you
can perceive. Musicians exploit this. MAGICians exploit this too in
the visual realm. Your INTENTION affects what you can perceive. Your
intention even affects how the lower neural circuits process raw
stimuli. For a discussion how your intention actually changes how you
process signals, see

http://www.nytimes.com/2005/11/22/science/22hypno.html?
ei=5070&en=968c0cb9ab848d69&ex=1185854400&adxnnl=1 &adxnnlx=1185702379-
xWx02NnMb9D0ytIAKCNlLw

and for a demonstration how you can completely miss important signals
if you are tied up looking for something else:

http://answers.google.com/answers/threadview?id=490946

In the face of evidence like this, the burden is on DBT advocates to
show that they _don't_ skew perception, if they wish us to believe
they are relevant to other listening contexts.

Prove a negative? No, thank you.

That's a cop out. You have to provide *some* evidence that DBT
conditions are valid.

valid for what? They have already proven themselves sufficient for
discriminative training to hear small differences.

-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[josko]

"Mike" wrote in message
...
Even a test based on longer listening segments
requires that the subject try to remember sound in a way that foreign
to listening for enjoyment.

So, if somebody listens for enjoyment, he can differentiate, say, a
silver speaker cable from a copper speaker cable of the same
construction or one CD player from another for example? And when one
reaches the conclusion that one of those cables/CD players "resolves
more inner detail" than another, is that a memory-based judgment, at
least in part, or that conclusion is reached instantaneously at a
particularly emotionally stirring moment while listening to Beethoven's
fifth symphony for example (or whatever moves that person)? If that's
the case, then I guess one registers somehow his emotional reactions
while he is listening to music, and then he attributes that reaction (or
reactions) to that moment (or moments) during Beethoven's fifth *as
reproduced* by that particular cable/CD player, but not by the other
cable/CD player, and to *nothing else*? And no retrieval of information
from long-term memory is going on at any point during this process? If
this is exclusively a stimulus-based judgment, what is the actual
stimulus being evaluated: music only, music as reproduced by a
particular piece of equipment (but not by other piece of equipment), or
the piece of equipment under evaluation itself, or some combination of
the two?

The problem is that when one says that he prefers one cable or CD player
to another cable or CD player, it is a safe bet that judgment is not
based *only* on sound *as reproduced* by those pieces of equipment, but
on bunch of other variables which have nothing to do with the sound as
reproduced by those pieces of equipment. And that is precisely a reason
why scientists (such as psychoacousticans) use DB ABX protocols.

Moreover, we have tremendously solid and empirically proven theory that
two cables/CD players shouldn't sound different in the first place. That
theory draws on physics, signal theory, electrical engineering,
psychoacoustics.... (Note: it is possible to design a cable or a CD
player that audibly distorts signal, but I'm not talking about those.)
At the same time we also have a number of potential explanations (all
theoretically and empirically sound) why audiophiles routinely report
that one cable/CD player is better/worse than another. They are
numerous, but they all boil down to the fact that when people reach the
conclusion that "X sounds better/worse than Y" (and there is no base for
that conclusion in *physical reality*), they do not *control* for a
large number of variables that are unrelated to the *reproduced sound*,
but can definitely influence their judgment. Hence, the use of DB ABX
protocols.

The example with gorilla that you talk about is all about the influence
of attentional focus on perception. Another reason why scientists (such
as psychoacousticans) use DB ABX protocols.

Also, registered emotions are particularly bad measure for evaluating a
piece of hi-fi gear.

[Arny Krueger]

"Mike" wrote in message


Neither do I call for abandoning blind protocols. I call
for understanding better their relationship to some of
the subtler qualities of sound, such as musical feeling.

I sense the presumption that the people who developed blind testing
protocols were somehow deficient in their appreciation of the subtler
qualities of sound, such as musical feeling.

Nothing of the sort is true.

Music is only one of the sounds and experiences provided
by an audio system, but it's a particularly interesting
one. The way that context affects perception of music,
something exploited by musicians and composers
constantly, leads to a particularly difficult situation
to test.

Are you aware of the concept of "preaching to the choir"?

[Arny Krueger]

"Mike" wrote in message

On Jul 29, 6:39 pm, Keith Hughes
wrote:

This infers, quite clearly, that the pattern processing
(i.e. your musical interpretation overlayed on the
sensory input) will be very difficult to overcome.
Rather puts paid to the notion that the DBT is *so*
disruptive that the *music* can't come through.

But context affects musical interpretation, and many DBT
protocols highly distort the context.

This misapprehension has been debunked in this discusison several times. Why
does it keep popping up?

Basic facts about
music suggest that interesting and subtle qualities won't
"come through".

Again, we have this strong sense of the belief that DBT tests were developed
by people who are deficient in their appreciation of the artistic and
aesthetic properties of music.

And, again, you conflate "quick-switching" with short
rapidly switched snippets of A and B (your next post).
Listen to each as long as you like, in any manner you
like, in any order you like, as relaxed or intent as you
like, in any setting you like, playing any music you
like, focusing on any parameter you like. *ALL* of
these are clearly doable in the context of DBT, ABX or
otherwise.

I'm making no assumption about the test conditions.

Then why all the mention of unecessary problems with the test conditions?

They can be anything we like, and in fact we should all be
working to find some set of conditions that work.

Depends on how you define "work". Many people believe that what we have now
is working just fine, and is doing a good job of exposing the truth about
the sound quality of variouis audio components.

However, it's important to mention quick-
switching/snippet because a large number of claims are
based on such tests (i.e. Arny's web site).

I challenge you to find one such claim on my web site.

What is on my web site are a collection of audio samples that have a good
track record for making audible differences as apparent as possible, given
that there are any audible differences to be apparent.

And I've been
told a few tests under other conditions were done---a
few---but there was no attempt to control the conditions
or how the subjects used their attention.

I'm sure there are all sorts of wild rumors being spread by people who face
stiff economic losses if their exceptional claims are not believed in their
markets.

Mike

[Mike]

On Jul 31, 3:36 pm, bob wrote:
On Jul 29, 11:03 am, Michael Mossey wrote:

Your INTENTION affects what you can perceive. Your
intention even affects how the lower neural circuits process raw
stimuli. For a discussion how your intention actually changes how you
process signals, see

http://www.nytimes.com/2005/11/22/science/22hypno.html?
ei=5070&en=968c0cb9ab848d69&ex=1185854400&adxnnl=1 &adxnnlx=1185702379-
xWx02NnMb9D0ytIAKCNlLw

Finally had a chance to go back to this article, and sure enough, your
statements about it are completely wrong. It's not about *intention*
at all. It's about expectation--something that has long been known to
affect perception.

Of course, DBTs are designed to minimize the effects of expectation,
compared to other listening contexts.

Actually a DBT such as ABX doesn't get away from expectation at all.
Let's say you're the test subject in an ABX test, and you listen to A,
then B, then X. Your immediate suspicion is that X=B. Who knows where
that impression came from? Who knows where your attention was? (These
things aren't controlled in a blind test.)

But the important point is: you suspect that X=B, but you switch back
to listen to A and B again, to check. You've already formed an
expectation, though!

The _assumption_ in an ABX test is that a "real difference" will
overcome the fluctuating expectations that happen during a blind test.
But the article points out how profoundly expectation can influence
conscious impressions. Therefore that's not a good assumption.

Let me point out, also, that "intention", "attention", and
"expectation" are overlapping concepts. You intend to attend to
something. You expect something, so your attention goes there. Etc.

Mike

So thanks for providing yet more
evidence for my side of the argument.

bob

[bob]

On Aug 1, 7:20 pm, Mike wrote:

Actually a DBT such as ABX doesn't get away from expectation at all.

This is nonsense. I mean, you can believe it if you want to, but it's
just plain false.

Let's say you're the test subject in an ABX test, and you listen to A,
then B, then X. Your immediate suspicion is that X=B. Who knows where
that impression came from? Who knows where your attention was? (These
things aren't controlled in a blind test.)

Nor do they have to be. The test is about what you can hear. It's not
interested in WHY you heard it.

Now, I'll grant you that subjects do better on ABX tests when they
focus in on the actual differences in sound, rather than letting their
minds wander off in some musical reverie. But good DBTs are designed
to encourage and help subjects to focus on the right things--by
choosing very revealing material (and using short snippets so that
only the most revealing bits are heard), and by training subjects to
hear differences before testing them.

In other listening contexts, like just sitting and enjoying music, you
don't have any of these aids, so your ability to hear differences
diminishes.

But the important point is: you suspect that X=B, but you switch back
to listen to A and B again, to check. You've already formed an
expectation, though!

So what? It's an expectation inside the test. Expectation bias is
about expectations formed outside the test. Those are the only ones we
need to preclude.

The _assumption_ in an ABX test is that a "real difference" will
overcome the fluctuating expectations that happen during a blind test.

And this is just drivel.

But the article points out how profoundly expectation can influence
conscious impressions. Therefore that's not a good assumption.

Whatever. It's not an assumption we make.

bob

[Arny Krueger]

"Mike" wrote in message


Many of us find
that music has a meaning only in context,

How does one *not* have a context when listening to music?

and that many DBT's will distort the context.

Are you saying that sighted listening does not distort the context?

Therefore, I'm interested
in finding some way of constructing a DBT that doesn't
distort the context.

First show me a sighted evaluation that doesn't distort the context.

It seems you've already made up your
mind there's no problem with DBT's (any DBT?).

DBTs are not perfect. Nothing in the real world is perfect. DBTs have fewer
critical flaws than sighted evaluations.

A DBT likewise encourages a restricted focus.

Sighted evaluations similarly encourage a restricted focus.

A test based on snippets
will restrict the focus to one short moment of sound.

That is either a good thing or a bad thing depending on how well the short
moment of sound relates to the sonic problem being investigated.

It is well known that some sounds cloud the listener's ability to hear some
audible differences.

It is equally well known that some sounds magnify the listener's ability to
hear some audible differences.

There are very few if any sonic differences that are equally audible with
all sounds.

A sound that makes it easy to hear some audible differences may be very
suboptimal for hearing other audible differences.

A test based on quick switching will restrict the focus to
the moment of switching.

That is either a good thing or a bad thing depending on how well the short
moment of switching relates to the sonic problem being investigated. It is
very common for the duration of sounds that are optimal for hearing audible
differences to be relatively short. If a sound that is optimal for hearing
a certain audible difference is short, then packaging it into a longer
segment of sound makes it harder for the listener to remember what he heard.

Even a test based on longer
listening segments requires that the subject try to
remember sound in a way that foreign to listening for
enjoyment.

It is well known that listening for enjoyment and listening for differences
between audio products are very often two very different things. Belaboring
this point sheds no light because it is simply part of life. Being paralyzed
by it makes no sense because we want to be effective, right?

[Arny Krueger]

"Mike" wrote in message


Actually a DBT such as ABX doesn't get away from
expectation at all. Let's say you're the test subject in
an ABX test, and you listen to A, then B, then X. Your
immediate suspicion is that X=B. Who knows where that
impression came from? Who knows where your attention was?
(These things aren't controlled in a blind test.)

You've missed some keys points about testing. The purpose of testing is to
ensure that the impression is reliable, and that its source is just
listening.

If the impression is not reliable, then there its cause is not audible
differences between the products being listened to because the differences
between the products are supposed to be constant during the test. For
example we generally don't want people playing with the tone and gain
controls of the equuipment being tested, during the test.

If the impression is reliable, but due to just listening, then our test is
not about the sound quality of the products being tested.

But the important point is: you suspect that X=B, but you
switch back to listen to A and B again, to check. You've
already formed an expectation, though!

The ABX test manages this situation quite effectively. Expectations average
out because the identity of the unknowns is randomized.

The _assumption_ in an ABX test is that a "real
difference" will overcome the fluctuating expectations
that happen during a blind test.

That's a pretty good assumption, one that has been found to be true all of
the time.

But the article points
out how profoundly expectation can influence conscious
impressions. Therefore that's not a good assumption.

You've ignored how ABX tests effectively manage this situation.

In fact there's no problem with listener expectations of audible differences
in ABX tests, because the randomizing of the hidden variable make that all
average out during the statistical analysis.

[Mike]

On Aug 2, 3:33 pm, "Arny Krueger" wrote:
"Mike" wrote in message


But the important point is: you suspect that X=B, but you
switch back to listen to A and B again, to check. You've
already formed an expectation, though!

The ABX test manages this situation quite effectively. Expectations average
out because the identity of the unknowns is randomized.

The _assumption_ in an ABX test is that a "real
difference" will overcome the fluctuating expectations
that happen during a blind test.

That's a pretty good assumption, one that has been found to be true all of
the time.

This sounds like circular reasoning.


But the article points
out how profoundly expectation can influence conscious
impressions. Therefore that's not a good assumption.

You've ignored how ABX tests effectively manage this situation.

In fact there's no problem with listener expectations of audible differences
in ABX tests, because the randomizing of the hidden variable make that all
average out during the statistical analysis.

The fluctuating expectations add noise. Since the result of each trial
is quantized, there's no way for it to average out.

In other words, the fluctuating expectations mean that certain real
differences cannot be perceived reliably, so the subject's answers
look like guessing. There's no way to "average that out".

Mike

[bob]

On Aug 5, 11:05 am, Mike wrote:
The fluctuating expectations add noise. Since the result of each trial
is quantized, there's no way for it to average out.

In other words, the fluctuating expectations mean that certain real
differences cannot be perceived reliably, so the subject's answers
look like guessing. There's no way to "average that out".

Again, this is gibberish. You haven't demonstrated that 'fluctuating
expectations" exist. And you certainly haven't demonstrated that any
"expectations" developed during a DBT, based only on the sound,
represent a bias.

The term 'fluctuating expectations" is very nearly an oxymoron. The
problem with expectation bias is that it fixes our idea of something
and causes us to ignore countervailing information. But if our
"expectations" are "fluctuating," well, they're not really fixed, are
they? So how much are they interfering with our ability to process the
sonic information we get?

This is what happens when you try to pontificate on a subject without
adequate knowledge. You've got no evidence AT ALL, so you just make
something up.

bob

[Arny Krueger]

"Mike" wrote in message
...
On Aug 2, 3:33 pm, "Arny Krueger" wrote:
"Mike" wrote in message


But the important point is: you suspect that X=B, but you
switch back to listen to A and B again, to check. You've
already formed an expectation, though!

The ABX test manages this situation quite effectively. Expectations
average
out because the identity of the unknowns is randomized.

The _assumption_ in an ABX test is that a "real
difference" will overcome the fluctuating expectations
that happen during a blind test.

That's a pretty good assumption, one that has been found to be true all
of
the time.

This sounds like circular reasoning.

How?

It is no reasoning at all - it is just an assertion.

But the article points
out how profoundly expectation can influence conscious
impressions. Therefore that's not a good assumption.

You've ignored how ABX tests effectively manage this situation.

In fact there's no problem with listener expectations of audible
differences
in ABX tests, because the randomizing of the hidden variable make that
all
average out during the statistical analysis.

The fluctuating expectations add noise.

Right. What's unclear about "average out"?

Since the result of each trial is quantized, there's no way for it to
average out.

Say what?

Anybody who knows anything at all about statistics knows that properly
quantizing data is not a hinderance to it being averaged out. It's done all
of the time.

In other words, the fluctuating expectations mean that certain real
differences cannot be perceived reliably,

Say what?

The science of statistics was developed to effectively deal with all kinds
of fluctuations. The presence of random fluctuations at worst means that
you might have to take more data.

so the subject's answers look like guessing.

If the subject is reliably perceiving something, then the superficial
appearance of guessing can be overcome.

There's no way to "average that out".

Say what?

Your comments make no sense at all. Ever take a course in statistics? Ever
read a book about it?

Mike

[Arny Krueger]

"bob" wrote in message
...

You've got no evidence AT ALL, so you just make something up.

Exactly. I don't have the world's best credentials in statistics, but even
a cursory understanding of statistics says extactly the opposite of whats
been claimed.