As I've said before, the environmental factors around carefully designed DBT (not necessarily ABX) are going to have some minimum resolution. Alls you can do is alls you can do.
Anything that's shouted from the highest heavens as obvious when not blinded should pop out like a sore thumb when blinded. If it doesn't what does that say about the ostensibly obvious thing?
Anything that's shouted from the highest heavens as obvious when not blinded should pop out like a sore thumb when blinded. If it doesn't what does that say about the ostensibly obvious thing?
Oops, I didn't edit my last post fast enough. But, I see the point.
However, I would like to offer a rough, and only partly developed possible explanation for what may be going on, as follows below.
To those who find it hard to imagine any possible way for sighted listening to be legit, please let me suggest as a thought experiment something along the following lines:
In WWII, the Navajo code takers communicated in a way obvious to them, but only strange noises to most people. To people in remote Chinese villages, their dialects and accents may obvious to them, but imagine if in each village they wore different clothing styles. Suppose you traveled through that area and learned to associate the sounds of dialects that were only noises to you with clothing styles. If someone played you sound recordings and asked you to name the clothes associated with the dialect you hear, it might be very difficult. But if someone showed you some clothes and asked you to listen to the noises used in the dialects, you might find that it helped remind you of certain subtleties to listen for. In that case, the subtleties might have been learned by your system 1, and without your system 2 even being aware of it. But showing your system 1 the clothes, might load the DSP into system 1 working memory that recognizes the sound subtleties your system 1 learned to associate with the clothes. Of course, once your system 1 has loaded the proper DSP, the results might seem quite obvious, as your system 2 just hears it clearly without knowing how system 1 has done it.
In addition, what your system 1 learned might not be bilateral, that is it might be a one-way function, clothes -> sounds. Learning the reverse might be a different task, with different neurons involved. And, there could be something about brain evolution that favors the forward function, we don't know. But based on things we have learned from cognitive psychology and neuroscience, it very well could turn out to work that way.
Anyway, it's probably possible to devise experiments to find out if some brain mechanism or ability exists, more or less in keeping with what I described. If it does exist, it might also turn out that people who do need to legitimately use sighted listening for best results could be differentiated from fakers though fMRI observations during listening tests. If so, that might make it rather inconvenient for DIY-medical-research audio engineers working alone to carry out the research. But, then again, maybe it would be a good thing. Having some medical researchers on the team might make possible a number of improvements to the overall research quality.
However, I would like to offer a rough, and only partly developed possible explanation for what may be going on, as follows below.
To those who find it hard to imagine any possible way for sighted listening to be legit, please let me suggest as a thought experiment something along the following lines:
In WWII, the Navajo code takers communicated in a way obvious to them, but only strange noises to most people. To people in remote Chinese villages, their dialects and accents may obvious to them, but imagine if in each village they wore different clothing styles. Suppose you traveled through that area and learned to associate the sounds of dialects that were only noises to you with clothing styles. If someone played you sound recordings and asked you to name the clothes associated with the dialect you hear, it might be very difficult. But if someone showed you some clothes and asked you to listen to the noises used in the dialects, you might find that it helped remind you of certain subtleties to listen for. In that case, the subtleties might have been learned by your system 1, and without your system 2 even being aware of it. But showing your system 1 the clothes, might load the DSP into system 1 working memory that recognizes the sound subtleties your system 1 learned to associate with the clothes. Of course, once your system 1 has loaded the proper DSP, the results might seem quite obvious, as your system 2 just hears it clearly without knowing how system 1 has done it.
In addition, what your system 1 learned might not be bilateral, that is it might be a one-way function, clothes -> sounds. Learning the reverse might be a different task, with different neurons involved. And, there could be something about brain evolution that favors the forward function, we don't know. But based on things we have learned from cognitive psychology and neuroscience, it very well could turn out to work that way.
Anyway, it's probably possible to devise experiments to find out if some brain mechanism or ability exists, more or less in keeping with what I described. If it does exist, it might also turn out that people who do need to legitimately use sighted listening for best results could be differentiated from fakers though fMRI observations during listening tests. If so, that might make it rather inconvenient for DIY-medical-research audio engineers working alone to carry out the research. But, then again, maybe it would be a good thing. Having some medical researchers on the team might make possible a number of improvements to the overall research quality.
Last edited:
Back when I worked in audio, we used to get a lot of consultants commenting or otherwise involved in the process. They pretty much all rubbished DBT for the usual reasons. However, 95% of them failed to identify which setting was which, though they were convinced they did - we used to have a bit of fun saying "this is setting A" while using setting B... We never let on though...
I have some intuition similar to what you say, but I don't know if it will turn out to be correct or not. I also have some intuition that my prior intuition is wrong.
More seriously, experimentation how shown that many plausible-seeming intuitions later turn out to be wrong. Have to be very careful about the risk of over-confidence with such things.
Last edited:
Markw4 can that really be called a thought experiment? Sounds more like a daydream, which is not the same thing. I also don't think it sheds much light on sighted listening evaluations, and to the extent that it does it validates the idea that sighted evaluation tends to reinforce previous experience or preconception.
I have no problem with sighted listening, I do it every day. I also have no problem with forming judgments based on sighted listening, one can't help forming judgments and we would be fools to ignore them. However, assigning a lot of value to judgments based entirely on sighted listening is equally foolish, and publishing comparative evaluations based only on such judgments is pure hubris.
I have no problem with sighted listening, I do it every day. I also have no problem with forming judgments based on sighted listening, one can't help forming judgments and we would be fools to ignore them. However, assigning a lot of value to judgments based entirely on sighted listening is equally foolish, and publishing comparative evaluations based only on such judgments is pure hubris.
Fashions do change/evolve - but a 'Drunkard's Walk' isn't 'Progress'
and I woud like to know how you cherry pick to show any direction in Audiophile taste over the years
ears are necessary, its the brain that mixes in all sorts of other info - leading to the need to control for the senosry experience modifiers in the brain if you want to get closer to what the ears alone recieved
and I woud like to know how you cherry pick to show any direction in Audiophile taste over the years
ears are necessary, its the brain that mixes in all sorts of other info - leading to the need to control for the senosry experience modifiers in the brain if you want to get closer to what the ears alone recieved
I've met several very serious audiophiles in Japan that think nothing has progressed beyond NOS WWII vintage 300B's. What they paid for them at Akihabara was stunning.
If it wasn't for digital, there would have been no progress at all in audio for the last 40 years. Obsoleted FETs, 40 year old tubes, 60-ies circuits.
Except maybe for speaker/driver technology, where you see now and then real progress leading to more realistic and transparent reproduction.
Jan
Except maybe for speaker/driver technology, where you see now and then real progress leading to more realistic and transparent reproduction.
Jan
I share some of the concern of others about some audiophiles, and in particular, some audiophile magazines.
However, I also think that serious people like Bob Ludwig may be onto something when they have concerns about double blind testing as they are familiar with it. Therefore, I would be in favor of new research to try and find out where people like Ludwig are having a problem, and what we might be able to do about it.
Its hard to see how we could get past rehashing the same old arguments over and over again otherwise.
One thing seems quite clear, in recent years we have seen publication of research such as that described by Daniel Kahneman which shows that our intuitions about how brains work can be very different from the reality.
In addition, many new research methods and technologies have come into play in recent years. For one thing, fMRI has proven much more useful than I ever expected when I first heard about it. Because of such developments, it seems reasonable to expect substantial new progress could be made in the area of hearing research. My vote would be to find some way to move forward with research, and in the meantime try to exercise some detached scientific deferral in forming premature judgments about what must be going on inside other people's heads. That is, at least for serious people such as Ludwig, and others like him.
However, I also think that serious people like Bob Ludwig may be onto something when they have concerns about double blind testing as they are familiar with it. Therefore, I would be in favor of new research to try and find out where people like Ludwig are having a problem, and what we might be able to do about it.
Its hard to see how we could get past rehashing the same old arguments over and over again otherwise.
One thing seems quite clear, in recent years we have seen publication of research such as that described by Daniel Kahneman which shows that our intuitions about how brains work can be very different from the reality.
In addition, many new research methods and technologies have come into play in recent years. For one thing, fMRI has proven much more useful than I ever expected when I first heard about it. Because of such developments, it seems reasonable to expect substantial new progress could be made in the area of hearing research. My vote would be to find some way to move forward with research, and in the meantime try to exercise some detached scientific deferral in forming premature judgments about what must be going on inside other people's heads. That is, at least for serious people such as Ludwig, and others like him.
Last edited:
Jan, I think you underestimate the impact of post-1969 integrated circuit designs, upon audio gear. H-bridge input current feedback amplifiers, folded cascode amplifiers, nested compensation amplifiers with both feedforward and feedback, floating current mirrors (a la AD797), and Complementary-MOSFET amplifiers all originated after 1969, all originated in ICs, and all got copied into audio products.
I also think some of the analog circuits published in Linear Audio, post Y2000, represent real progress in audio. The cube law amplifier by Ian Heggelund, and the class-AB bias controller by Dan Joffe are two examples that spring to mind.
I also think some of the analog circuits published in Linear Audio, post Y2000, represent real progress in audio. The cube law amplifier by Ian Heggelund, and the class-AB bias controller by Dan Joffe are two examples that spring to mind.
My take is his concern is almost always methodology. DBT != ABX. Sitting down and getting comfortable with two components can still be done totally double blind. This discussion has once again gone difference ignoring preference. I've seen phono pre-amps with 3-5dB deviation from RIAA, in this case what's the point in DBT of any kind?
Don't underestimate the power of DSP to muddy the waters. I found myself stuck for an evening last week waiting for a morning flight with nothing but my wife's Kindle loaded with the last few Star Trek movies. The tiny speakers put 360 degree convincing images in space behind me and way outside the device, all I suppose the THX processing. It became at moments quite disconcerting.
All else being equal, full discrete fet based amplifiers can match or beat anything else. It is true that fets are more difficult to buy, but then so are vacuum tubes, compared to the past. We just pay the price for the extra quality. To think that there is no progress in analog amplifiers just shows that nobody really knows amplifiers very well, except in the academic sense. Our best topologies still don't have an IC equivalent, because complementary jfets are impractical.
Digital amps are continually improving as well. The first examples that I tested, perhaps 15 years ago, were really marginal. Today, they are better, probably thanks to 'that young Dutch guy who designs so many of them'. I am still looking for someone to tell me the good, bad or 'ugly' facts about the AES paper that I put up. Anybody read it? Any comments?
Digital amps are continually improving as well. The first examples that I tested, perhaps 15 years ago, were really marginal. Today, they are better, probably thanks to 'that young Dutch guy who designs so many of them'. I am still looking for someone to tell me the good, bad or 'ugly' facts about the AES paper that I put up. Anybody read it? Any comments?
I wouldn't trust fMRI here as far as I could throw a 7T machine. The field is rife with issues in terms of publishing utter garbage (and I really mean that as critically as I can) in studies on psychological or preference problems. Yes, certain sections of the brain light up with greater blood flow during these tests.
No doubt it's a great toolset that can be used in so many diagnostic fashions and tell us about brain function from gross events and disease states, but the low N coupled with the terrible statistics (fMRI data is notoriously noisy and all brain imaging modalities are hampered by temporal resolution) around most studies makes them nigh useless. Gross reactions (e.g. something like meta-analyses of the brain's pathway through a fear-inducing event) have a lot more utility, but still have tractability issues. Identifying and localizing structural effects of TBI, siezures, drug use, etc are quite a bit more telling. Then again, I get this second hand from a number of neuroscience grad student friends who are as grouchy as I am, so take it all with a grain of salt--I am not an expert in the field.
And then there's these logistical problems. Never mind the masking issues of being in an MRI tunnel. And a whole lot more stressful than trying to do a DBT!
- Status
- Not open for further replies.