OK, now we're moving on. First, re postulates supported by scientific method, none of the postulates (usually called hypotheses) you mention are supported by scientific method at the time they're made. Taking your first example: "Postulate: amps with lower distortion sound better Verification: tests with a variety of listeners would be able to statistically verify this at levels of THD of 10% compared to 1%" At the time it is first proposed, the hypothesis "amps with lower distortion sound better" has no scientific support. You may have reasons for thinking so (personal experience, anecdotal evidence) but no-one has tried to find out whether or not this is a reliable statement. We have one problem with the hypothesis and that is the use of the word "better" which in this case amounts to a statement of preference. Much better initially to hypothesise that "amps with lower distortion sound different" so that what we're trying to verify is that people hear something different when listening to an amp with lower distortion. Now, where the scientific method comes in is at the stage of what you called "verification". You make a predicition that will be true if the hypothesis is correct and, preferably, only if the hypothesis is correct (if it can be true if the hypothesis is not correct you either need to find another prediction that is only true when the hypothesis is correct or, if necessary, a set of two or more predictions that can only all be true together when the hypothesis is correct) and then set up a test where the only variable is the variable in the prediction. Here you'd probably want to restate the test for verification as "listeners can hear a difference between 10% and 1% THD" and you design a system so that you can vary the THD only using the same source, amp and speakers (can't change the amp because that will change other factors besides THD) and run your tests. Actually what you do when testing is to set up a counter hypothesis - that listeners won't hear a difference - and then try to show that the counter hypothesis is false but we won't go into that sort of area here. Now that's fine for distortion levels of the sort you've mentioned but we have a hypothesis that says there's an audible difference as distortion is lowered. What we also might be interested in are a couple of more difficult questions - first, what's the smallest difference in THD that people can hear (ie can they tell the difference between 10% and 10.1% or 10.05% or…) and second, what is the lowest level of THD that is audible since lowering THD below that level should then produce no difference in sound whatsoever so customers would then have no reason to spend more in order to get something with a lower THD spec knowing they couldn't hear it. It's much harder to answer those 2 questions than the original one since now we're working at the limits of what people can hear and individual difference becomes critical. It's relatively easy to define levels that everyone with normal hearing can hear and even to define a level that nobody with normal hearing can hear, but in between there we have the grey area I spoke about where some people with normal hearing will hear it and some won't. Note my restriction to people of normal hearing which requirs you to test the subjects first. If you include everyone, you include people with substandard hearing including some who are almost deaf and who will not be able to hear any difference at all at levels where most people are clearly hearing things. If you set your audibility limit high enough to include people with sub-standard hearing, it becomes useless as a real guide to the sorts of things we're interested in. That answers one of your questions "is it valid to proceed with arguments about levels of distortion if only some individuals can detect them?" You answer the question based on individuals with normal hearing because they're the bulk of the population. There's no problem for the people with better than average hearing - apart from the fact that they may complain about hearing things most other people don't - and the people with sub-standard hearing won't be challenged any more than they are elsewhere in life. Now, to get back to your use of the word "better" in the hypothesis which I carefully edited out for the reasons I gave. Once you've proved a difference in sound exists, you can repeat the same sort of test and ask people whether they think the sound is better. Note that we are now asking a question about what people think, not about the sound. That's a sure sign of a preference and we know that different people have different preferences. In testing this, we have to ensure that the levels of THD involved are audible - if we drop below the threshold of audibility and the subjects can't hear a difference their answers are useless. Also, we're now not trying to predict whether people will prefer it or not - we want to know what they prefer so it's somewhat like holding an election. Sometimes the result is a landslide for one view and at other times it's very close. The important thing to note is that the question is no longer about an observation but about an attitude to what is observed, and that we have to ensure that the thing is observed in the test to be able to rely on the answers we get. Youre next point was "I guess what I am trying (perhaps unsucessfully) to say is that many arguments in the field of audio are so riddled with varying perceptions that "logical" argument is difficult if not impossible." I think the issue here is the confusion about observations and attitudes I discussed in the previous paragraph. As long as we're clear about what we're talking about and how we test it, there is no problem about following a logical process to get answers to the questions and the type of answer we get depends on the type of question. Of course, once we've got the answer people tend to use it for their own purposes and if they don't understand the differences observations and preferences, they're likely to try and use the answers in ways that they shouldn't be used. That definitely isn't logical but the problem there is not with science or the scientific method, or even with the results. It's to do with people not understanding of the nature of the information they're reading and using. You mentioned people understanding words differently and I'll deal with that a bit later but I don't think that's the problem as you'll eventually see. So the first big thing is to understand the difference between an observation, which relates purely to what can or can't be sensed by a sense organ, and an attitude or preference which is something that we think about our observations. That brings us back to one of JJs perennial points that preferences don't need the support or otherwise of tests. Preferences are just our preferences and questions like whether tubes sound better than SS are about preferences. No amount of tests can prove that one sounds better than the other. The most you can do by testing is to show what percentage of people prefer one and what percentage prefers the other, and there's no right or wrong about that. Even if 99% preferred tubes, there is nothing WRONG in someone preferring SS but, if 99% did prefer tubes the SS lover should definitely be prepared to be considered a social deviant or even worse :-) The next big thing is the subjective/objective distinction. People tend to interpret it as subjective meaning what an individual experiences and objective as what test instruments show. It's not quite that simple and we couldn't know what a test instrument showed if we didn't observe the meter so that would make reading of the meter subjective and we'd never have anything objective if that were the case. I'm probably not too clear on the distinction as I should be, but the way I'd define it is that objective relates to things that people can agree about without arguing definitions, so reporting whether you heard something 3 times or 5 times would be an objective report. On the other hand, preferences and attitudes are by definition subjective - they relate to the subject's personally and not to other people. Finally there's the situation where you're called on to describe an observation in terms of a perceptual quality like colour or, more importantly for us here, an auditory quality. Sometimes that's easy and sometimes, if it's a subtly distinction, it's very hard. The more subtle it is, the more the consistency of judgements by the test panel relies on them having precisely the same understanding of the nature of the quality and the more likely it is that responses will vary. So there's a subjective element in the characterisation or description of what is observed and this element will vary in scale and importance to the researcher depending on the distinction concerned. You keep talking about "subjective perceptions" as if they're observations but really, most of the time, I think you're really referring to attitudes or preferences. So John, I don't think we're in quite as much agreement as you think though there is definitely some agreement. I think, and the whole of science is founded on the belief that observation is verifiable. I don't think observations vary as much as you seem to think though I definitely agree that preferences and attitudes do. I also don't think we vary as much in what we think words mean - there's definitely a very strong commonality in that or communication would break down very quickly. I think the problem isn't so much with words as with concepts and people need training with concepts. If you don't have scientific training you tend not to understand scientific concepts just as if you don't have psychological or educational training you don't understand psychological or educational concepts. Technical words are really "jargon" and problems arise when a word has a common use which is well understood and a jargon use, or even worse, two or more different jargon uses relating to different fields. That's when we really start to get confused by what someone is saying. I think, if I remember correctly, your professional background is education so you'll have that jargon and probably some psych jargon. My background is in philosophy and psych via a BA and then in occupational health and safety by post graduate training so I've got some jargon in common and some different jargon. And here we're talking hearing and perception and test design plus the basis of scientific proof. I've got a bit of the jargon in some of those areas due to the sorts of research I had to study and do when doing my OH&S qualifications but I'm definitely not as fluent as someone like JJ who works in that sort of field, but I think Iunderstand a bit about the basics of scientific method and proof and I think what I've said above is a reasonable stab at responding to what you're saying from a hard science perspective. David Aiken
|