6.1 “It’s science, but not as we know it”

Imagine you are visiting a violin dealer, choosing a new instrument. Perhaps you have your teacher with you, or a colleague from your orchestra. You play musical fragments to each other on various instruments, and you both try to put into words what you hear and feel. You might make comments like “This one is more rich, a bit quiet under the ear, but a little weak on the A string”. Straight away, many questions are raised in a scientist’s mind. Does everyone really use words like “rich” to mean the same thing? Does the listener experience the sound the same as the player? If you came back the following day and tried the instruments again, perhaps with a different friend or in a different mood, would you make the same judgements? Would you even be confident of recognising which instrument was which, without being told?

There is more. The range of instruments you have tried will cost different amounts: possibly spectacularly different, as illustrated by Fig. 1 where we see three violins with market values ranging over 3 or 4 orders of magnitude. How is it possible for instruments looking so similar, made from similar materials, to be different enough to justify such a remarkable range of market values? In the extreme: “everyone knows” that the very best violins were made by Antonio Stradivari in Italy in the decades around 1700. So what is the secret of these magical instruments?

This is the wrong question. We should first ask things like: Can we really tell old Italian violins apart from more recent instruments? Do players or listeners really prefer them, even when they haven’t been told which is which? But then: What exactly do we mean by “prefer” here? Does every player prefer the same instrument, so that there really might be a valid concept of a “best violin”? Is a beginner looking for the same thing as a competent amateur player in a string quartet, or an international soloist? Even for a single violinist, is it obvious that the same instrument is “best” for different styles of music, or for performing in different acoustical settings?

We would like to bring some kind of scientific method to bear on all this, so that we could draw conclusions backed by solid evidence. This leads us into the world of psychoacoustics, a rather different kind of science from what we have been talking about so far. This discipline is not based primarily on physics and laboratory measurements. A typical experiment involves many volunteers, listening with headphones to many groups of barely distinguishable sounds, and trying to discern which is the “odd one out” of each group. The purpose is to build up enough data to map out statistically reliable thresholds for discrimination, or correlations between physically measurable quantities and judgements of such things as “richness”. The style and limitations of this kind of experiment will have important implications for addressing musical questions. Furthermore, the questions raised so far are not a good place to begin delving into the world of psychoacoustics. We need to start with simpler questions: but “simpler” will prove to be only relative. Everything to do with perception turns out to be complicated!

The purpose of a psychoacoustical experiment is to reverse-engineer an aspect of human perception of sound. In the usual way of scientists, the researcher will try to break the complicated totality down into small pieces, then look at each piece separately. But from the perspective of the brain of the test subject, this is a profoundly unnatural thing. As we saw back in Chapter 1, your brain is constantly trying to combine the output from a host of low-level and high-level feature detectors, and interpret the result in terms of what is going on in the world around you. The experimenter wants to focus on a single low-level detector, such as one to decide the relative loudness of two sine waves at different frequencies. But unintended features of the detailed procedure may bring in the influence of other detectors. From the experimenter’s perspective, this is a source of bias. But in the bigger picture, it is telling us something else about the processes of perception.