Scientific research can be thought of as a process of asking questions of Nature. Perhaps it’s worth exploring that concept in a little more detail.
It is true that many scientific advances have started with a question. And the process of research can be considered a way of asking Nature questions. But the kind of questions that we can ask Nature are very specific.
First, the obvious: Nature doesn’t have a voice. Interviews are out. So we need to look for evidence instead.
The language that I’m using resembles a criminal investigation, and that’s deliberate. Scientific research is in fact very much like forensic work. We look for evidence, we analyse things that we observe, we try to find patterns and unravel processes. Forensics is all about mechanisms: how the crime was perpetrated. However, there’s usually an accompanying part of a criminal investigation, and that is the literal question-and-answer stuff. By interviewing a suspect, the investigator can try to unravel the question of motive. Forensics, for all its strengths, is powerless to address “why” questions. This, again, is like science.
There are many kinds of questions that we can ask another person, but the questions that science can ask Nature are far more limited. It’s not just the “why” stuff, we also can’t ask general or open-ended questions. We can’t say, for instance, “What causes the tides?” That’s too vague. The things that we can ask Nature are more like the game of “20 Questions” – she can only answer “yes” or “no”. (Actually, this is not strictly true, she usually answers with “possibly” or “probably not”, but that’s a minor point).
So we can ask, as Galileo did, “Does the rotation of the Earth cause the oceans to slosh around like a bathtub, and thus cause the tides?”
…and Nature says, “No”.
Or we can ask, as Newton did, “Does the moon pull the water around with its gravitational attraction?”
…and Nature says, “Yes.”
“Cool,” we reply. “Does that allow us to predict the tides?”
…and Nature says, “No, but you’re getting warmer.”
Lord Kelvin said, “What if we take Laplace’s differential equations and add in bathymetry and coastal boundaries?”
…and Nature said, “…even warmer…”
And so on, until George Darwin finally asked just the right questions, and tides could be predicted.
Now, this is a fine little historical excursion, but the important point is to note the form of the questions: they’re all phrased in terms of “yes/no” alternatives. (The more science-y term for asking a yes/no question of Nature is hypothesis testing). This allows for a rigorous, disciplined procedure, but it has a vital limitation:
You can only ask a question of Nature if you’ve already thought of the answer.
That’s how hypothesis testing works: you start with the way you think that stuff works, and you test it to see if you’re right.
There are two steps here. First, need to think up the mechanism, and then you need to think up a way of demonstrating the your idea is right. But your test must also, in as far as possible, rule out other possibilities. Necessarily, it can only rule out possibilities that you’ve also already thought of. And when you test it, you also try to think of all the ways that your test results could be misleading you. Hypotheses and theories can be confirmed, but that does not mean that they are true. It means that they have passed all the tests that we can think to put to them. As the mathematician John Lennox has observed, “Even observations tend to be theory-laden; we cannot take a temperature without having an underlying theory of heat.”
There are ways to control for some of these problems. The whole idea of a laboratory is that you can do an experiment in some degree of isolation, so you reduce the potential complicating factors. But lab results can have their own problems. In my field of marine biology, if I want to know how fast a fish can swim, I can take it to the lab, put it in a tank, and observe. I can even try to entice it to swim at a high speed, perhaps by controlling a current through the tank. But this doesn’t actually tell me how fast the fish does swim in the wild, it just tells me how fast it can be made to swim in a lab. The answers to these two questions are related, but they are not the same..
If only I could just ask the fish!