The reason that testability is not enough is that prediction is not, and cannot be, the purpose of science. Consider an audience watching a conjuring trick. The problem facing them has much the same logic as a scientific problem. Although in nature there is no conjurer trying to deceive us intentionally, we can be mystified in both cases for essentially the same reason: appearances are not self-explanatory. If the explanation of a conjuring trick were evident in its appearance, there would be no trick. If the explanations of physical phenomena were evident in their appearance, empiricism would be true and there would be no need for science as we know it.
The problem is not to predict the trick’s appearance. I may, for instance, predict that if a conjurer seems to place various balls under various cups, those cups will later appear to be empty; and I may predict that if the conjurer appears to saw someone in half, that person will later appear on stage unharmed. Those are testable predictions. I may experience many conjuring shows and see my predictions vindicated every time. But that does not even address, let alone solve, the problem of how the trick works. Solving it requires an explanation: a statement of the reality that accounts for the appearance.
Some people may enjoy conjuring tricks without ever wanting to know how they work. Similarly, during the twentieth century, most philosophers, and many scientists, took the view that science is incapable of discovering anything about reality. Starting from empiricism, they drew the inevitable conclusion (which would nevertheless have horrified the early empiricists) that science cannot validly do more than predict the outcomes of observations, and that it should never purport to describe the reality that brings those outcomes about. This is known as instrumentalism. It denies that what I have been calling ‘explanation’ can exist at all. It is still very influential. In some fields (such as statistical analysis) the very word ‘explanation’ has come to mean prediction, so that a mathematical formula is said to ‘explain’ a set of experimental data. By ‘reality’ is meant merely the observed data that the formula is supposed to approximate. That leaves no term for assertions about reality itself, except perhaps ‘useful fiction’.
Instrumentalism is one of many ways of denying realism, the common-sense, and true, doctrine that the physical world really exists, and is accessible to rational inquiry. Once one has denied this, the logical implication is that all claims about reality are equivalent to myths, none of them being better than the others in any objective sense. That is relativism, the doctrine that statements in a given field cannot be objectively true or false: at most they can be judged so relative to some cultural or other arbitrary standard.
Instrumentalism, even aside from the philosophical enormity of reducing science to a collection of statements about human experiences, does not make sense in its own terms. For there is no such thing as a purely predictive, explanationless theory. One cannot make even the simplest prediction without invoking quite a sophisticated explanatory framework. For example, those predictions about conjuring tricks apply specifically to conjuring tricks. That is explanatory information, and it tells me, among other things, not to ‘extrapolate’ the predictions to another type of situation, however successful they are at predicting conjuring tricks. So I know not to predict that saws in general are harmless to humans; and I continue to predict that if I were to place a ball under a cup, it really would go there and stay there.
The concept of a conjuring trick, and of the distinction between it and other situations, is familiar and unproblematic – so much so that it is easy to forget that it depends on substantive explanatory theories about all sorts of things such as how our senses work, how solid matter and light behave, and also subtle cultural details. Knowledge that is both familiar and uncontroversial is background knowledge. A predictive theory whose explanatory content consists only of background knowledge is a rule of thumb. Because we usually take background knowledge for granted, rules of thumb may seem to be explanationless predictions, but that is always an illusion.
There is always an explanation, whether we know it or not, for why a rule of thumb works. Denying that some regularity in nature has an explanation is effectively the same as believing in the supernatural – saying, ‘That’s not conjuring, it’s actual magic.’ Also, there is always an explanation when a rule of thumb fails, for rules of thumb are always parochial: they hold only in a narrow range of familiar circumstances. So, if an unfamiliar feature were introduced into a cupsand-balls trick, the rule of thumb I stated might easily make a false prediction. For instance, I could not tell from the rule of thumb whether it would be possible to perform the trick with lighted candles instead of balls. If I had an explanation of how the trick worked, I could tell.
Explanations are also essential for arriving at a rule of thumb in the first place: I could not have guessed those predictions about conjuring tricks without having a great deal of explanatory information in mind – even before any specific explanation of how the trick works. For instance, it is only in the light of explanations that I could have abstracted the concept of cups and balls from my experience of the trick, rather than, say, red and blue, even if it so happened that the cups were red and the balls blue in every instance of the trick that I had witnessed.
The essence of experimental testing is that there are at least two apparently viable theories known about the issue in question, making conflicting predictions that can be distinguished by the experiment. Just as conflicting predictions are the occasion for experiment and observation, so conflicting ideas in a broader sense are the occasion for all rational thought and inquiry. For example, if we are simply curious about something, it means that we believe that our existing ideas do not adequately capture or explain it. So, we have some criterion that our best existing explanation fails to meet. The criterion and the existing explanation are conflicting ideas. I shall call a situation in which we experience conflicting ideas a problem.
The example of a conjuring trick illustrates how observations provide problems for science – dependent, as always, on prior explanatory theories. For a conjuring trick is a trick only if it makes us think that something happened that cannot happen. Both halves of that proposition depend on our bringing quite a rich set of explanatory theories to the experience. That is why a trick that mystifies an adult may be uninteresting to a young child who has not yet learned to have the expectations on which the trick relies. Even those members of the audience who are incurious about how the trick works can detect that it is a trick only because of the explanatory theories that they brought with them into the auditorium. Solving a problem means creating an explanation that does not have the conflict.
Similarly, no one would have wondered what stars are if there had not been existing expectations – explanations – that unsupported things fall, and that lights need fuel, which runs out, and so on, which conflicted with interpretations (which are also explanations) of what was seen, such as that the stars shine constantly and do not fall. In this case it was those interpretations that were false: stars are indeed in free fall and do need fuel. But it took a great deal of conjecture, criticism and testing to discover how that can be.
A problem can also arise purely theoretically, without any observations. For instance, there is a problem when a theory makes a prediction that we did not expect. Expectations are theories too. Similarly, it is a problem when the way things are (according to our best explanation) is not the way they should be – that is, according to our current criterion of how they should be. This covers the whole range of ordinary meanings of the word ‘problem’, from unpleasant, as when the Apollo 13 mission reported, ‘Houston, we’ve had a problem here,’ to pleasant, as when Popper wrote:
I think that there is only one way to science – or to philosophy, for that matter: to meet a problem, to see its beauty and fall in love with it; to get married to it and to live with it happily, till death do ye part – unless you should meet another and even more fascinating problem or unless, indeed, you should obtain a solution. But even
if you do obtain a solution, you may then discover, to your delight, the existence of a whole family of enchanting, though perhaps difficult, problem children . . .
Realism and the Aim of Science (1983)
Experimental testing involves many prior explanations in addition to the ones being tested, such as theories of how measuring instruments work. The refutation of a scientific theory has, from the point of view of someone who expected it to be true, the same logic as a conjuring trick – the only difference being that a conjurer does not normally have access to unknown laws of nature to make a trick work.
Since theories can contradict each other, but there are no contradictions in reality, every problem signals that our knowledge must be flawed or inadequate. Our misconception could be about the reality we are observing, or about how our perceptions are related to it, or both. For instance, a conjuring trick presents us with a problem only because we have misconceptions about what ‘must’ be happening – which implies that the knowledge that we used to interpret what we were seeing is defective. To an expert steeped in conjuring lore, it may be obvious what is happening – even if the expert did not observe the trick at all but merely heard a misleading account of it from a person who was fooled by it. This is another general fact about scientific explanation: if one has a misconception, observations that conflict with one’s expectations may (or may not) spur one into making further conjectures, but no amount of observing will correct the misconception until after one has thought of a better idea; in contrast, if one has the right idea one can explain the phenomenon even if there are large errors in the data. Again, the very term ‘data’ (‘givens’) is misleading. Amending the ‘data’, or rejecting some as erroneous, is a frequent concomitant of scientific discovery, and the crucial ‘data’ cannot even be obtained until theory tells us what to look for and how and why.
A new conjuring trick is never totally unrelated to existing tricks. Like a new scientific theory, it is formed by creatively modifying, rearranging and combining the ideas from existing tricks. It requires pre-existing knowledge of how objects work and how audiences work, as well as how existing tricks work. So where did the earliest conjuring tricks come from? They must have been modifications of ideas that were not originally conjuring tricks – for instance, ideas for hiding objects in earnest. Similarly, where did the first scientific ideas come from? Before there was science there were rules of thumb, and explanatory assumptions, and myths. So there was plenty of raw material for criticism, conjecture and experiment to work with. Before that, there were our inborn assumptions and expectations: we are born with ideas, and with the ability to make progress by changing them. And there were patterns of cultural behaviour – about which I shall say more in Chapter 15.
But even testable, explanatory theories cannot be the crucial ingredient that made the difference between no-progress and progress. For they, too, have always been common. Consider, for example, the ancient Greek myth for explaining the annual onset of winter. Long ago, Hades, god of the underworld, kidnapped and raped Persephone, goddess of spring. Then Persephone’s mother, Demeter, goddess of the earth and agriculture, negotiated a contract for her daughter’s release, which specified that Persephone would marry Hades and eat a magic seed that would compel her to visit him once a year thereafter. Whenever Persephone was away fulfilling this obligation, Demeter became sad and would command the world to become cold and bleak so that nothing could grow.
That myth, though comprehensively false, does constitute an explanation of seasons: it is a claim about the reality that brings about our experience of winter. It is also eminently testable: if the cause of winter is Demeter’s periodic sadness, then winter must happen everywhere on Earth at the same time. Therefore, if the ancient Greeks had known that a warm growing season occurs in Australia at the very moment when, as they believed, Demeter is at her saddest, they could have inferred that there was something wrong with their explanation of seasons.
Yet, when myths were altered or superseded by other myths over the course of centuries, the new ones were almost never any closer to the truth. Why? Consider the role that the specific elements of the Persephone myth play in the explanation. For example, the gods provide the power to affect a large-scale phenomenon (Demeter to command the weather, and Hades and his magic seed to command Persephone and hence to affect Demeter). But why those gods and not others? In Nordic mythology, seasons are caused by the changing fortunes of Freyr, the god of spring, in his eternal war with the forces of cold and darkness. Whenever Freyr is winning, the Earth is warm; when he is losing, it is cold.
That myth accounts for the seasons about as well as the Persephone myth. It is slightly better at explaining the randomness of weather, but worse at explaining the regularity of seasons, because real wars do not ebb and flow so regularly (except insofar as that is due to seasons themselves). In the Persephone myth, the role of the marriage contract and the magic seed is to explain that regularity. But why is it specifically a magic seed and not any other kind of magic? Why is it a conjugalvisits contract and not some other reason for someone to repeat an action annually? For instance, here is a variant explanation that fits the facts just as well: Persephone was not released – she escaped. Each year in spring, when her powers are at their height, she takes revenge on Hades by raiding the underworld and cooling all the caverns with spring air. The hot air thus displaced rises into the human world, causing summer. Demeter celebrates Persephone’s revenge and the anniversary of her escape by commanding plants to grow and adorn the Earth. This myth accounts for the same observations as the original, and it is testable (and in fact refuted) by the same observations. Yet what it asserts about reality is markedly different from – in many ways it is the opposite of – the original myth.
Every other detail of the story, apart from its bare prediction that winter happens once a year, is just as easily variable. So, although the myth was created to explain the seasons, it is only superficially adapted to that purpose. When its author was wondering what could possibly make a goddess do something once a year, he did not shout, ‘Eureka! It must have been a marriage contract enforced by a magic seed.’ He made that choice – and all his substantive choices as author – for cultural and artistic reasons, and not because of the attributes of winter at all. He may also have been trying to explain aspects of human nature metaphorically – but here I am concerned with the myth only in its capacity as an explanation of seasons, and in that respect even its author could not have denied that the role of all the details could be played equally well by countless other things.
The Persephone and Freyr myths assert radically incompatible things about what is happening in reality to cause seasons. Yet no one, I guess, has ever adopted either myth as a result of comparing it on its merits with the other, because there is no way of distinguishing between them. If we ignore all the parts of both myths whose role could be easily replaced, we are left with the same core explanation in both cases: the gods did it. Although Freyr is a very different god of spring from Persephone, and his battles very different events from her conjugal visits, none of those differing attributes has any function in the myths’ respective accounts of why seasons happen. Hence none of them provides any reason for choosing one explanation over the other.
The reason those myths are so easily variable is that their details are barely connected to the details of the phenomena. Nothing in the problem of why winter happens is addressed by postulating specifically a marriage contract or a magic seed, or the gods Persephone, Hades and Demeter – or Freyr. Whenever a wide range of variant theories can account equally well for the phenomenon they are trying to explain, there is no reason to prefer one of them over the others, so advocating a particular one in preference to the others is irrational.
That freedom to make drastic changes in those mythical explanations of seasons is the fundamental flaw in them. It is the reason that mythmaking in general is not an effective way to understand the world. And that is so whether the
myths are testable or not, for whenever it is easy to vary an explanation without changing its predictions, one could just as easily vary it to make different predictions if they were needed. For example, if the ancient Greeks had discovered that the seasons in the northern and southern hemispheres are out of phase, they would have had a choice of countless slight variants of the myth that would be consistent with that observation. One would be that when Demeter is sad she banishes warmth from her vicinity, and it has to go elsewhere – into the southern hemisphere. Similarly, slight variants of the Persephone explanation could account just as well for seasons that were marked by green rainbows, or seasons that happened once a week, or sporadically, or not at all. Likewise for the superstitious gambler or the end-of-the-world prophet: when their theory is refuted by experience, they do indeed switch to a new one; but, because their underlying explanations are bad, they can easily accommodate the new experience without changing the substance of the explanation. Without a good explanatory theory, they can simply reinterpret the omens, pick a new date, and make essentially the same prediction. In such cases, testing one’s theory and abandoning it when it is refuted constitutes no progress towards understanding the world. If an explanation could easily explain anything in the given field, then it actually explains nothing.
In general, when theories are easily variable in the sense I have described, experimental testing is almost useless for correcting their errors. I call such theories bad explanations. Being proved wrong by experiment, and changing the theories to other bad explanations, does not get their holders one jot closer to the truth.
The Beginning of Infinity Page 3
