What is 'science'?

The term 'science' has a certain gravitas to it. Scientists are the new prophets that tell us about the world, reveal its hidden workings, and tell us what we must do in order to get nature to cooperate with our desires and intentions. Science has a different quality to it than history, or religion, or ethics. But what is it?

Science is aimed squarely at revealing the truth about the way the natural world works, but this can also be claimed for many other disciplines. An early attempt to capture the essence of science argued in the following way. Scientists collect data and from a certain amount of high quality observations (data points) they use the process of induction to infer how the world works. This inductive process, so the argument goes, is at the heart of science. It allows us to draw conclusions about events we have not observed.

But induction is logically untenable. It cannot be justified. Observations of a previous regularity in nature in no way logically imply that this regularity will continue. In describing the problem of induction, Wesley Salmon remarks, "We know (“in our hearts”) that we have knowledge of unobserved fact. The challenge is to show how this is possible." [The Foundations of Scientific Inference, I. The Problem of Induction]

Karl Popper argues that it is precisely this formulation that has led us into error. If you begin by assuming that 'we know' in the face of the impeccable logical arguments to the contrary, you begin down a well worn path into insoluble problems. The truth is, we do not know. We speculate, or conjecture. The question 'how do you know' presupposes an erroneous assumption and it is this assumption that 'we know' that has led to all the difficulties with induction. This mysterious process of induction simply does not exist.

But if this is true, where does this leave science? What is science if it is not that special discipline that uses scientific induction? In Popper's words, "if you abandon induction, how can you distinguish the theories of the empirical sciences from pseudo-scientific or non-scientific or metaphysical speculations?" [from Realism and the Aim of Science pg 159, italics in the original]

The short answer is that scientific claims (theories, or speculations, or conjectures) must be formulated in such a way that it is possible to falsify them through empirical observation. The real power behind the scientific method is that it opens itself to criticism. This means that it opens itself to correction. If there is no way to show that a theory about the natural world is wrong, if a theory can always be made to incorporate any evidence, then it is a theory without content. Such a theory does not really tell us anything about the external world because it does not forbid any observation.

But the lines between metaphysical speculation and scientific claims are not as sharp as one might think. Take, for example, the theory that everything is made from atoms. When there is no test for such a claim--because for any test that fails to find atoms one can always claim that atoms are smaller still--this theory lies in the real of metaphysics. But if it is reformulated in such a way that it not only claims the existence of atoms but also their dimensions. Then there is, logically speaking, a test that could be performed that could show it to be false. This, incidentally, is what led Ernst Mach to disbelieve in atoms. It was metaphysical speculation that could not be tested at the time. The work of Maxwell, Boltzmann, and Einstein turned this metaphysical speculation into testable hypotheses [Realism and the Aim of Science pg 160].

Another example is the lunar theory of tides. This metaphysical speculation originated from astrological lore and was rejected as nothing but superstition. Newton's theory of gravitation, however, gave testable predictions and from within this framework the moon's influence on tides could be understood. [Realism and the Aim of Science pg 190] Nevertheless, Newton's theories involved 'occult' action at a distance which has a strong mystical flavor.

Thus, mysticism can turn into science if properly formulated and metaphysical speculation may be very helpful for generating ideas that at first seem strange or unbelievable. But the key ingredient is openness to falsification. Popper remarks:

Astrologers have, incidentally, always boasted that their theories were based on an enormous number of verifications--upon overwhelming inductive evidence. This claim has never been seriously investigated and exploded, and I do not see why it should not be true. Yet it is hardly interesting to know how often astrology has been verified; the question is whether it has ever been seriously tested, by sincere attempts to falsify it. [Realism and the Aim of Science pg 191]

What is so special about falsification?

There are those who object to Popper's falsificationist thesis. One objection is that no falsification can be absolutely certain because even the facts of observation, the 'basic statements', have to be interpreted within some theoretical framework which is itself uncertain. This is true. But there remains a logical asymmetry between verification and falsification.

Suppose we know with certainty that some set of basic statements were true. It does not logically follow that we can prove a universal law from these statements. A universal law, by virtue of being universal, will always outstrip the capability of any finite set of basic statements.

However, a finite set of basic statements can demonstrate a theory to be false.

Hence the asymmetry is that a finite set of basic statements, if true, may falsify a universal law; whereas, under no condition could it verify a universal law: there exists a condition wherein it could falsify a general law, but there exists no condition wherein it could verify a general law.

Thus, if we accept as true the statement, 'This swan here is black', then we are bound, by logic, to admit that we have refuted the universal theory 'All swans are white'; and if we accept as true the statement, 'This planet is now more distant from the sun than it was a month ago', then we are bound, by logic, to admit that we have refuted the theory 'All planets move in circles with the sun as their common centre.' Now it is true--especially in the second case--that we may have made a mistake when we accepted the singular statement in question; and for this reason the falsification of the theory is not 'absolutely certain'. But it is absolutely certain that, if we accept any singular statement ('basic statement') that contradicts a theory we have accepted, we must have made a mistake somewhere--a mistake that must be corrected. And it is absolutely certain that if we accept a basic statement that contradicts the theory we are testing then we are bound to reject this theory as falsified. And it is thus also absolutely certain (since every basic statement contradicts some theories) that, whenever we accept any basic statement, some theories are thereby implicitly declared to be falsified, so that we are logically committed to rejecting them. But no theory has been verified: there is none which we are bound to accept as true. Hence the asymmetry. [Realism and the Aim of Science pg 185-6, emphasis in the original]

The reason that falsification has so much more power than verification lies in the absolutely certain, logical character of falsification as opposed to verification.