Scientific Method

The classic scientific method taught in most schools–and even believed by many scientists–goes something like this: a scientist makes a hypothesis and then carries out an experiment to see if this hypothesis is confirmed; if so, the hypothesis has gained support and, with enough support, eventually becomes a scientific theory. This view that scientific ideas can be empirically verified was popular among the logical positivists but cannot be placed on a logical framework. Karl Popper, one of the biggest critics of logical positivism, proposed an alternative view of the advancement of scientific knowledge in terms of falsifiability rather than verifiability.

If I make a hypothesis that water boils at one hundred degrees Celsius, I will have no problem finding an infinite number of confirming experiments. After a thousand successes in which I experimentally verify that water boils at this temperature, will the thousand and first experiment make me any more confident in my initial hypothesis? The verificationist school of thought will view this repeated act of confirmation as the gain of knowledge, but this reasoning falls prey to Hume’s problem of induction. Logically, an increasing number of confirming cases does not–and cannot–prove the validity of a statement.

Instead, suppose I were to try and disprove my initial hypothesis by ascending a mountain and finding that the water actually boils at a different temperature. Rather than perform a thousand experiments that all tell me the same thing, in this case I will actually have learned something: there are some situations where water does not boil at one hundred degrees Celsius. I can re-state my initial hypothesis as “water boils at one hundred degrees Celsius at sea level”, for example, and proceed to experimentally seek additional ways of critiquing and disproving my hypothesis.

We can never say for certain, then, that a given scientific theory is true, for we can never logically verify any theory. Falsification, though, provides a mechanism by which we can gain knowledge and develop better and better ways of looking at the world. Newtonian mechanics provided an excellent (and still useful) way of viewing the Universe, yet this idea ultimately falls short in relativistic realms–and even Einstein’s theory is far from a complete physical law of the Universe. Even if such physical laws exist, we will never know we have arrived at such a law, for our current theories are the ones that have withstood the greatest attempts at falsification with success. We don’t know which theories are true; we simply know which ones have not yet been proven false.

Popper delves into these ideas much more in his books (see, for example The Logic of Scientific Discovery or Conjectures and Refutations), but this basic idea of almost trial and error is certainly more gratifying than the (illogical) paradigm of verification. Science is not a quest for Truth, then, for even if Truth exists it can never be recognized as such; science, instead, improves upon itself by systematically rejecting old ideas in favor of more complete ones–which themselves will eventually be falsified.