Kissing Frogs: Falsification, Induction and Parapsychology

Chatting to a high-powered academic philosopher back in 2023, I wanted to explain why the many negative outcomes of experiments in parapsychology do not disprove the case for it. So I scratched around for a counter-example and came up with the Higgs boson.

In that example we evolved speculative theories, built vast particle accelerators, and sampled thousands of high-energy collisions, sifting through the tracks of the debris in search of the fabled Higgs. No luck. Perhaps we were looking in the wrong place. Discard that theory and brew up another one, build a bigger accelerator, sift through millions of events. Still no luck. Refine our theory down a bit, lather, rinse and repeat. Occasionally we see a tiny blip in the statistics, but they are never repeatable. In due course we built a truly gargantuan Large Hadron Collider and sifted through billions of events. The first tiny blip in the statistics proved just another false alarm. But we did not give up, we cranked up the power and carried on. Another blip, this one at last repeatable. Discovery! Now we know where to look and are starting in on a Higgs factory to study it at our leisure.

Compare this with say card or book tests for precognition. The first testable theory was that of JW Dunne. With limited resources, a few trials were run on a few dozen cases. A few blips from badly-run trials, not enough to convince. Then came the Rhines' psi hypothesis. A few hundred cases, a few more blips later, and still no luck. The Rhines built more complex, more automated test machines and moved into the tens of thousands. Still no luck. Perhaps they were looking in the wrong place (as Dunne's theory would suggest). Time to discard the psi hypothesis, brew up a fresh theory and get testing again.

Sure, there are differences, the parallel is not perfect. For a start the maths of quantum theory is sound, whereas that of Dunne was unsound and of the Rhines confined only to the statistical analysis. We must be careful not to compare chalk with cheese. But the modern theory of the Higgs is sound only because the older ones were falsified. Where is the modern theory of, say, dream precognition to put to the test? Where is the respect for the few rational theorists willing to pursue the work?

Had our attitude to the Higgs boson been so prejudiced back in the 1980s, we would never have built the LHC, never have had our quantum prejudices exposed for what they were.

To draw a more homely analogy, you have to make, stack up, and then tease through, an awful lot of hay before you can be sure there is no needle to be found in it. Or, to paraphrase an old saw, you have to kiss an awful lot of frogs before you can be sure than none of them is a prince.

Arguing that something has been found to be the case on many occasions, therefore it is always true, is known as proof by induction. For the most part, it is a dangerous fallacy; a counter-example to falsify your claim may turn up at any time. This is sometimes known as the "black swan" effect; Europeans knew that all swans were white, until someone went to Australia and found a black one. If you are confident there can be no such counter-example, then you must have already proved the case to your satisfaction, by other means! Induction can help point a way forward, but it can never get you there on its own. If you cannot come up with a better test, the only way forward is proof by statistics. Sometimes you can gather enough data to gain an overwhelming probability (known as 5-sigma) that the result is true. This is good enough for scientists (though not always for philosophers).

But of course that has to be valid data, it's no good if you are still way back down the trail gathering the wrong numbers. My Higgs boson analogy was to illustrate that, in the field of parapsychology, we are still way back there. The sceptical critics of parapsychology may or may not have a case, but one thing they do lack is rigorous scientific thinking.

Updated 15 Apr 2024