De Facto and De Jure in the Practice of Induction

ABSTRACT

‘Simplicity’ comes up in different senses in scientific methodology. The simplicity criterion at issue here is relied on in all inductive inference, it’s argued. Therefore, it cannot be inductively learned—except by learners who already rely on it. The question whether one is warranted in relying on it is indeed in order, but we all are found on the affirmative side of the question in practice. On the one hand, there is arguably an epistemological lesson in this. On the other hand, there is the question what reality must be like objectively for reliance on simplicity to be so apt to pay off when it comes to forecasting the future.

Notes

1 I have argued for something similar when it comes to desires in Goldstick 2000, 423–436.

2 Hume 1962, 37: ‘[…] all inferences from experience suppose, as their foundation, that the future will resemble the past....’

3 Cf. Broad 1918, 396–397.

4 The last two paragraphs and the last but one before them are based on my book, Goldstick 2009, 142–143.

5 Cf. Goldstick and O'Neill 1988, 589–593.

6 Hume 1962, 38: ‘It is impossible […] that any arguments from experience can prove this resemblance of the past to the future; since all these arguments are founded on the supposition of that resemblance’.

7 E.g. Quine 1960, 20: ‘Conservatism, a favoring of the inherited or invented conceptual scheme of one’s own previous work, is at once the counsel of laziness and a strategy of discovery’. The assessment seems justified that Quine’s characterisation of conservatism as the counsel of laziness really was meant to count pragmatistically in conservatism’s favour.

8 E.g. G.W.F. Hegel writing in 1817 on physicists’ appeal to ‘convenience’ in explaining specific gravity: ‘If convenience be all that is desired, surely it would be more convenient to banish calculation and thought altogether’ (Hegel 1892, 193).

9 But Thomson goes on to say, ‘Some theories are cumbrous, limited in scope and arbitrary. They seldom live long’ (Thomson 1961, 17). The disparaging of ‘cumbrous’ theorising reminds us of economy of thought. The pejorative word ‘arbitrary’ implicitly invokes something like the principle of sufficient reason. And ‘They seldom last long’ suggests a scientific concern, after all, for truth, or at least empirical adequacy, for the future as well as the past.

10 Popper 1965, 278: ‘Our science is not knowledge (epistēmē): it can never claim to have attained truth, or even a substitute for it, such as probability’.

11 In a falsificationist spirit also, but quite differently argued, Lewis S. Feuer’s ‘The Principle of Simplicity’ (Feuer 1957, 109–122) insists that to keep adding hypothesis-saving epicycles or the like to a theory so as to fit observations made is a symptom of psychoanalytically diagnosable neurosis, an idée fixe resistant to the testability requisite for thought capable of coping effectively with the world. But what is needed mostly for effective coping is truth as far as attaining it is possible. It is unclear whether Feuer accepts simplicity as a predictor of truth, or instead opposes theory-complication independently of that, on the grounds, rather, of his psychoanalytic diagnosis. After all, neurotic or not, if such tinkering, at each stage, didn’t detract from, or even enhanced, the amended theories’ ability to predict new truths, what basis would there be for objecting scientifically?

12 There are epistemologists who say that somebody justified in holding a belief needn’t always have a cogent argument in its favour, so long as one is available. But how can such an argument justify somebody in believing its conclusion unless the believer at least has grounds for thinking some such argument in fact is available?

13 According to the best physical theory, for instance, a uranium₂₃₈ atom has a half-life of 4,468 million years—meaning, there is a 50% chance that within a period of 4,468 million years it will suddenly decay into thorium₂₃₄ by emitting an alpha particle, without any prior causative event that it would be physically possible to detect.

14 Indeed, even a not-scandalously-leaky argument to this effect will certainly require some further premises. See Goldstick 2009, 162–173.