Determinism

J. B. S. Haldane

Editor’s Note

Arthur Eddington’s essay on the decline of determinism provoked strong responses. The biologist J. B. S. Haldane here criticizes Eddington’s remarks about human behaviour — specifically, that if the behaviour of atoms is indeterminate, then the human mind must also be. It was then considered that behaviour may be correlated between identical twins. Haldane points out that if one such twin breaks the law during a certain period, the chance that his twin will do so may be as high as 0.875. (These data were almost certainly flawed.) Eddington was effectively claiming that no amount of extra information could turn such a prediction into a certainty. But Haldane complains that this assertion is little more than an a priori pronouncement, unsupported by evidence.　　中文

IN his address on the decline of determinism,1 Sir Arthur Eddington enunciates a very curious equation. “If the atom has indeterminacy, surely the human mind will have an equal indeterminacy; for we can scarcely accept a theory which makes out the mind to be more mechanistic than the atom.” This statement will not bear too close an examination even from a non-quantitative point of view. Thus an attempt by myself to solve even a simple wave equation might lead to any of a large number of results; a similar attempt by Sir Arthur Eddington would lead to the correct solution with a high degree of probability. I do not think that this proves that his mind is more mechanistic than my own, whatever that may mean. Actually it is generally regarded as a compliment to describe a person as reliable, that is, to suggest that his conduct is predictable.　　中文

Fortunately, however, quantitative data exist which seem to show that, as regards moral behaviour, some minds are decidedly more determined than are some atoms as regards radiative behaviour. Consider a given man M1, and the probability p that between times T1 and T2 he will commit an action such as to lead to his imprisonment for a breach of the law. If we have no further information regarding M1, p is in most communities a small number, less than 0.01. If, however, M1 has a monozygotic twin M2 brought up in the same environment up to the age of 14, and M2 is known to have been imprisoned for crime between the ages of 16 and 40, we can infer with a fairly high degree of probability that M1, who has an identical nature and a similar nurture, has been or will be imprisoned between the same ages. Judging from Lange’s2 results, p is about 0.875 in south Germany when we have the above amount of information. If we increase the amount of information, for example, by excluding cases where M2 has suffered from head injury, the value of p is raised still further. Now, if it could be shown that with sufficient information p became unity in a case of this kind, we should, I take it, have proved the determinacy of some kinds, at least, of moral choice. Actually the most that scientific method can do is to prove p>1–ε. If Sir Arthur Eddington is correct, then no matter how complete our information, ε tends to a finite limit which is not very small. Clearly no amount of observation could prove it to be zero. But if it could be shown to be less than 0.01, we could neglect it to a first approximation in ethical theory, and if it proved to be less than 10－6 we might hazard the guess that the behaviour of human beings showed no more indeterminacy than that of other systems composed of about 2×1027 atoms.　　中文

I think that it is a legitimate extrapolation from the existing data that if we used all the available data in the above case, ε would be less than 0.05. It seems unfortunate that any attempt should be made to prejudge, on philosophical or emotional grounds, the magnitude of a quantity susceptible of scientific measurement. But from the heuristic point of view the deterministic theory has the advantage that it could be disproved, and would be if ε tended to a finite limit as the amount of available information increased indefinitely. On the other hand, indeterminism cannot be disproved unless its supporters state the value of ε, which they have so far carefully avoided. When the truth about human behaviour is discovered, it will probably appear that philosophers of all schools had failed to predict it as completely as they failed to predict Heisenberg’s uncertainty principle. Human behaviour is a subject for scientific investigation rather than a priori pronouncements.　　中文

(129, 315-316; 1932)

J. B. S. Haldane: Royal Institution, Albemarle Street, London, W.

References:

1. Nature, 129, 240 (Feb. 13, 1932).

2. Crime as Destiny (1931).