On the Meaning and Measure of Uncertainty: I

This file is a single-author theoretical article in economics. Shackle’s essay attacks the assumption that economic choice can be modeled as choice among known satisfactions. Standard value theory, he argues, silently assumes perfect knowledge; but many economic goods, especially information itself, exist only because their contents are not already known.

But the theory of consumer’s behaviour assumes that we always know what we are going to get.

Against this, Shackle shifts the object of choice from satisfactions to actions. The chooser does not select outcomes directly; he selects courses of action whose consequences remain hypothetical. This move is decisive for the article’s theory of uncertainty.

The things amongst which a man is free to choose are not satisfactions themselves, but actions designed to secure for him some sort of satisfaction; and except when he acts as a mere spender of income on familiar things for immediate consumption, there is no knowing whether any given course of action that he embarks on will secure him the sort of satisfaction he looks to it for, either in kind or in quantity.

The essay then constructs its central vocabulary. A decision-maker faces rival, mutually exclusive hypotheses about what an “experiment” or course of action may bring. Each hypothesis has a face-value, the gain or loss if it proves true, and a further quality measuring its claim on the imagination. Shackle first asks whether orthodox probability can supply this second variable.

But it is almost self-evident that the frequency-ratios cannot be used to predict the result of any one, particular, individual trial.

His answer turns on the distinction between divisible or seriable experiments and those that are non-divisible and non-seriable. Probability works where a single decision can be absorbed into a repeatable class or pooled, as in insurance. But investment, life choices, and “crucial” decisions often alter the agent’s situation irreversibly and cannot be repeated as trials of the same kind.

Economic life consists to an important degree in having to choose amongst courses of action each of which is what I have called a non-divisible non-seriable experiment.

Shackle’s five objections to numerical probability follow from this point. Probability cannot represent perfect plausibility when several rival hypotheses are all genuinely possible; it changes merely because more alternatives are imagined; it cannot preserve comparable maxima across different distributions; in a unique event, any probability other than zero or one is retrospectively “wrong”; and it cannot express the state in which both a proposition and its contradictory seem equally possible. Subjective probability does not solve the problem, because it still treats belief as a limited quantity to be divided among rivals.

These difficulties arise essentially because we have been using as our second independent variable something which is limited in amount and has to be shared out amongst the distinct non-excluded hypotheses; we have been using something which measures or indicates degrees of positive belief.

The constructive proposal is to replace degrees of positive belief with degrees of disbelief, understood not as lesser certainty but as the felt presence of some disabling incongruity between a hypothesis and one’s experience, information, or conception of the world.

By disbelief I do not now mean the absence of perfect certainty, but the positive recognition of some disabling circumstance.

This becomes Shackle’s measure of uncertainty: potential surprise. Zero potential surprise marks an outcome whose truth would cause no shock; maximum potential surprise marks exclusion. Unlike probability, this scale allows many mutually exclusive hypotheses to be fully possible at once.

Thus for the word ‘disbelief’ in the quite special sense in which I have been using it I propose now to substitute potential surprise.

The essay’s relevance lies in this conceptual displacement. Shackle does not merely deny that some probabilities are knowable; he denies that probability has the right logical form for much economic uncertainty. In its place he offers a psychology of imagined futures, where choice is guided by anticipated gain or loss constrained by what the agent can entertain without surprise.

And there is, in general, no limit to the number of mutually exclusive hypotheses to all of which simultaneously a person can, without logical contradiction, attach zero potential surprise.

The article thus supplies a non-probabilistic foundation for decision under uncertainty. Its enduring importance is that it treats economic action, especially investment and irreversible choice, as imaginative commitment under rival possible worlds rather than calculation over a fixed probability distribution.