Strange to say, even the utilitarian school, which, at first sight, appears to be distinguished from the rest by recognizing natural causation, is, if not so far from complete recognition of it, yet very far. (§21 ¶1)

Conduct, according to its theory, is to be estimated by observation of results. When, in sufficiently numerous cases, it has been found that behaviour of this kind works evil while behaviour of that kind works good, these kinds of behaviour are to be judged as wrong and right respectively. Now though it seems that the origin of moral rules in natural causes, is thus asserted by implication, it is but partially asserted. The implication is simply that we are to ascertain by induction that such and such mischiefs or benefits do go along with such and such acts; and are then to infer that the like relations will hold in future. But acceptance of these generalizations and the inferences from them, does not amount to recognition of causation in the full sense of the word. So long as only some relation between cause and effect in conduct is recognized, and not the relation, a completely-scientific form of knowledge has not been reached. At present, utilitarians pay no attention to this distinction. Even when it is pointed out, they disregard the fact that empirical utilitarianism is but a transitional form to be passed through on the way to rational utilitarianism. (§21 ¶2)

In a letter to Mr. Mill, written some sixteen years ago, repudiating the title anti-utilitarian which he had applied to me (a letter subsequently published in Mr. Bain’s work on Mental and Moral Science), I endeavoured to make clear the difference above indicated; and I must here quote certain passages from that letter. (§21 ¶3)

The view for which I contend is, that Morality properly so-called—the science of right conduct—has for its object to determine how and why certain modes of conduct are detrimental, and certain others modes beneficial. These good and bad results cannot be accidental, but must be necessary consequences of the constitution of things; and I conceive it to be the business of Moral Science to deduce, from the laws of life and the conditions of existence, what kinds of action necessarily tend to produce happiness, and what kinds to produce unhappiness. Having done this, its deductions are to be recognized as laws of conduct; and are to be conformed to irrespective of a direct estimation of happiness or misery. (§21 ¶4)

Perhaps an analogy will most clearly show my meaning. During its early stages, planetary Astronomy consisted of nothing more than accumulated observations respecting the positrons and motions of the sun and planets; from which accumulated observations it came by and by to be empirically predicted, with an approach to truth, that certain of the heavenly bodies would have certain positions at certain times. But the modern science of planetary Astronomy consists of deductions from the law of gravitation—deductions showing why the celestial bodies necessarily occupy certain places at certain times. Now, the kind of relation which thus exists between ancient and modern Astronomy, is analogous to the kind of relation which, I conceive, exists between the Expediency-Morality and Moral Science properly so called. And the objection which I have to the current Utilitarianism is, that it recognizes no more developed form of Morality—does not see that it has reached but the initial stage of Moral Science. (§21 ¶5)

Doubtless if utilitarians are asked whether it can be by mere chance that this kind of action works evil and that works good, they will answer—No: they will admit that such sequences are parts of a necessary order among phenomena. But though this truth is beyond question; and though if there are causal relations between acts and their results, rules of conduct can become scientific only when they are deduced from these causal relations; there continues to be entire satisfaction with that form of utilitarianism in which these causal relations are practically ignored. It is supposed that in future, as now, utility is to be determined only by observation of results; and that there is no possibility of knowing by deduction from fundamental principles, what conduct must be detrimental and what conduct must be beneficial. (§21 ¶6)