What does Hume say about inductive reasoning?
Hume’s Mitigated Skepticism: The Bounds of Inductive Reason
In Hume’s Enquiry Into Human Understanding, the philosopher interrogates the “proper province of human reason”. Hume begins by drawing a distinction between impressions and ideas. The former are the direct result of sensory experience; the latter are the “less lively…copies” of those impressions, often a result of reflection on the corresponding sensations (II.12-13). Even the most complex ideas may be taken as a composition or transformation of many simple ideas, which are still copies “from a precedent feeling or sentiment” (II.13).
Hume then divides all objects of human enquiry into two kinds of things: relations of ideas and matters of fact (IV.1.20). The former are analytic. They are a priori truths and may be ascertained through reason alone (IV.1.20). The latter are synthetic propositions; they are a posteriori (necessarily contingent on some empirical evidence). For every matter of fact, there exists “a contrary,” which is equally conceivable, and one “should in vain…attempt to demonstrate its falsehood” (IV.1.21).
Nevertheless, humans subscribe to many matters of fact, often “beyond the present testimony of our senses, or the records of our memory” (IV.1.21). Hume thus examines how humans find assurance in certain matters of fact, and not others. He finds that “all reasonings concerning matter of fact… [is] founded on the relation of Cause and Effect” (IV.1.22). Any one matter of fact is believed to be true, thanks to another matter of fact (IV.1.22). For example, a man comes to think that his friend is in France because his friend promised him as such (IV.1.22). That “his friend is in France” and “his friend promised him he is in France” are themselves two distinct statements, but the relation of cause-and-effect binds the two together.
Hume appropriately asks from whence the knowledge of cause-and-effect came. He ventures that knowledge of any cause-and-effect relation is a matter of fact, a synthetic a posteriori statement that is prone to error (IV.1.23). Cause-and-effect is a label we put on events which always happen in conjunction, with one always following the other. In this way, Hume cautions that much of human knowledge is a product of experience, and not reason. His mitigated skepticism accepts that cause-and-effect relations as a set of customary generalizations— still useful, but not unfailing. The sun may very well not rise the next morning.
Hume splits his argument into three parts. First, he demonstrates that the relation between an object and its effect is discoverable only by experience.
P1.1: Objects do not reveal their causes or potential effects via sensory qualities (IV.1.23). For example, fire is light and warm; these qualities do not suggest that fire burns.
P1.2: Reason cannot draw inference concerning matters of fact (a posteriori propositions) without experience (IV.1.23).
P1.3: A statement concerning the effect of an object is not a priori truth, and thus a matter of fact.
Given these three premises, one can only conclude the effect of an object by some experience of that effect. Adam, the first man and a perfect rational thinker, would not know that fire burns until he sees or learns that fire burns.
Hume then considers the relation between one event and a subsequent, resultant event. For the sake of argument, he assumes that the cause-and-effect relation between events is discoverable prior to experience. In other words, the human mind may become presented with any event and, without consulting past observation, accurately pronounce its effect.
Hume scrutinizes this assumption. He notices that “every effect is a distinct event from its cause,” no matter how connected the cause and effect may appear (IV.1.25). A human mind which cannot consult past observation must then proceed in this fashion:
- The mind considers the first event (e.g. the mind considers what would happen if you let go of a ball midair).
- The mind conjures up an “entirely arbitrary” event (the mind thinks that the ball would fly to the left, or that the ball would disappear, etc.). This event is arbitrary because absolutely anything may follow the first, and there is no indication that any one particular event is more likely.
- The mind ascribes that arbitrary event to the first event as its effect (e.g. the mind decides that whenever one lets go of a ball midair, that the ball would fly to the left).
By such an arbitrary operation of the mind, it stands to reason that “the mind can never possibly find the effect in the supposed cause” (IV.1.25). Hume thus concludes that the relation between one event and a subsequent, resultant event is discoverable also only by experience.
Having thus shown that knowledge of cause-and-effect relations is a posteriori, Hume challenges the validity of inductive reasoning: even if one has observed the same cause-and-effect sequence countless times, one cannot be certain that the same cause will invariably lead to the same effect in the future. The argument follows as thus:
P3.1 Past experience only “give[s] direct and certain information” about the objects or events that it had directly encountered (IV.2.29).
P3.2 The cause-and-effect relation that one experiences best understood as one event followed by another subsequent event. Repeated experience of the same cause-and-effect relation ought to become described as one event in constant conjunction with the same subsequent event (IV.2.29). Given P3.1 and P3.2, Hume concludes that repeated experience of the same cause-and-effect is not evidence that the same cause-and-effect relation will persist in the future.
P3.3 To prove that the same cause-and-effect relation will persist in the future, one must demonstrate that the future will resemble the past in the case of this particular cause-and-effect (IV.2.30-31).
P3.4 That “the future will resemble the past” is not an analytic, a priori truth. This means that “the future will resemble the past” must be an a posteriori proposition. Now recall P3.1: one cannot employ an argument from past experience to prove that “the future will resemble the past.”
Hume therefore concludes that the proposition “the future will resemble the past” is an uncertain matter of fact. The connection between cause and effect and the feeling of “consequence” is a logical leap “not supported by any argument or process of the understanding” (IV.2.29, V.1.34). Hume ascribes this leap to the principle of custom. One might not have any understanding of the connection between two events “constantly conjoined together,” but one still finds some utility in making that connection. It allows humans to push enquiries beyond their present experiences, beyond the “narrow sphere of [human] memory and senses” (V.2.44). Hume advises that his readers embrace experience as the “authority” and “great guide of human life” (IV.2.31, V.1.36). He emphasizes that his philosophy ought never to “undermine the reasonings of common life” (V.1.34). Indeed, induction is already at the foundation of “almost all knowledge” (V.1.34).
Overall, Hume’s argument is valid, but not necessarily sound.
P1.2 and P3.4 follow from the starting premise that there exist only two types of propositions: analytic a priori and synthetic a posteriori. Immanuel Kant would later re-evaluate this division. He pronounces a third category of synthetic, a priori truths that includes the concept of causality: humans intuitive impose the idea of cause-and-effect on perceptions to acquire understanding. Hume’s argument also raises additional considerations. Could the feeling of consequence be more than a result of habit or custom? One might argue that all scientific enquiries founded on the assumption that the universe follows some pattern and behaves, at least in part, in a non-chaotic manner. Following this line of thought, the expectation that “the future will resemble the past” is necessary even when considered independent of the effects of habit or custom.
Finally, notice the contrast between Hume’s mitigated skepticism and other unmitigated forms.
In section XII, Hume first criticizes the Cartesian method of doubt, which he finds to be too extreme. He sympathizes with Descartes’s hope to avoid errors in judgment, but ultimately thinks that “no reasoning could ever bring [such a skeptic] to a state of assurance and conviction upon any subject” (XII.1.116). Hume also addresses skeptics who, in some way, calls the process of sense-perception into question. He alludes to Berkeley’s theory, which posits that all immediate sensible qualities exist not in the objects but in the mind of a perceiver (XII.1.122). Hume thinks that such a skeptic position takes abstraction too far. Absurd to think that an object’s quality of extension exists only in the mind because “an extension, that is neither tangible nor visible. Cannot possibly become conceived”. Similarly, one cannot conceive of an extended object that neither hard nor soft (XII.1.122).
Altogether, Hume insists that productive skepticism must inspire academic modesty, but not “carry its doubts so far as to destroy all action, as well as speculation” (V.1.34, XII.3.129).
What does Hume say about inductive reasoning?
