Hume's Irrationalism

David Hume - was critical of the works of Descartes (rationalist) and Berkley (empiricist). Asked the question can we have any epistemically (rationally) justified beliefs about the external world?

Justified True Belief: True - something corresponds to how things really are. Epistemology assumes there is a “way things really are” independent of our beliefs (i.e. there is such a thing as objective truth or objective facts).

Types of Epistemic Justification:

  • A Posteriori: justification based on experiential states; sensory (visual, auditory, tactile, etc.), introspective.

    • S is justified a posteriori in believing that P = definition; S is justified in believing that P, and S’s justification in dependent on some experience you’re having or you’ve had.

    • E.g., I would be a posteriori justified in believing that fish exist, that fish can swim, that I am thinking about fish right now.

  • A Priori: justification that’s independent of prior experience.

    • Justified a priori in believing; geometrical propositions (e.g., Pythagoras’ Theorem), arithmetic propositions (e.g., three times five is half of thirty), definitional truths (e.g., there are no married bachelors).

A major fault-line in epistemology is what kinds of propositions can be justified a priori.

  • Hume’s Answer: only “relations of ideas” can be justified a priori, no "matters of fact” or “real existence” (this is the doctrine of empiricism)

  • Empiricism: Any justified propositions concerning matters of fact are justifiable only on the basis of experience (a posteriori).

  • Rationalism: Some justified propositions concerning matters of fact are justifiable by pure reason (a priori).

Hume’s Question: what is the nature of that evidence which assures us on any real existence and matter of fact beyond the present testimony of our senses, or the records of our memory? Do we, in fact, have any propositional knowledge at all?

Prediction and Retrodiction

  • When we predict, we use our knowledge of what sorts of effects the observed phenomenon tends to cause.

    • Experience: lightning. Inference: I will soon hear thunder. (Prediction)

  • When we retrodict, we use our knowledge of what sorts of causes tend to bring about the observed effect.

    • Experience: parking ticket on windscreen. Inference: a parking inspector was here. (Retrodiction)

Our knowledge of absent matters of fact seem to be based on our knowledge of causal relations.

  • How do we get knowledge of causal relations?

  • Hume: not a priori (the answer is “no”, even if you guessed the right answer)

  • For every cause, many different effects are conceivable

Hume thinks we can have a priori justification of a proposition only if its negation is inconceivable - i.e., implies a contradiction (a bachelor who is married, a four-sided shape that has five sides).

  1. All propositions justifiable a priori are propositions whose negations are inconceivable.

  2. No propositions about causal relations are propositions whose negations are inconceivable.

  3. Therefore, no propositions about causal relations are justifiable a priori.

Our knowledge of cause and effect - if we have any - must be justifiable a posteriori.

  • Hume: All our knowledge of causal relations comes from experience

  • Specifically, he thinks we make inductive inferences on the basis of our past experiences

    • Inductive Inference: reasoning from a finite/limited set of observations for a broader conclusion (extrapolate)

  • Singular Inductive Inference:

    • 1. Every F observed thus far is a G

    • 2. X is an F

    • 3. Therefore, X is a G

  • Inductive Generalization:

    • 1. Every F observed thus far is a G

    • 2. Therefore, all hitherto unobserved F’s are G’s

  • In every past instance these two events went together, therefore all future events will follow the same pattern.

According to Hume, our beliefs about absent matters of fact are based on our beliefs about cause and effect - which are formed via induction.

  • When we observe that A events are consistently followed by B events, we form the belief that this will continue to happen

  • Are inductively formed beliefs about absent matters of fact epistemically justified?

Hume’s Problem of Induction

A Necessary Assumption

  • The assumption that “the future will conform to the past”

  • Without that, the premise provides little support for the conclusion; call this the Principle of the ‘Uniformity of Nature’

Principle of the Uniformity of Nature (PUN)

  • Since inductive inference presupposes PUN, for induction to be capable of providing epistemic justification, PUN must be epistemically justified

  • Hume argues that neither a priori or a posteriori will work to epistemically justify this

Not A Priori

  • A proposition is justifiable in this way only if its denial is inconceivable (because it involves a contradiction)

  • Problem: the denial of PUN is not like this

  • We can coherently conceive of the world behaving radically differently in the future

  • We may not think it likely that the laws of nature will change dramatically, but there is no logical contradiction in this idea (it doesn’t involve claiming that something both is and is not the case)

    • Therefore PUN cannot be justified a priori

Not A Posteriori

  • It has been confirmed over again by our experience - we have always found, so far, that the future does resemble the past

  • You could make the argument:

    • 1. On all past occasions that I observed nature, it was uniform

    • 2. Therefore, on all occasions, past or future, observed or unobserved will be uniform

  • Here you are engaging in circular reasoning; reasoning that presupposes the conclusion you’re trying to prove - logical fallacy of ‘begging the question’

  • This argument only works if we have reason to think that PUN is true

Hume’s Argument

  1. Induction is justified only if PUN is justified

  2. If PUN is justified then either it is justified a priori or it is justified a posteriori

  3. PUN is not justified a priori

  4. PUN is not justified a posteriori

  5. (SC1) PUN is not justified

  6. Therefore, induction is not justified

  • If Hume’s argument succeeds then the vast majority of our beliefs about the world are epistemically unjustified and have no rational basis

  • Why do we form beliefs this way?

    • Hume: “The principle is Custom or Habit”

    • “Custom is the great guide of human life”

  • Today we might suggest that the ‘custom’ or ‘habit’ to make inductive inferences is innate, with a genetic basis shaped by evolution (Hume says something close to this) - is this epistemically upshot?

  • Hume is often seen as a sceptic about absent matters of fact (i.e. as one who denies we have knowledge about this topic)

  • MOF may not be so simple, however…

  • He thinks our empirical, inductive beliefs about absent MOF are rationally unjustified, but he also thinks it’s a good thing that we make them

  • Perhaps he is an irrationalist: reason can’t give us knowledge, but we do have knowledge, so much the worse for reason

  • On this view, he is rejecting the necessity of ‘justified’ in ‘Justified True Belief’

One Possible Response

The Probabilist Reply

  • One response to Hume:

    • Maybe observations of constant conjunctions between A & B doesn’t justify the conclusion that all A’s are followed by B’s

    • But surely it justifies the conclusion that any given A will probably be followed by B

  • An example of a good inductive argument:

    • All past lightning flashes observed were followed by thunder

    • So, probably every lightning flash is followed by thunder

    • So, each hitherto unobserved lightning flash is such that it probably is/was/will be followed by thunder

Salmon’s Humean Response

  • Wesley Salmon (1975) thinks this kind of response to Hume cannot work

  • He poses a dilemma: What exactly does ‘probably’ mean? - What conception of probability are we operating with?

  • There are two dominant conceptions:

    • 1. Frequency

    • 2. Degree of Rational Confidence

  • Either way, Salmon says we’re still forced to assume PUN, and thus stuck with Hume’s problem

Two Types of Probability

  • To say that “this particular roll of the die will probably (83%) land on a face higher than1” might mean:

    • “If there were an infinity of die rolls, most rolls (83%) would in fact land on a face higher than 1” (frequentist interpretation of probability)

  • Or it might mean:

    • It is rational to have 83% confidence that this particular roll of the die will land on a face higher than 1 (degree of rational confidence interpretation)

Two Types of Induction

  • Frequentist version of the lightning induction:

    • 1. Every past flash of lightning was followed by thunder;

    • 2. So, if there were an infinity of lightning flashes, most of those would in fact be followed by thunder

    • Problem: this still relies on PUN (else why think 1 is a guide to 2?)

  • Degree of rational confidence version of lightning induction:

    • 1. Every past flash of lightning was followed by thunder;

    • 2. So, it is rational to have a high degree of confidence in the belief that the next flash of lightning will be followed by thunder

    • Problem: This still relies on PUN

  • It seems we are stuck relying on the Principle of the Uniformity of Nature!