The problem of induction is considered a problem because of the argument that ultimately the past does not necessarily predict the future. It could be true that the sun won't rise tomorrow, but it has always done so in the past. We make this assumption that the future will match the past.

It is claimed that this assumption is unjustifiable. But isn't this also an assumption? It's assuming that justification should only depend on certainty. Why should it? Why should a belief be justified only if you're certain of it?

Besides, what does it mean to be certain of something? It is often argued that you can't know anything for certain except experience. I'm not even sure you can be "certain" of experience. Almost all experiences technically happen in the past once you reflect upon it. Once you experience a painful event, at any time you even consider whether or not the painful event happened, it's already in the past. But that is assuming that your memory is not defunct.

So even for conscious experiences, we're ultimately making assumptions, aren't we? What's wrong with making assumptions in the first place?

  • all philosophy depends on an assumption, even if it is someone else's. HTH, it did me. the old riddle of induction is scary because almost everything depends on its answer
    – user63756
    Dec 9, 2022 at 19:09
  • 1
    No. It requires a positive argument to justify an assumption. It requires no argument to point out that such a positive argument does not exist. Dec 9, 2022 at 21:40
  • See "The Problem of Induction" and "Certainty". Dec 10, 2022 at 1:38
  • 1
    have you read the SEP article plato.stanford.edu/entries/induction-problem doing so would be preferable to quizzing this site, especially as only you really know what you are confused about
    – user63756
    Dec 11, 2022 at 6:43
  • 2
    Can we be 100% certain that the sun will rise tomorrow? No. Can we be 99.999% certain the sun will rise tomorrow? Realistically it depends on how fast you can travel West.
    – Brendan
    Dec 11, 2022 at 8:24

5 Answers 5


You are asking about the “assumption” that the future will be like the past and, in that context, about justification and certainty, particularly the certainty of experience. You also ask whether the sceptical argument against induction is “not self-refuting”.


I’m not sure what you are getting at in the last question. The classical problem of induction is not self-refuting, if you mean by that a rational counter-argument with a sound conclusion or a clear demonstration that the sceptical argument is invalid. You are right that the sceptical conclusion is, in a sense, unacceptable, but that doesn’t mean that it is self-refuting; it's a reason for trying to find a solution.

It is worth pointing out that David Hume, who is always cited as the originator of the argument in modern philosophy, partly accepted the sceptical conclusion; he accepts that there is no relationship of necessity between cause and effect as propounded by earlier philosophers, particularly the scholastics, and I think that there is general acceptance of that.

The question of induction is another matter. Neither Hume nor anyone else has accepted the conclusion that formulating empirical generalizations on the basis of past experience and relying on them as predictions of the future should be abandoned. Yet no-one directly refutes the argument; the name of the game here is finding a way round it.

The assumption that the future will be like the past

The sceptical argument treats the principle of the uniformity of nature as an empirical generalization like any other (such as “my car always starts first time”) and then argues that since it is not necessary it must be uncertain. But I think that this is a gross over-simplification. We do not always abandon a well-established generalization as soon as a problematic case appears. We have many ways of dealing with it. When the evidence is slender, we are quick to abandon it; when the evidence is substantial (and varied), we have ways of adapting it to preserve past experience and incorporate new experience into a revised edition. We can revise the scope of the generalization (we thought that all large water-fowl were white, but now we see that only all swans are white) or we can create a new category (we could define swans as white and invent a new category for black swan-like birds). To put the point another way, “The future will be like the past” allows that it will not be the same as the past, and we are perfectly capable of adapting to new cases that are unlike the old.

Nonetheless, we do believe that a given cause (or set of causes) will produce the same effects wherever and whenever it occurs. There’s no empirical evidence sufficient to prove this; nor is there an a priori argument for it. Does it follow that it is uncertain or unjustified? It depends on what you mean by justification and by certainty.


In general, what counts as justification depends on the context, specifically on what kind of proposition you are talking about and the language-game it is embedded in. Ideas like this mean that one can respond to the sceptical argument about induction by pointing out that the sceptic is applying inappropriate standards to inductive conclusions. Again, the Uniformity Principle is plainly not a simple empirical generalization and in my view needs to be regarded in a quite different light. It is a requirement on specific causal laws; any hypothesis that doesn’t conform to it will not count as a causal hypothesis; in a sense, it is part of the definition of “cause” and “effect”. Perhaps more accurately, it is a methodological principle to be applied in formulating hypotheses and testing them. So it will be true because we will make it true and any justification will depend on whether we find it useful in practice.


There are (at least) two different (but related) meanings of the word “certain”. One is psychological and is used when someone feels certain. It is used of a state of mind and known not on the basis of evidence and argument but by introspection, so there is no doubt of the feeling of certainty; unfortunately, that is no guarantee of the truth of what one is certain about. (If I feel certain that it is raining, you cannot be certain that it is raining.) The other sense is objective and applicable standards depend on the context or language-game of the proposition you are talking about. It is a guarantee of truth, but is not attributed to any particular person. Naturally, there are sceptical arguments that this kind of certainty is never or rarely found in life and the reply is that the sceptic is applying inappropriate standards.

Can we be certain of our experience?

If you are talking about introspective experience, even sceptics have to accept that we are each a final authority on our experience, in the sense that we are aware of our own experience. But if you mean experience in the standard sense, in which we experience something when we perceive it and such experiences are a basis for claims about the objective world, there is normally room for a doubt, provided there is a possibility of resolving it. If there is nothing that could resolve the doubt, it doesn’t make sense.

Further reading:- Certainty, Epistemic Contextualism, Language game (philosophy)


It's called Hume's fork. Everyone agrees that we cannot be "certain" the sun will rise tomorrow, and Hume's argument is not just mindless skepticism.

All reasonings may be divided into two kinds, namely, demonstrative reasoning, or that concerning relations of ideas, and moral reasoning, or that concerning matter of fact and existence. (E. 4.2.18)

Hume considers the possibility of each of these types of reasoning in turn, and in each case argues that it is impossible for it to supply an argument for the Uniformity Principle

It may be that our intuitive belief in the uniformity principle is stronger than Hume's argument, but that would be question begging.

  • it may be self refuting, i don't know, but that's how i read what was confusing you in your question
    – user63756
    Dec 11, 2022 at 8:22

I find Kant's 'answer to Hume' a satisfactory resolution. Immanuel Kant's answer to Hume's skepticism about causality was to argue that the concept of causality is not derived from experience, but is instead a necessary a priori principle imposed by the mind on experience. In other words, our experience of the world is organized and understood in terms of cause and effect, not the other way around - the concept of causation is not derived from experience, but is a pre-requisite of making sense of experience. So, yes, 'the problem of induction' is indeed self-refuting - just as Kant demonstrated.

  • I think you're just going to further confuse them by implying that they are not begging the question.
    – user63756
    Dec 11, 2022 at 7:17
  • explaining in what sense the question is being begged is appropriate.
    – Wayfarer
    Dec 11, 2022 at 7:38
  • well i'm not sure that denying synthetic a priori judgments is "self refuting", do you have a reference for it?
    – user63756
    Dec 11, 2022 at 7:39
  • "It seems that whether the denial of a judgment involves self-contradiction is to be determined by whether the predicate of that judgment goes beyond the subject of it... In the case of synthetical judgments the predicate goes beyond the subject."
    – user63756
    Dec 11, 2022 at 7:58

The point is that deductive reasoning is pretty neat. So if you have "true knowledge" about an issue (it is true and you know about it and know that it is really, actually true) idk something like "All men are mortal". And you come across an example like "Socrates is a man (element of men)" then you can deduce "Socrates is mortal".

...Seriously you don't have to kill him to find out if he's mortal and then deal with the corpse and the bloody mess and the fact that you haven't researched a revive option yet and so he's dead for good or, you know, deal with the philosophical mess of whether reviving a dead person means he was actually dead and whether that means that you had proven mortality to begin with... None of that nonsense, if you know for a certainty that all men are mortal then knowing Socrates is a man is sufficient to prove certainly, 1000%, factual: he's mortal.

So knowing general principles let's you assess specific situations. Seriously it's awesome, you can stack these information nicely on top of each other because if they are true and the logical reasoning is true (which is actually pretty straight forward, you just need to follow certain patterns), than the outcome MUST also be true. Like yeah it's a little mathy, but for your effort you'd be rewarded with deep knowledge that lets you understand, explain and predict everything or it least it would have the potential to do so. And I mean REALLY understand, explain and predict something! No educated guesses or bullshit like that, but what your magic mind computes would be the real deal!! Isn't that awesome?

There's just one problem with that and that is that we don't have this "true knowledge" thingy.

And like ... not even a little. We don't have it, we don't know how to get it and we're not even sure we could identify it when we see it.

Now sure you can go down the rabbit hole of mathematicians, religious people and philosophers and just pretend or believe that you have "true knowledge" like idk test it a few times until you're convinced it works and then just run with it "if you are truly convinced of and/or believe in an idea that can't be wrong, right"? But all that means is that you live in a phantasy world where everything is neat and organized, but which only roughly or not at all translates to reality. While mathematicians at least realize that they're operating in phantasy worlds.

Or you could try what other philosophers and scientists have attempted and look for patterns, infer rules pretend these rules are true and see where that would lead you to and when the outcome is different from the expectation, well back to square one, analyze the new pattern, infer new rules ... and so on you know the drill. Though abandoning the concept of "finding the truth" and be acutely aware that everything is just an assumption and that every result comes with margins of error and uncertainty intervals.

But either way our "knowledge" that we presuppose or obtain from experience is either way the result of an induction and induction is an assumption, not a certainty. And that's not a value judgement in terms of it's usefulness or it's accuracy it just means that it cannot proof that it is reliable in ALL possible or even impossible situations.

So the problem of induction is that you'd need to find true knowledge that is based on assumptions. And I'm not sure how you're attempt to erode even the perceptual method to obtain information would solve that problem and not further cement it.


Statistically speaking, induction csn be stated as the past is representative of the future.


The past is representative of the future

This is the circularity Hume is referring to in his problem of induction. The "argument" for induction is deductive, but is a petitio principii

Statistically speaking, again, there's, if memory serves, this notion of representative sample ( a subset of the population) that should give a fairly accurate idea of the entire population. Is there one for induction?

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .