Induction here is considered in light of the modern view, which is the practice of inferring from particulars to generals. Hume believes that such inference is very problematic since it holds two things not necessarily related (in the sense that such a regularity shall not always occur) to be related. In such an analysis of induction there seems to be presupposed a certain inference of a different kind. For example, take the regularity of 'event B' following upon 'event A'. It seems that we must question the relationship we have to the events themselves, for it seems that our experience of 'event A' and our experience itself must be causally related. This is said because Hume's criticism of induction is apparently that we can think about and conceive of the events in question independent of each other, and so we needn't think that the two are necessarily related; but the same cannot be said of our experience and that which is experienced (the events themselves). This is because if try to think or conceive of such an event independent of our experience, such is an impossibility (in that we only conceive of such an event through the experience of it). We would be left with the ludicrous position of conceiving of that which we have yet to conceive of. Furthermore, if we try to conceive of our experience independent of that which is experienced, such also results in self-contradiction, since we know of our experience only in the experience of that which is experienced. So it seems that the presupposition of induction of any kind, at least according to the widely accepted Humean analysis of it, is that we can conceive of such events and terms in simpler or distinct ways. But oddly enough, this very point seems to indicate that we are in fact causally related to the world, through our experience of it.

Now more could be said of Hume's analysis, such as the non sequitur that Hume seems to get away with in believing that simply because we can conceive of two things independent from the other, it follows that the regularity in question is not explained in the context existent by some contingently necessary relation between the two things in question. So it seems that Hume's analysis is altogether merely descriptive rather than explanatory. But this is beside the point. The question to be answered is regarding whether or not the typical connotation of 'induction' that the modern world is accustomed to really does presuppose more necessary inferences into its terms, and if this indicates causality itself, at least between our experience and the events experienced.

  • Let's say causation from regularity by induction does presuppose necessary inferences, in fact this is how Kant supplemented Hume. Such multilayered picture can be and is used to give more nuanced accounts of induction. But unless we take them to be innate a priori that universally apply to all experience these presuppositions also derive from regularities of said experience, and hence have no more justification value than what they purport to justify. Without a strong argument for secure non-empirical origin of these presuppositions their presence does not affect Hume's analysis.
    – Conifold
    Oct 27, 2015 at 0:37
  • 1
    Where is the question exactly? Also where is "here" that you mention in the first sentence?
    – virmaior
    Oct 27, 2015 at 2:36
  • @Conifold I think you misinterpret the argument. The argument is not meant to argue for a realistic a priori knowledge, and it doesn't assume that 'innate' knowledge is even feasible. Rather, its purpose is to illustrate that any talk of Humean induction ultimately must be founded in some causal relation by which the terms are related. This is because any induction is said about two or more terms that simpler things can be said about (or terms that can be conceived of distinctly). But as such, induction must assume that we in fact know simpler things about the terms in question.
    – Chosen One
    Oct 27, 2015 at 12:24
  • But as for experience, we cannot possibly say anything simpler about it without reference to that which is experienced. And for that which is experienced, we cannot say anything simpler without assuming that we experience it first. Thus by Hume's own standard there shouldn't be an inductive relation between our experience and that which is experienced, but Hume holds that there is. Thus, there is in fact a contradiction involved.
    – Chosen One
    Oct 27, 2015 at 12:25

1 Answer 1


I'm a bit lost in some of your argument, but will unwarily speak my piece.

First, as Conifold has indicated, you see to be anticipating some of Kant's arguments by alluding to "necessary inferences." But I don't think it is quite accurate to say that Hume has a "problem with induction." Induction works and is, by definition, "mostly" correct. Hume was only pointing out that certain "necessary truths" were far more inductive than people had suspected. The warrant of Newton mechanics was by no means absolute, as proved to be the case. It's just that the "laws" of science, however reliable, do not get us beyond induction.

It has been pointed out that Hume seems inconsistent in arguing against necessary "causality" in observed phenomena, while apparently accepting some Lockean object "I-know-not-what" that "causes" our experience of that object. I believe this is what you are saying, partly.

But here again, Hume is only pointing to apparent regularities. He cannot be arguing that the "I-know-not-what" necessarily "causes" our experiences in some incorrigible correlation. How would we know? As with induction in natural sciences, such a necessary "correlation" behind our experiences cannot be observed or proved. Hume does not think all of this is a big problem. It is the overextension of knowledge on the basis of unwarranted certainties that are a problem, as the financial industry would do well the remember.

I hasten to add that I am not a scholar of Hume or anything else, and I may have misunderstood your question. I hope others will intervene if so. In sum, there is no big problem of induction, as long as you know that is what you are doing. Science has come to agree with Hume (or with Kant--they are in some ways more similar than not), that "causality" is our mathematical determination of probabilities to be formulated according to experience on a case-by-case basis. Not a universal glue. Only a "necessary" prop or condition of being what we are.

You must log in to answer this question.

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