2

Unless a statement is logically necessarily true or false, are we ever justified to make an inference on anything?

No matter what one uses or what reasons one comes up with to justify X, one can further ask for reasons to justify that ad infinitum. Sooner or later, you either have to have a foundational unjustified belief or some form of circularity or whatever else philosophers have conjured up to address.

My question is what is so specially problematic about induction that isn’t problematic within the concept of justification in general? Sure, the past doesn’t necessarily imply the future.

But we have no reason to justify believing that the world is a simulation vs. not either, apart from just assuming that this world is all there is. We have no reason to justify that there is no invisible demon in front of us breathing undetectable fire towards us just because we can’t observe it. And so on and so forth.

It seems that induction is just a specific form of the more general problem that almost none of our beliefs, apart from perhaps the belief that we are conscious, can be justified.

Note that changing justification to mean probabilistic justification does not escape this problem. One must first define what it means, in an ontological sense, for evidence to support a certain hypothesis with a degree of 99% probability for example. This must be first defined and then justified which as far as I’m aware, no philosopher has managed to do so without resorting to circularity again.

So to summarize, is the problem of induction really just a problem of justification?

Improve this question
3
  • 1
    No. Usually, we, at least, know what the mode of justification is supposed to be, even if making good on it with certainty (or at all) is beyond our means ("undetectable" demons could be detected should we acquire requisite spiritual senses, for example). The problem of induction is a meta-problem, it is about justifying a mode of justification that we could otherwise use to justify other claims. However, in this case, that would be circular, and deductive justification "from first principles" does not apply at all. Not to mention non-justificational aspects, like Goodman's "new riddle".
    – Conifold
    Commented Aug 29, 2023 at 0:13
  • Where is the justification for "one must define what it means, in an ontological sense, for evidence to support a certain hypothesis with a 99%-or-other probability"? If one has rigged the language game of justification to meta-obliviate the distinction between justified and unjustified anyway, why even play that language game at all (why use the word "justification" when justification is supposedly unidentifiable)?
    – Kristian Berry
    Commented Aug 29, 2023 at 17:00
  • Accepting a field means accepting its limitations. Athletes can't fly, economists can't pick the next big stock, Philosophers can't... agree on some things... We don't have to call those things 'problems '.
    – Scott Rowe
    Commented Oct 29, 2023 at 22:35

2 Answers 2

Reset to default
1

"[ ... ] One must first define what it means, in an ontological sense, for evidence to support a certain hypothesis with a degree of 99% probability for example[ ... ]" ~ thinkingman

I might be able to help you with that. It puzzled me too at one point. Look at the inductive argument below.

Argument A
90% of dogs are friendly
Timmy is a (random) dog
Ergo,
There's a 90% probability that Timmy is friendly.

Ask yourself, what is the probability that some random dog is friendly, given that you know 90% of dogs are friendly?

This is empirical probability based on frequency.

As for ontology, all I can do is refer you to propositions, what are they?

Improve this answer
4
  • The problem is there’s no way to justify that Timmy is actually representative.
    – user62907
    Commented Aug 29, 2023 at 6:16
  • 2
    Timmy hasta be random ... I suppose.
    – Hudjefa
    Commented Aug 29, 2023 at 6:31
  • @thinkingman It's hard for a single individual to be representative of a distribution, like Timmy would need to be friendly 90% of the time or something like that. It's like complaining that a single dice is not representative of the distribution of a fair dice throw as it will show you 1 number and not a 1/6 probability.
    – haxor789
    Commented Sep 28, 2023 at 9:37
  • 1
    OK ... Ok .. ok.
    – Hudjefa
    Commented Sep 28, 2023 at 10:57
1

Unless a statement is logically necessarily true or false, are we ever justified to make an inference on anything?

What do you mean by justified? Like if you mean, infer with certainty, then no you can't do that. If you just mean give any reason for why you think that is a good heuristic, well sure there are methods to distinguish a guess and an educated guess, though neither claims certainty both are aware of ultimately being a guess.

The problem with the problem of induction is more or less that outside of the safe space of logic where we just set things to be axiomatically true and some trivial edge cases where we can check all possible options, pretty much everything else is subject to the induction and thus effected by that problem of induction.

So yeah even if we make up criteria like if something happens x times out of 100 times, it is likely going to keep up that ratio, is ultimately an assumption and we could just have witnessed an edge case so far. So they come with the implicit statement of "if our observations and assumptions so far hold: ...".

So if the problem of justification means that you can't justify with certainty that induction provides a certainty, then yes that is what the problem of induction says.

And if the problem of justification means that you can't find a motivation for applying induction and ranking it's results, then no you can make educated guess and that education is also an educated guess. The justification comes after the fact by whether the guess works or not.

Improve this answer

You must log in to answer this question.