The Principle of Inferential Justification
Here, I explain what’s wrong with the Principle of Inferential Justification (as defended by Richard Fumerton).*
(In case you haven’t noticed this, every other week, I post a summary of one of my academic papers, like this one. I’m working through them chronologically.)
1. The PIJ
The Principle of Inferential Justification says:
PIJ To be justified in believing P on the basis of E, you must be (i) justified in believing E, and (ii) justified in believing that E makes it probable that P.
Part (i) of that is uncontroversial. Obviously you can’t be justified in believing a conclusion on the basis of an unjustified premise. Part (ii), however, is controversial.
2. Motivation for PIJ
The motivation for (ii) is the idea that you have to see how an inference is cogent in order to be justified in making it.
Example: Say that you talk to an astrologer, who tells you that there’s going to be prosperity next year because Jupiter will be aligned with Mars. You think her conclusion is unjustified. But it’s not because you doubt that Jupiter will be aligned with Mars. It’s because you doubt that Jupiter’s being aligned with Mars makes it likely that there will be prosperity. You ask the astrologer for evidence of that. When she fails to provide any good evidence for that, you conclude that she wasn’t justified in inferring that there will be prosperity next year.
This looks like a good case for (ii) in PIJ: She’s not justified in her conclusion because she’s not justified in believing that her premise makes her conclusion probable.
Nevertheless, I’m going to argue that the PIJ is a conceptual mistake.
3. Level Confusions
A level confusion is a confusion between, e.g., knowing that P and knowing that one knows that P, or between justification for believing P and justification for believing that one is justified in believing P, etc.
The PIJ rests on a level confusion. Properly understood, claims about probability are claims about justification: the probability of a proposition is a measure of how much justification we have for believing it. To say something is probable is to say there is more evidence for it than against; to say something is certain is to say we have conclusive justification for it; etc.
(There are other interpretations of probability, but they don’t fit in this context – the PIJ would not be plausible using a frequency, propensity, or subjective interpretation of probability.)
So to say “E makes probable P” is just to say that E provides justification for P.
So PIJ really says: E justifies P (for you) only if: (i) you’re justified in believing E, and (ii) you’re justified in believing that E justifies P.
The second part is a level confusion. It’s very close to saying that to be justified in believing something, you have to be justified in believing that you’re justified in believing it, which leads to an infinite regress.
4. PIJ Imposes a Contradictory Demand
Allegedly, in order to be justified in believing P on the basis of E, you have to be justified in believing something else besides E: namely, [E makes probable P]. So your justified belief in [E makes probable P] is part of what makes P justified. But that means that P isn’t really justified (solely) on the basis of E; rather, it’s justified on the basis of the more complex proposition, [E & E makes probable P].
So PIJ says that E constitutes your justification for P only if E doesn’t really constitute your justification for P. So PIJ implies that inferential justification is impossible.
5. Explaining the Example
What about the astrologer example from section 2? How do we explain that?
Well, first, [Jupiter will be aligned with Mars] simply does not in fact (by itself) make it probable that there will be prosperity next year. So if that was the astrologer’s only premise, then the conclusion isn’t justified.
In order for you to be justified in believing P based on E, it’s not that you have to be justified in believing [E makes probable P]. It’s that it has to be in fact true that E makes probable P.
Second, the astrologer probably didn’t really infer [there will be prosperity] from [Jupiter will be aligned with Mars] alone. Rather, the astrologer probably has some other background beliefs that are (implicitly) part of her premises. E.g., maybe she thinks that when Jupiter aligns with Mars, this causes people to work harder.
In that case, the problem would be that the background beliefs are unjustified. In other words, the problem is that (part of) E is unjustified, not that [E makes probable P] is unjustified.
So instead of PIJ, we should say: To be justified in believing P on the basis of E, you need to be justified in believing E, and it has to be true that E makes probable P.
* Based on “Fumerton’s Principle of Inferential Justification,” Journal of Philosophical Research 27 (2002): 329-40.
Postscript: What I Didn’t Account For
So the above is what I said in 2002. Later, I realized that I hadn’t adequately addressed all of the motivation for PIJ. I didn’t account for cases like this:
Suppose that I justifiably believe E. Suppose that E in fact supports P, but I myself am unable to see how. (Maybe it requires a complex series of steps to get from E to P, and I don’t see those steps.) Nevertheless, I somehow just infer P from E anyway. Stipulate that I don’t have any reason for doubting that E supports P; I simply don’t see how it does.
In this case, it seems that my belief in P is unjustified. PIJ explains that. My own account didn’t.
Much later (2016), I gave a better account of how inference works in my “Inferential Appearances”.