It’s possible that, in my earlier posts (here and here), I run two distinct things together: (i) the “takings” posited by the Taking Condition and (ii) the belief (in the best case, the knowledge)––acquired in any self-conscious inference––that you’ve come to know your conclusion by inferring it from your premises. If I have, that’s because it has seemed to me that the latter sort of belief could do the work that “takings,” otherwise understood, were supposed to do. But it now seems to me that the kind of thing a “taking” is usually* understood to be is importantly distinct from the kind of self-knowledge (knowledge of your own inferences) that I’ve primarily been interested in. So I want to use this post to try to clear things up a bit.
In his book Self-Consciousness, Sebastian Rödl suggests that self-consciousness can be understood as a capacity to know that you’re in a state by being in it. And I think we can add that it’s equally a capacity to know that you’ve performed an act––e.g., an inference––by performing it. As I understand it, the idea here is that self-consciousness is a basic way of knowing, on a par with, for example, perception (and inference itself). Employing this conception of self-consciousness, we can then define an epistemically successful self-conscious act (of, for example, perceiving or inferring) as an act in which you both come to know (say) that p and, in the very same act, come to know that you’ve come to know that p, in whatever particular way you have (for example, by seeing that p) precisely by coming to know that p in that particular way) And we can define a self-conscious act generally as an act in which you both come to believe (say) that p and, in the very same act, come to believe that you’ve come to know that p, in whatever particular way you have, precisely by coming to believe that p in that particular way.
That’s a bit of a mouthful, of course. So it might help to put the point another way. An epistemically successful self-conscious inference, on this conception, is an inference in which you come to know that you’ve come to know (say) that p by inferring it from <q> and <r>. How do you come to know this? Well, by inferring <p> from <q> and <r>. That is, by coming to know that p on the basis of your prior knowledge both that q and that r. Similarly, if you didn’t come to know that p by inferring it from <q> and <r>––because, say, you didn’t know that q––then you’ll merely believe that you’ve come to know that p by inferring it from <q> and <r>.
My central interest, in thinking about inference, has been in (i) the problem of causal deviance and (ii) the nature of self-conscious inferences. And, in this context, what’s interesting is that, if we take up Rödl’s general conception of self-consciousness as a capacity to know that you’re in a state (that you’ve performed an act) by being in it (by performing it), we get a pretty neat account of self-conscious inferences. And it’s one that, it seems to me, doesn’t need the Taking Condition (its first clause, in particular) in order either to solve the problem of causal deviance or to explain the nature of self-conscious inferences.
Of course, it might be that there are other problems we need the Taking Condition to solve. (If so, what are they?) What’s interesting, though, is that it now looks to me like the issues surrounding the Taking Condition might be wholly independent of the issues concerning the nature of self-conscious inferences. So while I’ve been concerned, in the past, to stress the fact that my account seems to me to obviate the need to introduce the Taking Condition, it now strikes me as equally important to stress that, strictly speaking, my account is (or now seems to me to be) compatible with the Taking Condition. (More carefully: I’m not sure that my account isn’t compatible with the Taking Condition.)
I still suspect that there are problems with the Taking Condition, problems connected with the fact that it seems to require that “takings” be somehow constitutive of the spontaneity of inference, in something reasonably close to Kant’s sense (I discussed these issues briefly in my most recent post). But for now I’ll set those problems aside. I’m not yet sure how to articuate them anyway, other than by saying, as I did about Kant’s notion of spontaneity, that I’m not really even sure what it is that I’m supposed to be arguing against. Instead, I just want to indicate why I think that my account of self-conscious inferences might actually be compatible with the Taking Condition.**
To see this, think of what happens when you “see,” say, that <q> and <r> together support <p>. I argued in my earlier two posts (again, here and here) that to “see” such a thing is to come to know that someone who knows both that q and that r is in a position to come to know that p by inferring it from <q> and <r>. But, importantly, you can come to know such a thing without yourself inferring <p> from <q> and <r>, even if you yourself believe (even know) both that q and that r (and even if you know that you do). It may be true, as I argued, that our understanding of “takings” is thus parasitic on our understanding of (epistemically successful) inferences (i.e., that we can only explain “takings” in terms of inferences). But that’s only a problem for those philosophers who want to use “takings,” and the Taking Condition, in a reductive account of inference; and not everyone wants to do that. (Those who aim to provide non-reductive accounts of inference, in the relevant sense, include Ulf Hlobil and Eric Marcus.)
So we might offer the following account of “takings”: to take <q> and <r> to support <p> is to believe that someone who knows both that q and that r could come to know that p by inferring it from <q> and <r>. Then what the Taking Condition says is that, when you infer <p> from <q> and <r>, this belief is both a cause and a ground of your belief that p. (Again, arguably, it can’t be a cause and a ground in the way that the premise-beliefs are, on pain of regress and––I’ve claimed––related problems. But defenders of the Taking Condition claim to have a way of escaping this objection, so I’ll continue to ignore it.) And they think this, I think, in large part because it seems to them that otherwise you would draw your conclusion blindly, in a sense that strikes them as deeply problematic. (I do mean my rhetoric here to indicate that I don’t share these particular worries.)
Okay, I hope that was reasonably clear. The point, at any rate, is just that (i) I haven’t provided any kind of argument against this sort of view and (ii) it doesn’t seem to me that I need to reject it, in order to say the things I want to say about the nature of self-conscious inferences. To see (ii), notice that, even if you know (that you know) that q, that (you know that) r, and that someone who knows both that q and that r could come to know that p by inferring it from <q> and <r>, you can’t know that you’ve come to know that p by inferring it from <q> and <r> until you’ve actually performed the inference. And so even if the Taking Condition, as we’re now understanding it, is true, we still need some account of your knowledge that you’ve come to know that p by inferring it from <q> and <r>. And my Rödl-inspired account of self-conscious inference seems like an attractive account of that.
Now, at this point, there’s a lot to say, and I shouldn’t try to say it all here. But just a few observations.
First, it seems plausible that your “taking”––that is, now, your belief that someone who knows both that q and that r could come to know that p by inferring it from <q> and <r>––will play some role in grounding your belief that you’ve come to know that p by inferring it from <q> and <r>. The idea, I suppose, is that you have this general belief––that anyone who knows these premises, etc.––and you simply apply it to yourself. So you perform the inference because you believe that doing so will result in new knowledge, and, once you’ve performed the inference, you believe that you know the conclusion because you believe that inferring it from these premises is, if the premises are known, a way of coming to know the conclusion. (As I said, this is all plausible. But I don’t think it’s right. Saying why, however, is a task for another day.)
Second, although the present rendering of the Taking Condition and my Rödl-inspired account of self-conscious inferences are, I think, strictly compatible, they do seem to point in opposite directions. Partly this is just because each of them seems to make the other unnecessary. But it’s also because, if we accept the present account of the Taking Condition, my Rödl-inspired account of self-conscious inferences is likely to look confused. (Maybe the converse is also true. That would certainly explain why I’m still not inclined to accept the Taking Condition, even on the present rendering of it, despite thinking that it’s strictly compatible with my account of self-conscious inferences.) This is closely connected with the first point. The issue here is that, if you know (that you know) that q, that (you know that) r, and that someone who knows both that q and that r can come to know that p by inferring it from <q> and <r>, all you seem to need, in order to come to know that you’ve come to know that p by inferring it from <q> and <r> (and thus in order to meet the conditions for performing a self-conscious inference) is to infer <p> from <q> and <r>. No appeal to self-consciousness as a basic way of knowing is required. (Whether that’s really right is another issue for another day. For now, I’ll just say that I have my doubts.)
Finally, one thing that definitely seems right to me is that “takings,” on the understanding of them explained here, are genuinely distinct from inferences in a sense that would allow them to be both causally and epistemically prior to inferences (i.e., to the formation of the conclusion-belief). For you can clearly know that someone who knows both that q and that r could come to know that p by inferring it from <q> and <r> without yourself inferring <p> from <q> and <r>, even if you know both that q and that r, and know that you do. What’s less clear to me is that the inferences in which such “takings” are causally and epistemically prior to the formation of the conclusion-belief are going to be inferences of the form <q; r; therefore, p> rather than inferences of the form <q; r; <q> and <r> support <p>; therefore, p>. That said, I think a new option is coming into view here: the defender of the Taking Condition might be able to say that “takings” are additional premise-beliefs, but deny that this generates a regress. If this strategy is taken as part of a non-reductive account of inference, I don’t at present see any problem with it. (Well, okay, I think I do; but the problem hinges on my idiosyncratic account of belief. So, yet again, another day.)
But I’ll leave it there for now.
*There are exceptions. Some philosophers (in particular, Ram Neta [in “Basing as Conjuring”]) understand “takings” (as they figure in the Taking Condition) as being about the premise- and conclusion-beliefs involved in inferences, and about the causal connection between them, and not merely about relations of support between the propositions believed. So understood, I think, “takings” are not distinct from the kind of knowledge of inference that makes an inference self-conscious. And so I think that these philosophers misunderstand both the kind of knowledge that makes an inference self-conscious and the nature of the “takings” posited by the Taking Condition.
**It’s also relevant––though I’ll ignore this issue here––that the Taking Condition rules out the possibility of non-self-conscious inferences. In other words, it rules out the possibility of a non-self-conscious thinker (a non-rational animal) coming to know things on the basis of other things it already knows. But there’s a long history––crystallized in the famous example of Chrysippus’s dog, which apparently performs a disjunctive syllogism––of thinking that such a thing is possible. I take it to be a minor virtue of my account that it’s neutral on this issue. But, however that issue turns out, my account (it now seems to me) is compatible with the application of the Taking Condition to self-conscious inferences. And that’s really all that’s at stake in this post.