For the last twenty years, off and on, I’ve been writing papers on interest-relativity in epistemology. For the last six years, off and on, I’ve been trying to wrangle my various thoughts and writings on the subject into a book. And at very long last, I’ve finished a draft of it.
The main aim of this post is to say, Hey, there’s a draft of a book I’ve been working on for a while, go check it out! But that wouldn’t be much of a post, so I’ll also say a bit about what’s in the book, and a bit about the software I used to make the draft.
Interest-relative theories say that what one knows depends on what question one is interested in at a time. Proper inquiry, as the Nyāya philosophers stressed, starts with knowledge. But different inquiries have different appropriate starting points. When I’m thinking about what to cover in lectures next term, I start with the knowledge that I’ll be teaching 400-level Groups and Choices. When I’m thinking about whether to renew my life insurance policy, I don’t even start with the knowledge that I’ll be alive next term; if I knew that, it would clearly not make sense to buy life insurance. Those differences in interests, which in the book I equate with differences in inquiries one is undertaking, matter for what one can properly take for granted, and hence with what one knows.
So far, this is more or less common ground among the (dozen or so) philosophers who have endorsed interest-relative theories. But my particular version differs in a few ways from what some of them think, and from what I used to think, and (especially) from what critics of interest-relativity think that we think.
In my version of interest-relative epistemology, both practical and theoretical interests matter to knowledge. One of the works that launched this subject was Jason Stanley’s Knowledge and Practical Interests. I’m already off the boat with the title; I don’t think practical interests have any special connection to knowledge. One knows the same things when engaged in the practical inquiry, What do I do now? as when engaged in the theoretical inquiry If I were in that situation, what would I do?, where ‘that’ picks out my actual situation.
My version of interest-relative epistemology gives no special role to high stakes situations. What it does give a special role to are long odds situations. Changes of interests typically cause one to lose knowledge when they mean that one moves to a situation of considering a bet on something one previously knew at extremely unfavorable odds. Declining life insurance, for instance, is betting that one will stay alive at rather long odds. High stakes situations are often long odds situations, so this matters less in practice than in theory. But theoretically, stakes are irrelevant.
When one knows something, that means it is a proper place to start inquiry. (Or, more precisely, that starting inquiry with it cannot be a mistake for a particular reason.) I used to think that when one knows something, that means that inquiries that start with it will end in the right place. I now think that was back-to-front. That an inquiry which started at one spot would end at the wrong spot is evidence that it was the wrong place to start. But it’s not the only reason the start might be wrong. For example, an inquiry that started in some spot might get to the right result for the wrong reason, or might say that a conclusion is safely true when it is not. And what matters for knowledge is whether the inquiries one is engaged in permit starting at a particular spot, not that they permit ending at some spot.
When one is engaged in practical inquiry, it helps to remember that one is human, and one is engaged in inquiries that are suitable for a human. Like a good mathematical social scientist, I used to equate inquiries into what to do with inquiries into what would maximise expected utility. But that’s not true in general; it is only true for creatures who have zero computational costs. And humans are not creatures like that. The theory of knowledge should integrate nicely with the theory of non-ideal decision theory. And this turns out to be useful, because the theory I’d been developing all these years turns out to play more nicely with non-ideal decision theory than with ideal decision theory.
Finally, and this is perhaps the thing I’m least sure about in the book, on my view everything in epistemology is interest-relative, but it’s all interest-relative in slightly different ways. So I have theory-sketches in the book of how knowledge, evidence, and rational belief, are all interest-relative, and how the interest-relativity of each can come apart. This seems messy in a way that I think good theories are a bit cleaner, and I suspect that it’s here I’m most likely to look back and wish I’d done things a bit differently. But the arguments seem convincing for now, and maybe this is somewhere that we just need some mess.
What I’d originally wanted to do in the book, but it’s clear after all this time that I’m not going to be able to, is connect this theory to the history of epistemology, and more broadly to the history of science. I think mainstream Anglophone epistemology is historically in a fairly weird place. It endorses all three of the following views:
- Knowledge is important; what one knows makes a difference to what one can rationally do and think.
- Knowledge is sub-optimal; one does not have to be in an optimal epistemic position to know something, i.e., it’s possible to know something and still be capable of improving one’s epistemic position.
- Knowledge is interest-invariant.
Much of the book is arguing that nothing could satisfy all three constraints. Philosophers who accept this and respond by weakening 1 are, I think, tacitly assuming that something else (maybe evidence, maybe certainty, maybe higher-order knowledge) can satisfy all the constraints. But in fact nothing can. That’s something I do argue for at some length.
What I suspect, but don’t really have the receipts for, is that it’s really weird to have any notion in one’s epistemology that satisfies all the constraints. My picture of the history of epistemology is that people realised that any important notion would either be optimal (something like scientia), or interest-relative (something like moral certainty). And it’s only with the invention of the printing press, and then later with the invention of the scientific journal, that we even had a need for something that met all three constraints. Because when one starts a scientific journal, it’s plausible that one needs a concept PUBLISHABLE that satisfies all the constraints.
- Being publishable is important.
- Being publishable is sub-optimal.
- Being publishable is interest-invariant.
You buy 1 because you’re a journal editor, and you don’t think you make arbitrary decisions. You buy 2 because you’re a journal editor, and you want to publish things, and literally nothing you get is completely beyond all doubt. And you buy 3 because you’re sending your journal to people you don’t know, and so you don’t know what interests they have, and so you need standards for publication that are independent of the (unknown) interests of the readers. So now you need a concept that satisfies all 3 constraints. The unfortunate fact that no concept does can’t be admitted into discussion, because then you’d have to shut down the journal. So you try to develop concepts that do. And that’s how recent Anglophone epistemology is born.
The attentive reader might notice that the last couple of paragraphs features zero (0) bits of actual historical evidence. I really don’t have the receipts for this picture at all. I think starting with the discussion in Robert Pasnau’s amazing book After Certainty, and tracing back through the original sources he discusses, could be a way to make a start on that. And it would be even better to compare those sources to what you find in South Asian traditions, especially those that think 1-3 can be satisfied as long as you’re particularly careful about how you understand optimality. But I’ve regretfully decided that I can’t put off publishing my book until after I actually carry that research out. And I suspect carrying it out will need rather different skills to the ones I have, or am likely to acquire.
I wrote the book in Quarto, and that means that it has lots of nice features. What I’m thinking of as the main version of the book is readable in-browser, with all the advantages that in-browser reading involves. So you can change the font size, use a screen reader, and adjust the colors or fonts. The little switch in the top left changes the screen from light to dark, and the search bar lets you search the full text of the book. By right clicking anywhere in the book, you can leave comments or annotations - though I’ll turn these off if it turns out to be a vector for spam. It is easy to click through the menus to move between the book. And the footnotes in the right hand margin appear just when you’re at the relevant part of the book. It’s all much more pleasant I find than reading documents formatted for print, and I hope more work is published this way in the future.
The book can also be downloaded in PDF, Word, or Kindle format. (Though there are formatting quirks in each of these, particularly involving some of the diacritics.) And there is a GitHub repository, if you’re interested in either leaving an issue there, or seeing the code as a jumping off point for your own Quarto project. I don’t think my design choices are particularly good, but I always find it useful to see the code of other projects to see what choices are possible.
There is still a bit of LaTeX guff at various points, because I haven’t figured out how to do exactly everything in Quarto. But it is getting to be a mature enough platform that it is just about possible to produce nice looking PDFs without having to learn much LaTeX code. And that’s going to be really valuable for young academics, who won’t have to climb LaTeX’s steep learning curve.