No Virtues
I object to the idea that we do and should decide what philosophical positions to adopt by amalgamating our opinions about the various theories' virtues like empirical adequacy, simplicity, explanatory power, fruitfulness, consistency, etc. Ever since Kuhn's work (at least) this has been a popular idea about how scientists do or should go about deciding what theory to adopt, and my impression is that many in our field propose we adopt a similar practice in philosophy. (This idea has appeared in print a few times, but I don't want to make this about disagreeing with particular people -- I hear the idea informally quite a lot, and so I am responding to something that I take to be in the air and which I object to quite generally, rather than in any particular spelling out.) I object to this on both the individual and social level -- I don't think each or any of us do or should do this, and nor do I think we should collectively do this when making joint choices. Likewise some of these objections may well transfer to the case of science, but since on this blog I mainly do metaphilosophy I am not going to focus on that here.
There are some objections to the idea of theory choice by virtue amalgamation which, while I am sympathetic with, are not what I mean to object. First, the Okasha objection, that in general amalgamating opinions about stuff is very difficult and it looks like the same will go in this domain. Second, the Novick objection (Novick politely demurs and points out that this general idea is not original to him -- he's the person I know it from, though), that it may be that possessing those virtues is only truth conducive given domain specific facts about stuff natural scientists are interested in, and this will not transfer. (As applied specifically to the virtue of simplicity, I think lots of people have this worry even within the sciences. Sometimes people treat aesthetic properties as a virtue, and the same might go there.) While I broadly agree with both of these points, I am not going to focus on them here either.
Instead, I mean to press two points:
(1) For many of the virtues in question there is no agreed upon way of seeing whether they apply to a given theory or making comparisons among options. This goes for, say, explanatory power, simplicity, maybe empirical adequacy depending on how deep down the statistical method rabbit hole one goes. (Note that the disambiuguated versions of these virtues often run into the problems mentioned next, e.g. theories of explanation which require me to know what the consequences of my theory are to know what it explains will also run into the problems I outline below.) In so far as theory choice is meant to help us come to shared agreement about what is best and make a kind of epistemically responsible communal progress in discovering the true or best theory, then the fact that the virtues are as controversial as the theories they are meant to appraise stands in the way of that.
For some of the virtues not only is there no agreed upon way of applying them, but even just personally I don't know how to tell which of theories have them to what degree, or possess them at all, or possess them more or less than their rivals. This goes for fruitfulness and some of the aesthetic virtues especially. How fruitful is virtue ethics? Is it more or less fruitful than Kantian deontology? (Which is more beautiful or elegant?) I have no idea, and for once in my life i am not inclined to think this is a special defect in myself.
And even for the virtues which are most straightforward (consistency) to check for them they often require that I know what counts as an entailment of the theory to apply, and given the vagueness and ambiguity common to philosophical theorising it's the case that for many philosophical theories I don't know what entails what. Note that here I think a special problem arises with philosophy. I think that we in philosophy are often working with a kind of `double squishiness' - there is, on the one hand, the squishiness of some of the virtues which makes it hard to say how they apply. But there is also the squishiness of the theories we produce -- it's not clear what they entail or are entailed by, or are inductively supported by and tend to confirm, or anything of this sort. Without having much to support this beyond a feeling: I think each squishiness compounds, I think that this attempt to apply virtues whose scope or meaning isn't quite clear to theories whose scope or meaning isn't quite clear just results in a mess, no serious project of appraisal of theories is carried out, because we're just not in a position to do anything like that.
So, at the level of applying the particular virtues to particular theories (or comparisons between theories) I don't buy that people are able to do it in any reliable way -- certainly not in a way that could reasonably be expected to put us on the sure path of a science by allowing us to make communal progress, but even (given the latter problems) not at the level of allowing me to just decide what respective balances of virtues are possessed by various theoretical options.
(2) I am impressed by the fact that nobody ever actually tells me how they are carrying out the amalgamation procedure. In fact, I suspect, they are not doing any serious amalgamation procedure at all (beyond respect for dominance, which in practice doesn't often arise). Here I mean by this -- I expect a method of reasoning to be something that could, if one wanted to, be explained and, if followed, come to broadly the same kind of conclusion at a rate better than chance. And so here I suspect that, first, nobody could explain how they are amalgamating the virtues, and, second, that past use of the amalgamation procedure is no guide at all to future behaviour.
On this latter I mean: on any given instance, one could treat somebody's actual choice of theory given the virtues they say it possess (and to what degree or how well it compared to others in possessing them) as consistent with some broad class of amalgamation procedures. I don't think one should predict that people's future attempts at amalgamation will fall within this class, people are not actually expressing any preference for an amalgamation procedure, because they're not really amalgamating at all but just plumping for a fave based on quite idiosyncratic factors. If on one occasion somebody makes a choice which only makes sense if they think that fruitfulness matters more than explanatory power, that is no reason to expect their future choices to respect that constraint -- they may well (both descriptively and normatively) in future decide something that only makes sense if explanatory power matters as much or more than fruitfulness.
So whether or not the virtues are truth conducive in the context of philosophical reasoning, I don't think we actually do, or in many cases could even if we wanted to, really apply the individual virtues. And then even supposing we did, I don't think we actually even attempt to carry out an amalgamation; which might in any case turn out to be impossible. In short, I think this is a very bad model of philosophical theory choice.
In so far as I think these virtues are playing a role at all, I think it is just as a list of things to discuss when talking about why you like your favourite theory. Having a convention that one discusses one's proposal in light of some stereotyped set of virtues could, I think, have various benefits, and I am not opposed to that. It's also not the case that I think the project of theory choice by virtues is just doomed -- I can even interpret some of my own past work as an attempt to make it easier to assess theories by their virtues. Maybe we could do that more broadly. But I think we are kidding ourselves if we think there is any serious method of theory choice or comparison that we currently do or could work with in philosophy.
(Sorry for the somewhat half-formedness of these thoughts. I wrote up some notes based on a conversation I had with Aaron Novick and Katie Creel (each of them should absolutely be held responsible for any and all errors in what was just said) -- since my sheer busy-ness prevents me from being able to dedicate too much time to blogging nowadays, I thought I'd use them for a post!)
There are some objections to the idea of theory choice by virtue amalgamation which, while I am sympathetic with, are not what I mean to object. First, the Okasha objection, that in general amalgamating opinions about stuff is very difficult and it looks like the same will go in this domain. Second, the Novick objection (Novick politely demurs and points out that this general idea is not original to him -- he's the person I know it from, though), that it may be that possessing those virtues is only truth conducive given domain specific facts about stuff natural scientists are interested in, and this will not transfer. (As applied specifically to the virtue of simplicity, I think lots of people have this worry even within the sciences. Sometimes people treat aesthetic properties as a virtue, and the same might go there.) While I broadly agree with both of these points, I am not going to focus on them here either.
Instead, I mean to press two points:
(1) For many of the virtues in question there is no agreed upon way of seeing whether they apply to a given theory or making comparisons among options. This goes for, say, explanatory power, simplicity, maybe empirical adequacy depending on how deep down the statistical method rabbit hole one goes. (Note that the disambiuguated versions of these virtues often run into the problems mentioned next, e.g. theories of explanation which require me to know what the consequences of my theory are to know what it explains will also run into the problems I outline below.) In so far as theory choice is meant to help us come to shared agreement about what is best and make a kind of epistemically responsible communal progress in discovering the true or best theory, then the fact that the virtues are as controversial as the theories they are meant to appraise stands in the way of that.
For some of the virtues not only is there no agreed upon way of applying them, but even just personally I don't know how to tell which of theories have them to what degree, or possess them at all, or possess them more or less than their rivals. This goes for fruitfulness and some of the aesthetic virtues especially. How fruitful is virtue ethics? Is it more or less fruitful than Kantian deontology? (Which is more beautiful or elegant?) I have no idea, and for once in my life i am not inclined to think this is a special defect in myself.
And even for the virtues which are most straightforward (consistency) to check for them they often require that I know what counts as an entailment of the theory to apply, and given the vagueness and ambiguity common to philosophical theorising it's the case that for many philosophical theories I don't know what entails what. Note that here I think a special problem arises with philosophy. I think that we in philosophy are often working with a kind of `double squishiness' - there is, on the one hand, the squishiness of some of the virtues which makes it hard to say how they apply. But there is also the squishiness of the theories we produce -- it's not clear what they entail or are entailed by, or are inductively supported by and tend to confirm, or anything of this sort. Without having much to support this beyond a feeling: I think each squishiness compounds, I think that this attempt to apply virtues whose scope or meaning isn't quite clear to theories whose scope or meaning isn't quite clear just results in a mess, no serious project of appraisal of theories is carried out, because we're just not in a position to do anything like that.
So, at the level of applying the particular virtues to particular theories (or comparisons between theories) I don't buy that people are able to do it in any reliable way -- certainly not in a way that could reasonably be expected to put us on the sure path of a science by allowing us to make communal progress, but even (given the latter problems) not at the level of allowing me to just decide what respective balances of virtues are possessed by various theoretical options.
(2) I am impressed by the fact that nobody ever actually tells me how they are carrying out the amalgamation procedure. In fact, I suspect, they are not doing any serious amalgamation procedure at all (beyond respect for dominance, which in practice doesn't often arise). Here I mean by this -- I expect a method of reasoning to be something that could, if one wanted to, be explained and, if followed, come to broadly the same kind of conclusion at a rate better than chance. And so here I suspect that, first, nobody could explain how they are amalgamating the virtues, and, second, that past use of the amalgamation procedure is no guide at all to future behaviour.
On this latter I mean: on any given instance, one could treat somebody's actual choice of theory given the virtues they say it possess (and to what degree or how well it compared to others in possessing them) as consistent with some broad class of amalgamation procedures. I don't think one should predict that people's future attempts at amalgamation will fall within this class, people are not actually expressing any preference for an amalgamation procedure, because they're not really amalgamating at all but just plumping for a fave based on quite idiosyncratic factors. If on one occasion somebody makes a choice which only makes sense if they think that fruitfulness matters more than explanatory power, that is no reason to expect their future choices to respect that constraint -- they may well (both descriptively and normatively) in future decide something that only makes sense if explanatory power matters as much or more than fruitfulness.
So whether or not the virtues are truth conducive in the context of philosophical reasoning, I don't think we actually do, or in many cases could even if we wanted to, really apply the individual virtues. And then even supposing we did, I don't think we actually even attempt to carry out an amalgamation; which might in any case turn out to be impossible. In short, I think this is a very bad model of philosophical theory choice.
In so far as I think these virtues are playing a role at all, I think it is just as a list of things to discuss when talking about why you like your favourite theory. Having a convention that one discusses one's proposal in light of some stereotyped set of virtues could, I think, have various benefits, and I am not opposed to that. It's also not the case that I think the project of theory choice by virtues is just doomed -- I can even interpret some of my own past work as an attempt to make it easier to assess theories by their virtues. Maybe we could do that more broadly. But I think we are kidding ourselves if we think there is any serious method of theory choice or comparison that we currently do or could work with in philosophy.
(Sorry for the somewhat half-formedness of these thoughts. I wrote up some notes based on a conversation I had with Aaron Novick and Katie Creel (each of them should absolutely be held responsible for any and all errors in what was just said) -- since my sheer busy-ness prevents me from being able to dedicate too much time to blogging nowadays, I thought I'd use them for a post!)
I agree that we don't choose philosophical theories by explicitly amalgamating their virtues. I wonder, though, what you think is going on in our heads when we choose one philosophical theory over another. What is the psychology of philosophical theory choice?
ReplyDeleteAh good question, and one to which I have no answer! Lol I was about to say ``I read an interesting paper on this once, check it out and see what you think: https://www.cmu.edu/dietrich/philosophy/docs/scheines/Philosophical%20temperament.pdf '' -- and then I reread the author list. So I guess instead I should just ask: what do you think the psychology is here?
DeleteHa! Thanks. I certainly don't think I know much about how we choose our philosophical theories. What I'm most curious about in the context of your post here is whether a good computational model of how we in fact make these choices might involve amalgamating theoretical virtues. Or whether it would show that our philosophical theorizing doesn't even *implicitly* involve amalgamating theoretical virtues. The underlying thought is something like: "Well, we manage to make choices *somehow*. So there ought to be a computational story about how those choices get made." But I have a hard time seeing how rational choice theory or prospect theory or even formal learning theory connect nicely to *philosophical* theory choice. Maybe that's just a failure of imagination on my part.
DeleteGood and interesting point! I think doing a lot of work here for me is that perhaps I am committed to the thought that we are capricious in a certain kind of way -- that while it might be that on any one occasion a model of virtue amalgamation would do ok at predicting a philosopher's choice (if it could be developed in sufficient detail to apply, which as you note is certainly not a given) and we could thereby attribute to somebody some range of weights over virtues consistent with their behaviour, I do not think that this would have much predictive power for the next choice made.
DeleteBut that, of course, in some sense commits me to having a more controversial psychological account of philosophical theory choice. And I don't got the goods to back that up! So I should think on this more. Cheers for making me think more on this!