a challenge to Sam Harris on free will

Sam Harris claims that free will does not exist. Especially, he claims that causal determinism is true, and that, as a consequence, there can be no such thing as either free choice or moral responsibility. Describing a horrifically depraved crime in some detail, Harris refers to the perpetrators and writes:

As sickening as I find their behavior, I have to admit that if I were to trade places with one of these men, atom for atom, I would be him: There is no extra part of me that could decide to see the world differently or to resist the impulse to victimize other people. Even if you believe that every human being harbors an immortal soul, the problem of responsibility remains: I cannot take credit for the fact that I do not have the soul of a psychopath.

I agree with Harris halfway: given the truth of causal determinism, were he to trade places "atom for atom," he too would be a depraved criminal. But then, like the criminals, he too would be properly blameable and punishable for his crimes.

What could Harris say to this? If Harris means to dismiss or rule out this response (and he does), then an independent argument is needed in support of the substantive philosophical position that causal determinism categorically rules out moral responsibility. So far as I can tell from the above excerpt, Harris has only assumed this critical principle, not argued for it.

To be fair, I haven't read his book. Maybe he has the right sort of philosophical argument for the 'causal determinism rules out free will' background principle there. But I doubt it. From what I have read, he gives the impression that he thinks the issue is simply one of showing causal determinism to be true. But it is not enough simply to assume that if causal determinism is true, then moral responsibility is impossible. That premise of the argument is substantive and controversial and not at all obvious. Moreover, this philosophically substantive premise could not be established on empirical grounds; moral responsibility as such, if it exists at all, wouldn't be the sort of thing that exists as an object of empirical investigation or scientific study. If moral responsibility exists in some possible world, then, in that world, try as you might, you're not going to get moral responsibility into a test tube, or into an fMRI.

So then, a challenge for Sam Harris, and for anyone else tempted to his position: why in the first place should we think that if casual determinism is true, then moral responsibility is impossible?

a user's guide to avoiding the repugnant conclusion

Suppose it is possible to define a unit of goodness. Then, we might think that better-than judgments should be understood on a schema like:

A is better than B just if A has more units of goodness than B.

Suppose also that better-than judgments are transitive. So, for example, if A is better than B, and B is better than C, than A is also better than C.

Given the idea of "units of goodness,' I think that these all seem like very minimal and natural assumptions. It is natural to think that we would prefer A to B if we thought that A had more of the good stuff than B.

Surprisingly, however, when we apply this schema for better-than to some actual example cases, we can derive some repugnant conclusions. Before I give an illustrative example, it helps to define goodness in some intuitively appealing way. So let's say that goodness is pleasure; we might just as well have said it was preference-satisfaction, or utility, or even human happiness so long as human happiness can be understood to be quantifiable along a single measure. (Human happiness units?) I will stick with pleasure in my example, but you could substitute 'pleasure' for anything else you like.

So now picture and compare three possible worlds to see which is better than the other:

Mass Poverty World: 1 trillion people are born and inhabit planet earth, and, over the course of their lives, they each have a total of .1 units of pleasure, a very, very small amount of pleasure for each person to have by our present standards of pleasure-expectancy.

Happy Medium World: 1 billion people are born and inhabit planet earth, and, over the course of their lives, they each have 99 units of pleasure, a normal amount of pleasure to have by our presents standards of pleasure-expectancy.

Pig Prince World: 1 person is born and inhabits planet earth, and, over the course of his life, he has 101 billion units of pleasure, an unimaginable amount of pleasure for him to have by our present standards of pleasure-expectancy.

On the present schema for better than, we must say that Pig Prince World (101 billion total pleasure-units) is better than any other world. Also, we must say that Mass Poverty World (100 billion total pleasure-units) is better than Happy Medium World (99 billion total pleasure-units). But this just seems to be the wrong result! What we really want to say is that Happy Medium world is better than both of Pig Prince World and Mass Poverty World. The conclusion we are driven to is repugnant.

(These examples are my own take on a problem which goes by the name of the "Repugnant Conclusion" in the philosophical literature, and which was first articulated by Derek Parfit. Incidentally, Parfit tells a different story than I do here, and it might be debatable whether the problem I'm on about here is actually parallel to his. I think it is (we just need take Parfit's unit of goodness to be "life worth living"), but, if it isn't a parallel, then please just take the present problem in its own terms, for whatever it is worth.)

So, what has gone wrong? The assumptions we started with seemed so natural!

To solve this problem, I think that we should reject the idea that "better than" is an idea of measuring and comparing quantities of stuff. We surely don't have to understand "better than" as "greater than" for some defined unit of goodness. But then we don't have to say that Pig Prince World and Mass Poverty World are better than Happy Medium World.

However, we may still want to know why it seemed natural to us to think, initially, that more units of goodness are better than less units. The answer, roughly, is that in our actual world and in nearly possible worlds, increases in the quantity of goodness-units often enough are moral ('better-than') improvements. (And, generally, we only pay attention when they are such improvements.) The mistake is in taking this rough correlation to be an identity of the quantity and moral better-than notion; the mistake is in thinking that increases in goodness-units necessarily means improvement overall.

Just for fun, here's an alternate picture of 'better-than': 'better-than' is the idea of a kind of distance from some Ideally Perfect World--for short, let's say Heaven. A is better than B if it is closer to Heaven. A is worse than B if it is further from Heaven. A is neither better nor worse than B if A and B are, within the possibility space, equidistant from Heaven. So, we could say: Happy Medium is closer to Heaven than either Pig Prince or Mass Poverty.

I suppose I might be asked how we should measure distance from Heaven. The answer is: I'm not sure. (Nearer-my-God-to-thee units?) But also, I'm not sure it really matters to be able to say.

Metaphors aside, the deeper question really is one of how we make sense of the idea that something can be better or worse in terms of a standard or ideal. (And, incidentally, we shouldn't assume that standards couldn't be multi-faceted: an ideal pencil both holds its point and leaves an appropriately dark mark.) We want to say: surely there is something it is to think that something is better or worse in relation to a standard. What that is, I don't know. But I doubt it is going to turn out to be, at root, some measurement of a quantity of stuff.

Two Distinctions in Goodness

The subject line is the title to a 1996 essay by Christine Korsgaard which has rocked my philosophical world and which is highly recommended if you can get your hands on it. Here, I thought I would share my (quickly-written) summary of the article and its main points:

Korsgaard draws a pair of distinctions between a) intrinsic/extrinsic goodness and b) instrumental/non-instrumental goodness. (For my COW readers: it helps to picture Thomson's grid pairing off absolute/relative and objective/subjective, and then substitute these category pairs on that sort of grid, and you'll get the picture.) This distinction makes a new concept of value possible: the extrinsic, non-instrumental good. The essential idea of this category is that of things whose goodness depend upon meeting certain conditions, but which things, if goods (if meeting those conditions), are not instrumental goods for any further purpose. So, Korsgaard gives the example of a painting: a painting can be a good, but it is not a good that is for something-- if it has value, it has value just as the aesthetic object that it is. So its goodness, if any, is non-instrumental. But a painting's having its particular goodness is nevertheless, Korsgaard thinks, conditional on its being viewable, or on not being hidden away forever in a dark closet, or some such (or so Korsgaard would say--not sure I would articulate the condition quite as she does, but I'm giving her example, not the half-dozen of my own I've thought of since).

 

The beauty of all this is that it does two things we want out of a theory of value, which two things are often enough thought to be at odds: first, this account preserves space for our sense that the goodness of goods somehow has to do with the sorts of creatures we (contingently) are -- the conditions of the goodness of some x in a particular case often enough has something to do with features of our creaturehood. Second, this account allows us to say that the goodness of good x's are complete in themselves-- we do not have to try to make every good thing the servant of some further end. We do not have to say, for example -- and always implausibly -- that the goodness of any good depends upon whether it gives pleasure or maximizes preference-satisfaction, as if we could have no further opinion on what it was appropriate to take pleasure in (or to prefer), or as if we were hedonists who actually thought that our pleasures were fungible, or as if we would recommend getting into the experience machine. And neither do we have to insist that x's are good in abstraction from the conditions of human experience. So the concept of the extrinsic, non-instrumental good is a powerful one that potentially bridges the gap between two heretofore competing visions of goodness.

 

As if this wasn't progress enough, Korsgaard exploits this category by then adding in an additional Kantian twist. She teaches that it is a sufficient condition for the goodness of any x that it is rationally willed. (We'll want to know what account to give of "rationally willed," but leave that aside and assume we have a suitable independent account-- I actually don't know the answer here.) Any such x will of course be in the category of extrinsic, non-instrumental good. And this fuller picture, if adopted, now shows / explains several otherwise puzzling features of moral experience:

 
1) How it could be proper that people can have inconsistent personally chosen ends (people rationally choose differently, and rationally choosing differently is the ultimate sufficient condition of something's being good) 

2) Why it is that an x can be special just in virtue of someone caring for it

3) Why it can be so hard to explain of one's own chosen-ends what it is about that end which demands its choosing (in short, there may be no feature of the thing which demands its choosing-- but, still, the thing may be good, and non-instrumentally so)

4) Nevertheless, why we have the sense that our choices are, well, ends, i.e., non-instrumental goods (because, on this picture, they are not for anything)

 

More needs to be said before we have the full picture of morality and ethics. But, like I said, I think that this is progress. It is the Kantian stuff that worries me the most, although I will confess to being drawn towards it-- it explains a lot that needs explaining, and does so fairly parsimoniously. But I have not yet finished my reflective scrutiny.

 

Anyway, whether or not you go the full Kantian way with Korsgaard, just in having suggested the possibility of extrinsic-cum-non-instrumental good, it seems to me that Korsgaard delivers into your hands a powerful conceptual tool for use in your own philosophical reflection.