Why would Mother Nature bother?

Been feeling bad about being a thing? Been feeling that the laws of nature are pushing you around? Here’s a book-length dose of Daniel Dennett’s Cold Comfort Cure. According to Dennett, ‘naturalism is no enemy of free will; it provides a positive account of free will.’ Sound too good to be true? Well, so it is. Proposals for ‘compatibilist’ resolutions of the problem of determinism aren’t new to philosophy, of course. But they always turn out to be a sort of Chinese lunch: there’s the lurking sense that what you got isn’t quite what you ordered, and half an hour later you’re hungry again.

So, for example, the treatment Dennett has on offer won’t work unless you’re prepared to break ‘your bad habit of yoking determinism with inevitability . . . determinism is the friend, not the foe, of those who dislike inevitability.’ And what you get if you swallow this pill is only the ‘proper concept’ of freedom; which is to say, not quite the concept of freedom you’d thought you wanted, but something prosthetic that the doctor says you’d better learn to live with. In a crucial (but parenthetical) passage, Dennett sort of ‘fesses up to this: ‘hard determinists . . . may find in subsequent chapters that [their] considered view is that whereas free will – as [they] understand the term – truly doesn’t exist, something rather like free will does exist, and it’s just what the doctor ordered for shoring up your moral convictions . . . [This] is perhaps only terminologically different from . . . the view that I am defending in this book.’ But surely that gives the game away right at the start. Being ‘only terminologically different from’ is the kind of relation that cuts both ways. If what I call being determined is only terminologically different from what you call being free, then what I call being free is only terminologically different from what you call being determined. That’s what I meant about the comfort being cold.

Dennett’s main polemical tactic in Freedom Evolves is what he describes elsewhere as ‘bait and switch’: advertise one thing but sell something else. ‘Bait and switch’ combines what philosophers call ‘paradigm case arguments’ with a soupçon of changing the topic. Thus we might imagine Bishop Berkeley: ‘But my dear Dr Johnson of course stones are real. What you just kicked was a paradigm; do feel free to kick another. It’s just that stones aren’t quite what you probably thought they were. Actually, they’re Ideas.’ Likewise Dennett: ‘My view is that free will is indeed real, but just not quite what you probably thought it was.’ Dennett’s way of getting agents into a deterministic world depends on selling you an instrumentalist account of agency. Roughly, instrumentalism is the view that theories and explanations are (just) devices for making predictions; their predictions are the only claim they make to correspondence with the world. Accordingly, their predictions exhaust their content. I think Dennett’s (usually tacit) instrumentalism is close to the heart of his philosophy. It is ubiquitous in the present book, and it takes a variety of forms. Consider, before we turn to the main business, a sketch of his explanation of why a ‘meme’ (read ‘idea’ in English-language editions) can become widely accepted in a culture:

It’s because the meme is beneficial.

And to whom does the benefit accrue?

Why, to the meme. ‘In the domain of memes, the ultimate beneficiary . . . is: the meme itself.’

And what is the benefit that the meme bestows on itself?

Why, that it becomes widely accepted in the culture.

The vacuity of this parody of Darwinism doesn’t need remarking on; if you’ve been following the literature, you’re probably inured to it. But what’s striking is Dennett’s completely ignoring the possibility that, now and then at least, or maybe here or there, an idea becomes widely accepted because it is true. Likewise, mutatis mutandis, when Dennett discusses moral issues: he simply doesn’t consider the (after all, not vastly implausible) view that the reason so many people think it’s wrong to rape and pillage is that, as a matter of fact, it is wrong to rape and pillage. All that is Dennett’s instrumentalism at work; it doesn’t allow him the robust notion of truth (of correspondence to fact) that such suggestions require.

Well, given the instrumentalism, Dennett is able to argue that some creatures are agents, but only in the thin sense that viewing them as such is a reliable heuristic for predicting their behaviour. Agency doesn’t, therefore, require creatures that actually act out of their preferences. It doesn’t even require creatures that actually have preferences. All it requires is creatures that behave as though they had and whose behaviour is therefore interpretable ‘from the intentional stance’. There is, however, the usual price to be paid when theories, psychological or otherwise, are instrumentally construed: there’s no logical space for a distinction between true ones and the ones that merely save the appearances. That higher organisms seem to be free agents is, after all, no news; it’s common ground in the present discussion. What worries determinists is that the appearance is illusory. If this possibility keeps you awake at night, Dennett’s defence of freedom won’t cure your insomnia.

So how, according to Dennett, is all this supposed to work? Just how, assuming instrumentalism, do free agents get into a deterministic world? The gist of the story is familiar from Richard Dawkins, E.O. Wilson, earlier work of Dennett’s, and many other current sources. You can even read it in the New York Times, where Steven Pinker has recently urged teaching it to innocent children in grade school. O, brave new world! It’s a dubious brew of neo-Darwinism and blatant anthropomorphism, and it indulges talk of ‘selfish’ genes and redwood trees with ‘points of view’. The stuff about ‘memes’ remarked on above is an offshoot.

Here’s Dennett in full flight: ‘The wisdom inherent in the design of multicellular life forms can best be understood by adopting the intentional stance to the whole process of evolution. From this perspective we can discern the free-floating rationales that have guided the evolutionary . . . process to ever more sophisticated rational agents.’ In brief, there are multicellular organisms because that’s the kind of creature that ‘Mother Nature, otherwise known as natural selection’ prefers there to be.

Tell that to the paramecia. They, too, successfully reproduce their kind (quite nicely, thank you; they’ve been around much longer than we have and they will certainly live to bury us). So, is it that they ‘solved’ a different evolutionary ‘problem’ from the one we ‘solved’? Or is it that they ‘solved’ the same ‘problem’, only in a different way? Was their evolution ‘guided’ by the same ‘free floating rationale’ that ‘guided’ ours, except that it somehow ‘guided’ them in the opposite direction? Such questions would have a grip if we really (not just instrumentally) ‘adopted the intentional stance to the whole process of evolution’ in order to explain how genes, paramecia, redwood trees and agents came to be. But to see how silly it would be for us to do so, you need only ask the same questions without the scare quotes. To just which evolutionary problem was the survival of the paramecium the solution? The survival of the paramecium, perhaps?

Instrumentally speaking, there’s no more to a creature being an agent than its behaving like an agent. Put that together with the thought that creatures that behave like agents are better ‘solutions’ to ecological problems (oops, ‘problems’) than creatures that don’t and you can leave the rest to evolution. For to behave like an agent is to behave in a way that would co-ordinate one’s actions (if they really were actions) with one’s preferences (if one actually had preferences) in light of the perceptual information that one’s environment provides (if, strictly speaking, one were able to see or hear). Well, it’s clear enough how such subtle creature-world co-ordinations might be advantageous, so you can see how evolution might have selected them. Or, if you can’t actually see how, you’re free to make it up – we are now well into the world of the Just So Stories. Because creatures that appear to be agents (actors, perceivers, thinkers) are such good ‘solutions’ to her ‘problems’, Mother Nature increases their relative frequency in their breeding group. So here we are. And here’s the paramecium, too. (One wonders, in passing, why Mother Nature bothers with this elaborate charade. Wouldn’t the best way for her to make a creature that acts just like an agent be for her to make a creature that is an agent? Such are the puzzles instrumentalists are prone to.)

The important thing to notice – Dennett’s particular contribution to this line of thought – is that if, by instrumentalist assumption, the evolution of agency is just the evolution of sufficiently agent-like behaviour, then whether or not you are an agent is independent of how your behaviour is caused. A fortiori, it’s independent of whether your behaviour is caused deterministically. With regard to the evolution of avoidance behaviour, for example, ‘the fact that the whole process is determined doesn’t detract from the fact that as time passes it generates more and more of something that looks for all the world [my emphasis] like avoidance.’ And since there’s a continuum of degrees to which your behaviours can be like (or unlike) what they would be if they were really caused by a desire to avoid something, there can be ‘a seamless blend of intermediate steps’ between things that count instrumentally as agents and things that don’t count as agents at all. The rock that went away when you kicked it didn’t, of course, decide to go away. Still, it was, one might say, on the right path. For, though there are clear cases at both ends, there’s no principled distinction between the rock that ‘wants to go away’, the cat that ‘wants its dinner’ and the author who really does want to finish the review that he’s writing. There’s no place where the scare quotes must come off.

You’re an agent pro tem if it pays an interpreter of your behaviour to treat you that way; and agents pro tem are the only kinds of agent there are. So, then: a little instrumentalism (all right, a lot of instrumentalism), some do-it-yourself Darwin, et voilà!: agents, believers and thinkers in a deterministic world (all right, as-it-were-agents, as-it-were-believers and as-it-were thinkers in a deterministic world). What’s doing the work is, to repeat, a tacit instrumentalism that identifies my being an agent with my behaviour being such that you could reliably predict it ‘from the Intentional Stance’.

But how could that be right? Whether I’m an agent isn’t about you, surely; it’s about me. In particular, it’s about the aetiology of my actions; you and your predictions don’t come into it. It’s true, I know, that Dennett is by no means the only philosopher who thinks that you and your predictions must come into it somehow; but frankly, that strikes me as nutty. My thought is of course prior to your interpretation of it. The interpretation is required to conform to the thought, not vice versa. How on earth could those philosophers have got things so backwards?

I fear these remarks will seem excessively emphatic. But there is, in Freedom Evolves, a recurrent and disturbing suggestion that Dennett’s critics are arguing in bad faith; they are, he thinks, well-intentioned but disingenuous. ‘I am trying to plant the seed of suspicion in you that some of these eminent critics of ours may even know in their hearts that we are right.’ I have therefore searched my heart as well as I am able. I wish to make clear, vehemently, that my objection is not that Dennett’s sort of view is unsalubrious but that it is, as far as I can tell, plain false.

Sometimes Dennett speaks of relativising attributions of agency not to the heuristic requirements of behavioural prediction, but rather to ‘levels’: ‘At one level nothing ever changes . . . At another level we see different kinds of worlds.’ But he takes the two ways of talking to be equivalent, as far as I can tell. ‘The confusion arises when one tries to maintain two perspectives on the universe at once. From the timeless, God’s-eye perspective nothing ever changes . . . From the engaged agent’s perspective, things change over time, and agents change to meet those changes.’ So, the essence of Dennett’s compatibilism is that freedom (according to one perspective) is compatible with determinism (according to a different perspective). Notice it doesn’t follow that freedom (without the parentheses) is compatible with determinism (without the parentheses).

Dennett regularly equivocates between the two theses. If you don’t keep that in mind, some of the things he says sound awfully peculiar. Could I really make you free just by changing my perspective on your behaviour? Come to think of it, can you make yourself free by changing your perspective on your own behaviour? Dennett writes as though positive thinking may indeed do the trick. ‘Isn’t it at least probable that having free will depends on believing you have free will?’ (I suppose he could have caught this meme in California.) But put the caveats back in and you see that what he’s urging is truistic: changing your viewpoint on whether you are free can make you free according to your new point of view. Bait and switch.

Why, anyhow, would you care whether you are free according to Dennett’s perspectival construal? If the kind of freedom that determinism threatens is worth having, that’s exactly because it’s a matter of fact whether or not you’ve got it. What one wants, precisely, isn’t freedom from a perspective. What one wants is metaphysical freedom; freedom tout court. One wants to be what tradition has it that Eve was when she bit the apple: perfectly free to do otherwise. So perfectly free, in fact, that even God couldn’t tell which way she’d jump. I do doubt that you can have that kind of freedom in a deterministic world, and I guess that determinism is very likely true; if so, so be it. Rather bad news than bait and switch.

To be sure, it may be that the metaphysical notion of freedom is confused; but Dennett doesn’t argue that it is. The closest he gets is Chapter Four, where he claims, rightly I think, that if there is such a thing as metaphysical freedom, it must involve a kind of causation (‘agent causation’) that science as yet tells us nothing of. Quantum mechanical randomness, for example, doesn’t help to save the bacon. It appears that Dennett takes this consequence to be a reductio, but I can’t think why. It is entirely plausible, indeed entirely self-evident, that there is a very great deal about how the mind works that we do not understand. Here and elsewhere, Dennett writes as though science is over and we know already what’s in it and what’s not. (‘Can’t we dismiss the whole sorry lot’ of hypotheses about decision making on the grounds that they ‘are wildly unrealistic oversimplifications of what is known about how decision-making works in the brain? Yes indeed, we could, and we should.’) Dennett’s faith in ‘Artificial Intelligence’, ‘Cognitive Neuroscience’ and ‘Evolutionary Psychology’ is touching but it isn’t warranted. In fact, the three barely exist. The working assumption must be that the science of the mind (to say nothing of the mind/brain) hasn’t more than begun; everything is up for grabs and is likely to remain so for a very long time.

But ‘aside from metaphysical curiosity, what interest should we take in [freedom] in your sense?’ We are, as Dennett rightly says, free in lots of respects that have nothing much to do with whether the world is deterministic. There are indeed paradigm cases: there’s the freedom to assemble that the Bill of Rights and the cop on the corner purport to protect, for example; and it would be silly to say that you can’t be free to assemble if determinism is true. But if you think that having that kind of freedom makes you an agent, you may have to break your bad habit of yoking being free to assemble with being free to choose to assemble (or to choose not to). It’s the Chinese lunch effect once again: what you get isn’t quite what you ordered.

Who cares about metaphysical freedom? Well, I do for one. I am metaphysically curious. I’m professionally interested in whether metaphysical freedom is compatible with determinism; that’s the sort of question philosophers make their exiguous livings by asking. Why isn’t that OK? To be sure, I worry about metaphysical determinism a deal less than Dennett does. That may be because he runs the free will issue together with a number of others, including naturalism, dualism, consciousness, the unity of the self and how much of a parallel processor the brain is, to which I doubt that it’s importantly connected. Also, I’m interested in how the mind works; I’d therefore like to know whether only instrumentalists can make sense of agency, action, belief and the like. I’d be sad if that were so. I strongly suspect that, though it may be heuristically useful for predicting behaviour, the kind of agency that instrumentalists can certify is useless for explaining it. In particular, it’s much too thin to explain how the mind causes behaviour; heuristic fictions don’t cause anything.

If ‘how does the mind cause behaviour?’ isn’t an interesting question, I can’t imagine what would be. I therefore take the better path to be the very one that Dennett hopes to avoid: admit straight out that if determinism is true, then there’s an important sense in which agents aren’t free and compatibilism can’t be maintained. What you get in exchange is that you can kick away all the stuff about levels, perspectives and interpreters. Agents act out of their beliefs and desires, and they do so tout court, not just at some level of description, or from someone’s perspective. And that is so whether or not determinism is true.

I think you should take this line, too; at very worst it can’t lose you much. Dennett’s instrumentalist kind of compatibilism may soothe your metaphysical anxieties for a time, but cold comforts don’t last; you’re hungry again in half an hour.