How many philosophers does it take to write a dialogue?

by Jeroen Bouterse

“The constant direct mode of address was a chore. No one will enjoy having this read to them.” Quoting from a referee report on the Nicomachean Ethics misses the point of James Warren’s hilarious rejection letter, but I looked it up because I remember thinking that the fictional critic was onto something, and not just about Aristotle. “I had the impression at times that some kind of conversational or dialectical background was being assumed but this is not at all marked in the text.” Indeed! Why hide the fun part?

I owe several very happy moments to well-executed philosophical dialogues: Imre Lakatos’ Proofs and Refutations, Larry Laudan’s Science and Relativism, and Aristotle’s former supervisor come to mind. I will be ever so grateful to anybody who can point me to similarly exciting conversations. The dialogue form draws and holds the attention: it can let worldviews clash in the abstract, but it can simultaneously delve into matters of detail without becoming boring – these details having been established, after all, to flow not just from the idiosyncratic preoccupations of one contingent mind, but from larger intellectual interests common to at least two separate perspectives.

Still, they are usually written by one person. I was thrilled last year to find out about a philosophical dialogue where both positions were written by people actually holding them: in Just Deserts (2021), Gregg Caruso and Daniel Dennett debate the implications of their ideas about free will, especially via the question whether people ‘deserve’ blame, praise, punishment and reward (henceforth ‘BPPR’) for their actions. Caruso’s position is that there is an important sense in which they don’t, and that this ought to be reflected in the way in which we deal with bad or criminal behavior.

Dennett’s position is that they do, albeit in a commonsensical, pragmatic sense that sometimes threatens to deflate the entire controversy, so that this dialogue is simultaneously a philosophical discussion and a curious case-study of the limits of philosophy. There are stages in the discussion where our whole social arrangement seems to depend on the truth or falsehood of free will skepticism; there are also passages where the stakes seem to collapse to the mere question whether we should add the semantics of insult to the pragmatics of injury. I am speaking out of turn now, however, commenting upon a discussion before even semi-accurately representing it. Let’s look at the main themes.

Caruso’s opening move is the claim that since what we do is ultimately the result of factors beyond our control, we never truly deserve BPPR. In response, Dennett makes a distinction between causation and control: the past does not ‘control’ us, and if people have agency (whatever the causal history that led to them becoming agents), then they can be held responsible. The discussion then shifts to the question whether this kind of control or agency actually justifies assigning BPPR to people for their actions. Dennett maintains that it does: assigning judgment and consequences to people’s past actions is a necessary feature of the Moral Agents Club, of which most people are part and which sustains our society. Although this system is itself justified in a forward-looking (consequentialist) way, within the system it is important to assign BPPR based on merit or desert: it needs to be clear that agents who break the rules deserve certain consequences, just like sports players who cheat.

Caruso finds Dennett’s view hard to distinguish from retributivism: the view that it is intrinsically good if people suffer negative consequences for their bad actions. Both Dennett and he reject this view, but Dennett retains deserved BPPR as an inextricable part of the-best-of-all-social-arrangements, so that Caruso suspects his arguments against it may apply to Dennett as well. It is precisely the validity of this BPPR-based social system that is in question, relying as it does on so strong a view of free will. Because people obviously don’t control their own causal histories, Caruso says, their growing into moral agents is a matter of luck, as is their behavior in specific circumstances. A system acknowledging this will be better than a system that is preoccupied with assigning BPPR based on what rule-breakers deserve; a desert-based system will tend to treat them too harshly, and tend to ignore the systemic causes of bad behavior.

In order to illustrate his case against compatibilism, Caruso proposes a thought experiment involving a woman (Elizabeth) being manipulated by neuroscientists into killing somebody, but in such a way that the causal chain leading to the murder meets the conditions for control in a compatibilist sense: Elizabeth’s thought processes might as well have been normal, caused in the end by impersonal forces. Caruso’s point is that if we don’t hold Elizabeth responsible in the case where she is manipulated, then we should also not hold her responsible in the case where the same causal processes occur in a purely natural way. To her inner states, after all, it makes no difference whether there is an external agent in the causal chain or not.

To Dennett, this hypothetical case suggests a misunderstanding of what it means to be self controlled, and of the limits of manipulation. Just being at the start of a complicated causal process like this does not enable you to control its outcome; that would require constant monitoring of other causes intervening in Elizabeth’s history, thus undermining the analogy with impersonal forces. If the neurosurgeons control Elizabeth and she has no way to counteract their manipulation, then she is not responsible, but this proves nothing about the actual world; if they don’t, the thought-experiment resembles the normal case in which she is indeed unproblematically responsible.

I found it very difficult to keep score in this part of the conversation. It seems to me that Dennett’s criticism of the manipulation analogy is mostly pertinent, and that he wins this part of the debate on points; but I cannot help but think that his putdowns also cloud a reasonable intuition that a more charitable reading might have drawn out. ‘Agency’ is a big, abstract word; and though Dennett and Caruso both agree that it is a meaningful one, Caruso seems to be making the case here that things happen on its fuzzy boundaries which can behave chaotically and lead to morally radically divergent outcomes. A small nudge here, a flap of a butterfly’s wing there, and the same person will develop in a different way, all within the confines of agency-broadly-speaking but tipping the scales into her committing a murder that otherwise she might just have shied away from attempting. If causal histories are this complicated or chaotic, if it is possible for big things to hinge on a coin-flip at a moment when no-one was looking, then what is it about the person that we are judging? Does Elizabeth deserve such a major leap in BPPR-space, admittedly based on a major difference in actions but tracing back to small steps and little bumps in her causal history, of which each in itself she is innocent?

I read Caruso as saying that treating Elizabeth as if at every one of those small steps and little bumps there is a malevolent demon tipping the scales in favor of a disastrous outcome (or indeed its secular equivalent the evil neurosurgeon) is a good reminder not to take normal luck for granted in our judgment. Dennett is still right to stipulate that none of these actors can be allowed to Actually-Control her, in a way that takes away her Dennett-style free will; that would be cheating. But Caruso is not cheating, and I think something of his point should have remained that got buried in Dennett’s (in itself fair-minded) critique. In the actual world the malevolent agent drops out anyway: the point is to argue that it is relevant that people don’t make their own causal histories.

In any case, after a while Caruso gives up on this line of argument. He now takes on the way in which people deserve BPPR in Dennett’s model. Dennett claims that people may not be ‘intrinsically’ or ‘absolutely’ responsible, but that within the system, they really are and can be held responsible – much like money can have real value without having intrinsic value. Caruso happily runs with the metaphor: it confirms his suspicion that responsibility is assigned by society without being grounded in real properties of people; that free will is a construct that, like money, fulfils a social function but can lose its value or be replaced by other institutions. Thus, the discussion shifts to the question whether an alternative system without BPPR is feasible and sustainable (as Caruso claims) or not (as Dennett claims).

Before this really gets off the ground, however, Caruso and Dennett pick up the thread of luck. Against Caruso’s contingentist view of moral development, Dennett maintains that many people are provided with strong stimuli to develop into mature agents. This is the rule rather than the exception, because their environment in turn has robust incentives to provide these stimuli: parents, for example, want their children to do well in an environment that is built around rules governing the fair assignment of BPPR.

In Dennett’s system, BPPR are indeed fair, to the extent that they are assigned only to members of the Moral Agents club, who understand the rules and have bought into a game where they will be held responsible if they cheat. Caruso suspects that a free will skeptic could go along with Dennett’s system but still refrain from saying that people really ‘deserve’ punishment. This is one point where the discussion almost collapses into semantics, and it is hard to decide who to blame (or whose position to attribute this to). Of course, there is substantive disagreement underneath these terminological skirmishes, too, and now that the outlines of Dennett’s system are clear, Caruso also sketches his own, non-retributive, non-judgmental, free-will-skeptical approach to bad behavior. This is his ‘public health-quarantine’ model: when we quarantine potentially contagious people, this is not in retaliation, but in order to protect ourselves and others. We can approach bad behavior with a similar attitude: not as something to be punished, but as something to prevent, to protect ourselves against, and to reply to in a way that best serves the legitimate interests of everyone involved. BPPR drop out of this system, as do any notions grounded in the assumption of free will, like desert.

Dennett’s main objection to this model is that it is vulnerable to exploitation: how does Caruso enforce his system without the threat of punishment? Though the question whether a non-BPPR model is sustainable in practice is reasonable enough, Dennett seems to assume that Caruso’s system does not allow for any kind of coercion. Coercion and punishment are such different concepts that it puzzles me Dennett slides so easily from one into the other. Caruso gives a straightforward example to illustrate the difference: rules, including liberty-restricting rules, are enforced in nursing homes for people with Alzheimer’s, but surely not with the intention to punish. The discussion remains interesting from here on, but doesn’t really shed this misunderstanding: every now and again, Dennett exclaims that Caruso’s proposals will have no more than the status of a recommendation, because anybody can opt out without consequences.

In general, Dennett is worried that Caruso is presenting a “philosopher’s sort of theory” (161), while he himself is trying to provide a framework that can account for the stability of existing morality and law. From a Darwinian perspective, he predicts that a general policy of excuse is going to be unsustainable. Since the people whose behavior we are trying to influence are language-using agents, we simply cannot treat them as if they are not responsible for anything: they will quickly find and exploit the loopholes gaping in this consequence-free system. We can also not treat everyone as an agent, however, and this is precisely where concepts like free will, responsibility and desert come into play: we need to distinguish between the morally competent and incompetent, between those who deserve to be punished for bad acts and those who don’t. Caruso in turn emphasizes that skeptics can also distinguish between the morally competent and incompetent, and adapt the consequences of wrongdoing to the reason-responsiveness of the agent – all in the light of future protection, reconciliation, and future moral formation, however; not because of “basic-desert moral responsibility”.

This exchange seems to me the high point of the debate. The internal coherence of Dennett’s and Caruso’s positions is becoming ever clearer to me: though it remains hard to classify Dennett’s views on punishment, it is clear what he means when he says punishment is deserved; and though Caruso’s radical view has the rhetorical disadvantage of going against deep-seated common sense and status quo bias, it is starting to take root in my mind as a live option. Where could we go from here?

As it happens, the conversation unravels over the last thirty pages. A sample to suggest the tone:

“That’s unfair […]” (183)

“So you say.” (183)

“It seems to me that you are beating on an unlocked door” (190)

“Rather, what has been ‘getting in the way’ is your failure to distinguish between …” (193)

“Until now, you’ve danced around a direct answer.” (193)

“Seeking to diagnose our failures to communicate […]” (197)

“Your position […] is like wrestling an eel” (198)

“I found myself in a thicket of contradictions, ambiguities, and questions. Unfortunately, I have very little hope, especially at this late hour, that we’ll ever be able to untangle the thicket and come to an agreement.” (201)

“If you adopt this view, you are heroically biting yet another bullet, or maybe a hand grenade […]. Then what are we doing […] and why?” (203)

“From my perspective, you’ve failed to spell out your view clearly and sufficiently enough for me to understand the details of your account” (205)

“I’m disappointed that I haven’t been able to explain to you my rather straightforward and detailed proposal” (206)

Similar expressions of frustration and exasperation occur throughout the dialogue, but they increase in frequency and intensity as we approach the end. How to diagnose them?

One ingredient is surely a more general frustration with the opponent’s way of arguing: Dennett feels not without reason that Caruso keeps trying to box him in with positions that he doesn’t want to defend (“you want to give the notion of desert an important justificatory role in your account of punishment” (184)). Caruso in turn may well feel that he is doing all the philosophical work while Dennett is kicking stones, pointing at the robustness of existing practices as if thus the possibility of philosophically grounded alternatives is refuted. Susy drew on the wall with crayons and Tommy didn’t, so Susy deserves a time-out but Tommy doesn’t, and that is that (190) – as if nothing has been achieved at all in the dozens of pages before this point; as if none of the questions and criticisms aimed at these commonsensical reflexes have stuck.

I am not accusing anyone of either literal misreading or willful sophistry, on the contrary. It goes without saying that Dennett and Caruso are both in the truth-seeking business, defending genuinely held positions that are well thought through. But precisely as I, as a reader, started to understand those positions and to predict accurately how they would respond to certain objections, they started claiming louder and louder that the other, in failing to realize important distinction X or mistakenly focusing on unimportant distinction Y, was making further productive debate impossible.

This is puzzling, but perhaps explicable by how this text came to be. In a conversation that reaches a semi-adversarial stage – where both parties have stated positions that are at odds with each other and that they want to defend and protect – people will retreat to the most objection-proof versions of their own ideas, and leave charitable versions of alternative views to fend for themselves; why do your opponent’s work for him, after all? When you are writing on behalf of an absent other, you will need to inhabit their ideas simply in order to render them convincingly; when you are debating a living opponent, those ideas already have a champion.

This leaves me wondering what makes a better philosophical dialogue: one written by two thinkers, or one written by one? For the largest part of this book, I learned a lot about both Dennett’s compatibilism and Caruso’s free-will-skepticism, owing to the fact that they were formulated by thinkers with an interest in giving those positions a fair chance. Furthermore, my understanding of both was sharpened along the way by the objections and refutations of the other, in a much more efficient and exciting way than a monograph would have done – I hugely enjoyed reading their exchange (and would enjoy having it read to me). Yet where the ideal philosophical dialogue proceeds via some kind of dialectical structure to reach some kind of synthesis, or even just a shared understanding of the disagreements, a real conversation between interested participants can be chaotic, contain loose ends to be picked up at will throughout the discussion, and circle back to earlier arguments without real progress.

I guess Aristotle was onto something, too.