Ought, is, and logic

young Immanuel Kant

Last month I published an essay on alleged empirical evidence that Kant’s idea that ought implies can (OIC) is false. To refresh your mind, the paper I discussed was published by Vladimir Chituc and co-workers, who claimed that — because a good number of random folks say that someone ought to do X when it is plain impossible for X to actually be carried out — then Kant’s famous dictum from the Critique of Pure Reason: “The action to which the ‘ought’ applies must indeed be possible under natural conditions,” must be wrong. I suggested instead that the folks used as subjects by Chituc and colleagues simply didn’t understand basic logic. An epic Twitter battle ensued.

Much of the exchanges ended up being among yours truly, Vladimir Chituc himself, Wesley Buckwalter (a philosophy postdoctoral student), and Moti Mizrahi (an assistant professor at Florida Tech).

One of the strangest arguments advanced by my critics (Buckwalter in particular) was in response to an obvious question I posed: please provide any case at all, outside of ethics, where ought does not imply can (e.g., you ought to defy the law of gravity). They couldn’t, obviously, so they argued that, you see, ethics is a special domain, with its own logic…

Speaking of logic, Moti actually published a paper in Philosophical Frontier against the OIC principle, arguing by counter-example, in the old fashioned way of philosophers. You can check it here. Turns out, in fact, that a mini cottage industry has developed over the past few decades of philosophers challenging Kant on this point (here is a good summary of the literature), an industry that has so far failed to convince the majority of practicing philosophers, who stubbornly insist in teaching the principle at both the undergraduate and graduate levels.

But there is more. After having been accused of daftness and even of simply stomping the ground rather than providing arguments (remember, Twitter “arguments” are limited to 140 characters at a time), I was even given formal logical proof that the OIC is wrong. This was published by Paul Saka in the American Philosophical Quarterly, and it surely ought (ah!) to be convincing. After all, it’s logic, right?

You could go through the 13 pages of formal logical argumentation, but I wouldn’t bother, if I were you. This, I hasten to say, is not because Saka’s reasoning is incorrect (though, as all formal logic, it quite crucially depends on the particular assumptions being made about what Kant may or may not have meant; also, having been published in 2000 it too has so far failed to sway the majority of practicing philosophers). Rather, it is for two other reasons.

First, because it is just as easy (thank Google!) to find equally tight logical proofs that the OIC is correct. One such is provided by Harry J. Gensler in his book, Formal Ethics (particularly sections 3.3 and 3.4, pp. 43-48). I have no more reason to think that Gensler is a bad logician than I have to doubt Saka’s abilities, but the two reach diametrically opposite conclusions. So there.

The second reason is more important, and it is the real topic of this post. I think it is a mistake to look for proof (or disproof) of the OIC because it is not meant (or it ought not to be understood) as a principle of formal, but rather of informal logic.

The distinction is well established in philosophy, and in fact typically philosophy departments teach two types of logic courses: those that largely deal with formal logic (i.e., the sort of thing that can be done exclusively using formalisms and symbols) and those having to do mostly with informal logic (also known as “critical thinking”).

To make the distinction concrete, consider the difference between logical fallacies of the formal vs informal variety.

Here is a typical formal logical fallacy, known as affirming the consequent:

P1. If P then Q
P2. Q
C. Therefore P

Even if both premises (P1 and P2) happen to be true, the conclusion, C, does not logically follow. To see this, let us establish that P = “it rained” and Q = “the sidewalk is wet.” While it is true that if it rains then the sidewalk is going to be wet, one cannot infer from the fact that the sidewalk is wet that it rained, since it may have been made wet by someone who sprayed water on it in order to clean it (as it often happens in New York, where I live).

Contrast the above with an instance of informal fallacy, the argumentum ad hominem:

P1. Joe is known to be an unreliable car salesman

C. You shouldn’t buy a car from Joe

If understood as a formal argument, i.e., as something that is formally valid, the above won’t fly. Just because Joe has conducted shady deals in the past it doesn’t necessarily follows that it will swindle you on the current deal.

And yet, you’d certainly be a fool not to take the possibility into consideration and look for a more reputable dealer, or at the very least being extra careful in scrutinizing whatever car Joe is attempting to sell you.

Indeed, Maarten Boudry, Fabio Paglieri and I have written an entire paper, published in Argumentation, in which we propose that so-called logical fallacies are not fallacies at all, but rather instances where a given course of action (e.g., to distrust Joe the car salesman) may or may not be reasonable, given background conditions, conditions that need to be investigated and properly evaluated, pretty much case by case. Here is the abstract of the article:

“Philosophers of science have given up on the quest for a silver bullet to put an end to all pseudoscience, as such a neat formal criterion to separate good science from its contenders has proven elusive. In the literature on critical thinking and in some philosophical quarters, however, this search for silver bullets lives on in the taxonomies of fallacies. The attractive idea is to have a handy list of abstract definitions or argumentation schemes, on the basis of which one can identify bad or invalid types of reasoning, abstracting away from the specific content and dialectical context. Such shortcuts for debunking arguments are tempting, but alas, the promise is hardly if ever fulfilled. Different strands of research on the pragmatics of argumentation, probabilistic reasoning and ecological rationality have shown that almost every known type of fallacy is a close neighbor to sound inferences or acceptable moves in a debate. Nonetheless, the kernel idea of a fallacy as an erroneous type of argument is still retained by most authors. We outline a destructive dilemma we refer to as the Fallacy Fork: on the one hand, if fallacies are construed as demonstrably invalid form of reasoning, then they have very limited applicability in real life (few actual instances). On the other hand, if our definitions of fallacies are sophisticated enough to capture real-life complexities, they can no longer be held up as an effective tool for discriminating good and bad forms of reasoning. As we bring our schematic ‘fallacies’ in touch with reality, we seem to lose grip on normative questions. Even approaches that do not rely on argumentation schemes to identify fallacies (e.g., pragma-dialectics) fail to escape the Fallacy Fork, and run up against their own version of it.”

Why is this pertinent to the OIC? Because the best way to understand it is as a statement in informal, not formal, logic. I’m not just making this up to deflect criticism. On p. 50 of the above mentioned Formal Ethics the author cites a much older entry in the literature, going back to 1954, by William Frankena:

“Ought implies can need not be construed as asserting a strict logical implication. It may plausibly be understood as saying: (a) moral judgments ‘presuppose’ or ‘pragmatically imply’ that the agent is able to act; or (b) the point of uttering moral judgments disappears if the agents involved are not able to act; or (c) it would be morally wrong to insist that an agent ought to do a certain action, if he is or is thought to be unable to do it.”

All three interpretations are eminently sensible and entirely bypass any (formal) proof or disproof of OIC. I particularly like (b), as to insist that someone ought to do X if it is in fact impossible to do X amounts to stomping one’s foot, which, I have been told on Twitter, is not a good argument.

It is also well worth noting that none of the above papers (except for the one by Chituc and collaborators, of course) — either pro or against OIC — depends at all on empirical data. Whether we should think of OIC as a statement in formal or informal logic is a philosophical, not empirical, question. If we do consider it a statement in formal logic, whether it can be formally proven or disproven to be valid is a matter for formal logic, not empirical data. Which reinforces my original contention that to attempt to “show” that OIC is invalid by see how a bunch of people with no philosophical or logical training use ought language is a non starter.

The more general point here, which goes well beyond the OIC, is that formal logic is of rather limited value when it comes to human affairs (works like a charm with computers, though!), because human decisions are “logical” (in the sense of being rational, sensible) depending on a number of assumptions and background conditions whose variability defies any attempt to find the sort of silver bullet mentioned by Maarten, Fabio and myself.

As for the OIC itself, it seems perverse — to paraphrase Frankena — to blame someone for failing to do what he could not possibly do, and I would hope that this principle is clear in the mind of judges in courts of law, if not in those of more or less experimental philosophers. One could say that ought may or may not imply can, depending on what specific meanings and logic one adopts. But if we are talking ethics, it surely ought to.