Welcome to what we expect will be a very interesting and productive discussion of Abe Roth‘s “Intention, Expectation, and Promissory Obligation.” The paper is published in the most recent edition of Ethics and is available through open access here. Sarah Stroud has kindly agreed to contribute a critical précis, and it appears immediately below. Please join in the discussion!
Roth’s interesting and wide-ranging article covers a lot of ground. It may be helpful if I disaggregate some of the bricks in the wall, sequentially identifying what I take to be the important themes and moves in the account.
Giving the promisee a more important role. Roth’s opening move seems to me of interest even independent of what he later does with it. His suggestion is that we should beef up the distinctly secondary role which has always been assigned to the promisee in the creation of a promissory obligation binding on the promisor. Roth notes that some promises may in fact be unwelcome to the promisee. If promissory obligation were generated automatically, even when the promise is unwelcome, this would in effect harness “the resources of moral motivation, obligation, blame, sanction, etc. associated with promising … in a way that works against the promisee” (p. 89). Surely it would be more natural if promisees had available to them “a mechanism to block the generation of an obligation when the promise is unwelcome” (ibid.).
Roth maintains that we do in fact possess such a mechanism, namely rejecting a promise (of which accepting a promise is the mirror image). Granting promisees the normative power to accept or reject a promise gives promisees some control over which promissory obligations are generated, a desirable result. And Roth argues that it would not work as well to make the creation of promissory obligation dependent, not on an act of acceptance or rejection by the promisee, but merely on the promisee’s psychological states (such as whether he wants the promisor to ϕ). Once we focus on the promisee’s option to reject or accept a promise, we will see (says Roth) that rejection of a promise blocks the generation of a promissory obligation, while acceptance of a promise “help[s] to ensure that a promissory obligation is generated” (p. 89). This picture thus “gives a more intimate and active role for the promisee in generating promissory obligation” (p. 110); it portrays promissory obligation as a joint creation of the promisor and the promisee, rather than a solo act by the promisor.
Understanding the acceptance of a promise in terms of an intention that the promisor ϕ. Roth’s next move is to offer a novel account of what constitutes acceptance of a promise. He proposes that for B to accept A’s promise to ϕ is for B to (i) intend A’s ϕing and (ii) communicate that intention to A. Roth notes that this makes acceptance of a promise partly but not wholly an “internal” matter. Clause (i) of the proposal refers to an “internal” psychological state of B’s; but since clause (ii) of the proposal adds an “external” or public component, B’s psychological states alone do not suffice for acceptance. In this sense Roth’s proposal could be considered a hybrid model of acceptance: it is partly an “internal” mental state and partly a public act.
All the same, Roth’s subsequent analysis and arguments turn mostly on clause (i). (This is reflected in the name he gives his proposal: Accepting as Intending.) And while Roth thinks clause (ii) may in the end drop out as redundant (p. 93), he insists on the necessity of (i): saying (for instance) that you accept a promise does not constitute your accepting it, if your speech act is not in fact accompanied by an intention on your part that your promisor ϕ. That is insincere acceptance, says Roth, which is not a subspecies of acceptance and does not generate promissory obligation (n. 17, n. 33). This worried me a bit: I wondered how well the “internal” component of Roth’s hybrid conception squares with his argument that due to epistemic and other difficulties, “internal” psychological states are ill-suited to be the basis for promissory obligation. Roth’s hybrid conception of promise acceptance also seems to contrast with the dominant understanding of normative powers in general—even though the former would appear to be a species of the latter, if it does the normative work Roth assigns to it. Exercising a normative power is typically understood, I take it, as something we do, via “performative” speech or other public acts. I would not have thought it is generally conceived as a hybrid of what we do and what we are thinking “internally.” This would have been an interesting point to address.
Any worries one might have about (i), though, must be balanced against the arguments Roth offers for the idea that accepting a promise can best be understood in terms of intending the promisor’s ϕing. 1) Just as acceptance by a promisee settles a certain matter that prior to acceptance was up in the air, a salient aspect of intention as a mental state is that it too settles a practical question that previously was open. 2) It is possible for acceptance of a promise to lead to fulfillment of the promise via an intuitively deviant route. The similarity with the problem of “deviant causal chains” should strike us. 3) The acceptance of promises is subject to norms, e.g. of consistency, to which intentions (but not other conative or desiderative states) are also subject. One can’t, Roth says, reasonably and consistently accept two promises which are inconsistent (in the sense that A’s fulfilling his promise is incompatible with C’s fulfilling hers). If accepting promises involved intending their fulfillment, we would have an elegant explanation of this fact, since it is well known that one cannot reasonably and consistently intend two incompatible things.
Roth gives a fourth reason in favor which I’ll mention later. He also spends some time on why it is not problematic to think that someone can intend a different person’s action. We should not, he argues, retreat to a weaker content for the intention involved in promise acceptance.
Cognitivism about intention as entailing non-evidential warrant for belief. By general consensus, while one can perhaps desire to do something which one knows or believes one cannot do, one cannot intend to do such a thing. Many philosophers have concluded from observations such as these that intention involves belief: if I intend to ϕ, I thereby believe that I will ϕ. (Contrapositively: if I don’t believe I will ϕ, I cannot be said to intend to ϕ.) Roth calls this idea Cognitivism about Intentions (although some writers use that label to refer to a much stronger thesis, according to which intention somehow reduces to or is merely a species of belief). Roth points out a little-remarked-upon implication of cognitivism about intention: one which (again) I take to be of interest independent of what Roth does with it. When I decide to ϕ, and thus form an intention to ϕ, I thereby come to believe that I will ϕ, since intention (according to cognitivism) implies belief. But notice that I do not form the belief in question on the basis of evidence. Nor does the warrant for my belief derive from evidence: rather, in such cases my belief “has a warranted status by default” (p. 109). Cognitivism about intention thus seems to imply that some of our beliefs, at least, possess a non-evidential form of warrant.
Putting the above points together to respond to a well-known worry about Scanlon’s account of promissory obligation. On T. M. Scanlon’s account, the key trigger for promissory obligation is A’s leading B to expect that A will ϕ. (Some further conditions are also required.) Let us suppose that A leads B to expect this by, specifically, promising (as opposed to via some other route). Still, it may seem mysterious how exactly A’s promising to ϕ leads B to expect that A will ϕ, and thus how promissory obligation can ever get triggered. If A can only be expected to ϕ because or insofar as she is obligated to ϕ, but she cannot be obligated to ϕ unless and until B expects her to ϕ, a worrying circularity looms. Much ink has already been spilled (notably, by Kolodny and Wallace) about this problem; Roth points out that the constellation of views he has put on the table offers a novel response to it. Suppose A promises B that she’ll ϕ. As we have seen, A is not yet under a promissory obligation to ϕ: that will become the case only if and when B accepts A’s promise. Suppose he does. Then, as per Accepting as Intending, B now intends that A ϕ. Since—by Cognitivism about Intentions—intention involves belief, B now believes that A will ϕ, albeit not on evidential grounds. We could summarize this process by saying that A has led B to expect that she’ll ϕ. (She initiated the whole process, after all, by promising.) Thus Scanlon’s theory will predict that promissory obligation has indeed been triggered, as of course it has.
I want to close by mentioning a fifth theme which runs through various of Roth’s arguments: “directed” obligation and the special status of the promisee. One of the distinctive features of promissory obligation is that it seems to be a “directed” obligation, with respect to which the promisee is in a special position. A promisor is not merely under an obligation to ϕ; she owes it, specifically, to the promisee to ϕ. If she doesn’t ϕ, she does not merely act wrongly; she wrongs the promisee. And only the promisee can release the promisor from her obligation. Roth thinks the considerations his theory puts forward—and especially his emphasis on acceptance by the promisee—can explain the above features, and more generally the “directed” character, of promissory obligation. The promisee, says Roth, is in a special position to object to non-performance because of his power of acceptance, which as we have seen gives him effective authority to settle (to the extent an intention ever settles) whether the promisor ϕs. I found Roth’s explanations of this idea rather dark and hard to follow, although this may just reflect how difficult it is to explain or illuminate the source of directed obligation. What is clear is that Roth thinks acceptance by the promisee is necessary for the generation of a “directed” promissory obligation. Absent (or prior to) acceptance, he suggests, any residual obligations around the penumbra of a promise would not be “directed” (see, e.g., n. 9, n. 18). This is a surprising claim in relation to the literature on promising, since everyone seems to think promissory obligation is directed, while not everyone thinks promisee acceptance is necessary for promissory obligation, and fewer still give it the prominent role Roth does.3