We're reading: Rationality Through Reasoning (Wiley-Blackwell, 2013)
Here is a full history of Broome's work on these topics.
From the Preface about how his views evolved on the way towards the book:
First, I disentangled rationality from normativity in general. Many philosophers think of rationality as a sort of enforcer for normativity: it is your rationality that makes you do what you have a reason to do, or at least what you believe you have a reason to do. I now think that rationality is much less tightly connected with normativity than that. Second, I disentangled reasoning, which is something a person does, from rationality, which is a property of a person and her mental states.
My account of reasoning has gone through several revolutions, each correcting an initial mistake of mine. At first I was deceived by a similarity between the contents of instrumental practical reasoning and the contents of theoretical reasoning by modus ponens. I thought that the two were somehow fused together. I have now concluded that their similarity is only superficial. A second mistake was to assume that, when reasoning is correct, it is made correct by requirements of rationality. I now realize that reasoning is made correct by permissions, not requirements. Correct reasoning is not reasoning you are required to do by rationality, but reasoning you are permitted to do by rationality. This seems intuitively obvious, but I understood it properly only as a result of facing up to an objection to my previous account of reasoning that was shown me by Kieran Setiya. A third mistake was to assume that reasoning --- at least when it is conscious and something we do --- has to be conducted in language. This may be true, but a discussion with Paul Boghossian persuaded me it is best not to assume it.
Core theses of the book:
Rationality requires you, among other things, to: intend to do what you believe you ought to do.
Reasoning is a process by which you can bring yourself to comply with rationality's requirements (when that doesn't happen automatically), without necessarily deploying higher-order beliefs.
Chapter 12: rejects that reasoning must involve higher-order belief, that you ought to have some attitude Chapter 13: FO theoretical reasoning Chapter 14: practical reasoning, where your premises/conclusions aren't always beliefs Chapter 15: explicit reasoning: do attitudes have to be expressed in language to reason with them? Chapter 16: enkratic reasoning (pp. 288-294, short)
Broome identifies the "central" meaning of "ought" through the Enkratic Principle, it's the meaning under which (he says) it's true that:
Rationality requires: B(you ought to F / that you F) ⊃ you intend to F (pp. 22-3, 34, 89-90, 136, 170ff)
Schroeder's 2011 Phil Review paper is a good counterpoint to Broome's §2.3, on owned oughts and their complements. Schroeder distinguishes epistemic uses of "ought" (meaning roughly "likely"), evaluative uses wrt some ideal, and deliberative uses. Broome's core use excludes Schroeder's first two uses, and includes his third use, but is somewhat broader in ways that Schroeder criticizes.
Broome says there may be "unowned" normative oughts, as in "Life ought not be so unfair" (p. 18). He reserves the grammar "It ought not be the case that ..." to explicitly signal these. These sound like Schroeder's "evaluative" uses.
Broome's central use is an "owned", "normative" ought, having to do with normative reasons had by the ought's owner. (Later Broome will say that "you ought to F" is synonymous with "your reasons require you to F.") Deontic logicians call these "personal obligations." The ought is owned by the person of whom something is required, and who is responsible/at fault if the requirement is not met. An "ought" claim may describe an owned ought without ascribing the ownership, as in "M ought to get punished" = "Someone (e.g., the judge) ought that M gets punished".
Linguists treat the following as semantically equivalent (where "PRO" is a silent pronoun that gets its reference from the subject of the verb it's embedded under):
N expects PRO to get a hat
N expects N to get a hat
N expects that N will get/gets a hat
Broome wants us to think of "ought" as working similarly, so that these are equivalent:
N ought PRO to get a hat
N ought N to get a hat
N ought that N will get/gets a hat
Schroeder challenges this analogy; instead he likens the deliberative use of "ought" to verbs like "neglect," "forget," and "aspire," all of which also take infinitival complements but don't permit arbitrary propositional arguments.
When the ought's owner is not the agent of what ought to be done, as in "The judge ought that M gets punished," is this the same as: "The judge ought to see to it that M gets punished"? Broome says no, the judge might ought that M gets punished, but doesn't need to see to it that it happens, because her subordinates discharge the responsibility for her (p. 16). Is it the same as: "The judge ought not let it be false that M gets punished"? Broome isn't sure (pp. 16-17).
The content of (even an owned, normative) ought might be that something be a certain way; it needn't always directly specify what agents ought to do. Broome wants to say you ought: not believe in fairies, know your own name, be at work by nine, not prefer your own lesser good to your greater, intend to keep a promise when you make it, be careful in the dark, wear a hat when it's sunny, not break the law just because you'd benefit, either leave your husband or stop mistrusting him (pp. 8, 17, 23).
Others, like von Wright's early paper "Deontic logic" (Mind 1951), Geach "Whatever happened to deontic logic?" (Philosophia 1982), and the Schroeder paper linked above, restrict the relevant use of ought to complements that specify act-types (including refrainings as well as performings), rather than arbitrary propositions.
In the other direction, some like Chisholm "The ethics of requirement" (APQ 1964) and Williams "Ought and moral obligation" (in his Moral Luck 1981) attempt to reduce all the relevant oughts to unowned oughts: "N ought to F" = "It ought to be the case that N Fs." Broome argues against this at pp. 18-20: if A promises to be Fer than B, and B promises to be Fer than A, and promises generate oughts, then we'd have here two conflicting unowned oughts, but deontic conflict if possible at all shouldn't be so easily come by. If the oughts are owned by different agents, there is no problematic conflict. Also (p. 24) Broome says: only owned oughts engage with practical rationality through the Enkratic Principle. (Though they may be "agent-neutral" oughts, owned by everyone.)
On pp. 20-21, Broome discusses a more complex reduction due to John (Jeff) Horty. Here "N ought to F" = "It ought to be the case that N sees to it that N Fs." Broome's main reservation here is a desire to distinguish ownership of an ought from agency. It may be that the captain owns the ought that the hatch be closed, but that the deckhand is who he ought to have close it. (We're not supposed to hear the claim "The captain ought to close the hatch" as true in this scenario. Broome is thinking that the claim "The deckhand ought to close the hatch," understood not to ascribe ownership, would be wrongly analyzed by Horty.)
Unlike Wedgwood, Broome thinks the central ought can also apply to mental states you can't control (even indirectly) through intentions and plans.
In §2.4, Broome argues that his central ought is "unqualified," or the target of your all-things-considered judgment, rather than restricted as in "you ought, so far as prudence goes, ..." For the latter idea, Broome will prefer the locution "Prudence requires you to ..."
§§3.1-3.2 are critical of some attempts to distinguish "objective" from "subjective" oughts. In §§3.3-3.4, Broome argues that "in cases where consequentialism applies," the central ought is one that pays attention to your evidence/reasonable expectations, not just to the values of the possible outcomes. See our discussion of the Miner case.
Some Rs explain why the normative fact that N ought that N Fs obtains. Broome calls these "pro toto" reasons for N to F (in earlier work like his "Reasons" (2004), he called them "perfect" reasons). They have to explain and not merely entail the normative fact. Pro toto/perfect reasons may (but needn't) themselves be normative facts.
Other Rs participate in "weighing" explanations of whether N ought that N Fs. Broome calls these "pro tanto" reasons for N to F/not-F.
Broome allows that ought facts might not always have weighing explanations, e.g., the fact that you ought not have contradictory beliefs (pp. 58-60).
Can there be "enticing" reasons as well as "peremptory" ones, where only the latter contribute towards making it the case that something is impermissible? Broome accommodates this idea to some extent, but insists that when the balance of reasons supports Fing, you still ought to F (even if you are in some sense permitted not to F) (pp. 60-61).
If part of the explanation of why you ought to F is that, though you have promised not to F, your friend released you from your promise, that is an example of a non-reason role in the explanation of why you ought to F (p. 62).
§4.4: Some philosophers define "You ought to F" as "The reasons for you to F outweigh the reasons for you not to F." But what if you ought to F, and there is no weighing explanation of that? We might still call this "having most reason to F."
Unlike with "ought," Broome thinks there is a grammatical difference between reasons-claim that do and those that don't ascribe ownership:
There is a reason [for N] to F = For N, there is a reason to F = N has a reason to F
There is a reason for [N to F] = Someone has a reason for it to happen that N Fs
Broome doesn't accept the usage of "N has a reason to F" that's meant to explicitly signal that the reason is epistemically accessible to N.
Owned reasons can be agent-neutral (owned by everyone), or agent-relative (pp. 66-69).
Broome considers the theses that: You are rational if (Sufficiency) and only if (Entailment) you respond "correctly" to your reasons (that is, the totality of normative reasons owned by you). In other words, you F whenever you ought to F / your reasons require you to F. (For Sufficiency, we should also add an extra clause that you F because of your reasons; see TODO counterfactual.)
There is a "Quick Objection" to Entailment. Broome considers and rejects some rebuttals to it, in the end leaving it standing, and so rejecting Entailment. The Quick Objection says: What if your reasons require you to F, but you believe (and so do) otherwise? Can't you in at least some such cases be rational, despite not doing what your reasons require?
One rebuttal (p. 75): a rational person must have correct beliefs about what their reasons require. This may be buttressed by a view where reasons always consist of attitudes rather than external facts. But Broome thinks a rational person can have attitudes they are mistaken or ignorant about.
Another rebuttal (pp. 75ff): mightn't we be "strictly liable" to our reasons, so that ignorance doesn't exculpate us for failing to comply with them? (There's a different use of "strict" to contrast with "slack", merely pro tanto or commendatory reasons.) Here too, Broome suggests, the view may be buttressed by thinking of reasons generated by the mere having of certain attitudes. Eg., Bp may be a reason to ¬B¬p; Bp and B(p⊃q) may be a reason to Bq; intending F may be a reason to ¬intend(¬F); and so on.
Why should we think attitudinal reasons can impose strict liability? Because you can respond correctly to them without believing that they obtain, or believing that they are reasons (pp. 77-8). (I'm not clear on why that supports the claim. Also, are we supposed to think that these things are only true for attitudinal reasons? Broome's discussion of basing at pp. 188-9 may be relevant.)
Can attitudinal reasons support limited versions of the initial theses against the Core Objection?
First (pp. 79-81), Broome argues that responding correctly to attitudinal reasons would not suffice for being rational. Such reasons may oppose each other. When they do, Broome says that responding to their totality will "inevitably" involve acting in a way that some of them oppose. (I pointed out in seminar that this isn't right: responding to the attitudinal reason generated by Bp would involve abandoning your conflicting B¬p, so you wouldn't have conflicting attitudinal reasons any more. But presumably some cases would work as Broome imagines.) If you act in a way that some of them oppose, you'd be irrational. (For Broome, the (alleged) attitudinal reasons correspond to non-negotiable constraints on rationality.) As he summarizes the point on p. 81, reasons are "slack" (can be outweighed), but the necessary conditions of being rational are "rigid."
There's an interesting reply considered on pp. 80-81, which says that two attitudinal reasons can't oppose each other unless you've failed to respond correctly to some other attitudinal reason, and hence are already irrational.
Second (pp. 81-82), Broome argues that there are no attitudinal reasons. When your evidence for q falls short, you can't by your mere belief in (p and p ⊃ q) bootstrap a new reason for q into existence. (Consider the extreme case when q = p.) Similarly, intending to F does not bootstrap any new reason for you to F. I observed in seminar that these bootstrapping arguments used to play a much larger role in Broome's arguments for wide-scoping. Other bootstrapping arguments occur on pp. 144-6, 179, 184.
So the Quick Objection to Entailment still stands.
pp. 83ff discuss Sufficiency: whether responding correctly to all one's reasons (not just attitudinal ones, if there are such) suffices for being rational.
One argument for Sufficiency: Perhaps (a) you have reason to be rational, and (b) therefore reason to satisfy every necessary condition for being rational. Chapter 11 will examine (a). Even if (a) is granted, does (b) follow? In the first weeks of seminar, we discussed how there might be mere enabling conditions for you to be justified in believing p, upon which your belief need not be based for it to be well-founded. Similarly, perhaps there might be necessary conditions for being rational that you don't have reasons to satisfy? (see also Broome p. 119, 135)
Even if (b) is true, it would only follow that you have some reason(s) to satisfy the necessary conditions of rationality (Broome calls these "reasons of rationality", and asserts on pp. 83-4 that any such reason would be "state-given"). You may also have other reasons not to do so so, and so "responding correctly" to the totality of your reasons might involve failing to comply with one of the conditions of rationality (p. 83).
A different argument for Sufficiency (inspired by Raz, Kolodny): if your mind is "properly adjusted" to the world, it will automatically be internally coherent and so rational. Broome counters: you can have some reasons not to be rational (e.g., being known to be irrational can be an advantage in a nuclear showdown). Who's to say that the totality of your reasons won't sometimes support your not being fully rational?
For beliefs, maybe evidentialism would support the Raz/Kolodny position. Their opponents would instead argue that evidence can sometimes make it permissible to Bp and also permissible to B¬p. In such cases, Broome insists it would be incompatible with rationality to simultaneously have both beliefs, and this won't have been explained by the constraint of "responding correctly to the evidence" (pp. 84-5).
Broome is more confident that your practical reasons (at least the "worldly" ones, meaning: ignoring any "reasons of rationality," if there be such) can permit you to intend F and also permit you to intend ¬F; but as with beliefs, Broome insists that if you intended both, you'd be irrational; and the Raz/Kolodny position doesn't explain that (pp. 85-6).
Broome doesn't accept that responding correctly to beliefs about reasons suffices for being rational (pp. 88, 90), but in some senses (explored in Chapter 6) he agrees it is entailed by being rational.
There's a quick discussion of the "Direct Enkratic Condition" (DEC) which says:
If you're rational, then you will F when you believe you ought to F / your reasons require you to F.
at pp. 88-89 (see also p. 32), and then a longer discussion of limited versions of it in §6.2. Broome's objection to the unlimited version is that he thinks "rationality supervenes on the mind" (pp. 89, 151-2); so a subject may be rational but fail to F solely because of external obstacles.
This suggest limiting the DEC to cases where F is adopting some attitude or mental state (pp. 92 ff; this is roughly thesis C+ in Kolodny 2005).
At pp. 93-4, Broome argues that even if such a limited DEC were true, it couldn't replace the other conditions of rationality. Why should "strict liability" be confined to just this one condition?
At pp. 94-6, Broome argues that this limited DEC is not true. In Kavka's Toxin Puzzle, you may be rational while not having the intention to drink, e.g., if you believe you won't drink when the opportunity comes, and that prevents you from having the intention. Yet you may believe your reasons require you to have that intention. Similarly, you may rationally belive there is no God in response to your evidence, while believing that Pascalian considerations give you all-things-considered reason to believe in God. Broome makes the following observations about what drives these examples:
We discussed in seminar which of these features were essential, given Broome's other commitments.
pp. 96ff discusses the further, Scanlon-inspired limitation that When Fing is adopting some attitude or mental state, being rational entails that you F when you believe that your object-given reasons require you to F. Broome allows that this may be true, but says there are few cases where subjects have beliefs about what their object-given reasons require. So as above, there must also be other necessary conditions for rationality.
Broome is more sympathetic to the (Non-Direct) Enkratic Condition:
Being rational entails that you will intend to F when you believe you ought to F / your reasons require you to F.
He accepts this, not merely as a necessary condition for being rational, but also as articulating part of what rationality requires/prescribes (pp. 22-3, 34, 89-90, 136, 170ff). But again, there are other requirements of rationality too, that being rational also entails complying with.
§6.3 considers the Parfitian view that being rational entails responding correctly to beliefs whose contents if true would be reasons ("P-beliefs"). This is akin to a view of Schroeder's.
Can you respond correctly to just a P-belief? Doesn't responding correctly to reasons have to be more holistic? Perhaps you can respond to a P-belief whose content if true would be a pro toto/perfect reason, and would impose strict liability. That buttresses the case that failing to respond would constitute being irrational.
Broome discusses an epistemic example at pp. 102-3. What if, in addition to the P-belief, you also had a normative belief that you should not respond it? In these cases, Broome think you can rationally ignore the normative belief.
Broome discusses practical examples at pp. 103-5. In these cases, he thinks it's less defensible that you could rationally ignore a bizarre normative belief that said you shouldn't respond to the P-belief. The key argument is on p. 105, and I didn't find it convincing. Broome argues that you may be rational in having the bizarre normative belief, because your beliefs aren't inconsistent, and it may be supported by your (e.g., testimonial) evidence. He says "we are not entitled to impugn your rationality just because you hold a bizarre normative theory." OK, but this doesn't settle the question that paragraph is meant to address, whether the subject may in virtue of her total state --- which includes not just the rationally-held normative belief, but also the P-belief, and thus the inevitability of either violating the Enkratic Condition wrt the normative belief, or failing to respond to the P-belief, or having conflicting intentions --- whether this subject may in fact have no fully rational response. Unlike Broome, I am more sympathetic to such possibilities. Also, unlike Broome, I am not so sure that having conflicting intentions must always be irrational.
Chapter 7 begins by disambiguting the verb "requires": (i) one use is synonymous with "needs," as in "The patient requires/needs attention"; (ii) another concerns what's required in order to have some property; (iii) a third concerns what some (real or alleged) authority prescribes. §7.2 discusses different ways of analyzing sense (ii), and suggests that the appeal of Standard Deontic Logic (SDL) may derive from notion (ii)'s "friendliness" to that formalization. The rest of Broome' discussion concerns requirements in sense (iii), which are much less friendly to SDL. Later in the discussion he also allows for rational prohibitions and permissions.
For Broome, the order of explanation is: an authority's requirements/prescriptions come first, and the property of satisfying/complying with them is derivative (p. 117-18).
What an authority requires/prescribes may differ in different worlds, in ways that are sensitive to our contingent imperfections (our not having the property of fully complying). "What prudence requires of you in your actual imprudent state may not be what it would require of you if you were prudent" (pp. 118-19). Broome describes this as the requirements being (modally) "local." This theoretical choice will affect how one handles the "deontic puzzles" we described in previous weeks.
Evaluating what deontic logic might be supported by "Authority so-and-so requires that p": Broome rejects axiom K (R(φ ⊃ ψ) ⊃ (Rφ ⊃ Rψ)), since you might be required to eat heartily if you exercise, and required to exercise, but (since you won't exercise) not required to eat heartily --- at least not if the requirements are to respect your contingent imperfections. Similarly, he rejects rule RM (From ⊢ φ ⊃ ψ, infer ⊢ Rφ ⊃ Rψ), since it might be that you're required to exercise and eat heartily, but (since you won't exercise) not required to eat heartily. However, Broome does accept rule RE (From ⊢ φ ⊂⊃ ψ, infer ⊢ Rφ ⊂⊃ Rψ).
In rejecting K, Broome rejects "deontic detachment" for this logic, and he also rejects "factual detachment." But he does accept the principle of "necessary detachment" (R(φ ⊃ ψ) ⊃ (Nec φ ⊃ Rψ)), interpreting "unalterable" facts (such as facts about promises you've already made in the past) as "necessary" (pp. 123-5). He doesn't extend the same treatment to mere psychological necessity (p. 145).
Broome isn't prepared to accept axiom D (Rφ ⊃ ¬R¬φ) for a general logic of requirements, but he is sympathetic to this axiom being true of rationality, and perhaps of all normative sources of requirement. He's especially sympathetic to its being true for the notion of "the totality of your reasons require φ / you ought to φ" (pp. 122-3, 128, 136-8). Another way that the effect of axiom D is sometimes achieved would be with the axiom ¬R⊥, and Broome expresses skepticism that rationality requires any "analytic or metaphysical" impossibilities (pp. 156-7).
Another theorem of SDL would be R⊤; Broome doubts that this could be true of rationality's requirements (p. 157).
Chapter 8 explores relations and differences between "wide-scope" and "narrow-scope" conditional requirements. In the first case, when the antencedent of the conditional fails, you satisfy the requirement; in the second case, you avoid it. Since the requirements would be violated in the same circumstances, they have the same upshot for who has the property of fully complying with the requirements (p. 133-4, 148; this supposes that the conditional in question is just ⊃).
Kolodny prefers a narrow-scope Enkratic Requirement. If that were correct (and if rationality were normative, which Kolodny denies), it would follow that there can be normative conflicts, because you could BOp and BO¬p. (Perhaps believing that would require you to already be irrational.) Kolodny's Enkratic Requirement would also entail that there can be objectionable bootstrapping (pp. 136-8, 144-6).
Schroeder, Kolodny, and Price complain that wide-scope requirements have symmetric compliance conditions. So far as such requirements are concerned, you might refrain from BOφ on the basis of intending not to φ; or refrain from B(m is a necessary means for e), on the basis of intending e but not m. Broome agrees that to do those things would be irrational, but argues that their irrationality should be explained by your violation of other principles of rationality, such as respecting your evidence or respecting "basing prohibitions" (pp. 138-40, 186-9). Moreover, even if our original Enkratic or Means-End requirements had non-symmetric compliance conditions, that wouldn't yet explain the negative requirement not to base attitudes in these ways (pp. 142-3).
Broome thinks rationality also issues some basing permissions (pp. 189-91), including permission to form some intentions on no basis.
Broome says in general, he's inclined to formulate requirements "strictly," in ways that don't yield (much) to human cognitive limitations (p. 153-4). But in fact his application of this seems half-hearted: see the rejection of Consistency, Deduction, and some formulations of a Closure/Modus Ponens Requirement (and his comments on p. 159).
The fundamental Instrumental/Means-End Principle says:
Rationality requires: If you intend that e, believe that m is a (circumstantially, if not indepensably) necessary means to e, and believe that m is up to you, then you intend that m (pp. 159ff).
Broome's formulation also includes some ms that would not naturally be called "means" for e, which doesn't bother him. Also, the means may fail to be "believed up to you" in the sense involved here because it's a believed side-effect of something else you intend.
Broome denies that "cognitivist" construals of the Instrumental/Means-End Principle (Wallace, Setiya, Ross) capture the whole truth (pp. 163-8).
The Instrumental/Means-End Principle still applies to conceptually unsophisticated people (pp. 168-9).
There arguably is also a requirement for you to take the (believed) best means to your ends, but Broome doesn't know how to formulate it (p. 169-70). Further, there is sometimes a requirement for you to choose a means, if not doing so will prevent you from achieving the end (p. 170).
The official Enkratic Requirement (using some of Broome's shorthand) says:
Rationality requires: if you believe you ought that p, and believe that it's up to you whether or not p, then you intend that p (pp. 170ff)
Often this requirement won't apply, because you know that your intentions don't control the relevant p (e.g., if p is you forming a belief, desire, or intention). Broome implicitly rejected more "direct" versions of Enkratic Requirement in the §6.2 discussion of the DEC. The latter only explicitly concerns necessary conditions for being rational, not what rationality requires.
Broome counters objections to the Enkratic Requirement at pp. 173-5, 291-4.
§10.1 discusses a rational requirement that "you do not forget" your intentions, that is, they persist until you deliberatively revisit them (Broome calls this "(re)considering whether to F"), or come to believe that it's no longer up to you whether you F. On the other hand, §10.3 suggests rationality may permit you to forget beliefs about matters you have no interest in.
§10.2 considers whether there is some further requirement for you to do what you intend to do. Broome argues no, his Instrumental/Means-End Requirement already requires you to take believed means to what you intend; and that and the Persistence of Intention Requirement capture all the truth there is in the neighborhood. Broome criticizes views like Velleman's and Korsgaard's that claim intending to F always generates a reason to F (pp. 183-5). One of his complaints is that we need to explain why failing to carry out your intention to F can sometimes result in you being irrational, even when you ought not to F.
Broome distinguishes three questions, of decreasing ambition:
He considers some arguments that address the middle question. In the end he thinks the answer to the middle question is yes, and that the reason in question is non-derivative, but he knows of no compelling argument for that answer, or explanation of why it is the answer.
At pp. 195-6, Broome reconstructs an argument from Dreier that it makes no sense to ask the middle question. Going by Broome's reconstruction, Dreier's view seems to be: you can't be given a reason to do what rationality requires if you aren't already rational, so asking the question makes no sense. Dreier relies on the premise that if something is a reason for you to F, it must be able to motivate you to F; but Broome rejects this.
Broome argues against the idea that we always have derivative reasons to do what rationality requires, since e.g., you can falsely believe you ought to F even when you have no reason to F. In such a case rationality would require you do what you have no (other) reason to do (pp. 197-8).
Also, it's hard for a derivative account of rationality's normativity to deliver a yes question to the middle question (where rationality's requiring F is itself a reason for you to F), as opposed to a yes question to the weaker question (where it merely entails you have a reason to F) (pp. 198-9).
Next Broome considers whether we might have reason to "have a rational disposition," since this may be the best means of getting much of what we ought (for other reasons) to achieve. Even if that's so, though, he argues that we wouldn't thereby get any reason to comply with the individual requirements of rationality. The disposition would merely cause us to generally so comply (pp. 200-1).
Next Broome considers whether we ought to be fully rational, and as a result have reason to satisfy the necessary coditions of being fully rational. He expresses doubt about both this premise and the inferential step (pp. 201-2).
Finally, Broome considers whether we might merely have some reason to be rational (p. 203). His complaint here seems to be that the most defensible version of that premise won't be strong enough to give us what we're after. It would only entail that we have reason to satisfy some of the requirements of rationality.