Sorites ISSN 1135-1349
Issue #18 -- February 2007. Pp. 56-59
An Argument against External Reasons
Copyright © by Jonathan Anomaly and Sorites

An Argument against External Reasons
Jonathan Anomaly

With the publication of «Internal and External Reasons» (1981) Bernard Williams breathed new life into an old debate about normative practical reasons. The question to which internalists answer `yes' and externalists answer `no' is deceptively simple: do all reasons for action depend on desires?Foot note 1 As stated, the question is doubly ambiguous. First, according to some, the question at issue is whether the judgment by agent `A' that she has a reason to do action `φ' entails a desire to act on that perceived reason.Foot note 2 Others argue, following Williams, that the issue concerns not the relation between desires and judgments about reasons, but the relation between desires and the existence of reasons.

The second ambiguity concerns the entailment relation in the generic internalist slogan that `reasons entail desires'.Foot note 3 The entailment relation is ambiguous between whether, necessarily, reasons give rise to desires or, necessarily, reasons owe their existence to antecedent desires. In the first case the desire is explained by the existence of a reason; in the second case the reason is explained by the existence of a desire. Existence internalists such as Williams lay claim to the second view.Foot note 4

On this interpretation, which I call constitutive existence internalism (CEI), practical reasons are constitutively tied to the contents of an agent's current desire set. Although he is often misinterpreted, Williams' endorsement of CEI is clear: «If A has a reason to φ, then ... there must be a sound deliberative route to Φ-ing which starts from A's existing motivations» (1995: 187).Foot note 5

Externalism is the denial of this conditional. Consider an example. According to externalists, an instrumentally rational psychopath may have a reason to treat all human beings with respect even though he lacks the capacity to be moved by moral considerations.Foot note 6 In other words, some externalists claim that from the fact that a psychopath lacks the capacity to see other human beings as more than mere objects, it does not follow that he has no reason to treat others with respect.

I shall now argue that all such externalist claims are incompatible with a plausible version of the ought-implies-can principle (OIC). The intuitive case for OIC is straightforward: if we reject OIC, we must accept the existence of obligations that cannot be fulfilled. Although such obligations may be found in ordinary moral thought and discourse,Foot note 7 they make little sense in an account of normative practical reasons. To suggest, for example, that a quadriplegic has a reason to run the Boston Marathon -- that he ought to run the race despite his inability to do so -- strikes us as absurd. Such a suggestion seems, at best, an expression of a wish that things were different than they actually are and, at worst, evidence of an incoherent belief -- namely, the belief that someone might act in a way that is physically impossible.

Before deploying OIC in an argument against externalism, its constituent parts must be parsed. For the purposes of the following argument, I take the ought component of ought-implies-can to apply to normative practical reasonsFoot note 8, the implication relation to be one of conceptual entailmentFoot note 9, and the can component to indicate motivational possibility (since CEI primarily concerns the relation between motivations and practical reasons).Foot note 10

Many externalists would claim that a psychopath -- such as Jeffrey Dahmer -- has a reason to treat people with respect, despite his inability to be motivated by moral considerations. Externalists are thus committed to an argument of the following sort:

1) J.D. ought to φ.

2) J.D. cannot φ.

3) It is impossible for one to do what one cannot do.

4) It is impossible for J.D. to φ.


5) J.D. ought to do something that he cannot do.

As the conclusion suggests, externalists are committed to the claim that there are cases in which people ought to act in ways that are impossible, given their motivational makeup or psychological profile. Internalists are skeptical, and their tacit endorsement of OIC is a plausible explanation of their skepticism. If this is right, internalists might revise the above argument (using OIC as a premise) as follows:

1*) J.D. ought to φ.

2*) J.D. cannot φ.

3*) Ought implies can.


4*) Either 1* or 2* is false.

5*) By assumption, 2* is true.


6*) 1* is false (It is not the case that J.D. ought to φ).

This argument may lead us to reject all strong externalist claims -- claims that fail to take into account the different capacities of people to be motivated by considerations that underlie reasons claims.Foot note 11 Perhaps, though, we can still accept weak externalist claims which apply only to people who can, in principle, be motivated by such considerations.Foot note 12

However, if weak externalists are characterized as those who both deny internalism and affirm OIC, they are compelled by theoretical consistency to acknowledge other factors that further limit the ability of individual deliberators to be moved by considerations that underlie reasons claims. These include contingent limitations in the availability of information, the potentially prohibitive cost of gathering information, and cognitive limitations in one's ability to imagine alternatives or acquire beliefs that would help one solve a practical problem. But once weak externalists accept such factors as conceptual constraints on practical reasons, the battle is won. Externalism collapses into some form of internalism.Foot note 13

Some internalists have defended their position by arguing that externalism fails to explain rational action, since external reasons claims do not refer to anything distinctive about an agent, such as her current beliefs, desires, or other psychological states.Foot note 14 But if the foregoing argument is correct, the decisive case against external reasons ultimately rests on the plausibility of the ought-implies-can principle. Unless we can find an argument that vitiates OIC as a theoretical constraint on practical reasons we are forced to reject externalism.


Jonathan Anomaly
Tulane University
<Anomaly [at] cal [dot] berkeley [dot] edu>

[Foot Note 1]

`Desire' is a philosophical term of art. Williams uses `desire' and `motivation' interchangeably, and includes under their rubric «dispositions of evaluation, patterns of emotional reaction, personal loyalties, and various projects...embodying commitments of the agent» (1981: 105).

[Foot Note 2]

Korsgaard's `internalism requirement' (1986) can be interpreted as a form of judgment internalism.

[Foot Note 3]

On this ambiguity see Hurley (2001).

[Foot Note 4]

I borrow the distinction between `judgment' internalism and `existence' internalism from Darwall (1992).

[Foot Note 5]

Williams clearly presupposes a `procedural' rather than `substantive' conception of rational deliberation. On this distinction see Hooker (1987) and Parfit (1997).

[Foot Note 6]

Whether we think of the capacity the rational psychopath lacks as psychological or motivational, the point remains the same. Externalists deny that any such idiosyncratic facts limit either the scope of reasons or the class of people to whom reasons apply. See Scanlon (1998: 363-73).

[Foot Note 7]

On this, see Sinnott-Armstrong (1984: 260) and Fischer (2003: 248).

[Foot Note 8]

As Parfit argues (1997: 121), to say that we ought rationally to φ is just another way of saying that we have a reason to φ.

[Foot Note 9]

Although an excellent case can also be made for logical or linguistic entailment, I ignore these issues here.

[Foot Note 10]

Internalists about theoretical reasons, in contrast, would focus primarily on epistemic possibility, as well as the cost and availability of information relevant to belief formation.

[Foot Note 11]

Considerations such as the fact that an action will harm X or help Y.

[Foot Note 12]

McDowell (1995) seems to endorse this position.

[Foot Note 13]

Scanlon (1998: 371-3) comes close to acknowledging this.

[Foot Note 14]

See especially Williams (1993).

Back to the SORITES portal