Posted 10/5/99

"A Problem For Expressivists" Analysis, Vol. 58, No. 4 (October, 1998) 239-251, by Frank Jackson and Philip Pettit (Australian National University),

Reviewed by Michael Ridge (Australian National University)

(Click here for Jackson and Pettit's reply)

e-drop.gif - 1.3 Krank Jackson and Philip Pettit (henceforth "JP") offer an interesting critique of expressivism. On their account, expressivism makes two claims:

(a) Ethical sentences are not truth-apt.
(b) Ethical sentences do not express beliefs; they express (but do not report) attitudes.

JP argue that expressivists must admit that becoming competent with ethical utterances involves learning to make them only when one believes one has the relevant attitude. For expressivists hold that communicating our attitudes is the function of ethical utterances, in which case sincerity demands that we not utter an ethical sentence unless we believe we have the relevant attitude. So (b) is false, as long as we suppose that this commitment, as reflected in well-entrenched and clear-cut (henceforth, 'robust' abbreviates 'well-entrenched and clear-cut') conventions, means that ethical utterances express one's belief that one has the relevant attitude. This, in turn, means that (a) is false, if we grant that a belief expressed in virtue of such conventions provides the utterance's truth-conditions.

Properly understood, expressivism is immune from the JP's critique. Grant that their objection refutes expressivism as they define it. It remains possible that expressivism might be more sympathetically characterized as follows:

(i) If ethical utterances are truth-apt, then they are not truth-apt in virtue of expressing beliefs that provide their truth-conditions.
(ii) Ethical sentences do not express ethical beliefs, but express ethical attitudes.

As some expressivists allow that ethical utterances are truth-apt, whereas other expressivists deny this, we should characterize expressivism in a way that is neutral on this question. Still, insofar as they allow that ethical utterances are truth-apt, even sophisticated expressivists do not think that they are truth-apt in virtue of expressing mental states that are representational. Grossly oversimplifying, representational states like beliefs are "to fit the world," whereas attitudes are "to have the world fit them." I shall here use 'belief' to refer to representational mental states and 'judgment' to refer to mental states that are not necessarily representational. Rather than appealing to the truth-conditions of beliefs expressed by ethical utterances, sophisticated expressivists hold that such utterances are truth-apt in virtue of a "quasi-realist" account of truth or in virtue of a minimalist account of truth and truth-aptitude. Hence, I have softened JP's (a) into the conditional claim of (i) to allow such views to count as expressivist. Finally, on JP's account, the expressivist denies that ethical utterances express any beliefs whatsoever, whereas my expressivist denies only that they express any ethical beliefs. Hence the shift from (b) to (ii).

Suppose we grant that ethical utterances express beliefs. From this it follows that (b) is false. However, it does not follow that (ii) is false, for it remains an open question whether the beliefs are ethical. On any plausible construal, the beliefs expressed are the agent's beliefs about her attitudes and such beliefs are famously not good candidates as ethical judgments. As any expressivist shall insist, a belief with the content, "I approve of X" is not an ethical judgment, as one might always intelligibly admit that one approves of X and still wonder, "Is X really good?" The ethical judgment remains the relevant attitude and not the belief about the attitude, according to my expressivist. Nothing in JP's argument challenges this.

So JP's argument does not undermine (ii). Does it undermine (i)? If we were to grant that the beliefs expressed by ethical utterances provide the truth-conditions for those utterances, it would. However, if the beliefs expressed by ethical utterances are not ethical beliefs we should not grant this. For it would be perverse to suppose an ethical utterance's truth-conditions are provided by a non-ethical belief.

So it is a mistake to infer from "Utterances of type U express beliefs of type B in virtue of robust conventions" to "B-beliefs provide the truth-conditions for U-utterances." Actually, we have good independent reason to reject this form of argument. For example, JP seem right in holding that in uttering, "Shut the door!" a speaker expresses her belief that she is commanding her interlocutor to shut the door in virtue of robust conventions. However, it is implausible to conclude from this, as JP do, that the "Shut the door!" is truth-apt. Such an inference moves directly from platitudionous common-sense to highly revisionary theory.

JP argue that it is not to the point that it is "crook" English to say such utterances are true or false, on the grounds that if we take ordinary language as sacrosanct, then the expressivist automatically loses, as in ordinary language we certainly do speak of ethical utterances as being true or false. However, sophisticated expressivists grant that ethical utterances are truth-apt, though not in virtue of expressing ethical beliefs. Further, ordinary language is in any case to be given only defeasible weight.1

To take an even clearer counter-example, some interrogatives express beliefs in virtue of robust conventions. In asking, "Have you stopped smoking?" I express my belief that my interlocutor has smoked. Again, this emerges from a robust convention. However, it would be implausible to infer that questions have truth-values. When we start talking of questions being true or false, language really has gone on holiday.

Properly understood, expressivism is compatible with the contention that ethical utterances express beliefs. Specifically, it is compatible with this contention so long as these beliefs are not ethical, in which case they do not provide the truth-conditions for ethical utterances. If expressivism is mistaken, we must look elsewhere to find its mistake.2

1. JP also argue that there is no difference between, "I command that the door be shut" and "Shut the door!" that is relevant to truth-aptitude, and conclude that since the former obviously has truth conditions, then so must the latter. However, there is a difference between the two that is plausibly taken as relevant- the former is in the indicative mood; the latter is not.

2.Thanks to Frank Jackson, Simon Blackburn, Philip Petitt, and Michael Smith for useful discussion.

[To BEARS Homepage]