Jason Turner

Joshua Knobe (2003b) has recently argued that, according to the folk concept of intentional action, the answer depends on B's moral significance. In particular, if B is reprehensible, people are more likely to say that S intentionally brought it about. Knobe defends this position with empirical facts about how ordinary people use the adjective 'intentionally.'

Knobe's experiments involved people in a Manhattan public park. He first provided them with a version of the following scenario:

They were then asked to rate how much praise or blame (on a 0 to 6 scale) the chairman deserved for his actions and to state whether or not they thought the chairman intentionally affected the environment.

The only variable between subjects was the program's effect on the environment. In half the surveys, the program helped the environment, and in the other half the program harmed the environment. Surprisingly, although only 33% of respondents given the help condition said the chairman intentionally helped the environment, a whopping 82% of respondents given the harm condition claimed the chairman intentionally harmed the environment (Knobe 2003b: 191­92). The apparent conclusion is that the folk are more willing to call a side effect intentional when they think that effect is bad than they will when they think the effect is good. (A second study confirmed the first's results; see Knobe 2003b: 192­93.)

It appears that this result is not a consequence of pragmatic features of the word 'intentional.' Fred Adams and Annie Steadman (2004: 178­80) have suggested that, if the folk think someone can be blameworthy for bringing about B only if he intended to bring about B, they may classify the chairman's environment-harming as intentional in order to conversationally imply that his actions were blameworthy. A later study using the same vignette (Knobe 2004: 184­185) produced a similar asymmetry when respondents were asked, 'Did the chairman [affect] the environment in order to increase profits?' Since an affirmative answer implies the chairman affected the environment intentionally, but does not imply blameworthiness, this result suggests that the asymmetry stems from the folk concept of intentional action rather than pragmatic considerations.

If the conclusion is merely that there are cases in which the folk's intuitions about the intentionality of a side effect is somehow affected by the perceived moral content of the side effect, then there is little to dispute. There is, however, a deeper conclusion lurking in the background ­ the conclusion that the folk's concept of intentional action is fundamentally tied to objective moral considerations (Bratman 1984; Knobe 2003a, esp. 311­12). We could state this as the conclusion that moral facts are 'at the root of theŠ asymmetry in people's application of the concept intentional' (Knobe 2003b: 193, emphasis altered). The idea is that a correct analysis of the folk concept of intentional action would include irreducible reference to objective ethical facts. This is a striking conclusion, for it entails that a correct analysis of intentional action cannot be given in purely psychological and action-theoretic terms. If judgments of the intentionality of an agent's act are sensitive to the judge's evaluation of the moral worth of the act and independent of the actor's evaluation, then intentional actions are

If this grander conclusion is intended by Knobe, though, it is too hasty. To be sure, Knobe's results are consistent with the thesis that the concept of intentional action is fundamentally evaluative. There is an alternative hypothesis, however, which can account for Knobe's data and which keeps the concept of intentional action within the purview of action theory. In what follows I first present this hypothesis and then suggest ways to test it empirically.

The hypothesis takes Gilbert Harman's (1976) discussion of side effects as its point of departure. Harman considers a sniper who knows that, if he shoots at his target, he will alert his enemies to his presence (433). Harman claims that, if the sniper shoots, he intentionally alerts his enemy to his presence. Suppose this is correct ­ that is, suppose that the folk would call the sniper's alerting his enemy intentional. From this, Harman concludes:

Two modifications are required. First, Harman's thesis is distinctively evaluative ­ whether a side effect is intentional depends on 'what ought to be a reason not to do it.' In order to bring it into conformity with the view that the concept of intentionality is not fundamentally evaluative, we should replace this clause with 'what the agent takes to be a reason not to do it.' This factor will then depend only on the agent's psychological state with respect to the side effect, as opposed to any objective facts (moral or otherwise) about the effect.

There is an ambiguity in the phrase 'a reason not to A.' It may indicate either a factor which is decisive against A-ing or one which merely provides some motivation towards not A-ing. The latter is the intended interpretation; agents who satisfy the condition under the former interpretation are criticizably irrational.

The second modification is inspired by a case suggested by Alfred Mele and Steven Sverdlik:

Doris performs the operation in the face of a factor that counts against it ­ its risk to Pat. Did Doris intentionally kill Pat? What would ordinary people say?

Mele guesses that people would not say Doris intentionally killed Pat, but confesses that he has been wrong about such guesses in the past (2003: 329). Let us suppose, for now, that his guess is right. The problem appears to lie in the oddity of saying one intentionally did what one took pains to keep from happening. Since Doris made an effort to keep Pat from dying, it sounds strange to say that her failure in these efforts was intentional.

Taking these considerations into account, I suggest that the following conditions are jointly sufficient for a side effect E, produced by S's action A, being intentional:

(i) S knows that E will (or is likely to) occur as a result of A-ing,

(iii) S does not try to keep E from occurring.

Clearly, in the harm condition, Knobe's chairman satisfies conditions (i) and (iii). Does he satisfy (ii)? We are supposed to think that he does not ­ after all, he explicitly states that he does not care about the environment. But do respondents reading the vignette interpret the chairman's claim that he does not care about the environment as evidence that the program's negative environmental impact (in the harm condition) does not count against the program from the his point of view?

It seems unlikely that respondents would interpret this claim so literally. The locution 'S cares about y' implies that y has some intrinsic value for S. Yet it is entirely plausible that the environment has no intrinsic value for the chairman and yet the policy's negative environmental impact counts against it from the chairman's point of view. Not harming the environment is likely to have instrumental value for the chairman. Implementing the environmentally harmful policy incurs the risk of protest, legal investigation, dissent from the lower echelons of the company, etc. Thus respondents are likely to see the policy's environmental harmful effects as counting against it from the chairman's point of view.

This plausibility is increased if we think ordinary people employ a principle of moral charity when interpreting others' actions. It is a matter of considerable debate whether one can believe that A-ing is wrong while being entirely unmotivated to avoid A-ing. It is not improbable, though, that the folk conception of morality supports a negative answer to this question. In this case, the folk may interpret 'I don't care at all about harming the environment' as meaning 'harming the environment has no intrinsic disvalue for me,' while still thinking that environment-harming carries, for the chairman, an extrinsic disvalue that comes from its being morally wrong.

If this is correct, then we can see how known immoral side effects will always, from the folk's perspective, satisfy condition (ii) of the hypothesis above. Thus the hypothesis supports the weak conclusion that moral asymmetries in side effects can result in asymmetries in folk ascriptions of intentionality without entailing that these latter asymmetries are rooted in the former ones.

The hypothesis offered above is an empirical one about how ordinary folk use the term 'intentionally' and is therefore open to falsification. Since the hypothesis proceeds from assumptions about Harman's sniper and Mele and Sverdlik's surgeon, these cases should be tested against ordinary people's intuitions. (Note that if people do think Doris intentionally killed Pat, the hypothesis above may be repaired merely by removing condition (iii).)

Supposing these assumptions prove correct, though, the hypothesis makes a number of testable predictions. The most striking is that Knobe's asymmetrical results should be attainable in non-moral cases. Consider a vignette along the following lines:

The contrast case replaces 'criticized' with 'praised.' If the above hypothesis is correct, we should expect respondents to say the producer intentionally incurred the criticism of the media but did not intentionally gain their praise, even if studies do not show a significant difference in respondents' moral evaluations of the producer's act. Furthermore, even if the hypothesis under consideration is incorrect, a study like the one above may produce an asymmetry in ratings of intentionality without a corresponding moral asymmetry. In this eventuality, we should feel compelled to look for a unified explanation of the asymmetry. Since this explanation cannot employ moral asymmetry, we will need to reject the evaluative hypothesis.

Adams, F., and A. Steadman. 2004. Intentional action in ordinary language: core concept or pragmatic understanding? Analysis 64: 173­181.

Bratman, M. Two faces of intention. The Philosophical Review 93: 375­405.

Harman, G. 1976. Practical reasoning. Review of Metaphysics 79: 431­63.

Knobe, J. 2003a. Intentional action in folk psychology: an experimental investigation. Philosophical Psychology 16: 309­24.

Knobe, J. 2003b. Intentional action and side effects in ordinary language. Analysis 63: 190­94.

Knobe, J. 2004 Intention, intentional action and moral considerations. Analysis 64: 181-187.

Mele, A. 2003. Intentional action: controversies, data, and core hypotheses. Philosophical Psychology 16: 325­29.

Mele, A., and S. Sverdlik. 1996. Intention, intentional action, and moral responsibility. Philosophical Studies 82: 265­87.