Intentionality and Moral Judgments in Commonsense Thought about Action

Steven Sverdlik[1]

Southern Methodist University

The concept of intentional action occupies a central place in commonsense or folk psychological thought. Philosophers of action, psychologists and moral philosophers all have taken an interest in understanding this important concept. One issue that has been discussed by philosophers is whether the concept of intentional action is purely ‘naturalistic’, that is, whether it is entirely a descriptive concept that can be used to explain and predict behavior. (Of course, judgments using such a concept could be used to support moral or evaluative judgments about responsibility, praise and blame.) A related question is whether speakers’ views about moral and evaluative issues at least affect their judgments about intentionality, even if their explicit concept of intentional action is not itself evaluative.

The folk psychological concept of intentional action has a number of components. These include the agent’s desire or interest in producing an effect, her control over the production of this effect, and her belief about the likelihood of her producing the effect. (Mele and Moser 1994) We can ask whether moral or evaluative considerations influence how people interpret each of these components. Consider the component of the agent’s desire or interest in performing the action or producing a certain effect. This component is related to a distinction we commonly make about the effects of action. We distinguish between effects that an agent produces intentionally and other effects of her action that she may foresee but does not produce intentionally. For example, an agent could intentionally start her car by turning the ignition key, and foresee that the car will make noise as a result of the ignition, where she has no desire or interest in making this noise. In this case we say that she knew that her action would produce the noise, but that she did not produce it intentionally. It might be suggested here that speakers say that a foreseen effect that is not desired—a so-called ‘side’ or ‘double’ effect like the noise from the engine’s starting---was produced intentionally if they believe that the agent is morally responsible for it. Mele and Sverdlik (1996) examined the arguments in the philosophical literature purporting to show this sort of influence on people’s use of ‘intentional’. They contended that none of these arguments is convincing. This means that speakers are perfectly capable of saying that an agent did not produce a bad effect intentionally, but is morally responsible for it, and that an agent did produce a bad effect intentionally but is not responsible for it.

Joshua Knobe, in an interesting series of papers, began empirical investigations of this and related questions about the concept of intentional action. (Knobe 2003a; Knobe 2003b; Knobe 2004; Knobe and Mendlow 2004) When he asked subjects questions about the agents in described scenarios—one of which is presented below---he found evidence that the subjects’ views about responsibility and value do, in fact, affect their judgments about intentionality. Where subjects believe that an agent is morally responsible for a bad side effect they are inclined to say that she produced it intentionally. Knobe also found that there is an asymmetry in the way that subjects connect responsibility and intentionality in cases of good and bad side effects: speakers are more likely to say that a bad side effect is produced intentionally than they are to say that a good side effect is produced intentionally.

This paper describes two psychological experiments designed by the author and Joshua Knobe. The experiments investigate further some questions that arose from his work on responsibility and intentionality beliefs in folk psychology. They show that there is reason to doubt that subjects’ beliefs about the intentionality of side effects are simply a product of their beliefs about the agent’s responsibility for these effects. The author also considers how the experimental results bear on Knobe’s most recent views about the relation of subjects’ value judgments about side effects and their intentionality judgments. What the experimental results suggest is that subjects do not simply use either their belief that a side effect is bad, or that the agent is responsible for it, to determine their view about the intentionality of its production.

Experiment 1

Design Knobe (2003a) found that subjects are influenced in their descriptions of side effects as intentionally produced or not by their beliefs about whether the agent described is to be blamed (or is responsible) for those effects. One of the scenarios he presented to his subjects is the following, which will be called Indifference.

The vice-president of a company went to the chairman of the board and said, “We are thinking of starting a new program. It will help us increase profits, but it will also harm the environment.”

The chairman of the board answered, “I don’t care at all about harming the environment. I just want to make as much profit as I can. Let’s start the new program.”

They started the new program. Sure enough, the environment was harmed.

Knobe found that 82% of his subjects said that the chairman brought about the side effect intentionally. This is surprising, given the fact that the scenario gives us no reason to suppose that he desired to harm, or had an interest in harming, the environment. A related surprising finding involved the production of good side effects. Knobe found that when the same subjects were presented with a scenario exactly like this one, except for the substitution of ‘help’ for ‘harm’’, 77% said that the chairman did not produce the side effect intentionally.

The present experiment was designed to investigate further some of Knobe’s findings. One question that arose about Indifference was whether the response to it was due to the specific mental attitude that the chairman is implied to have. The chairman seems to be (outrageously) indifferent to the harm that will result from his action. It is possible that subjects would be less likely to say that a foreseen harm is produced intentionally when it is made clear that the agent is reluctant to produce it, and regrets it, after the fact. This sort of situation is captured in the following scenario, which will be called Regret.

Jones normally mows his lawn every week. But during the summer he and his family take a two-week vacation. The day before they leave, Jones realizes that he has not mowed the lawn. In order to do that before they leave for the airport, he has to mow at 7 AM on Sunday.

Jones thinks: “If I mow the lawn at 7 AM, I’ll be waking up the neighbors, but I can’t see leaving the lawn unmowed for two weeks. Well, I really hate to wake them up, but I guess there’s just no way around it.”

Jones wakes up early on Sunday and does the mowing. While he is on his vacation, Jones feels bad about having woken up the neighbors.

Regret and the questions about it were also designed to make the subjects consider the general possibility that an agent is morally responsible for a bad result that she does not produce intentionally. We wondered whether subjects in Knobe (2003) who read Indifference were reasoning that the chairman must have produced the harm intentionally because he was responsible for it.

In this experiment the control condition consisted of Indifference, along with three questions about it. The experimental condition consisted of two scenarios, the first being Regret, with the same three questions, the second being Indifference, along with the three questions. The questions asked whether the agent described believed that he was producing the relevant effect; whether he intentionally produced that effect; and whether he was responsible for producing it.

We had two hypotheses. We predicted that fewer subjects would say about Jones that he intentionally awoke his neighbors than would say that the chairman intentionally harmed the environment. We also predicted that after the subjects in the experimental condition encountered (in Regret) the conceptual possibility of foreseeing a harmful result while being reluctant to produce it, they would be less likely than subjects in the control group to say of the chairman (in Indifference) that he produced the harm intentionally. On the other hand, it was hypothesized that the subjects in the second condition would be just as likely as the control subjects to say that the executive was responsible for the harm.[2]

Results of Experiment 1

Part I – Control

Number of subjects = 34

Indifference Yes % No %

Chairman believed he would harm environment 21 62 13 38

Chairman intentionally harmed environment 15 44 19 56

Chairman responsible for harm to environment 34 100 0 0

Subjects who said ‘yes’ to ‘responsible’ and to ‘intentional’ 15 (45% of 34)

Subjects who said ‘yes’ to ‘responsible,’ ‘no’ to ‘intentional’ 19 (55% of 34)

Part II – Experimental

Number of subjects = 51

Regret Yes % No %

Jones believed he would wake neighbors 48 94 3 6

Jones intentionally woke neighbors 12 24 39 76

Jones responsible for waking neighbors 40 78 11 22

Subjects who said ‘yes’ to ‘responsible’ and to ‘intentional’ 9 (22% of 40)

Subjects who said ‘yes’ to ‘responsible’ and ‘no’ to ‘intentional’ 31 (78% of 40)

Indifference

Chairman believed he would harm environment 42 82 9 18

Chairman intentionally harmed environment 31 60 20 39

Chairman responsible for harm to environment 48 94 3 6

Subjects who said ‘yes’ to ‘responsible’ and to ‘intentional’ 29 (60% of 48)

Subjects who said ‘yes’ to ‘responsible’ and ‘no’ to ‘intentional’ 19 (40% of 48)

Discussion There are some noteworthy features of our results.

First, we found that a much smaller percentage of subjects who read Indifference (in the control condition) answered that the chairman produced the harm intentionally than did subjects in Knobe (2003a). Knobe found 82% of his subjects said that the chairman produced the harm intentionally. We found that only 44% of the subjects in the control condition said this.

Second, it is indeed noteworthy that a smaller percentage of the subjects in the experimental condition said that Jones produced the bad side effect intentionally than said (in the control condition) that the chairman produced the side effect intentionally. 24% of the subjects in Regret said that Jones produced the result intentionally; 44% in the control condition said that the chairman produced the harm intentionally. This supports our first hypothesis. It is not simply an agent’s foreseeing of a bad side effect that is playing a role in subject’s responses, but also whether the agent is indifferent to it, or is reluctant to produce it.

Third, if we consider the subjects who say that the agent in the scenario is responsible for the side effect, there is a significant difference in the two cases. A higher percentage of the subjects in the control condition of Indifference said that the chairman was responsible and produced the effect intentionally than said of Regret that Jones produced the effect intentionally and was responsible for it. In the control condition of Indifference, 45% of the subjects who gave a ‘yes’ answer to the responsibility question also said that the executive produced the result intentionally. Hence, 55% of those saying he was responsible for the effect said he did not produce it intentionally. Only 22% of the subjects who said that Jones was responsible for waking the neighbors also said that he did this intentionally. Thus, 78% of those saying he was responsible for a bad side effect said he did not produce it intentionally. This result certainly suggests that it is incorrect to assert that subjects who believe that an agent is morally responsible for a bad side effect also suppose that this agent produced the effect intentionally. It does, however, support the conclusion that subjects are more likely to say that an agent produces a bad side effect intentionally when he seems indifferent to it, as compared to an agent who produces a bad side effect with reluctance.

Fourth, our second hypothesis was disconfirmed. The subjects in the experimental condition were more, not less, likely to say of Indifference that the chairman produced the harm intentionally. 44% of the control subjects said that the chairman produced the harm intentionally, whereas 60% of the subjects in the experimental condition said this. Also, the percentage of subjects who said the chairman was responsible that also said he produced the result intentionally went from 45% of the control subjects to 60% of the subjects in the experimental condition. Perhaps this could still be attributed to the effect of the factor under consideration, though. That is, we hypothesized that once the subjects were clearly confronted with the possibility that an agent is responsible for a result that he does not desire to produce—and this possibility is evident in a case where the agent is reluctant to produce a foreseen effect—they would be more likely to say that the chairman did not produce the harm intentionally, but was responsible for it. But the description in Regret, where Jones is reluctant to wake up his neighbors, may actually have highlighted the fact that the chairman does not have such reluctance, and thereby increased the subjects’ inclination to say that the chairman produced the harm intentionally. Note, however, that even after considering Regret 39% of the subjects said that the chairman in Indifference did not produce the harm intentionally. And 40% of the subjects who said ‘responsible’ of the chairman said that he did not produce harm intentionally. So the connection between responsibility and intentionality is not tight, even when the agent foresees a bad side effect and is indifferent to it.

A further point can be made with some hesitation. This experiment investigated the relationship between the subjects’ beliefs about intentionality and responsibility with respect to side effects. Recently, Knobe and Mendlow (2004) have defended a somewhat different claim from that in Knobe (2003a). In response to Nadelhoffer they have suggested that the idea that influences subjects’ responses about intentionality is not responsibility (including praise and blame judgments), but rather simply beliefs about the goodness or badness of the foreseen effects. The two views are different, since it is possible, for example, that an agent foresees a bad effect and produces it, but deserves no blame for doing so. The example that Knobe and Mendlow use to exemplify this involves an executive who makes a decision knowing it will reduce sales in one area, but will also increase sales in another area by a greater amount. They found that subjects do not assert that the executive is to be blamed for reducing the sales, and yet they largely say that she intentionally reduced the sales. Knobe and Mendlow conclude that what is influencing the subjects’ responses is the belief that the reduction is bad. Now the experiment we ran did not ask the subjects about their beliefs about whether waking the neighbors in Regret is a bad thing. But it is surely reasonable to suppose that they did believe this. Indeed it is plausible to think that judgments asserting blame or ‘negative’ responsibility for the production of an effect entail judgments that the result is a bad thing. (Examples such as the executive in Knobe and Mendlow (2004) show that the converse entailment does not hold.) Therefore, it seems reasonable to hold that the results from Regret in our experiment tell against an unqualified version of the claim in Knobe and Mendlow (2004). That is, they tell against the claim that subjects who believe that an agent knowingly produces an effect that the subjects believe is bad will say that the agent produced it intentionally.

Experiment 2

Design In this experiment we further investigated the role of the state of mind of the agent described in the scenario. There is a distinction between indifference to a foreseen bad effect and reluctance to produce it. But there is at least one other possible attitude that an agent can have towards a foreseen bad result: she can actively try to prevent or minimize it.[3] (In contrast, of course, she can welcome it.) The author hypothesized that subjects who read a description of an agent who exhibits this sort of attitude towards a foreseen effect of her action will be even less inclined to describe her production of the effect as intentional than they would an agent who is reluctant to produce the effect but does nothing to prevent it.

The experiment had two conditions. The control condition consisted of Regret, followed by two questions, instead of three. The first asked whether Jones intentionally woke up his neighbors; the second asked whether Jones was to blame for waking up his neighbors. The experimental condition consisted of a new scenario about a dentist, Smith, along with two questions. Let us call this scenario Prevention. With Prevention we had some subjects answer the question about blame first, and the question about intentionality second. Other subjects answered these questions in reverse order. Having both orders could clarify whether subjects who first describe an agent as ‘to blame’ (or responsible) are more likely to describe her also as intentionally producing the result, than subjects who consider intentionality first. We found no significant difference in the two groups, so their responses are amalgamated in the following table. Here is Prevention.

Dr. Smith is a dentist. Mr. Taylor is one of her patients. One day Dr. Smith realizes that Taylor has a certain condition that normally requires some very painful surgery. She puts the issue before Taylor, informing him of the risks and benefits of the procedure. He tells her that he is prepared to undergo it if she thinks it is advisable.

Dr. Smith thinks, “If I operate, I’ll take care of the problem, but the effects of the surgery will be very painful. If I don’t operate, eventually his condition could mean he loses all his teeth. I’m sure that Taylor would feel pain after the procedure, so if I do it I should give him the largest possible safe dose of pain reliever before I begin. I hate to cause him pain, but I can’t leave his condition untreated. I’ll perform the surgery.”

She gives Taylor the dosage of pain reliever she considered, and goes on to perform the surgery successfully. Taylor does feel some pain.

Results of Experiment 2

Part I – Control

Regret

Number of subjects = 33

(Note: some subjects did not answer all questions, or did not simply answer ‘yes’ or ‘no.’)

Yes % No %

Jones intentionally woke neighbors 14 45 17 55

Jones to blame for waking neighbors 24 77 6 23

Subjects who said ‘yes’ to ‘to blame’ and to ‘intentional’ 10 (42% of 24)

Subjects who said ‘yes’ to ‘to blame,’ ‘no’ to ‘intentional’ 14 (58% of 24)

Part II – Experimental

Prevention

Total number of subjects in IIA and IIB = 80

Yes % No %

Smith intentionally produced Taylor’s pain 33 41 47 59

Smith to blame for Taylor’s pain 21 26 59 74

Subjects who said ‘yes’ to ‘to blame’ and to ‘intentional’ 6 (29% of 21)

Subjects who said ‘yes’ to ‘to blame,’ ‘no’ to ‘intentional’ 15 (71% of 21)

Discussion One notable feature of this experiment is that the percentage of subjects who say that Jones intentionally woke up his neighbors is 45%, whereas in the first experiment only 24% said this. There seems to be some significant variation on this point. On the other hand, 22% of the subjects in the first experiment who considered Regret said that Jones was responsible for waking up his neighbors and did so intentionally (i.e., gave two ‘yes’ answers), while 30% of the subjects in this experiment who considered that scenario gave two ‘yes’ answers. On this point, the responses of the two sets of subjects are close.

In the experimental condition it is striking that 26% of the subjects said that Smith is to blame for Taylor’s pain. The response about blame or responsibility attributed to Smith is markedly different from the blame attributed to Jones. 77% of the subjects in the control section of this experiment said of Jones that he was ‘to blame’ for waking up his neighbors. In the first experiment, 78% of the subjects asked about this said that Jones was ‘responsible’ for waking them up. This might be seen as confirming the suggestion that subjects have less inclination to attribute blame to an agent who takes steps to prevent a foreseen bad effect. But there are other differences between the two cases. For one thing, Smith secures Taylor’s consent to operate; Jones does not secure his neighbors’ consent to mow.

The results of the second experiment confirm subjects’ ability to separate the ideas of responsibility and intentionality, although in some surprising ways. In Regret most subjects said that Jones was to blame for waking his neighbors. 58% of those who said that Jones was to blame for waking his neighbors said that he did not wake them intentionally. Most of the subjects in the Smith scenario said that Smith was not to blame for Taylor’s pain. But 71% of the subjects who said that Smith is to blame for Taylor’s pain said that she did not produce it intentionally. So in both experiments the majority of subjects who attribute blame or responsibility also say that the agent did not produce the relevant effect intentionally. But a number of subjects in the second condition of this experiment make a different sort of conceptual separation. 33 subjects said that Smith produced Taylor’s pain intentionally, and 27 of these (81%) also said that she was not to blame for it. You could certainly say that these 27 subjects separate the questions of intentionality and responsibility.[4] The author was interested in cases where agents affirm the presence of responsibility but deny the presence of intentionality. He had focused on Regret as a way to bring home to the subjects that it is possible to be responsible for a bad side effect without intentionally bringing it about. And the two presentations of Regret showed a significant number of subjects making the distinction in this way. The author would not have predicted that many subjects who considered Prevention would have said that Smith intentionally produces Taylor’s pain but is not to blame for it.

The question of the separability of intentionality and responsibility is related to the new proposal found in Knobe and Mendlow (2004). They suggest that the operative factor in subjects’ thinking about intentionality in these sorts of cases is the badness of the side effects rather than the issue of the agent’s blameworthiness or responsibility. Again, there is a problem with their proposal. Presumably all the subjects regard Taylor’s pain as bad. But 59% of them said that Smith did not produce Taylor’s pain intentionally, which disconfirms their hypothesis.

There are some other surprising results of this experiment. One is that 26% of the subjects in Prevention said that Smith was to blame for Taylor’s pain. We would not have supposed that any subjects would have given this response. The author is inclined to think that one of the sources for these results was the use of the word ‘blame’. We had intended to use a more colloquial synonym for ‘morally responsible’, but it may be that the causal meaning of the word ‘blame’ was the one employed by some of the subjects. In the sentence ‘The storm was to blame for the power outage’ the word ‘blame’ has a causal meaning. (The word ‘responsible’ also has this meaning.) So perhaps some of those subjects who agreed with the statement that Smith was to blame for Taylor’s pain only meant that Smith was the cause of it. We might have gotten a different response rate if we had asked whether Smith deserved any blame for Taylor’s pain.

The most surprising result was that 41% of the subjects in Prevention said that Smith produced Taylor’s pain intentionally. Compare this to the 44% of the subjects in the control condition of the first experiment who said that the chairman in Indifference produced the harm intentionally, and the 45% of the subjects in the control condition of this experiment who said that Jones woke up his neighbors intentionally. (On the other hand, as was noted, only 24% of the subjects in the experimental condition of the first experiment said that Jones woke up his neighbors intentionally.) We expected the percentage of subjects considering Prevention who said that Smith produced Taylor’s pain intentionally to be smaller than the percentage of subjects considering Jones who said that he woke up his neighbors intentionally. Thus, our hypothesis was disconfirmed. Smith takes steps to avoid a bad side effect whereas Jones is reluctant to produce a bad side effect, but takes no steps to avoid it, and yet about the same percentage of subjects said that Smith produced the bad effect intentionally as said that Jones produced the bad effect intentionally. The author has no explanation for the high percentage of subjects who say that Smith produced Taylor’s pain intentionally. It is important to remind ourselves, however, that 59% said that she did not produce it intentionally.

Summary

It is clear that subjects are able, in general, to separate the idea of responsibility for an effect from the idea that it is produced intentionally. That is, they are able to ascribe one of these concepts to an agent’s behavior while withholding the other, even when the issue is a bad side effect. In the case of a bad side effect, we found cases where subjects will say that an agent did not produce the effect intentionally but is responsible for it. So, in Regret in the first experiment, 76% of the subjects said that Jones did not produce the result intentionally, but 78% said that he was responsible for it. And, furthermore, only 22% of the subjects who did say that Jones was responsible for waking his neighbors thought that he did this intentionally. There was a surprising confirmation of this general point about conceptual separability in the responses to Prevention. 21 subjects said that Smith was to blame for Taylor’s pain. 71% of these subjects said that she did not produce his pain intentionally. (But it is possible that the ‘blame’ here is not a moral notion.)

It seems reasonable to conclude, further, that subjects are able to distinguish the question of the badness of a foreseen effect from the question of the intentionality of the production of it. We did not ask our subjects whether they regarded harm to the environment, waking up one’s neighbors on Sunday morning, or pain as bad. But it is reasonable to think that they all believe that these effects are bad. Yet 56% of the subjects in the control condition of the first experiment said that the chairman’s production of the harm was not intentional; 76% of the subjects in the experimental condition of this experiment said that Jones’ waking of the neighbors was not intentional; 55% of the subjects in the second experiment said that Jones’ waking of the neighbors was not intentional; and 59% of the subjects in the second experiment said that Smith’s production of Taylor’s pain was not intentional. (It was only in the experimental condition of the first experiment that a bare majority, 60%, said that the executive harmed the environment intentionally.)

The author had supposed that a critical variable affecting the subjects’ views about intentionality would be the attitude ascribed to the agent in the scenario. This seemed to be only partly confirmed. In the first experiment a substantially smaller percentage of subjects said that Jones woke his neighbors intentionally than said the chairman harmed the environment intentionally (24% versus 44%), and this difference might well be explained by saying that Jones is described as reluctant to wake his neighbors while the chairman is indifferent to the harm to the environment. But in the second experiment what seems like an analogous difference had virtually no effect. One might suppose that efforts to prevent a bad side effect bespeak an even stronger aversion to it than does mere reluctance to produce it, and yet 41% of the subjects said that Smith produced Taylor’s pain intentionally and 45% said that Jones woke his neighbors intentionally. So we can either conclude that the attitudinal factor reaches a sort of threshold of effectiveness when subjects conclude that the agent is reluctant to produce the side effect, or that they are not seeing efforts to prevent the side effect as an augmentation of the aversion found in reluctance.

In only one of the experimental conditions did a majority of the subjects say that the agent intentionally produced the bad side effect. So it is safe to say that we did not find Knobe’s hypotheses about the influence of responsibility and badness judgments on intentionality judgments to be strongly confirmed. The author’s hypothesis that the subjects’ beliefs about the agent’s attitude toward the side effect influence their intentionality judgments was not strongly confirmed, either. It is, of course, likely that all three factors have some influence on the responses of some subjects in some cases, since a purely ‘naturalistic’ notion of intentionality would probably not lead anyone to say that the chairman in Indifference produces the harm intentionally.

Department of Philosophy

Southern Methodist University

Dallas, Texas 75275

USA