2.1. Ordinary understanding and scientific theory

To begin with, we will need to distinguish two different aspects of the claim that people's ordinary understanding is analogous to a scientific theory. First, there is the claim that human thought might sometimes take the form of a theory. To assess this first claim, one would have to pick out the characteristics that distinguish theories from other sorts of knowledge structures and then ask whether these characteristics can be found in ordinary cognition. This is certainly a worthwhile endeavor, but it has already been pursued in a considerable body of recent research (e.g., Carey & Spelke Reference Carey and Spelke1996; Goldman Reference Goldman2006; Murphy & Medin Reference Murphy and Medin1985), and I will have nothing further to say about it here. Instead, the focus of this target article will be on a second claim, namely, the claim that certain facets of human cognition are properly understood as scientific.

To begin with, it should be emphasized that this second claim is distinct from the first. If one looks to the usual sorts of criteria for characterizing a particular knowledge structure as a “theory” (e.g., Premack & Woodruff Reference Premack and Woodruff1978), one sees immediately that these criteria could easily be satisfied by, for example, a religious doctrine. A religious doctrine could offer systematic principles; it could posit unobservable entities and processes; it could yield definite predictions. For all these reasons, it seems perfectly reasonable to say that a religious doctrine could give us a certain kind of “theory” about how the world works. Yet, although the doctrine might offer us a theory, it does not appear to offer us a specifically scientific theory. In particular, it seems that religious thinking often involves attending to different sorts of considerations from the ones we would expect to find in a properly scientific investigation. Our task here, then, is to figure out whether certain aspects of human cognition qualify as “scientific” in this distinctive sense.

One common view is that certain aspects of human cognition do indeed make use of the very same sorts of considerations we find in the systematic sciences. So, for example, in work on causal cognition, researchers sometimes proceed by looking to the statistical methods that appear in systematic scientific research and then suggesting that those same methods are at work in people's ordinary causal judgments (Gopnik et al. Reference Gopnik, Glymour, Sobel, Schulz, Kushnir and Danks2004; Kelley Reference Kelley and Levine1967; Woodward Reference Woodward2004). Different theories of this type appeal to quite different statistical methods, but these differences will not be relevant here. The thing to focus on is just the general idea that people's ordinary causal cognition is in some way analogous to a scientific inquiry.

And it is not only the study of causal cognition that proceeds in this way. A similar viewpoint can be found in the theory-of-mind literature (Gopnik & Meltzoff Reference Gopnik and Meltzoff1997), where it sometimes goes under the slogan “Child as Scientist.” There, a central claim is that children refine their understanding of the mind in much the same way that scientists refine their theories. Hence, it is suggested that we can look at the way Kepler developed his theory of the orbits of the planets and then suggest that children use the same basic approach as they are acquiring the concept of belief (Gopnik & Wellman Reference Gopnik and Wellman1992). Once again, the idea is that the cognitive processes people use in ordinary life show a deep similarity to the ones at work in systematic science.

It is this idea that we will be taking up here. Genuinely scientific inquiry seems to be sensitive to a quite specific range of considerations and seems to take those considerations into account in a highly distinctive manner. What we want to know is whether certain aspects of ordinary cognition work in more or less this same way.

2.2. Refining the question

But now it might seem that the answer is obvious. For it has been known for decades that people's ordinary intuitions show certain patterns that one would never expect to find in a systematic scientific investigation. People make wildly inappropriate inferences from contingency tables, show shocking failures to properly detect correlations, display a tendency to attribute causation to whichever factor is most perceptually salient (Chapman & Chapman Reference Chapman and Chapman1967; McArthur & Post Reference McArthur and Post1977; Smedslund Reference Smedslund1963). How could one possibly reconcile these facts about people's ordinary intuitions with a theory according to which people's ordinary cognition is based on something like a scientific methodology?

The answer, I think, is that we need to interpret that theory in a somewhat more nuanced fashion. The theory is not plausibly understood as an attempt to describe all of the factors that can influence people's intuitions. Instead, it is best understood as an attempt to capture the “fundamental” or “underlying” nature of certain cognitive capacities. There might then be various factors that interfere with our ability to apply those capacities correctly, but the existence of these additional factors would in no way impugn the theory itself.

To get a rough sense for the strategy here, it might be helpful to return to the comparison with religion. Faced with a discussion over religious doctrine, we might say: “This discussion isn't best understood as a kind of scientific inquiry; it is something else entirely. So if we find that the participants in this discussion are diverging from proper scientific methods, the best interpretation is that they simply weren't trying to use those methods in the first place.” This would certainly be a reasonable approach to the study of religious discourse, but the key claim of the person-as-scientist theory is that it would not be the right approach to understanding certain aspects of our ordinary cognition. Looking at these aspects of ordinary cognition, a defender of the person-as-scientist view would adopt a very different stance. For example, she might say: “Yes, it's true that people sometimes diverge from proper scientific methods, but that is not because they are engaging in some fundamentally different sort of activity. Rather, their underlying capacities for causal cognition and theory-of-mind really are governed by scientific methods; it's just that there are also various additional factors that get in the way and sometimes lead people into errors.”

Of course, it can be difficult to make sense of this talk of certain capacities being “underlying” or “fundamental,” and different researchers might unpack these notions in different ways:

1. 1. One view would be that people have a domain-specific capacity for making certain kinds of judgments but then various other factors intrude and allow these judgments to be affected by irrelevant considerations.

2. 2. Another would be that people have a representation of the criteria governing certain concepts but that they are not always able to apply these representations correctly.

3. 3. A third would be that the claim is best understood counterfactually, as a hypothesis about how people would respond if they only had sufficient cognitive resources and freedom from certain kinds of biases.

I will not be concerned here with the particular differences between these different views. Instead, let us introduce a vocabulary that allows us to abstract away from these details and talk about this approach more generally. Regardless of the specifics, I will say that the approach is to posit an underlying competence and then to posit various additional factors that get in the way of people's ability to apply that competence correctly.

With this framework in place, we can now return to our investigation of the impact of moral considerations on people's intuitions. How is this impact to be explained? One strategy would be to start out by finding some way to distinguish people's underlying competencies from the various interfering factors. Then one could say that the competencies themselves are entirely scientific in nature, but that the interfering factors then prevent people from applying these competencies correctly and allow moral considerations to affect their intuitions. This strategy is certainly a promising one, and I shall discuss it in further detail later. But it is important to keep in mind that we also have open another, very different option. It could always turn out that there simply is no underlying level at which the relevant cognitive capacities are purely scientific, that the whole process is suffused through and through with moral considerations.

3.1. Intentional action

Perhaps the most highly studied of these effects is the impact of people's moral judgments on their use of the concept of intentional action. This is the concept people use to distinguish between behaviors that are performed intentionally (e.g., hammering in a nail) and those that are performed unintentionally (e.g., accidentally bringing the hammer down on one's own thumb). It might at first appear that people's use of this distinction depends entirely on certain facts about the role of the agent's mental states in his or her behavior, but experimental studies consistently indicate that something more complex is actually at work here. It seems that people's moral judgments can somehow influence their intuitions about whether a behavior is intentional or unintentional.

To demonstrate the existence of this effect, we can construct pairs of cases that are exactly the same in almost every respect but differ in their moral status.Footnote ¹ For a simple example, consider the following vignette:

Faced with this vignette, most subjects say that the chairman intentionally harmed the environment. One might initially suppose that this intuition relies only on certain facts about the chairman's own mental states (e.g., that he specifically knew his behavior would result in environmental harm). But the data suggest that something more is going on here. For people's intuitions change radically when one alters the moral status of the chairman's behavior by simply replacing the word “harm” with “help”:

Faced with this second version of the story, most subjects actually say that the chairman unintentionally helped the environment. Yet it seems that the only major difference between the two vignettes lies in the moral status of the chairman's behavior. So it appears that people's moral judgments are somehow impacting their intuitions about intentional action.

Of course, it would be unwise to draw any strong conclusions from the results of just one experiment, but this basic effect has been replicated and extended in numerous further studies. To begin with, subsequent experiments have further explored the harm and help cases to see what exactly about them leads to the difference in people's intuitions. These experiments suggest that moral judgments truly are playing a key role, since participants who start out with different moral judgments about the act of harming the environment end up arriving at different intuitions about whether the chairman acted intentionally (Tannenbaum et al. Reference Tannenbaum, Ditto and Pizarro2009). But the effect is not limited to vignettes involving environmental harm; it emerges when researchers use different cases (Cushman & Mele Reference Cushman, Mele, Knobe and Nichols2008; Knobe Reference Knobe2003a) and even when they turn to cases with quite different structures that do not involve side-effects in any way (Knobe Reference Knobe2003b; Nadelhoffer Reference Nadelhoffer2005). Nor does the effect appear to be limited to any one particular population: It emerges when the whole study is translated into Hindi and conducted on Hindi-speakers (Knobe & Burra Reference Knobe and Burra2006) and even when it is simplified and given to 4-year-old children (Leslie et al. Reference Leslie, Knobe and Cohen2006a). At this point, there is really a great deal of evidence for the claim that people's moral judgments are somehow impacting their intuitions about intentional action.

Still, as long as all of the studies are concerned only with intuitions about intentional action specifically, it seems that our argument will suffer from a fatal weakness. For someone might say: “Surely, we have very strong reason to suppose that the concept of intentional action works in more or less the same way as the other concepts people normally use to understand human action. But we have good theories of many of these other concepts – the concepts of deciding, wanting, causing, and so forth – and these other theories do not assign any role to moral considerations. So the best bet is that moral considerations do not play any role in the concept of intentional action either.”

In my view, this is actually quite a powerful argument. Even if we have strong evidence for a certain view about the concept of intentional action specifically, it might well make sense to abandon this view in light of theories we hold about various other, seemingly similar concepts.

3.2. Further mental states

As it happens, though, the impact of moral considerations does not appear to be limited to people's use of the word “intentionally.” The very same effect also arises for numerous other expressions: “intention,” “deciding,” “desire,” “in favor of,” “advocating,” and so forth.

To get a grip on this phenomenon, it may be helpful to look in more detail at the actual procedure involved in conducting these studies. In one common experimental design, subjects are randomly assigned to receive either the story about harming the environment or the story about helping the environment and then, depending on the case, are asked about the degree to which they agree or disagree with one of the following sentences:

When the study is conducted in this way, one finds that subjects show moderate agreement with the claim that the chairman harmed intentionally and moderate disagreement with the claim that he helped intentionally (Knobe Reference Knobe2004a). The difference between the ratings in these two conditions provides evidence that people's moral intuitions are affecting their intuitions about intentional action.

It appears, however, that this effect is not limited to the concept of intentional action specifically. For example, suppose we eliminate the word “intentionally” and instead use the word “decided.” The two sentences then become:

Faced with these revised sentences, subjects show more or less the same pattern of intuitions. They tend to agree with the claim that the agent decided to harm, and they tend to disagree with the claim that the agent decided to help (Pettit & Knobe Reference Pettit and Knobe2009).

Now suppose we make the case a little bit more complex. Suppose we do not use the adverb “intentionally” but instead use the verb “intend.” So the sentences come out as:

One then finds a rather surprising result. People's responses in both conditions are shifted over quite far toward the “disagree” side. In fact, people's intuitions end up being shifted over so far that they do not, on the whole, agree in either of the two conditions (Shepard Reference Shepard2009; cf. Cushman Reference Cushman2010; Knobe Reference Knobe2004b; McCann Reference McCann2005). Nonetheless, the basic pattern of the responses remains the same. Even though people's responses don't go all the way over to the “agree” side of the scale in either condition, they are still more inclined to agree in the harm case than they are in the help case.

Once one conceptualizes the issue in this way, it becomes possible to find an impact of moral considerations in numerous other domains. Take people's application of the concept in favor. Now consider a case in which an agent says:

Will people say in such a case that the agent is “in favor” of bringing about the outcome?

Here again, it seems that moral judgments play a role. People disagree with the claim that the agent is “in favor” when the outcome is morally good, whereas they stand at just about the midpoint between agreement and disagreement when the outcome is morally bad (Pettit & Knobe Reference Pettit and Knobe2009). And similar effects have been observed for people's use of many other concepts: desiring, intending, choosing, and so forth (Pettit & Knobe Reference Pettit and Knobe2009; Tannenbaum et al. Reference Tannenbaum, Ditto and Pizarro2009).

Overall, these results suggest that the effect obtained for intuitions about intentional action is just one example of a far broader phenomenon. The effect does not appear to be limited to the concept intentionally, nor even to closely related concepts such as intention and intending. Rather, it seems that we are tapping into a much more general tendency, whereby moral judgments impact the application of a whole range of different concepts used to pick out mental states and processes.

3.3. Action trees

But the scope of the effect does not stop there. It seems also to apply to intuitions about the relations that obtain among the various actions an agent performs. Philosophers and cognitive scientists have often suggested that such relations could be represented in terms of an action tree (Goldman Reference Goldman1970; Mikhail Reference Mikhail2007). Hence, the various actions performed by our chairman in the help case might be represented with the tree in Figure 1.

Figure 1. Action tree for the help case.

Needless to say, ordinary folks do not actually communicate with each other by writing out little diagrams like this one. Still, it seems that we can get a sense of how people are representing the action tree by looking at their use of various ordinary English expressions, for example, by looking at the way they use the expressions “in order to” and “by.”

A number of complex issues arise here, but simplifying slightly, the key thing to keep in mind is that people only use “in order to” for relations that go upward in the tree, and they only use “by” for relations that go downward. Thus, people are willing to say that the chairman “implemented the program in order to increase profits” but not that he “increased profits in order to implement the program.” And, conversely, they are willing to say that he “increased profits by implementing the program” but not that he “implemented the program by increasing profits.” Looking at people's intuitions about simple expressions like these, we can get a good sense of how they are representing the geometry of the action tree itself.

But now comes the tricky part. Experimental results indicate that people's intuitions about the proper use of these expressions can actually be influenced by their moral judgments (Knobe Reference Knobe2004b; forthcoming). Hence, people are willing to say:

but not:

And, similarly, they are willing to say:

but not:

One natural way of explaining these asymmetries would be to suggest that people's moral judgments are having an effect on their representations of the action tree itself. For example, suppose that when people make a judgment that harming the environment is morally wrong, they thereby come to represent the corresponding node on the action tree as “collapsing” into a lower node (see Fig. 2).

Figure 2. Action tree for the harm case.

The asymmetries we find for “in order to” and “by” would then follow immediately, without the need for any controversial assumptions about the semantics of these specific expressions. Although the issue here is a complex one, recent research does seem to be supporting the claim that moral judgments are affecting action tree representations in this way (Knobe, forthcoming; Ulatowski Reference Ulatowski2009).

3.4. Causation

All of the phenomena we have been discussing thus far may appear to be quite tightly related, and one might therefore suspect that the effect of morality would disappear as soon as one turns to other, rather different cases. That, however, seems not to be the case. Indeed, the very same effect arises in people's intuitions about causation (Alicke Reference Alicke2000; Cushman Reference Cushman2010; Hitchcock & Knobe Reference Hitchcock and Knobe2009; Knobe, forthcoming; Knobe & Fraser Reference Knobe, Fraser and Sinnott-Armstrong2008; Solan & Darley Reference Solan and Darley2001).

For a simple example here, consider the following vignette:

Faced with this vignette, most subjects say that the professor did cause the problem but that the administrative assistant did not cause the problem (Knobe & Fraser Reference Knobe, Fraser and Sinnott-Armstrong2008). Yet, when we examine the case from a purely scientific standpoint, it seems that the professor's action and the administrative assistant's action bear precisely the same relation to the problem that eventually arose. The main difference between these two causal factors is just that the professor is doing something wrong (violating the departmental rule) while the administrative assistant is doing exactly what she is supposed to (acting in accordance with the rules of the department). So it appears that people's judgment that the professor is doing something wrong is somehow affecting their intuitions about whether or not the professor caused the events that followed.

Now, looking just at this one case, one might be tempted to suppose that the effect is not at all a matter of moral judgment but simply reflects people's intuitive sense that the professor's action is more “unusual” or “strange” than the administrative assistant's. But subsequent studies strongly suggest that there is something more afoot here. People continue to show the same basic effect even when they are informed that the administrative assistants never take pens whereas the professors always do (Roxborough & Cumby Reference Roxborough and Cumby2009), and there is a statistically significant effect whereby pro-life subjects are more inclined than pro-choice subjects to regard the act of seeking an abortion as a cause of subsequent outcomes (Cushman et al. Reference Cushman, Knobe and Sinnott-Armstrong2008). All in all, the evidence seems strongly to suggest that people's moral judgments are actually impacting their causal intuitions.

3.5. Doing and allowing

People ordinarily distinguish between actually breaking something and merely allowing it to break, between actually raising something and merely allowing it to rise, between actually killing someone and merely allowing someone to die. This distinction has come to be known as the distinction between doing and allowing.

To explore the relationship between people's intuitions about doing and allowing and their moral judgments, we used more or less the same methodology employed in these earlier studies (Cushman et al. Reference Cushman, Knobe and Sinnott-Armstrong2008). Subjects were randomly assigned to receive different vignettes. Subjects in one condition received a vignette in which the agent performs an action that appears to be morally permissible:

These subjects were then asked whether it would be more appropriate to say that the doctor ended the homeless man's life or that he allowed the homeless man's life to end.

Meanwhile, subjects in the other condition were given a vignette that was almost exactly the same, except that the doctor's internal monologue takes a somewhat different turn:

These subjects were asked the same question: whether it would be more appropriate to say that the doctor ended the man's life or allowed it to end.

Notice that the doctor performs exactly the same behavior in these two vignettes, and in both vignettes, he performs this behavior in the hopes that it will bring about the man's death. The only difference between the cases lies in the moral character of the doctor's reasons for hoping that the man will die. Yet this moral difference led to a striking difference in people's intuitions about doing versus allowing. Subjects who received the first vignette tended to say that the doctor “allowed” the man's life to end, whereas subjects who received the second vignette tended to say that the doctor “ended” the man's life. (Moreover, even within the first vignette, there was a correlation whereby subjects who thought that euthanasia was generally morally wrong were less inclined to classify the act as an “allowing.”) Overall, then, the results of the study suggest that people's moral judgments are influencing their intuitions here as well.

It would, of course, be foolhardy to draw any very general conclusions from this one study, but the very same effect has also been observed in other studies using quite different methodologies (Cushman et al. Reference Cushman, Knobe and Sinnott-Armstrong2008), and there is now at least some good provisional evidence in support of the view that people's intuitions about doing and allowing can actually be influenced by their moral judgments.

3.6. Additional effects

Here we have discussed just a smattering of different ways in which people's moral judgments can impact their intuitions about apparently non-moral questions. But our review has been far from exhaustive: there are also studies showing that moral judgments can affect intuitions about knowledge (Beebe & Buckwalter, forthcoming), happiness (Nyholm Reference Nyholm2009), valuing (Knobe & Roedder Reference Knobe and Roedder2009), act individuation (Ulatowski Reference Ulatowski2009), freedom (Phillips & Knobe Reference Phillips and Knobe2009), and naturalness (Martin Reference Martin2009). Given that all of these studies were conducted just in the past few years, it seems highly probable that a number of additional effects along the same basic lines will emerge in the years to come.

4.1. The motivational bias hypothesis

Think of the way a District Attorney's office might conduct its business. The DA decides to prosecute a suspect and hands the task over to a team of lawyers. These lawyers then begin looking at the case. Presumably, though, they do not examine the evidence with perfectly unbiased eyes. They have been hired to secure a conviction, and they are looking at the evidence with a view to achieving this goal (cf. Tetlock Reference Tetlock2002). One might say that they are under the influence of a motivational bias.

A number of researchers have suggested that a similar mechanism might be at the root of the effects we have been discussing here (Alicke Reference Alicke2008; Nadelhoffer Reference Nadelhoffer2006a). Perhaps people just read through the story and rapidly and automatically conclude that the agent is to blame. Then, after they have already reached this conclusion, they begin casting about for ways to justify it. They try to attribute anything they can – intention, causation, et cetera – that will help to justify the blame they have already assigned. In essence, the suggestion is that the phenomena under discussion here can be understood as the results of a motivational bias.

This suggestion would involve a reversal of the usual view about the relationship between people's blame judgments and their intuitions about intention, causation, and so forth. The usual view of this relationship looks something like what's shown in Figure 3

Figure 3. Traditional account of the process underlying blame ascription.

Here, the idea is that people first determine that the agent fulfilled the usual criteria for moral responsibility (intention, cause, etc.) and then, on the basis of this initial judgment, go on to determine that the agent deserves blame. This sort of model has a strong intuitive appeal, but it does not seem capable of explaining the experimental data reviewed above. After all, if people determine whether or not the agent caused the outcome before they make any sort of moral judgment, how could it be that their moral judgments affect their intuitions about causation?

To resolve this question, one might develop a model that goes more like the one shown in Figure 4

Figure 4. Motivational bias account of blame ascription.

In this revised model, there is a reciprocal relationship between people's blame judgments and their intuitions about intention, causation, et cetera. As soon as people observe behavior of a certain type, they become motivated to find some way of blaming the agent. They then look to the evidence and try to find a plausible argument in favor of the view that the agent fulfills all of the usual criteria for responsibility. If they can construct a plausible argument there, they immediately blame the agent. Otherwise, they reluctantly determine that the agent was not actually blameworthy after all. In short, the hypothesis says that people's intuitions about intention and causation affect their blame judgments but that the causal arrow can also go in the other direction, with people's drive to blame the agent distorting their intuitions about intention and causation.

One of the main sources of support for such a hypothesis is the well-established body of theoretical and experimental work within social psychology exploring similar effects in other domains. There is now overwhelming evidence that motivational biases can indeed lead people to interpret evidence in a biased manner (for a review, see Kunda Reference Kunda1990), and, within moral psychology specifically, there is a growing body of evidence suggesting that people often adopt certain views as part of a post hoc attempt to justify prior moral intuitions (Ditto et al. Reference Ditto, Pizarro, Tannenbaum, Bartels, Bauman, Skitka and Medin2009; Haidt Reference Haidt2001). So the motivational bias hypothesis is perhaps best understood as the application to a new domain of a theoretical perspective that is already quite well supported elsewhere.

More importantly, the hypothesis makes it possible to explain all of the existing results without supposing that moral considerations actually play any role at all in any of the relevant competencies. The thought is that people's competencies are entirely non-moral but that a motivational bias then interferes with our ability to apply these concepts correctly. (An analogous case: If John sleeps with Bill's girlfriend, Bill may end up concluding that John's poetry was never really any good – but that does not mean that Bill's criteria for poetry actually involve any reference to sexual behavior.)

All in all, then, what we have here is an excellent hypothesis. It draws on well-established psychological theory, provides a clear explanation of existing results, and offers a wealth of new empirically testable predictions. The one problem is that when researchers actually went out and tested those new predictions, none of them were empirically confirmed. Instead, the experimental results again and again seemed to go against what would have been predicted on the motivational bias view. At this point, the vast majority of researchers working on these questions have therefore concluded that the motivational bias hypothesis cannot explain the full range of experimental findings and that some other sort of psychological process must be at work here (Hindriks Reference Hindriks2008; Machery Reference Machery2008; McCann Reference McCann2005; Nichols & Ulatowski Reference Nichols and Ulatowski2007; Turner Reference Turner2004; Wright & Bengson Reference Wright and Bengson2009; Young et al. Reference Young, Cushman, Adolphs, Tranel and Hauser2006).

4.2. The conversational pragmatics hypothesis

Let us turn, then, to a second possible alternative hypothesis. When people are engaged in ordinary discussions, their use of words does not simply serve as a straightforward reflection of the way they apply the corresponding concepts. Instead, people strive to act as helpful conversation partners, following certain complex principles that enable them to provide useful information to their audience. The study of these principles falls under the heading of “conversational pragmatics,” and researchers engaged in this study have illuminated many puzzling aspects of the way people ordinarily use language in communication. A number of researchers have suggested that this approach might also serve to explain the phenomena we are trying to understand here (Adams & Steadman Reference Adams and Steadman2004a; Reference Adams and Steadman2004b; Driver Reference Driver and Sinnott-Armstrong2008a; Reference Driver and Sinnott-Armstrong2008b).

To get a sense for this hypothesis, it might be helpful to start out by looking at a potentially analogous case in another domain. Imagine that you have a bathroom in your building but that this bathroom is completely nonfunctional and has been boarded up for the past three years. And now imagine that someone hands you a questionnaire that asks:

It does seem that your actual concept bathroom might correctly apply to the room in your building, but when you receive this question, you immediately have an understanding of what the questioner really wants to know – namely, whether or not you have a bathroom that actually works – and you might therefore choose to check the box marked “No.”

With these thoughts in mind, consider what might happen when subjects receive a questionnaire that asks whether they agree or disagree with the sentence:

It might be thought that people's concept of intentional action does not, in fact, apply to cases like this one; but that, as soon as they receive the questionnaire, they form an understanding of what the questioner really wants to know. The real question here, they might think, is whether the chairman deserves to be blamed for his behavior, and they might therefore check the circle marked “definitely agree.”

Similar remarks might be applied to many of the other effects described above. Thus, suppose that subjects are asked whether they agree or disagree with the sentence:

It might be thought that people's concept of causation does apply in cases like this one, but it also seems that subjects might quite reasonably infer that the real point of the question is to figure out whether the administrative assistant deserves blame for this outcome and that they might therefore check the circle marked “definitely disagree.”

Before going on any further, it might be helpful to take a moment to emphasize just how different this pragmatic hypothesis is from the motivational bias hypothesis we discussed above. The motivational bias hypothesis posits an error that affects people's understanding of certain morally relevant events. By contrast, the pragmatic hypothesis does not involve any error or even any effect on people's understanding of events. It simply suggests that people are applying certain kinds of conversational rules. The basic idea is that moral considerations aren't actually affecting people's understanding of the situation; it's just that moral considerations do sometimes affect people's view about which particular words would be best used to describe it.

In any case, although the two hypotheses are very different in their theoretical approaches, they have proved remarkably similar in their ultimate fate. Like the motivational bias hypothesis, the pragmatic hypothesis initially looked very promising – a clear and plausible explanation, backed by a well-supported theoretical framework – but, as it happened, the actual empirical data just never came out the way the pragmatic hypothesis would predict. Indeed, the pragmatic hypothesis suffers from many of the same problems that plagued the motivational bias hypothesis, along with a few additional ones that are all its own.

4.3. Summary

Looking through these various experiments, one gradually gets a general sense of what has been going wrong with the alternative explanations. At the core of these explanations is the idea that people start out with an entirely non-moral competence but that some additional factor then interferes and allows people's actual intuitions to be influenced by moral considerations. Each alternative explanation posits a different interfering factor, and each explanation thereby predicts that the whole effect will go away if this factor is eliminated. So one alternative explanation might predict that the effect will go away when we eliminate a certain emotional response, another that it will go away when we eliminate certain pragmatic pressures, and so forth.

The big problem is that these predictions never actually seem to be borne out. No one has yet found a way of eliminating the purported interfering factors and thereby making the effect go away. Instead, the effect seems always to stubbornly reemerge, coming back again and again despite all our best efforts to eliminate it.

Now, one possible response to these difficulties would be to suggest that we just need to try harder. Perhaps the relevant interfering factor is an especially tricky or well-hidden one, or maybe there are a whole constellation of different factors in place here, all working together to generate the effects observed in the experiments. When we finally succeed in identifying all of the relevant factors, we might be able to find a way of eliminating them all and thereby allowing people's purely non-moral competence to shine through unhindered.

Of course, it is at least possible that such a research program would eventually succeed, but I think the most promising approach at this point would be to try looking elsewhere. In my view, the best guess about why no one has been able to eliminate the interfering factors is that there just aren't any such factors. It is simply a mistake to try to understand these experimental results in terms of a purely non-moral competence which then gets somehow derailed by various additional factors. Rather, the influence of moral considerations that comes out in the experimental results truly is showing us something about the nature of the basic competencies people use to understand their world.

5.1. General approach

At the core of the approach is a simple and straightforward assumption that has already played an enormously important role in numerous fields of cognitive science. Specifically, I will be relying heavily on the claim that we make sense of the things that actually happen by considering other ways things might have been (Byrne Reference Byrne2005; Kahneman & Miller Reference Kahneman and Miller1986; Roese Reference Roese1997).

A quick example will help to bring out the basic idea here. Suppose that we come upon a car that has a dent in it. We might immediately think about how the car would have looked if it did not have this dent. Thus, we come to understand the way the car actually is by considering another way that it could have been and comparing its actual status to this imagined alternative.

An essential aspect of this process, of course, lies in our ability to select from among all the possible alternatives just the few that prove especially relevant. Hence, in the case at hand, we would immediately consider the possibility that the car could have been undented, and think: “Notice that this car is dented rather than undented.” But then there are all sorts of other alternatives that we would immediately reject as irrelevant or not worth thinking about. We would not take the time, for example, to consider the possibility that the car could have been levitating in the air, and then think: “Notice that this car is standing on the ground rather than levitating in the air.”

Our ability to pick out just certain specific alternatives and ignore others is widely regarded as a deeply important aspect of human cognition, which shapes our whole way of understanding the objects we observe. It is, for example, a deeply important fact about our way of understanding the dented car that we compare it to an undented car. If we had instead compared it to a levitating car, we would end up thinking about it in a radically different way.

A question now arises as to why people focus on particular alternative possibilities and ignore others. The answer, of course, is that all sorts of different factors can play a role here. People's selection of specific alternative possibilities can be influenced by their judgments about controllability, about recency, about statistical frequency, about non-moral forms of goodness and badness (for reviews, see Byrne Reference Byrne2005; Kahneman & Miller Reference Kahneman and Miller1986; Roese Reference Roese1997). But there is also another factor at work here that has not received quite as much discussion in the existing literature. A number of studies have shown that people's selection of alternative possibilities can be influenced by their moral judgments (McCloy & Byrne Reference McCloy and Byrne2000; N'gbala & Branscombe Reference N'gbala and Branscombe1995). In other words, people's intuition about which possibilities are relevant can be influenced by their judgments about which actions are morally right.

For a simple illustration, take the case of the chairman who hears that he will be helping the environment, but reacts with complete indifference. As soon as one hears this case, one's attention is drawn to a particular alternative possibility:

This alternative possibility seems somehow to be especially relevant, more relevant at least than many other possibilities we could easily imagine. In particular, one would not think:

Of course, one could imagine the chairman having this latter sort of attitude. One could imagine him saying: “I don't care at all whether we make profits. What I really want is just to make sure that the environment is harmed, and since this program will help the environment, I'm going to do everything I can to avoid implementing it.” Yet this possibility has a kind of peculiar status. It seems somehow preposterous, not even worth considering. But why? The suggestion now is that moral considerations are playing a role in people's way of thinking about alternative possibilities. Very roughly, people regard certain possibilities as relevant because they take those possibilities to be especially good or right.

With these thoughts in mind, we can now offer a new explanation for the impact of moral judgments on people's intuitions. The basic idea is just that people's intuitions in all of the domains we have been discussing – causation, doing/allowing, intentional action, and so on – rely on a comparison between the actual world and certain alternative possibilities. Because people's moral judgments influence the selection of alternative possibilities, these moral judgments end up having a pervasive impact on the way people make sense of human beings and their actions.Footnote ³

5.2. A case study

To truly spell out this explanation in detail, one would have to go through each of the different effects described above and show how each of these effects can be explained on a model in which moral considerations are impacting people's way of thinking about alternative possibilities. This would be a very complex task, and I will not attempt it here. Let us proceed instead by picking just one concept whose use appears to be affected by moral considerations. We can then offer a model of the competence underlying that one concept and thereby illustrate the basic approach. For these illustrative purposes, let us focus on the concept in favor.

We begin by introducing a fundamental assumption that will guide the discussion that follows. The assumption is that people's representation of the agent's attitude is best understood, not in terms of a simple dichotomy between “in favor” and “not in favor,” but rather, in terms of a whole continuum of different attitudes an agent might hold. So we will be assuming that people can represent the agent as strongly opposed, as strongly in favor, or as occupying any of the various positions in between. For simplicity, we can depict this continuum in terms of a scale running from con to pro. Footnote ⁴ (See Fig. 6.)

Figure 6. Continuum of attitude ascription.

Looking at this scale, it seems that an agent whose attitude falls way over on the con side will immediately be classified as “not in favor,” and that an agent whose attitude falls way over on the pro side will immediately be classified as “in favor.” But now, of course, we face a further question. How do people determine the threshold at which an agent's attitude passes over from the category “not in favor” to the category “in favor”?

To address this question, we will need to add an additional element to our conceptual framework. Let us say that people assess the various positions along the continuum by comparing each of these positions to a particular sort of alternative possibility. We can refer to this alternative possibility as the default. Then we can suggest that an agent will be counted as “in favor” when his or her attitude falls sufficiently far beyond this default point. (See Fig. 7.)

Figure 7. Criteria for ascription of “in favor.”

The key thing to notice about this picture is that there needn't be any single absolute position on the continuum that always serves as the threshold for counting an agent as “in favor.” Instead, the threshold might vary freely, depending on which point gets picked out as the default.

To get a sense for the idea at work here, it may be helpful to consider a closely analogous problem. Think of the process a teacher might use in assigning grades to students. She starts out with a whole continuum of different percentage scores on a test, and now she needs to find a way to pick out a threshold beyond which a given score will count as an A. One way to do this would be to introduce a general rule, such as “a score always counts as an A when it is at least 20 points above the default.” Then she can pick out different scores as the default on different tests – treating 75% as default on easy tests, 65% as default on more difficult ones – and the threshold for counting as an A will vary accordingly.

The suggestion now is that people's way of thinking about attitudes uses this same sort of process. People always count an agent as “in favor” when the agent's attitude falls sufficiently far beyond the default, but there is no single point along the continuum that is treated as default in all cases. Different attitudes can be treated as default in different cases, and the threshold for counting as “in favor” then shifts around from one case to the next.

Now we arrive at the crux of the explanation. The central claim will be that people's moral judgments affect their intuitions by shifting the position of the default. For morally good actions, the default is to have some sort of pro-attitude, whereas for morally bad actions, the default is to have some sort of con-attitude. The criteria for “in favor” then vary accordingly.

Suppose we now apply this general framework to the specific vignettes used in the experimental studies. When it comes to helping the environment, it seems that the default attitude is a little bit toward the pro side. That is to say, the default in this case is to have at least a slightly positive attitude – not necessarily a deep or passionate attachment, but at least some minimal sense that helping the environment would be a nice thing to do. An attitude will then count as “in favor” to the extent that it goes sufficiently far beyond this default point. (See Fig. 8.)

Figure 8. Representation of the continuum for the help case.

But look at the position of the agent's actual attitude along this continuum. The agent is not even close to reaching up to the critical threshold here – he is only interested in helping the environment as a side-effect of some other policy, and people should therefore conclude that he does not count as “in favor” of helping.

Now suppose we switch over to the harm case. There, we find that the agent's actual attitude has remained constant, but the default has changed radically. When it comes to harming the environment, the default is to be at least slightly toward the con side – not necessarily showing any kind of vehement opposition, but at least having some recognition that harming the environment is a bad thing to do. An agent will then count as “in favor” to the extent that his or her attitude goes sufficiently far beyond this default (Fig. 9).

Figure 9. Representation of the continuum for the harm case.

In this new representation, the agent's actual attitude remains at exactly the same point it was above (in Fig. 8), but its position relative to the default is now quite different. This time, the attitude falls just about at the critical threshold for counting as “in favor,” and people should therefore be just about at the midpoint in their intuitions as to whether the agent was in favor of harming – which, in fact, is exactly what the experimental results show.

Notice how sharply this account differs from the alternative hypotheses discussed above. On those alternative hypotheses, people see that the agent harmed the environment, want to blame him for his behavior, and this interest in blame then shapes the way they conceptualize or describe various aspects of the case. The present account says nothing of the kind. Indeed, the account makes no mention at all of blame. Instead, it posits a role for an entirely different kind of moral judgment – a judgment that could be made even in the absence of any information about this specific agent or his behaviors. The claim is that before people even begin considering what actually happened in the case at hand, they can look at the act of harming the environment and make a judgment about what sort of attitude an agent could be expected to hold toward it. This judgment then serves as a standard they can use to make sense of the behavior they actually observe.

5.3. Extending the model

What we have here is a model of the competence underlying people's use of one particular concept. The key question now is whether this same basic approach can be applied to the various other concepts discussed above. In a series of recent papers, I have argued that it can be used to explain the impact of moral judgment on people's intuitions about freedom, knowledge, and causationFootnote ⁵ (Hitchcock & Knobe Reference Hitchcock and Knobe2009; Pettit & Knobe, Reference Pettit and Knobe2009; Phillips & Knobe Reference Phillips and Knobe2009). But new studies are coming out all the time, and we may soon be faced with experimental results that the model cannot explain. At any rate, one certainly should not expect that this model will turn out to be correct in every detail. Presumably, further work will show that it needs to be revised or expanded in various ways, and perhaps it will even have to be scrapped altogether.

In the present context, however, our concern is not so much to explore the details of this one model as to use it as a way of illustrating a more general approach and the contrast between this approach and the one we saw in the alternative explanations described above. The alternative explanations start out with the idea that the relevant competencies are entirely non-moral, but that some additional factor then interferes and allows people's intuitions to be influenced by moral considerations. These explanations therefore predict that it should be possible, at least in principle, to eliminate the interfering factors and examine the judgments people make in the absence of this influence.

By contrast, in the approach under discussion here, moral considerations are not understood as some kind of extra factor that gets added in on top of everything else. Instead, the whole process is suffused with moral considerations from the very beginning. Hence, in this approach, no real sense can be attached to the idea of eliminating the role of morality and just watching the basic process unfold in its pure, non-moral form.