Knobe presents us with a false dilemma on the level of the metaphors he uses: maybe we are neither scientists nor moralists. But he also presents us with a false dilemma when it comes to the two explanatory schemes he considers: The first one is that the competences that underlie our mental capacities (to attribute intentionality or to spot causal relevance) are influenced by moral considerations. The second is that these competences are themselves non-moral, but there is some additional factor that makes it the case that our attribution of intentionality is influenced by moral considerations. I will focus on the attribution of intentionality that Knobe considers to be the strongest case in favor of his claims.
The two options Knobe offers are not exhaustive. In fact, they share a premise that we have good reasons to doubt: that is, the premise that the attribution of intentionality is influenced by moral considerations. Knobe's reason for holding this claim is that in two very similar scenarios, the “harm” and the “help” scenarios (Knobe Reference Knobe2003a; see also sect. 3.1 of the target article) that differ only in their moral overtones, our attribution of intentionality also differs. As he says, “the only major difference between the two vignettes lies in the moral status of the chairman's behavior” (sect. 3.1, para. 2).
But that is definitely not the only major difference (see Nanay [2010] for an overview). One striking feature of the experiments Knobe and his collaborators conducted on this topic is that they all share the same structure. To put it very simply, in one scenario, the agent has two reasons for performing a certain action and ignores one of these; in the other, the agent has a reason for and a reason against performing an action and ignores the reason against. Thus, in Knobe's most famous helping/harming experiment (Knobe Reference Knobe2003a; see also target article, sect. 3.1), we have the following two scenarios:
(a) In the harm case, the chairman has a reason (R1) for introducing the plan (i.e., to increase profit) and a reason (R2) against (i.e., to avoid harming the environment).
(b) In the help case, in contrast, the chairman has two different reasons to introduce the plan: he had a reason to increase the company's profit (R1) and he also had a reason to help the environment (R3).
In short, the difference between scenario (a) and scenario (b) is that in (a) the chairman has R1 for and R2 against introducing the plan, whereas in (b) he has R1 and R3 both in favor of performing this action. Importantly, the chairman chooses to ignore the environmental considerations: R2 and R3, respectively. This leaves R1 in both scenarios, which is a reason for introducing the plan. There is no difference between (a) and (b) in the actual reason the chairman is acting on.
But there is a modal difference between (a) and (b): a difference in what would happen if the chairman did not ignore R2 and R3, respectively. Contrast the original scenarios (a) and (b) with another pair of cases where the chairman chooses not to ignore the environmental considerations:
(a*) The chairman chooses not to ignore R2 (i.e., a reason against introducing the plan). Then his action would, or at least it could, be different, as now he has a reason for (R1) and a reason against (R2) introducing the plan.
(b*) The chairman chooses not to ignore R3 (i.e., a reason for introducing the plan). His action would still be the same, as now he has two reasons (R1 and R3) in favor of introducing the plan.
So an important difference between case (a) and case (b) is a modal one: The outcome would be different if the chairman didn't ignore the environmental considerations. In (b), ignoring that the plan helps the environment would make no difference, as there are two independent reasons in favor of introducing the plan: The chairman's actions in (b) and in (b*) will be the same. In (a), on the other hand, ignoring that the plan harms the environment would make (or at least it could make) a difference: The chairman's actions in (a) and in (a*) will be (or at least can be) different.
Thus, what this experiment shows is that in (b), introducing the new scheme does not depend counterfactually on ignoring the environmental considerations, whereas in (a), there is counterfactual dependence between ignoring the environmental considerations and introducing the new scheme. This counterfactual dependence in (a) is not very strong, as not ignoring will not guarantee that the chairman's action will be different, but it is an instance of counterfactual dependence nonetheless. In (b), we have no counterfactual dependence, weak or strong.
What I have said so far shows that the experimental data Knobe uses can be explained with the help of an alternative hypothesis, where the attribution of intentionality does not depend on our moral judgments. In other words, we have two ways of explaining Knobe's original experiments: one appeals to moral judgments, the other one does not. The fact that my explanatory scheme is consistent with Knobe's findings in itself casts doubt on his conclusion.
But we can say something even stronger. My explanatory scheme is in fact preferable to Knobe's for two reasons. First, my explanatory scheme is more robust than Knobe's: it can explain cases of the attribution of intentionality that Knobe's cannot. There are several scenarios where we get differences in the attribution of intentionality without any moral difference (Machery Reference Machery2008; Mallon Reference Mallon2008; Nanay Reference Nanay2010; Nichols & Ulatowski Reference Nichols and Ulatowski2007; maybe even Knobe Reference Knobe2007). As these cases all follow the modal asymmetry I identified, I can account for them (Nanay Reference Nanay2010). Knobe cannot.
Second, those of us with naturalist leanings prefer to explain our complex mental capacities in simple terms. When explaining the mental capacity of attributing intentionality to others, the (broadly) naturalistic way to proceed would be to account for this mental capacity with reference to simple mental processes. This is exactly my strategy: If we can explain the attribution of intentionality with reference to mental capacities that nonhuman animals also possess, plus some further ability to be sensitive to counterfactual situations (which at least some nonhuman primates may also possess; see Suddendorf & Whiten Reference Suddendorf and Whiten2001), we should not rely on any further, uniquely human higher-order phenomena, such as morality.
You have
Access
Knobe presents us with a false dilemma on the level of the metaphors he uses: maybe we are neither scientists nor moralists. But he also presents us with a false dilemma when it comes to the two explanatory schemes he considers: The first one is that the competences that underlie our mental capacities (to attribute intentionality or to spot causal relevance) are influenced by moral considerations. The second is that these competences are themselves non-moral, but there is some additional factor that makes it the case that our attribution of intentionality is influenced by moral considerations. I will focus on the attribution of intentionality that Knobe considers to be the strongest case in favor of his claims.
The two options Knobe offers are not exhaustive. In fact, they share a premise that we have good reasons to doubt: that is, the premise that the attribution of intentionality is influenced by moral considerations. Knobe's reason for holding this claim is that in two very similar scenarios, the “harm” and the “help” scenarios (Knobe Reference Knobe2003a; see also sect. 3.1 of the target article) that differ only in their moral overtones, our attribution of intentionality also differs. As he says, “the only major difference between the two vignettes lies in the moral status of the chairman's behavior” (sect. 3.1, para. 2).
But that is definitely not the only major difference (see Nanay [2010] for an overview). One striking feature of the experiments Knobe and his collaborators conducted on this topic is that they all share the same structure. To put it very simply, in one scenario, the agent has two reasons for performing a certain action and ignores one of these; in the other, the agent has a reason for and a reason against performing an action and ignores the reason against. Thus, in Knobe's most famous helping/harming experiment (Knobe Reference Knobe2003a; see also target article, sect. 3.1), we have the following two scenarios:
(a) In the harm case, the chairman has a reason (R1) for introducing the plan (i.e., to increase profit) and a reason (R2) against (i.e., to avoid harming the environment).
(b) In the help case, in contrast, the chairman has two different reasons to introduce the plan: he had a reason to increase the company's profit (R1) and he also had a reason to help the environment (R3).
In short, the difference between scenario (a) and scenario (b) is that in (a) the chairman has R1 for and R2 against introducing the plan, whereas in (b) he has R1 and R3 both in favor of performing this action. Importantly, the chairman chooses to ignore the environmental considerations: R2 and R3, respectively. This leaves R1 in both scenarios, which is a reason for introducing the plan. There is no difference between (a) and (b) in the actual reason the chairman is acting on.
But there is a modal difference between (a) and (b): a difference in what would happen if the chairman did not ignore R2 and R3, respectively. Contrast the original scenarios (a) and (b) with another pair of cases where the chairman chooses not to ignore the environmental considerations:
(a*) The chairman chooses not to ignore R2 (i.e., a reason against introducing the plan). Then his action would, or at least it could, be different, as now he has a reason for (R1) and a reason against (R2) introducing the plan.
(b*) The chairman chooses not to ignore R3 (i.e., a reason for introducing the plan). His action would still be the same, as now he has two reasons (R1 and R3) in favor of introducing the plan.
So an important difference between case (a) and case (b) is a modal one: The outcome would be different if the chairman didn't ignore the environmental considerations. In (b), ignoring that the plan helps the environment would make no difference, as there are two independent reasons in favor of introducing the plan: The chairman's actions in (b) and in (b*) will be the same. In (a), on the other hand, ignoring that the plan harms the environment would make (or at least it could make) a difference: The chairman's actions in (a) and in (a*) will be (or at least can be) different.
Thus, what this experiment shows is that in (b), introducing the new scheme does not depend counterfactually on ignoring the environmental considerations, whereas in (a), there is counterfactual dependence between ignoring the environmental considerations and introducing the new scheme. This counterfactual dependence in (a) is not very strong, as not ignoring will not guarantee that the chairman's action will be different, but it is an instance of counterfactual dependence nonetheless. In (b), we have no counterfactual dependence, weak or strong.
What I have said so far shows that the experimental data Knobe uses can be explained with the help of an alternative hypothesis, where the attribution of intentionality does not depend on our moral judgments. In other words, we have two ways of explaining Knobe's original experiments: one appeals to moral judgments, the other one does not. The fact that my explanatory scheme is consistent with Knobe's findings in itself casts doubt on his conclusion.
But we can say something even stronger. My explanatory scheme is in fact preferable to Knobe's for two reasons. First, my explanatory scheme is more robust than Knobe's: it can explain cases of the attribution of intentionality that Knobe's cannot. There are several scenarios where we get differences in the attribution of intentionality without any moral difference (Machery Reference Machery2008; Mallon Reference Mallon2008; Nanay Reference Nanay2010; Nichols & Ulatowski Reference Nichols and Ulatowski2007; maybe even Knobe Reference Knobe2007). As these cases all follow the modal asymmetry I identified, I can account for them (Nanay Reference Nanay2010). Knobe cannot.
Second, those of us with naturalist leanings prefer to explain our complex mental capacities in simple terms. When explaining the mental capacity of attributing intentionality to others, the (broadly) naturalistic way to proceed would be to account for this mental capacity with reference to simple mental processes. This is exactly my strategy: If we can explain the attribution of intentionality with reference to mental capacities that nonhuman animals also possess, plus some further ability to be sensitive to counterfactual situations (which at least some nonhuman primates may also possess; see Suddendorf & Whiten Reference Suddendorf and Whiten2001), we should not rely on any further, uniquely human higher-order phenomena, such as morality.