Fiery Cushman's account of rationalization involves two main claims: that the beliefs and desires appealed to in rationalizing explanations are constructed post hoc, and that these constructions can help improve future reasoning. While we agree with Cushman regarding the roles that non-rational systems such as social norms (sect. 3.3) and instincts (sect. 3.2) play in rationalization, we believe that he extends his account too far with his discussion of habits (sect. 3.6). Cushman gives two examples to motivate the claim that rationalization extracts implicit information from habitual actions to guide future behavior. However, it seems that these examples are either better explained as intentional actions or they do not involve rationalization at all. In either case, the role of habitual action in Cushman's account of rationalization needs revision.
Post hoc rationalization is effective so long as the agent can construct a belief and desire that favor that particular behavior token. For example, suppose you are on a diet but decide to eat the last cookie anyway, thereby experiencing a lapse in self-control. When asked why you ate the cookie, you are asked to state the beliefs and desires that would thereby make your action rational. In other words, you are asked to give a means-end justification. In response, you might construct a post hoc justification for yourself that makes your action seem rational rather than weak-willed. For example, “it would have gone to waste if I didn't eat it.” If not letting food go to waste is a reasonable end, then, ceteris paribus, eating the cookie is a justified means to achieve that end. You might even believe that this was the operative reason for your behavior. However, it was your desire to eat the cookie that played the operative role.
Means-end explanations are likewise constructed when we rationalize behavior influenced by social norms and instincts. In each case of rationalization, we construct a belief and desire that favor the behavior in question – thus attributing reasons to non-rational processes. However, it is less clear when and why we would rationalize our habitual actions.
When we act out of habit, the resulting behavior typically either (1) will map onto the right circumstances to produce the intended result or (2) will “misfire.” If the habitual action maps onto the right circumstances, then the action accomplishes the very end(s) for which the habit was formed in the first place. In such cases, the habitual behavior token is supported by a belief and desire. But these were not constructed post hoc. Rather, the habitual behavior token is best explained as being caused by a standing belief and desire (see Mele Reference Mele2007). There is no rationalization of the behavior because the relevant beliefs and desires were there all along.
Now consider Cushman's primary example given to explain the post hoc rationalization of habits: turning on lights in a sleeping baby's room (sect. 3.6). Suppose that I turn on the lights in this scenario out of habit. Suppose further that I developed this habit because it is typically useful to turn lights on when I enter dark rooms. However, in this particular case, it was not useful. Indeed, turning on the lights directly opposes my supposed end: namely, making sure the baby is still asleep. This example thus fits into the second category of habitual action. The habitual action token is a “misfire,” which means it did not map onto the right circumstances. I might have some end met by turning on the light (e.g., being able to see clearly), but that end still does not make my behavior rational. It is not rational because my end of keeping the baby asleep is more important to me than being able to see clearly. So, instead of constructing new beliefs or desires that were not operative at the time of turning on the light, the natural response would be to say something like “I forgot the baby was sleeping.”
“Misfire” cases like the sleeping baby example leave little room for post hoc rationalizations because they are not supported by a belief and desire. My habitual behavior token did not map onto the right circumstances, and the underlying cause of the behavior was in opposition to my end. Instead of using means-end language, I simply admit my mistake. If rationalizing habits does “provide new information to a system of reasoning” (p. 26), this example does not seem to provide much evidence for that claim. Indeed, it is still unclear what sort of habitual action would provide that cognitive benefit.
Let us now turn to Cushman's other example of habitual action. This is the example of the person who habitually eats cake (sect. 3.6.2). If our cake eater does not recognize any salient reasons to stop eating cake, it is unclear what role rationalization could play. However, he might rationalize his behavior if he does recognize salient reasons to stop eating cake. But if he continues to eat cake despite recognizing these reasons, there appears to be more going on here than mere habitual action. In particular, our cake eater seems to be suffering from a lack of self-control. Once akrasia is introduced as the cause of his behavior, it is no longer a clear example of behavior generated by non-rational processes. This is because akrasia is generally understood as intentionally doing x even though the agent has judged it to be all-things-considered better to do y than x (e.g., see Davidson Reference Davidson and Feinberg1970). In other words, akrasia is the result of at least some rational processes rather than non-rational ones. If Cushman's goal is to identify scenarios in which “the rationalization of habitual action could provide new information to a system of reasoning” (sect. 3.6, para. 2), then the cake-eating example seems to miss the mark as well.
Perhaps habitual action does play an important role in rationalization, but revision and clarification are needed this aspect of Cushman's account to be useful. As it stands, it is unclear when – if ever – we rationalize habitual behavior in the first place.
You have
Access
Fiery Cushman's account of rationalization involves two main claims: that the beliefs and desires appealed to in rationalizing explanations are constructed post hoc, and that these constructions can help improve future reasoning. While we agree with Cushman regarding the roles that non-rational systems such as social norms (sect. 3.3) and instincts (sect. 3.2) play in rationalization, we believe that he extends his account too far with his discussion of habits (sect. 3.6). Cushman gives two examples to motivate the claim that rationalization extracts implicit information from habitual actions to guide future behavior. However, it seems that these examples are either better explained as intentional actions or they do not involve rationalization at all. In either case, the role of habitual action in Cushman's account of rationalization needs revision.
Post hoc rationalization is effective so long as the agent can construct a belief and desire that favor that particular behavior token. For example, suppose you are on a diet but decide to eat the last cookie anyway, thereby experiencing a lapse in self-control. When asked why you ate the cookie, you are asked to state the beliefs and desires that would thereby make your action rational. In other words, you are asked to give a means-end justification. In response, you might construct a post hoc justification for yourself that makes your action seem rational rather than weak-willed. For example, “it would have gone to waste if I didn't eat it.” If not letting food go to waste is a reasonable end, then, ceteris paribus, eating the cookie is a justified means to achieve that end. You might even believe that this was the operative reason for your behavior. However, it was your desire to eat the cookie that played the operative role.
Means-end explanations are likewise constructed when we rationalize behavior influenced by social norms and instincts. In each case of rationalization, we construct a belief and desire that favor the behavior in question – thus attributing reasons to non-rational processes. However, it is less clear when and why we would rationalize our habitual actions.
When we act out of habit, the resulting behavior typically either (1) will map onto the right circumstances to produce the intended result or (2) will “misfire.” If the habitual action maps onto the right circumstances, then the action accomplishes the very end(s) for which the habit was formed in the first place. In such cases, the habitual behavior token is supported by a belief and desire. But these were not constructed post hoc. Rather, the habitual behavior token is best explained as being caused by a standing belief and desire (see Mele Reference Mele2007). There is no rationalization of the behavior because the relevant beliefs and desires were there all along.
Now consider Cushman's primary example given to explain the post hoc rationalization of habits: turning on lights in a sleeping baby's room (sect. 3.6). Suppose that I turn on the lights in this scenario out of habit. Suppose further that I developed this habit because it is typically useful to turn lights on when I enter dark rooms. However, in this particular case, it was not useful. Indeed, turning on the lights directly opposes my supposed end: namely, making sure the baby is still asleep. This example thus fits into the second category of habitual action. The habitual action token is a “misfire,” which means it did not map onto the right circumstances. I might have some end met by turning on the light (e.g., being able to see clearly), but that end still does not make my behavior rational. It is not rational because my end of keeping the baby asleep is more important to me than being able to see clearly. So, instead of constructing new beliefs or desires that were not operative at the time of turning on the light, the natural response would be to say something like “I forgot the baby was sleeping.”
“Misfire” cases like the sleeping baby example leave little room for post hoc rationalizations because they are not supported by a belief and desire. My habitual behavior token did not map onto the right circumstances, and the underlying cause of the behavior was in opposition to my end. Instead of using means-end language, I simply admit my mistake. If rationalizing habits does “provide new information to a system of reasoning” (p. 26), this example does not seem to provide much evidence for that claim. Indeed, it is still unclear what sort of habitual action would provide that cognitive benefit.
Let us now turn to Cushman's other example of habitual action. This is the example of the person who habitually eats cake (sect. 3.6.2). If our cake eater does not recognize any salient reasons to stop eating cake, it is unclear what role rationalization could play. However, he might rationalize his behavior if he does recognize salient reasons to stop eating cake. But if he continues to eat cake despite recognizing these reasons, there appears to be more going on here than mere habitual action. In particular, our cake eater seems to be suffering from a lack of self-control. Once akrasia is introduced as the cause of his behavior, it is no longer a clear example of behavior generated by non-rational processes. This is because akrasia is generally understood as intentionally doing x even though the agent has judged it to be all-things-considered better to do y than x (e.g., see Davidson Reference Davidson and Feinberg1970). In other words, akrasia is the result of at least some rational processes rather than non-rational ones. If Cushman's goal is to identify scenarios in which “the rationalization of habitual action could provide new information to a system of reasoning” (sect. 3.6, para. 2), then the cake-eating example seems to miss the mark as well.
Perhaps habitual action does play an important role in rationalization, but revision and clarification are needed this aspect of Cushman's account to be useful. As it stands, it is unclear when – if ever – we rationalize habitual behavior in the first place.