2. The Normative Phenomena and Their Dominant Explanation

Let us start by observing some normative phenomena associated with intention. Normally, an agent is irrational if she intends to do A, believes that doing B is the only way to do A, but doesn't intend to do B. Anna is irrational if she intends to be in Vancouver by midnight, believes that taking the next flight is the only way to be in Vancouver by midnight, but doesn't intend to take the next flight. Also normally, an agent is irrational if she intends to do A, intends to do B, but believes that doing B would prevent her from doing A. Bill is irrational if he intends to meet Carla on the west side of town, intends to meet Dan on the east side of town, but believes that meeting Dan would prevent him from meeting Carla. And normally, an agent is irrational if she intends to do A and, without reason, reconsiders whether she should do A. Ellie is irrational if she concludes that, all things considered, she should buy the blue dress, and thereby intends to buy the blue dress, but without reason, reconsiders once again whether she should buy the blue dress.

One explanation of these normative phenomena goes like this. Intentions, as such, are subject to a set of rational requirements. If an agent violates these requirements, then she is irrational. And there is a requirement that corresponds to each of the above irrationalities.Footnote ⁴ Anna is irrational because an agent must not intend to do A, believe that doing B is the only way to do A, but not intend to do B. Bill is irrational because an agent must not intend to do A, and intend to do B, while believing that doing B would prevent him from doing A. Ellie is irrational because an agent must not intend to do A and, without reason, reconsider whether she should do A. The irrationality does not arise from the reasons for action and belief that bear on the agent but from having the intentions themselves, given the kind of thing an intention is.

This explanation is important because it functions as an argument for a certain theory of intention. According to this theory, an intention is a special state of mind individuated (partly) by the distinctive rational requirements to which it is subject. It looks plausible that an intention is such a state of mind because assuming so gives us an explanation of the normative phenomena. Having an intention puts one in a state of mind that is subject to the relevant rational requirements. How else could one explain them?

3. The Alternative Explanation of the Normative Phenomena

Through a series of articles by Raz, Kolodny, and Tenenbaum, an alternative explanation has emerged. According to this explanation, there are no (or need not be any) special rational requirements that apply to intentions as such. Rather, the irrationality arises from the reasons for action and belief that bear on the agent.

Consider Anna. She intends to be in Vancouver by midnight, believes that taking the next flight is the only way to be in Vancouver by midnight, but doesn't intend to take the next flight. Why is this irrational? According to this alternative explanation, it is because if there is decisive reason to be in Vancouver by midnight, then there is decisive reason to intend to be in Vancouver by midnight. So, an agent who is responsive to the reasons for action will form the intention to be in Vancouver by midnight. Now, if there is also decisive reason to believe that taking the next flight is the only way to be in Vancouver by midnight, then there is decisive reason to take the next flight. And if there is decisive reason to take the next flight, then there is decisive reason to form the intention to take the next flight. So, an agent who is responsive to the reasons for action and belief will not only form the intention to be in Vancouver by midnight and the belief that taking the next flight is the only way to be in Vancouver by midnight but also the intention to take the next flight. Anna is irrational because, given that she is responding appropriately to the reasons in forming her initial intention and her belief, she is failing to respond appropriately to certain reasons for action — namely, the decisive reason to take the next flight. Thus, her irrationality derives not from special requirements on intention as such but from the reasons for action and belief in the present circumstances.Footnote ⁵

Now consider Bill. He intends to meet Carla on the west side of town, intends to meet Dan on the east side of town, but believes that meeting Dan would prevent him from meeting Carla. Why is this irrational? According to the alternative explanation, there are two possibilities. Let us suppose that there is decisive reason to believe that meeting Dan would prevent him from meeting Carla. First, if there is more reason to meet Carla than Dan and there is decisive reason to meet Carla, then there is decisive reason to not intend to meet Dan. This is because there is decisive reason to believe that merely having the intention to meet Dan lowers the probability of meeting Carla. Having that intention will just get in the way of doing what Bill has most reason to do. On this possibility, Bill is irrational because, given that he is responding appropriately to the reasons in forming his intention to meet Carla and his belief that meeting Dan would prevent him from meeting Carla, he is failing to respond appropriately to the reasons for action in maintaining his intention to meet Dan.

On the second possibility, if there is equal reason to meet both Carla and Dan, then (all other courses of action being worse) there is decisive reason to not have both intentions. Each intention is getting in the way of the opposing action. There is decisive reason to give up one of the intentions. So, Bill is irrational because, given that he is responding appropriately to the reasons in forming his belief that meeting Dan would prevent him from meeting Carla, he is failing to respond appropriately to the reasons for action in maintaining both intentions. In each possibility, Bill's irrationality derives not from special requirements on intention as such but from the reasons for action and belief in the present circumstances.Footnote ⁶

Finally consider Ellie. She concludes that, all things considered, she should buy the blue dress, and thereby intends to buy the blue dress, but without reason, reconsiders once again whether she should buy the blue dress. Why is this irrational? First, let's note that if there was decisive reason to buy the blue dress at the conclusion of her deliberation and there is no reason to believe that reconsidering her conclusion would prevent her from buying the blue dress, then there is no reason against reconsidering her conclusion (even if there is decisive reason to believe that the reasons haven't changed). In this scenario, reconsidering doesn't seem irrational. However, normally, there is decisive reason to believe that reconsidering an intention too much will prevent one from achieving what's intended. Thus, if there was decisive reason to buy the blue dress at the conclusion of her deliberation and there is decisive reason to believe that reconsidering her conclusion too much will prevent her from buying the blue dress, then there is decisive reason to not reconsider too much. Ellie is irrational, in this scenario, because, given that she is responding appropriately to the reasons in forming her intention to buy the blue dress and her belief that reconsidering her conclusion too much will prevent her from buying the blue dress, she is failing to respond appropriately to the reasons for action in reconsidering her conclusion if she is disregarding how much is too much. Again, her irrationality derives not from special requirements on intention as such but from the reasons for action and belief in the present circumstances.Footnote ⁷

Thus, this alternative explanation undermines three reasons for thinking that an intention is a special state of mind individuated (partly) by the distinctive rational requirements to which it is subject. For many, at this point, this will not undermine the thought that an intention is a state of mind but merely the thought that such a state of mind is subject to and individuated by distinctive rational requirements. In other words, many will continue to believe that intention is a mental state, while believing that such rational requirements are ‘a myth.’ It is worth noting, however, that nothing in this alternative explanation suggests that intention is a mental state. We can explain the normative phenomena without assuming that intention is subject to and individuated by distinctive rational requirements and without assuming that intention is a mental state. So, this alternative explanation does not supply us with reason to believe that intention is a mental state.

4. The Descriptive Phenomena and Their Dominant Explanation

One might think that although the alternative explanation does not make reference to intentions as mental states, it does presuppose certain descriptive phenomena and we need to assume that intentions are mental states in order to explain those descriptive phenomena. In particular, the alternative explanation presupposes that having an intention to do A makes it more likely that one will do A, and this seems to be due (at least) to the following three descriptive phenomena.

(1) Normally, if an agent intends to do A, she is more likely to reason about how to do A than otherwise. If Frank intends to order something in for dinner, he is very likely to reason about which restaurant to order from. He is more likely to reason about how to order in dinner than if he had no intention at all to order in. He is also more likely to reason about how to order in than if he merely wanted to order in. (2) Normally, if an agent intends to do A, she is more likely to avoid intending or doing conflicting actions than otherwise. If Gertrude intends to spend the next hour swimming laps, she is very likely to put her book away and stop reading — much more likely than if she had no intention to swim or merely wanted to swim. And (3) normally, if an agent intends to do A, she is more likely to not deliberate about whether to do A than otherwise. If Harry intends to vacation in Thailand, he is likely (all else remaining equal) to no longer consider whether to vacation in Thailand — at least, much more likely than if he had no intention to vacation in Thailand or merely wanted to. Having an intention, in some way, makes it more likely that an agent will reason how, avoid conflicts, and close the question about whether to act.

Even if the alternative explanation of the normative phenomena doesn't presuppose these, they are (at least) independent reasons to believe that intention is a mental state. Or so one might think. Why? Because the claim that intention is a mental state explains these descriptive phenomena. An intention, the explanation goes, is a mental state that plays a certain causal role. Namely, it causes an agent to reason about how to execute it and avoid conflicts and close the question. This is the way that having an intention makes it more likely that an agent will reason how, avoid conflicts, and close the question. Frank is in a state that causally influences him to reason about which restaurant to order from. Gertrude is in a state that causally influences her to put her book away. And Harry is in a state that causally influences him to not consider whether to vacation in Thailand. The claim that intention is a mental state explains the descriptive phenomena. So, we think it is a mental state.

5. The Alternative Explanation of the Descriptive Phenomena

But there is an alternative explanation of the descriptive phenomena. According to this explanation, intention is not a mental state but an action. In particular, a token intention to do A is a token action of type A. So, a token intention to bake bread is a token bread-baking.Footnote ⁸

But an intention is also more than this. First, to intend to do A, it is not enough that the agent is engaged in some process that is bringing about the change that defines the completion of an action of type A. For the agent may be engaged in this process inadvertently, involuntarily, or accidentally. The agent must be engaged intentionally. Thus, an intention to do A is an intentional action of type A. So, a token intention to bake bread is a token intentional bread-baking.Footnote ⁹

But this is also not enough. For, if it were, there would be no difference between intending the action and doing it. It would be enough to be baking bread that one intends to bake bread, yet this is false. The difference between intentions and doings is in their sub-actions. Roughly, intentions characteristically have preparatory sub-actions, and doings have productive sub-actions. Somewhat more precisely, intentions have sub-actions that enable the completion of the intention, and doings have sub-actions that satisfy the satisfaction conditions of that type of action. Thus, an intention to do A is an intentional action of type A such that it is currently composed of a preparatory sub-action. For example, an intention to bake bread may be an intentional bread-baking currently composed of buying some flour at the supermarket. The agent merely intends to bake bread and is not baking bread because her action (a token of the bread-baking type of action) is currently composed of a preparatory sub-action: the buying of some flour. All else being equal, when the agent is kneading the dough, then her bread-baking is a doing: she is baking bread intentionally.Footnote ¹⁰

According to this explanation, an action is not made intentional by being caused by or explained by an intention. Rather, an action is made intentional by being the explanation of its sub-action — by being the whole that explains the part. Thus, when a token action of type M is a sub-action of some token action of type E, the latter is thereby performed intentionally and is thereby the agent's motivating reason for the former. That is, when a token dough-kneading is a sub-action of some token bread-baking, the bread-baking is performed intentionally, and the agent's motivating reason for kneading dough is her baking of bread.

But what makes it the case that a token action of type M is a sub-action of some token action of type E? It is (this explanation continues) the exercise of instrumental knowhow. Instrumental knowhow is a specific kind of knowhow. Where knowhow in general is knowledge how [to perform E], instrumental knowhow is knowledge how [to perform E by performing M]. When such knowledge is exercised, one performs an action of type E and an action of type M but also the latter is thereby a sub-action of the former, and thereby the former is performed intentionally and is the agent's motivating reason for performing the latter. In the above case, for instance, one is exercising her knowledge of how [to bake bread by kneading dough].

It is worth emphasizing how different this explanation is from the dominant one. On the dominant view, if one is baking bread intentionally, the bread-baking is intentional because one has an intention or some other state of mind that explains the bread-baking. In particular, it is often thought that the bread-baking is made intentional by being caused by a mental state of intention — let's suppose an intention to eat bread. This intention to eat bread is thereby (that is, because it is the right kind of explanation of the bread-baking) the agent's motivating reason for baking bread. On the alternative view, the facts are not like this at all. If one is baking bread intentionally, the bread-baking is intentional because the bread-baking explains some sub-procedural part of it — let's suppose, as above, a dough-kneading. This bread-baking is thereby (that is, because it is the right kind of explanation of the dough-kneading) the agent's motivating reason for kneading dough. And the bread-baking is this explanation of the dough-kneading and thus intentional because it is the exercise of the agent's knowledge of how [to bake bread by kneading dough].

We can now use this alternative view to explain our descriptive phenomena. Normally, having an intention makes it more likely that one will reason how, avoid conflicts, and close the question because, normally, if one has an intention, then she is performing an intentional action currently composed of some preparatory sub-action and thus exercising knowledge how [to perform that type of action by performing this type of preparatory sub-action]. Reasoning how and avoiding conflicts are preparatory actions. Deliberating about whether to perform an action is not.

Consider Frank. If Frank intends to order something in for dinner, he is very likely to reason about which restaurant to order from. He is more likely to reason about how to order in dinner than if he had no intention at all to order in. He is also more likely to reason about how to order in than if he merely wanted to order in. This is because Frank's intention to order in is an intentional action of ordering in. Therefore, this action is composed of some preparatory action, and he is exercising knowledge how [to order in by performing this preparatory action]. Reasoning about which restaurant to order from is a preparatory action (a preparatory mental action). Thus, it is more likely that Frank is reasoning about which restaurant to order from than were he not performing the action of ordering in. Gertrude and Harry will have parallel explanations.

Having an intention to do A makes it more likely that one is also reasoning about how to do A, and more likely that one is avoiding intending and doing actions that conflict with doing A, and more likely that one is not deliberating about whether to do A not because the intention is a mental state that plays just that causal role. Rather, it is because the intention is an action that is characteristically composed (or not) of just those types of action.

More broadly, (normally) if an agent intends to do A, then she is more likely to do A because (normally) if she intends to do A, then she is taking and knows how to take some preparatory step toward doing A, thereby enabling the satisfaction of some satisfaction-condition for doing A and thereby making such satisfaction more likely than it would otherwise be.

Therefore, the descriptive phenomena can be explained without reference to a state of intending, and if our alternative explanation of the normative phenomena presupposes these descriptive phenomena, then it is also free of reference to a state of intending. Furthermore, just as our alternative explanation of the normative phenomena draws on resources that should already be accepted — namely, the presence of reasons for belief and action — so too does our alternative explanation of the descriptive phenomena — namely, the presence of instrumental knowhow. That is to say, the reasons for belief and action in the alternative explanation of the normative phenomena are not something that we should deny. It should be common ground to accept that there are reasons for belief and action, and the specific reasons referenced are very plausible. This is the power of the alternative explanation: it explains the phenomena without positing special requirements but instead by drawing on already existing resources. This is why so-called ‘myth theorists’ (e.g., Raz, Kolodny, Tenenbaum) think that these special requirements are a myth.

Similarly, the instrumental knowhow in the alternative explanation of the descriptive phenomena is not something that we should deny. It should be common ground to accept that there is instrumental knowhow, and the specific knowhow referenced — knowing how to do an action by reasoning how or avoiding conflicts or not deliberating — is very plausible. This is the power of this alternative explanation: it explains the phenomena without positing a special state of the agent but instead by drawing on already existing resources. This is why we should also think that this special state — a state of intending — is a myth. Rather, an intention to do A is an intentional action of type A such that it is currently composed of a preparatory sub-action.Footnote ¹¹

6. An Objection

One might object that the alternative explanation of the descriptive phenomena fails to eliminate reference to a state of intending. After all, it still refers to knowhow, and knowhow is a state. So, it looks like we should identify an intention with the knowhow rather than the action. But if we do this (the objection goes), a state of intending is not a myth. It is simply a state of knowhow.

However, this objection fails because an intention should not be identified with knowhow. In particular, an intention to do E should not be identified with knowledge of how [to perform E] or knowledge of how [to perform E by performing M]. This is because one can know either without the intention to take the corresponding action. One can know how to play basketball without an intention to play basketball. One can know how to drive downtown by reasoning about which route to take without an intention to drive downtown. Agents have plenty of knowhow about plenty of actions that they have no intention to take. So, the state of knowhow is not a state of intending.

The alternative explanation does imply that the act of intending is the exercise of a certain instrumental knowhow. But this does not imply that intending is a state or is characteristically mental. It is not a state because the state of knowhow is a capacity of some sort (whether intellectualist or anti-intellectualist), and the exercise of a capacity is not a state. It is not characteristically mental because, although in many cases preparatory steps are mental actions, in many other cases preparatory steps are physical actions. Frank intends to order in because he is performing a mental preparatory step: he is reasoning about which restaurant to order from. While Gertrude intends to swim, because she is performing a physical preparatory step: she is putting her book away. Sometimes an intention consists in a mental action; sometimes it consists in a physical action.

But what makes it the case that Gertrude is putting her book away rather than any number of behaviourally identical physical actions that are not putting her book away? It is the fact that she is exercising instrumental knowledge of how [to swim by putting her book away]. But what makes that the case? Good question! We should like an answer to that question, but that is beyond the scope of this paper. (I can't explain everything.) The alternative explanation does imply, however, that an answer to this question will not be given by reference to the agent's intention. We must give some other kind of explanation of what it is to exercise instrumental knowhow. But it is not at all obvious that we should want such an explanation to make reference to intending.

7. An Advantage of This Explanation of Intention

Not only does this alternative explanation of the descriptive phenomena undermine the thought that intention is a mental state, it also has a significant advantage. Luca FerreroFootnote ¹² has argued for a condition of adequacy for any theory of intention: it should not make forming an intention a kind of self-manipulation. Here are two paradigmatic cases of self-manipulation. (1) Isabella wants to take a nap at noon but knows that she will be too distracted and excited to remember, so she takes a strong sleeping pill in the morning that will activate at noon and cause her to fall asleep. (2) Jacques is weak-willed when it comes to eating but is determined to lose five pounds by the end of the month, so he makes a serious bet with a serious man for $10,000 that he can do it.

In the former case, an agent at time t creates a state that at t+1 will have some causal effect such that the agent at t+1 will be more likely to do what the agent at t wants to do. The earlier self causally manipulates the later self. In the latter case, an agent at t creates a state that alters the stakes at t+1 such that it will be more reasonable (and thus more likely in the normal case) for the agent at t+1 to do what the agent at t wants to do. The earlier self manipulates the later self by changing the stakes. Notice that these general descriptions don't make reference to the impediments in the examples. Even if Isabella wouldn't be distracted at noon (and didn't believe so), taking the sleeping pill would be self-manipulation. Even if Jacques wasn't weak-willed (and didn't believe so), placing the bet would be self-manipulation. Forming an intention is not like this. It is not manipulation. It is a more direct expression of one's autonomous choice.

It is difficult to see how the dominant explanation of the descriptive phenomena can satisfy Ferrero's condition. According to it, when an agent at t forms an intention to do A, she is putting herself in a state with certain causal effects — namely, causing the agent at t+1 to reason how to do A or avoid conflicts with doing A or refrain from reconsidering doing A. On the face of it, this looks like self-manipulation: it looks a lot like what Isabella does. But what she does is self-manipulation, and forming an intention is not self-manipulation. We could fix this problem by specifying the causal role differently. Perhaps the causal role of the state of intending is to change the stakes such that it is more reasonable for the agent at t+1 to reason how to do A or avoid conflicts with doing A or refrain from reconsidering doing A. But this would make forming an intention like what Jacques does, which is also self-manipulation. Perhaps there is a way to specify the causal role of intention such that forming one would not be like taking the sleeping pill or placing the motivational bet, but it is not clear what that would be. And whatever it could be would not be straightforward.

The alternative explanation avoids this difficulty. When an agent at t forms an intention, she is not putting herself in a state that makes it more likely that she at t+1 will do what she wants (whether by immediate causal influence or by changing the stakes). Rather, it is more likely that the agent at t+1 (when she intends) will do what the agent at t (prior to intending) wants because reasoning how and avoiding conflicts and refraining from reconsidering are themselves actions that make it more likely that she will do what she wants. She is not being caused to or normatively influenced to perform these actions. She is performing the actions (if she is) as part of exercising her instrumental knowhow. When an agent intends to do A, she is more likely to do A not because intention is a state with a special sort of influence but because intending is a special sort of acting intentionally (exercising a certain sort of instrumental knowhow that enables future action). Thus, forming an intention involves no form of self-manipulation. Rather, it is simply initiating a special stage of acting. The intention is itself an intentional action and thus a direct expression of one's autonomous choice. Ferrero's condition is satisfied.

Together, the alternative explanation of the normative phenomena and the alternative explanation of the descriptive phenomena undermine some of the most significant reasons for initially believing in a state of intending. And believing so is not without its problems. Thus, we should consider the possibility that this belief is philosophical dogma. A state of intending is a myth.