Willpower or internal self-control is the psychological function that resists present temptations (by contrast with forestalling future temptations, which is not considered a variety of willpower). Temptations arise when the agent is confronted with a choice between a smaller reward that is closer in time, and a larger reward that is temporally more distant. The agent will act according to the option which is the most rewarding. Temptations arise because the closer the sooner-smaller alternative gets, the more rewarding it appears.

There are, Ainslie maintains, two methods for resisting temptations: one can confront the temptation head on by inhibiting other options or keeping one's attention away from them (suppression); alternatively, one can reinforce the intention by interpreting the current choice as a more general test case for some desired pattern of behavior (resolve). Ainslie argues furthermore that suppression is necessarily effortful, whereas resolve may be effortless. He identifies effort with the operational costs of implementing a willpower tactic. According to him, resolve involves operational costs only when temptations generate risks associated with giving in. By contrast, resolve involves no operational cost – and is, therefore, effortless – when agents are confident that they will not give in to temptations, as in the case of flight attendants who feel no urge to smoke while in flight.

Although we agree with the suppression/resolve distinction we maintain that there is no such thing as effortless resolve. The view that exertion of willpower is always effortful is largely accepted in the literature (see, however, Fujita, Reference Fujita2011). Why does Ainslie think that in the case of resolve, that orthodoxy has to be rejected?

Ainslie's reason is this: there are documented cases in which agents are entirely confident that they will not give in to temptation, so that no effort is needed to resist the temptation. Reporting Dar, Stronguin, Marouani, Krupsky, and Frenk studies (Reference Dar, Stronguin, Marouani, Krupsky and Frenk2005), Ainslie writes: “Orthodox Jews who never smoke on the Sabbath and flight attendants who never smoke during flights have no urge to smoke during those times, while still having strong urges at other times.” These, he maintains, are cases of effortless resolve.

We do agree that such cases are effortless, but we reject that they are cases of resolve and more generally cases of willpower. Willpower is defined as the resistance to present temptation; but such cases are explicitly described as cases where there is no urge. In the absence of temptation or risk to give in, there is no need to employ any willpower tactics. It would, in fact, be irrational to enter into complex intertemporal bargaining if not under the pressure of temptation.

Perhaps, “no urge” is an overstatement here; perhaps, the idea is that urges in such cases are very low. But the point remains: either the urge is too low to justify the implementation of a self-control method, in which case there is no effort, but no resolve either. Or the urge constitutes a “marginally permissible temptation,” in which case we need to employ resolve, which, as a willpower tactic is going to be (marginally) effortful.

Thus, on closer scrutiny, the cases put forward by Ainslie provide no reason to admit effortful resolve. Moreover, there is a strong reason to admit that resolve is effortful:

• P1 Resolve entails a process of cognitive abstraction.

• P2 Cognitive abstraction processes always have operational costs.

• P3 Having operational cost entails being effortful.

• C Resolve is effortful.

P1 follows from Ainslie's definition of resolve as interpreting a particular decision context as an instance of more general pattern or rule. That requires cognitive abstraction. P3 follows from Ainslie's definition of effort. P2 seems highly plausible – it is hard to see how cognitive abstraction could be implemented at no cognitive cost at all. Hence, the view that resolve may be effortless is incompatible with the view that cognitive abstraction is costly, a view that appears to stand on firm ground.

If correct, alleged cases of effortless resolve are indeed cases in which resolve is absent. More generally, they are not cases of willpower. But if so, what are they?

After having discussed suppression and resolve, Ainslie moves on to habits, which according to him are not methods of self-control but results thereof. We suggest that alleged cases of effortless resolve are in fact cases of habits.

More specifically, we propose that such cases are either cases of good habits or cases of routine habits. Good habits are more fragile than routines. Having a reason not to go running today comes with a rush of pleasure and endangers the habit that one has effortfully managed to create. Therefore, good habits are tightly dependent on resolve: When temptation kicks in, resolve has to come to the rescue to preserve them. Ainslie concludes that good habits are therefore “forms of willpower.” But this does not follow – to depend on something does not entail being a species of it. If true, alleged case of effortless resolve might be cases of good habit without being cases of resolve or willpower more generally.

The alleged cases of effortless resolve mentioned by Ainslie are perhaps even better seen as cases of routine habits, which, Ainslie rightly insists, are not forms of willpower. In routine habits, the temptation that was regularly resisted in the good habit ends up disappearing. Routine habits no longer need the protection of resolve for they are no longer challenged. Tellingly, Ainslie also mentions the case of a person who never thinks of smoking during a flight as an example of routine habit. However, because resolve is a mechanism and routine habit is a result, the same example cannot be at once an instance of resolve and an instance of routine habit.

Instead of subsuming such cases under the dubious category of effortless resolve, Ainslie's position would be stronger if they were subsumed under the category of – good or routine – habits.