Are routine actions rational?
In “Actions, Reasons, and Causes” (1963) Donald Davidson famously argued that actions are both ‘rationalized’ and caused by the agent’s reasons. Here is the tritest illustration of this “standard account” of actions: When Brett wants a beer and believes there is beer in the fridge, this gives him a reason to open the fridge and causes him to do so. The usual interpretation of such a case is that, when Brett opens the fridge to take a beer, he does so as the practical conclusion of a syllogistic inference, the two premises of which are his desire for a beer and his belief that there is beer in the fridge. Taken jointly these two premises constitute Brett’s reason for opening the fridge (this is a simplified description, of course, but here it will do).
Imagine now that looking for beer in the fridge is something Brett, who likes beer, has been doing several times a day over many years. Quite plausibly, he doesn’t have to go through the steps of this syllogism anymore; his behaviour has become routinized. In other terms, he has developed a little modular mechanism that guides his opening of the fridge when he want a beer. Brett does, of course, believe that there is beer in the fridge (which is almost always true in his house), but this belief doesn’t have to play a role as a premise in a syllogistic inference for the action to occur. In philosophical jargon, this belief may well remain ‘dispositional’ (as opposed to ‘occurrent’) when he performs the action. The routine is triggered by Brett’s occurrent desire for beer and causes him to open the fridge. This routine is well-adapted to the fact that the fridge is a reliable source of beer. That much shouldn’t be too controversial.
One evening, Brett takes out the last bottle of beer in the fridge. He realises – forms an occurrent belief, if you prefer – that there is no beer left in the fridge and decides to buy some the next day. The next morning, he fails to remember that there is no beer in the fridge. It is not that the belief he formed the night before has been erased from his mind; if Brett were asked whether there still is beer in the fridge, his belief would become occurrent and causally potent and Brett would say that, no, there is no beer left. It is just that, when Brett comes to want a beer that morning, the belief that there is no beer left in the fridge remains dispositional; it doesn’t play any causal role and in particular it fails to inhibit the routine, as it probably would have if it had been an occurrent belief.
Most of us sometimes look for something where it ordinarily is but where, as we knew and should have remembered, it currently is not. Such behaviour is, I would argue, evidence that the behaviour is caused by a routine that by-passes the relevant belief. In such cases, the Davidsonian account doesn’t strictly apply.
When your action is caused by a reason, then the rationality of the action is, at least in principle, easy to assess: if the reason is a good one, so is the action (again, this is very simplified, but it will do).
A routine action, on the other hand, is caused not by a full reason but by a mere component of a reason, a desire in our example. The other component, the belief that there is beer to be found in the fridge, doesn’t play any causal role in the routine action (and if anybody wants to object that the belief plays an “implicit” causal role in an “enthymematic” syllogistic inference, I will gladly answer their arguments).
So, how rational is a routine action, which is not caused by a mentally represented reason? This question can be asked in two ways: as a normative question about the facts of the matter (if there are such facts) or as a descriptive social-psychological question about people’s intuitive assessment of rationality (which might vary across cultures).
One possible way to answer the normative question would be to argue that routine actions, because they are not caused by a reason, are not to be assessed in terms of their rationality (just as, arguably, reflex actions such as a withdrawal movement when touching a hot surface shouldn’t be assessed in terms of rationality either). Another possible answer, which I favour, would consist in arguing that, as long as the routine emerged (in evolution or in development) because it is well-adapted to a goal of the agent (satisfying Brett’s desire for beer, for instance), then following this routine is rational, even if the routine action itself is not caused by a reason and even if the routine itself has not been adopted as a result of a reason-based decision. If so, Brett can be said to act rationally when he routinely opens the fridge, even on the rare occasion where, as he should have remembered, there is no beer in the fridge. The difference between these two answers to the normative question may hinge on the way you define rationality: a terminological issue.
Regarding the descriptive social-psychological question, one would need empirical evidence regarding the way people assess routine actions. Do they view them only as successful or unsuccessful (depending, for instance, on whether there is beer in the fridge); or do they also view them as rational or irrational (or reasonable or unreasonable, or justified or unjustified, and so on) depending on whether the routine is or is not well-adapted to cause successful action with sufficient frequency (what counts as sufficient frequency being a cost-benefit issue)? In other terms, how strictly Davidsonian are people in their intuitive judgments of the rationality of actions?
Pierre Jacob 24 October 2018 (12:02)
Are reflex actions rational?
The target of Dan’s intriguing blog is Davidson’s highly influential philosophical picture of rational action according to which an agent’s action (e.g. Brett’s opening the fridge) is rational if and only it is caused by the agent’s reason, where the agent’s reason consists of his motivation (e.g. his desire for beer) paired with his epistemic state (e.g. his belief that there is beer in the fridge).
Dan argues that routinized actions constitute a counterexample to the Davidsonian picture because they too are rational, but they don’t meet Davidson’s conditions for rationality.
In Dan’s example, Brett’s occurrent desire for beer is a cause of his opening the fridge, but he has no occurrent belief that there is beer in the fridge, since when he last opened the fridge he saw that there was no beer left. This is an instance of a routinized action: Brett routinely opens the fridge whenever he has a desire for beer. Brett’s routine action counts as rational to the extent that his routine is well-adapted to his local environment in which his fridge is a reliable source of beer. Brett’s occurrent desire is only a constituent of a Davidsonian, i.e. subjective, reason.
If Brett’s desire for beer and his belief that there is beer in the fridge were both causes of his opening the fridge, then Brett would have a subjective reason for his action. And his subjective reason would be an accurate representation of his objective reason (i.e. the fact that his fridge is a reliable source of beer). If Brett had an occurrent false belief that there is beer in the fridge (and an occurrent desire for beer), then his subjective reason for opening the fridge would misrepresent his objective reason. But then his opening the fridge would not be a routinized action. What is crucial to Dan’s argument is that if Brett’s action is routinized, then it is likely that his fridge is a reliable (or regular) source of the presence of beer. And Brett need not have an occurrent mental representation of the regular presence of beer in his fridge for his action to be rational.
Brett’s routine action fails the Davidsonian criterion because an occurrent desire is only half a subjective reason. The fact that the fridge is a reliable source of beer provides an objective reason for Brett to perform his routinized action. His objective reason need not be mentally represented at all for his action to count as rational.
Dan’s discussion explicitly rests on the following tripartite taxonomy of actions. First, there are rational actions that fit Davidson’s criterion of rationality: they are caused by a pair of occurrent mental states (an agent’s desire and an agent’s belief). Secondly, there are routine actions that don’t fit the Davidsonian picture: they are caused only by the agent’s occurrent desire, not by any of his occurrent beliefs. Finally, there are reflex actions (e.g. Brett’s scratching his skin because it itches or withdrawing his hand from a hot surface), which arguably need not have any of the agent’s occurrent desires or occurrent beliefs among their causes.
There is an interesting relevant difference between reflexes and routine actions: while routine actions emerge from an individual’s ontogenetic development, reflexes have a phylogenetic origin. But I’m not quite sure what to make of this difference.
I expect that Davidson would argue that reflexes are paradigm cases of actions which should not qualify as rational and also that routine actions are more like reflexes than genuine rational actions. Dan argues both that the regular presence of beer in Brett’s fridge is an objective reason for Brett’s routine action and that it is sufficient to make Brett’s routine action rational.
My question to Dan is: are you willing to call an agent’s reflex action rational on the ground that it is well-adapted to some regularity that need not be mentally represented at all by the agent?