On the Nonexistence of Practical Withholding

In this post, I wish to highlight what I take to be an important but under-appreciated disanalogy between rationally permissible belief and rationally permissible action. I will say more about why I think this disanalogy is under-appreciated in later posts. For now, I will simply attempt to identify the disanalogy, and say something about why it exists.

Consider the following example of an agent engaged in a piece of practical deliberation:

Example 1: Lorry Driver Intention

A lorry driver, Jesse, is trying to decide between continuing to drive down a long stretch of road or stopping for a break.  After weighing all the considerations in favour of continuing to drive (e.g., she is more likely to make her delivery on time) and all the considerations in favour of taking a break (e.g., she would be able to get some much deserved shut-eye), she comes to the conclusion that the total evidence available is inconclusive either way.  She has just as much reason to continue driving as she does to stop and take a break. Jesse decides to flip a coin: ‘head-side-up’ for continuing to drive, and ‘tail-side-up’ for stopping for a break.  When the coin lands ‘tail-side-up’, Jesse pulls the lorry off to the side of the road and removes the key from the ignition. 

I take the following three claims to be uncontroversially true about the agent is Example 1:

(1)  The fact that Jesse decided to flip a coin and the coin landed tails-side-up explains why she decided to stop driving. 

(2)  The fact that Jesse decided to flip a coin and the coin landed tail-side-up does not justify her decision to stop driving. 

(3)  Jesse is not guilty of any irrationality for basing her decision to stop driving on the fact that she decided to flip a coin and the coin landed tail-side-up. 

Now, consider the following modified version of the Lorry Driver Intention example:

Example 2: Lorry Driver Belief

Jesse, the lorry driver, is trying to determine if the following the proposition is true: (H): “My lorry will run out of petrol before the next station”.  However, suppose that all the evidence available is inconclusive.  Jesse has just as much evidence in favour of (H) as she does against (H).  She therefore decides to flip a coin.  Assigning the value of ‘true’ to head-side-up, and ‘false’ to tail-side-up, and proceeds to flip a coin.  When the coin lands head-side-up, Jesse comes to believe (H).  Jesse radios her dispatching and reports that she believes her lorry will run out of petrol before the next station.  When she is asked why she believes this, she explains that since all the evidence she has available was inconclusive, the decided to flip a coin.  Incredulous, the dispatcher points out that a coin flip fails to constitute evidence one way or the other.  Jesse agrees, but insists that the coin flip remains the basis of her belief.

I take the following three claims to be uncontroversially true about the agent in Example 2:

(1)  The fact that Jesse decided to flip a coin and the coin landed heads-side-up explains why she believes the lorry will run out of petrol. 

(2)  The fact that Jesse decided to flip a coin and the coin landed heads-side-up does not justify her belief that the lorry will run out of petrol. 

(3)  Jesse is guilty of irrationality for basing her belief that she will run out of petrol on the fact that she decided to flip a coin and the coin landed heads-side-up. 

In contradistinction to the first example, the agent in the second example appears to display gross irrationality.  This suggests an important disanalogy between rationally permissible belief and rationally permissible action. While an agent’s actions may be rationally permissible even though the explanation of her action fails to justify her action, the same cannot be said of an agent’s beliefs.  

I will now attempt to offer an explanation of why the above disanalogy between rationally permissible belief and action exists:  

Theoretical deliberation always involves a choice between three doxastic attitudes: believing, disbelieving, or withholding belief and disbelief.  This means failing to believe that P does not entail disbelieving P and failing to disbelieve P does not entail believing P.  Withholding P always remains an option. Withholding P, as I am using the expression, is not the same as failing to adopt an attitude of belief and disbelief towards P.  One may fail to adopt an attitude of belief and disbelief towards P because one has simply not considered P.  In such a case, one neither believes nor disbelieves P.  But one is not withholding P either.   One simply has not taken any attitude towards P.  In short, withholding P is as much an attitude towards P as believing or disbelieving P.  

I believe that the disanalogy between reasons for belief and reasons for actions, highlighted in the previous section, is tied to the absence of a practical analogue to withholding. Consider once again the lorry driver who has to decide between continuing to drive and stopping for a break.  The lorry driver’s decision has a zero-sum structure. If she adopts the intention to continue driving, she has ipso facto adopted the intention not to take a break.  The same is true, mutatis mutandis, if she adopts the intention to take a break. There is no attitude of practical withholding that she may adopt as an end point of her deliberative process. In sum, while there are three possible doxastic attitudes one can take towards P—believing P, disbelieving P, and withholding P—there are only two possible volitional attitudes one can take towards an outcome—intending to bring about P and intending not to bring about P.  On the present analysis, not intending to bring about P is not an attitude towards bringing about P but rather the absence of an attitude towards bringing about P.

In cases in which the weighing of evidence is relevant, one must have net evidence in favour of believing or disbelieving in order for either attitude to be rationally permissible.  Believing P is rationally permissible only if one’s net evidence is in favour of P.  Disbelieving P is rationally permissible only if one’s net evidence is against P.   If one lacks any evidence for or against P or if one has equal amounts of evidence both for and against P, then one is rationally obligated to withhold P. Hence, since the agent in the second example has equal amounts of evidence both for and against the truth of the claim that her lorry will run out of petrol, she is rationally obligated to withhold belief.  By contrast, she could have no analogous obligation when it comes to the question of whether she should continue or stop driving since withholding from both is simply not an option.  Hence, even if she has equal amounts of evidence in favour of continuing to drive and stopping for a break, she still has to do one or the other.  Simply put, practical withholding is not an option.  The upshot is that, unlike belief, there cannot be a requirement that an agent's net evidence favour a course of action in order for that course of action to be rationally permissible.