Bernie, The Petrol, and The Negative Reason Existential Fallacy
Consider the unfortunate case of Bernie (this case is originally given in Williams’s “Internal and External Reasons”). Bernie is at a cocktail party. He is thirsty and his favorite drink is gin and tonic. He thus orders a gin and tonic. The bartender gives him a glass with a clear liquid in it. Bernie forms the belief that his glass contains gin and tonic. Bernie forms an intention to drink from the glass. Unfortunately for Bernie, the glass doesn’t contain gin and tonic, it contains petrol. There are at least five interesting questions to ask about Bernie: (1) Does he have a reason to believe that his glass contains gin and tonic? (2) Does he have a reason to intend to drink from the glass? (3) Is he rational in believing that the glass contains gin and tonic? (4) Is he rational in intending to drink from the glass? (5) What’s the relationship between (1) and (3) and (2) and (4)?
Almost everyone has the intuition that Yes is the answer to (3) and (4). We will assume, then, that Bernie’s belief and intention are rational. The interesting questions, then, are (1), (2), and (5).
The common answer to (2) amongst metaethicists is No–Bernie does not have a reason to drink from the glass. Let’s call this The Answer. The Answer, I think, is spurious for many reasons.
First, let’s talk about what the best answer to (1) is. I take it that people will have a very strong intuition that Yes must be the answer to (1). This just follows from our intuition that Yes is the answer to (3). That is, if Bernie has a rational belief that the glass contains gin and tonic–i.e. if Bernie has a justified belief that the glass contains gin and tonic–then there must be sufficient reasons to believe that the glass contains gin and tonic. For example, that there is a clear liquid in his glass, that he ordered his drink from a reliable bartender, that there are others at the party drinking potable alcoholic beverages etc. It is a conceptual truth about epistemic justification that one is justified in believing p iff there is sufficient reason to believe that p. You aren’t justified by arbitrarily believing that p–there must be evidence that p (and good evidence). This seems right. Moreover, it seems like Bernie is justified in believing that his glass contains gin and tonic.
But if that is true, then why say Bernie has no practical reason to drink from the glass? It seems like, given his background desire for gin and tonic, the types of considerations that will be reasons to drink from his glass are the types of considerations that make it likely that the glass contains gin and tonic. In other words, they are the types of considerations that count in favor of believing that his glass contains gin and tonic. We already decided that there have to be these types of considerations in Bernie’s situation if we are going to conclude that Bernie is rational in believing that his glass contains gin and tonic. Thus, if the practical reasons for drinking from the glass are just the considerations that make it more likely than not that his glass contains gin and tonic, Bernie does have some reasons to drink from the glass. Some examples are the fact that his glass contains a clear liquid, the fact that he got his drink from a reliable bartender, and the fact that others are drinking potable alcoholic beverages.
Another problem with The Answer is that it appears inconsistent with what some metaethicists want to say about the connection between reasons and rationality. The prime example is Parfit. Parfit claims that rationality has something to do with correctly responding to reasons. He gives The Answer to (2). He also argues that Bernie (or someone in an analogous situation to Bernie) is rational. But it is hard to see how all of those three things can be true together. If Bernie doesn’t have a reason to drink and being rational has something to do with responding to reasons, then how can Bernie be rational by intending to drink? In the end, Parfit denies that rationality has something to do with correctly responding to reasons. Despite his denial of this, he still likes to say it a lot. That’s because it seems true.
So, it seems as if there are some problems with answering No to (2). Is there any motivation for such an answer? I don’t see any beyond the intuition that it doesn’t seem like Bernie should have a reason to drink. This intuition is driven, I think, by a variant what Mark Schroeder calls The Negative Reason Existential Fallacy. I think that people have the intuition that Bernie has no reason because there is such a huge, obvious reason not to drink, viz. that the glass contains petrol. That is, people think about the case and notice that there is such a huge reason not to drink from the glass. They conclude that since this reason is so overriding, it must be the case that there is no reason to drink from the glass. But just because there is such an overriding reason not to drink, why conclude that there is no reason whatsoever to drink? Especially if holding (the quite intuitive position, I think) that there are some reasons to drink can help us explain (1) why Bernie is rational and (2) why rationality has something to do with correctly responding to reasons.
I’ll have more to say about reasons and rationality, but I thought I’d briefly address this:
I don’t see any beyond the intuition that it doesn’t seem like Bernie should have a reason to drink. This intuition is driven, I think, by a variant what Mark Schroeder calls The Negative Reason Existential Fallacy. I think that people have the intuition that Bernie has no reason because there is such a huge, obvious reason not to drink, viz. that the glass contains petrol. That is, people think about the case and notice that there is such a huge reason not to drink from the glass. They conclude that since this reason is so overriding, it must be the case that there is no reason to drink from the glass. But just because there is such an overriding reason not to drink, why conclude that there is no reason whatsoever to drink? Especially if holding (the quite intuitive position, I think) that there are some reasons to drink can help us explain (1) why Bernie is rational and (2) why rationality has something to do with correctly responding to reasons.
This diagnosis might explain some people’s attitudes, but not mine. Agreed, we sometimes say there’s no reason to X when and because there’s really, really good reason to refrain from X-ing. The goodness of floor mats is no reason to spend $1,000,000,000 on a car. But, it’s a reason. I guess when we say ‘No reason’, it’s sort of idiomatic. However, since it is idiomatic, I think it’s implausible to think that this is how philosophers come to the view. They’ve just mixed up their idiomatic ways of speaking with their settled judgments? Unlikely.
Here’s why I personally go in for the view. I’m well aware of how reasons defeat one another, weigh on each other, etc… I’d never be tempted to say that it logically follows from the fact that there’s exceptionally strong reasons to X that there’s no reason not to. I’m tempted to say there’s no reason to drink the contents of this glass because I think we use the concept of guiding reason to do work. Suppose I’m a pretty hard core Humean and I really want to smoke a cigarette that will kill me. I also really want to live. As I try to figure out which desire is stronger, weightier, gives better reason, etc…, I see that there are two facts, each fact favors a different course of action, and what I should do is dependent upon which fact gives the better reason.
There’s not two things occupying the same ontological category playing opposite roles in the gin/petrol example. So, either I gotta say there’s nothing that plays the guiding reasons role in the gin/petrol case (good view) or what plays the guiding reasons role in the cigarette case are mental states. I think we use the concept of guiding reason in such a way that we cannot have two guiding reasons, R and R’, such that R favors one course of action, R’ favors an incompatible course of action, but R and R’ occupy different ontological categories because R is a psychological state and R’ is a proposition, fact, or state of affairs.
Since I see no good reason to think of guiding reasons as psychological states (wait for post about reasons and rationality), I’m led to the view I have about reasons by something that seems different from the negative existential reasons fallacy.
Clayton,
I’m not entirely sure what you mean by guiding reason. But your main point seems to be this: There is no reason for Bernie to drink because there are no true propositions that count in favor of him drinking. It’s not as if there are two (or more) true propositions that are competing like in the case of the smoker. In Bernie’s case, there are only psychological states, and psychological states aren’t (perhaps in-themselves) reasons.
I agree that psychological states aren’t reasons. And I agree that if there weren’t true propositions in Bernie’s case that counted in favor of drinking, then he wouldn’t have a reason. But the main part of my argument is that there are true propositions that count in favor of him drinking. For example, that there is a clear liquid in his glass, that he got the glass from a reliable bartender, that others are drinking potable alcoholic beverages etc. Given his background desire to drink a gin and tonic, those are all reasons to drink. Of course, those are all outweighed reasons since the fact that the glass contains petrol is a much weightier reason not to drink than the reasons he has to drink. Unfortunately for him, he is not in a position to know that it contains petrol.
So, perhaps you still have committed the negative reasons existential fallacy. That is, perhaps your judgment that the fact that the glass contains petrol is such a weighty reason not to drink has led you to conclude that those other true propositions aren’t reasons to drink. Of course, I’m not inside your head and I don’t know if that is the case.
Sometimes I use ‘guiding reason’. Sometimes I use ‘normative reason’.
The truths you cite don’t strike me as reasons for Bernie to drink the glass. That the bartender served him the stuff seems like evidence of a reason rather than a reason. That the glass contains clear stuff is either also evidence of a reason or a reason only insofar as he had additional desires to drink something clear. I don’t think I’ve committed the NREF b/c I think they aren’t reasons because they aren’t the kinds of considerations that favor drinking, that would make it right to drink in the absence of weightier reasons not to, etc… If Bernie drank the stuff, you couldn’t say that there was a silver lining–the bartender told him the stuff was gin, the liquid was clear, etc… I take it that if one acts on what one learns later to be an overridden reason, the defeated reason is nevertheless a pro, a silver lining, etc…
About reasons and rationality, let me say this. I’m somewhat attracted to a view Raz seems somewhat attracted to on which the judgment that some action or attitude is rational commits you to nothing stronger than claims about the rational agent’s use of the faculty of reason in response to what the agent _takes_ her reasons to be. This gives us _a_ connection between reasons and rationality without committing ourselves to the view that for every rational attitude or action there was a genuine reason to form that attitude or perform that action.
One of the reasons I like this view is that I think it has the resources for dealing with three kinds of case whereas a view that forces us to say there’s a genuine reason to A whenever an agent was rational in A-ing does not seem to have the resources. Think about three shootings: Sam shoots Bob in self-defense, Sam shoots Bob on the mistaken belief that this is necessary for self-defense, and Sam shoots Bob on the mistaken belief that this is what his dog asked him to do.
Depending on how you fill out the details of the case, I think we have three cases in which it would be improper to blame Sam. In the first, I wouldn’t blame him because he acted rightly. In the second, he acted wrongly but ought to be excused. In the third, Sam’s not the kind of character that possesses the rational capacities he must to be properly blamed.
In cases of ‘imperfect’ self-defense (i.e., cases in which someone acts on the mistaken belief that someone is an aggressor and takes violent means against this person), I think that we ought to say someone is blameless only if we can say that they are rational for having done what they did. (If they were irrational but had the capacity for rational decision making, blame would be appropriate). I think we need go no further than saying that the subject responded in the way we’d judge he ought if his perception of the reasons were correct to warrant the claim that Sam is both excused and rational. We do not need to find any good reasons for him to have pulled the trigger.
Now, here’s the important part, it seems that a necessary condition on offering a proper excuse is that we can offer a rational explanation of the wrongful action. If you can perform an action A that is excusable but wrongful when there was no reason that favored A-ing that figured in the agent’s reasoning, it seems that since the subject’s A-ing is excusable only if rational, we have a case of A-ing rationally without there being reasons to A. To maintain the difference between excuses (as typified by the case of imperfect self-defense) and exemptions (what we offer in the case of the insanity defense), I think it’s helpful to sever the connection between rationality and reasons. I think there’s some good stuff by Michael Moore and John Gardner on this. You might think it’s wrong, but it’s certainly worth a read.
You identify two tests for whether something is a reason. First, The Undefeated Test:
the undefeated test: Consideration C is a reason to A only if one ought to A if there are no considerations against A-ing.
I grant that this is common test to see whether something is a practical reason. Clearly it isn’t a test to see whether something is an epistemic reason. Moreover, I think that there are decent counterexamples. Consider supererogatory reasons. Let’s say that you have a supererogatory reason to do A. Let’s say there are no reasons against A-ing. Nevertheless, by the definition of supererogatory reasons, you aren’t required to A.
The second test is The Silver Lining Test:
the silver lining test: If C is a reason to A, then if I perform A when C is overridden, there will be a silver lining to A-ing.
I’m not certain this is true either. I guess it depends on what you mean by silver lining. A practical silver lining? If that, I think there are counter-examples. Consider the reason to pay attention to your evidence. Imagine that I am in a position to know that my wife is cheating. I have argued that you have a practical reason to pay attention to your evidence (as you know). If the silver lining test is true, then if my reason to pay attention to my evidence is overridden, then there is a practical silver lining to paying attention to my evidence. But that seems false if paying attention is going to cause me great distress. So I guess epistemic silver linings count, and my knowledge of my wife’s cheating is a silver lining. But if that is the way to read it, then why not think that Bernie knowing that the glass contains petrol is a silver lining?
Now, let me offer my own test–The Changing Conditions Test.
the changing conditions test: If C is a reason to A for P and P is in a position to know C, then changing C will change the agent’s rational status if the agent does what she did when C was the case.
So, make the liquid in Bernie’s glass brown. Let’s say that he still believes that it’s gin and tonic and still intends to drink. Is he rational? I doubt it. Let’s say he knows that the bartender wants to kill him. But he still believes that the glass contains gin and tonic and intends to drink. Is he rational? Again, I doubt it. Let’s say that other people at the party are drinking and then immediately throwing up. But Bernie still believes that the glass contains gin and tonic and intends to drink. Still rational? Again, I doubt it.
On to the reasons/rationality. Lots of people like your view (e.g. Broome, Parfit, Schroeder). My view of rationality is conjunctive, meaning I think that both structural and substantive rationality is important. I also think that not being blameworthy (or criticizable) is a mark of being rational. So, in your case of the imperfect self-defense, it might be that Sam is structurally rational but substantively irrational. However, without the details being worked out, I can’t say. I think that there might really be reasons to shoot. Certainly overridden reasons, but reasons nonetheless. But I don’t know without the details. It could be a counterexample to my view.
Finally, why do we need the third case? How is that a problem for any view of rationality? Since Sam is ex hypothesi not a rational agent, why would him shooting Bill be a problem for any theory of rationality?
[...] from Blame: Excused v. Justified In the comments thread on my last post, Clayton mentioned the distinction between being immune from blame because one was justified and [...]