Plains Reasoning

I. The Puzzle

Suppose Jack wants a gin and tonic. Jack believes that what is in the bottle labeled ‘gin’ is gin, but what is in the bottle labeled ‘gin’ is petrol: odorless and colorless petrol. Since Jack is justified in believing that what’s in the bottle is gin and he really wants a gin and tonic, he pours himself a glass. As reason ascribers, what should we say that Jack has reason to do? Here are our options:
(1) Jack has reason to drink what’s in the glass.
[What is in the glass is petrol.]
(2) Jack has reason to drink the petrol.
The most popular option is to say that (1) and (2) are false . The reason is that Jack acts on a reason based on a justified, but false, belief. Moreover, it is claimed that justified false beliefs are detrimental to an agent’s rationality; hence Jack has no reason to drink what’s in the glass because what is in the glass is petrol: something he wants not to drink in the first place. Call this the third personal reasons ascription account, or TPRA.
A second option is to say that (1) is true, while (2) is false by way of opacity. Given that Jack does not know that what is in the glass is petrol, he would not assent to (2) as a reason in favor of drinking the liquid in the glass. Jack would assent to (1) because his belief that what is in the glass is gin, and that belief is justified, but false. Call this the opacity of reasons ascription, or ORA.
A third option, the very option I advocate is that we should say that (1) and (2) are true. The reason is that, from my perspective, Jack has reason to drink what is in the glass and he has reason to drink the petrol, but given that Jack does not know that the glass is full of petrol, Jack has reason to drink what’s in the glass, but has reason not to drink the petrol. Call this view the context dependence of reasons ascription, or CDRA.
I support this view because TPRA maintains that justified false beliefs are detrimental to an agent’s rationality, even though this is false. We cannot blame those who are justified in believing something false when they cannot have access to the feature of the case that causes their belief to be false. Moreover, ORA conflates the third-personal perspective and the first-person perspective to derive the conclusion that reasons ascriptions are opaque. In what follows, I elaborate on these objects to the two rivals of CDRA. I then offer a defense of CDRA.
II. Third-personal reasons ascription
Advocates of TPRA give the following argument in favor of the analysis of reasons ascriptions.
1. A reason in favor of acting must be derived from a sound deliberative route.
2. If a reason is derived from a sound deliberative route, then it is rational to act on that reason.
3. Justified false beliefs are not part of a sound deliberative route to a reason for action.
4. ∴ All reasons derived from a justified false belief are irrational and cease to be reasons for action. [Williams 1981: 102-103]
If we apply this argument to the case presented in Section 1, we notice that neither (1) nor (2) is a reason for action. Both ascriptions depend on Jack’s justified false belief that the glass is full of gin. Since Jack’s belief was false, it would be irrational for Jack to drink what is in the glass, provided that Jack does not have a reason for action. Had Jack known that the bottle was filled with petrol, he would have no reason to drink what is in the glass. Hence, both (1) and (2) are false.
I object to premise three, and thus deny the conclusion of the argument in favor of TPRA. My argument is the following. If all one can know are the features of the case accessible to the agent, then it seems like we occasionally act on justified false beliefs. It seems all too difficult to call such an agent irrational. It certainly seems rational, given what is accessible to Jack, to conclude that what is in the bottle is in fact gin because of the evidence he can have access to at the time. Moreover, it seems unlikely that any further deliberation that Jack could partake in could not further the evidence he has in front of his face. We should not accuse people of irrationality because they deliberated correctly from all that they could know. Had there been some other means that Jack could take to deliberate to the fact that the bottle is full of petrol, yet failed to recognize this feature, then it seems like we could rationally blame Jack by saying that he has no reason to drink what is in the glass.
Second, if Jack collected all of the evidence available to him and concluded still that what is in the bottle is gin, it appears that his belief that the bottle is filled with gin can function satisfactorily in a deliberative route to a reason for action. We cannot blame an agent for failing an omniscience test in concluding what action they should take. Jack’s deliberation is sound, given what Jack can access with respect to the context he’s in.
I have shown in this section that the possibility of justified false beliefs playing a rational role in an agent’s deliberation is intuitive. This argument opens the door for both ORA and CDRA. I will explore the motivation of ORA, and then reject it because it relies on a conflation of context.
III. The Opacity of Reasons Ascriptions
Advocates of ORA maintain that Jack has reason to drink what is in the glass, yet fails to have reason to drink the petrol. Reason ascriptions are opaque; meaning that while one ascription may be assented to by the agent we ascribe the reason to. The agent will deny the other ascription, given the accessibility of the features of the case .
This claim is analogous to the view that belief ascriptions are opaque. Take the following example.
(3) Lois believes that Superman is strong.
[Superman is Clark Kent]
(4) Lois believe that Clark Kent is not strong.
Passing knowledge in comic books tells us that Clark Kent is Superman, yet Lois does not know that Clark Kent and Superman are the same person. The evidence she has in favor of her beliefs is that whenever she sees Superman, she sees him in blue tights and a cape, while Clark Kent is wearing glasses and a collared shirt. So, if one were to ask Lois (3), she would assent, while if one were to ask Lois if she believed that Clark Kent was strong, she would laugh in your face. Hence, one can conclude that Lois believes that Superman is strong and that Clark Kent is not strong, even though they are the same person.
If were to apply this to a case of reason ascriptions like Jack’s, ORA would tell us that (1) is true, while (2) is false. What Jack can know in his position is that he has reason to drink what is in the glass because he believes it is gin. In his position, he cannot know that he has reason to drink the petrol. If we were to ask Jack for his reason for acting, Jack would assert (1) and deny (2). This gives one reason to believe that reason ascriptions are opaque.
I maintain that reasons ascriptions are not opaque. ORA theorists support opacity through conflating different contexts of reason ascriptions. I elaborate on this objection below and provide a further analysis of reason ascriptions that claim that (1) and (2) are true of Jack.
IV. The Context Dependence of Reason Ascriptions
CDRA maintains that (1) is true and (2) is true, while maintaining that (1) is true for Jack and that (2) is false for Jack.
One very good question is how CDRA theorists can do this? That Jack has reason to drink what from the glass is ascribed to Jack by concluding that from his perspective, that is given what Jack can access about the world (given his perspective), he has reason to drink what is in the glass. In my mouth, or from the third-personal perspective, Jack has reason to drink what is in the glass and has reason to drink the petrol, because what I have access to. In other words, the scope of what I have access to about Jack’s position is far wider than Jack’s own scope, which is hindered by the features he is in a position to know. Given what I have access to, I can ascribe both that he has reason to drink what’s in the glass, and Jack has reason to drink the petrol.
Furthermore, I can assert that Jack has reason to drink what is in the glass and that he has no reason to drink the petrol. One might object and claim that by asserting both that Jack has reason to drink the petrol and Jack has no reason to drink the petrol is contradictory, but this is not what I am asserting. I am simply stating that from my perspective, Jack has reason to drink what is in the glass and Jack has reason to drink the petrol, given what I have access to in my evaluation of Jack’s position. If I were to think about the case from Jack’s perspective, he would certainly ascribe to himself that he has reason to drink what is in the glass, but has no reason to drink the petrol because the fact that what is in the glass is petrol is inaccessible to him. In order to derive a contradiction, I must be think to merge the contexts together, but notice that they are clearly different with respect to the agents.
The problem with ORA is that it conflates my context of reason ascriptions with Jack’s context of reason ascriptions to conclude that reason ascriptions are opaque. Notice that in one sense reason ascriptions are clearly transparent, given my context, but merely appear to be opaque in Jack’s context. Reason ascriptions are not opaque. One needs to take care in discussing from which position are we ascribing the reason.

I’ve been thinking quite a bit lately about Niko Kolodny’s Transparency Account and whether it can accommodate for all forms of rational advice.

 

All rational requirements take a certain form. Kolodny calls this form the ‘core requirements’. If there is any seemingly normative force of rational requirements, then they must all manifest from the same core. The ‘core requirements’ take the following form.
C+: If one believes one has conclusive reason to have A, then one is rationally required to have A.

 

C-: If one believes that one lacks sufficient reason to have A, then one is rationally required not to have A. (Kolodny 2005: 557)

 

Let A be any attitude under consideration by an agent. This is general enough so that all rational requirements stem from these two requirements. If this is the case, then we have a possible explanation of the seeming normative force of rationality. In order to capture the normative force, let’s look at the consequences of the ‘core requirements’ from the outside first (or what it’s like to receive advice) and the inside second (or what it’s like when one receives advice from the inside) .

Suppose that we tell someone in the register of advice that he ought to have attitude A. It would be irrational of him not to have A. When we tell them that they ought to have attitude A, we are pointing out that he satisfies the antecedent of C+. He believes he has conclusive reason for A (Kolodny 2005: 557). In other words, the advisor explains to the advisee that from the advisee’s perspective, he has conclusive reason to have the attitude. We are not offering the advisee a further reason to have A, we are pointing to a reason he already has. That reason is his fulfillment of the antecedent of C+. The advisor merely plays a reminding role. Advisors merely remind the advisee that they have conclusive reason for that attitude and since they have conclusive reason for A, they ought to adopt A.

According to an initial formulation of the ‘core requirements’, the role of the advisor is unclear. It is difficult to distinguish whether the advisor points out that the advisee has reason for the attitude because he already has conclusive reason for that attitude, or the advisor points to an attitude in support of adopting a different attitude. If the former is the case, the role of the advisor would be trivial and there is virtually no need for advising. To illustrate this point, consider the following case.

Logic: Tina is in the middle of a quiz over Modus Ponens: the logical rule that allows one to derive the consequent of a conditional if the antecedent of that conditional is assumed true. Tina knows the rule and notices that she has a conditional. She also knows that one of the premises is the assumption of the antecedent, but has failed to see what conclusion should be drawn.

 

In Logic, Tina can reason from her belief that this proof applies to the rule she knows and derive what she ought to do merely on her own, without an advisor. If Kolodny intends the advisor to function in this way, then we may never need advisors. Moreover, while there are cases of advice where advisors do play a trivial role such as reminding one that they already possess the attitude, there are further cases of rational advice that are not covered by this interpretation of the ‘core requirements’.

One other consequence of this interpretation is that an advisor has some kind of mental telepathy. Suppose you have a case where a person wants to eat at a new restaurant, yet has no idea what kind of food the restaurant serves, but knows that people genuinely enjoy the cuisine.

Pad Thai: Sid is going to Molly Woo’s, a new restaurant in town. Sid’s friend, Nancy, has attended the restaurant. Sid asks Nancy what he should order at the restaurant. Nancy tells him that he ought to want to order the Pad Thai.

 

In Pad Thai, Nancy reminds Sid he wants Pad Thai. However, according to the ‘core requirements’, even if Sid had no knowledge of the kind of restaurant, yet knows that most people genuinely enjoy the cuisine, Sid wanted to order the Pad Thai all along. If the ‘core requirements’ allow the advisor to play such a role, Nancy must have telepathy. It’s clear that we should not allow that.

In both Logic and Pad Thai, we see two cases of unintuitive results that stem from the structure of the ‘core requirements’ as such. In Logic it seems like the advisor may play no role. If the advisee thought just hard enough, they would derive the proper rule to follow. In Pad Thai, we see that the advisor knows what he will order all along, even if he has no knowledge of his circumstances. It must be the case, then, that the advisor points to the advisee’s conclusive reason for an attitude as a reason for adopting a new attitude. I will turn now to a revised version of the ‘core requirements’.
In order to capture the proper function of the ‘core requirements’, I must make the distinction between the attitude possessed by the advisee and the attitude that the advisee ought to adopt.

C+: If one believes one has conclusive reason to have A, then one is rationally required to have A*.

 

C-: If one believes that one lacks sufficient reason to have A, then one is rationally required not to have A*.

 

It is a minor, but important revision. Here, we have illustrated the distinction between the attitudes possessed by the advisee and the attitude they are rationally required to adopt. Moreover, this formulation permits the advisor to play a crucial role in the advisee’s adoption of the attitude. The advisor points out that within the advisee’s attitudinal set, the advisee has a reason for adopting a further attitude. Furthermore, it avoids the confusion provided by the previous ‘core requirements’ under consideration. A case better illustrates the revised version of the ‘core requirements’ .

Veggie: Trudy and Denise disagree over whether one ought to be a vegetarian. Trudy believes that she has conclusive reason for believing that it is wrong to cause unnecessary suffering to animals. Trudy believes she has conclusive reason to believe that consuming meat perpetuates unnecessary suffering to animals, yet continues to consume meat. Denise reminds Trudy that she holds these attitudes and that she has conclusive reason to intend to be a vegetarian. It would be irrational of her not to intend to be a vegetarian.

 

In Veggie, Denise reminds Trudy that her beliefs entail that she’s rationally required to intend to be a vegetarian. Without Denise, Trudy may not have known that she had reasons for being a vegetarian. Furthermore, it seems unlikely that Trudy, through reasoning alone, could reason from the attitudes she possesses to the attitude that she is ought to adopt. Denise does not provide Trudy with a new reason to adopt her intention to be a vegetarian; she already has this reason within her attitudinal set. This case illustrates what I take to be the role of the ‘core requirements’.

Kolodny favors this view because of the failures of the views mentioned in section one. It avoids bootstrapping worries because the advisor merely points to a reason we already have for adopting a new attitude. The attitude one believes they have conclusive reason to have is justified by other pre-existing attitudes within the advisee’s attitudinal set. This attitude justifies adopting a new attitude.

Second, it avoids any worries about the nature of our reasons to be rational. There only appears to be a seeming normative force when one provides rational advice. From the outside, it appears as if an advisor provides the advisee with further reason for adopting an attitude. From the inside, it is merely the ‘ought’ of reasons: the attitude one believes one has conclusive reason to have plays the role of the reason one already has for adopting the attitude. In the case of rational advice, the seeming normative force stems from the transparency of the ‘ought’ of reasons. The ‘ought’ of reasons appears to be the ‘ought’ of rationality. In Veggie, it appears from the outside that Denise offers Trudy some new reasons that she ought to be a vegetarian. The reasons for Trudy to be a vegetarian are already possessed by her and justified by some fact of a previous situation she encountered. Thus, it appears that there is some reason to be rational, but we’ve been rational all along. Trudy has already possessed her reasons for being rational.

The question at hand is whether this picture is accurate. I will now turn to my argument against Kolodny’s conception of rational advice. The ‘core requirements’ do not accurately capture all forms of rational advice, and in fact, there is more than just a seeming reason to be rational, we really have them. I will conclude with an argument against the Transparency Account because of its failure to correctly capture all cases of rational advice. Additionally, I take this as a strong reason to reject the non-reductionist attempt to divide our normative requirements.

Against the Transparency Account

The ‘core requirements’ do not capture all accounts of rational advice. There are situations where an advisor does provide us with further reason using the ‘ought’ of rationality, which is clearly inconsistent with Kolodny’s ‘core requirements’. I will present a case and argue that this particular case is not handled well by the ‘core requirements’. Finally, I explain why it is an instance of a rational ‘ought’.

Directions: Stewart is new to Lincoln and he believes he wants to get to I-80, so that he can drop off his rental car and to catch his flight home. Stewart also wants to know the fastest way to the airport. He stops at the nearest gas and asks the attendant, Bertha, for directions. Bertha tells Stewart that he ought to intend to take a right at the next light.

 

In Directions, Bertha plays the role of the advisor to Stewart, the weary traveler. Stewart believes he has conclusive reason for wanting to get to I-80, which fulfills the antecedent of C+. Does Stewart have conclusive reason for taking a right? In other words, is Stewart rationally required to take a right at the next light by merely believing that he has conclusive reason for wanting to get to I-80? I’m inclined to think that he does not. While Stewart has a reason for adopting the means to his end, it is not clear from his adopted attitudes that he has reason to take one route over another until the moment Bertha moves her lips to speak and gestures in the direction that Stewart ought to go. There are a plethora of options, all of which Stewart is uncertain of until that magic moment when Bertha utters the proper directions. It appears Bertha not only understands and points to an attitude Stewart has, but provides a further reason for an attitude Stewart ought to adopt.

I suspect that one response to this example is the following: Bertha does not provide Stewart with a further reason; it merely appears that she provides Stewart with a reason for adopting the attitude. Bertha merely fulfills the consequent for Stewart by point out that he ought to take a right because of his desire to get to I-80.

Suppose that Bertha is a local and knows about the traffic patterns and construction situations in Lincoln. Bertha knows that under ordinary circumstances, continuing to go straight is the fastest way to I-80. If Stewart were to continue to travel straight, he would not make it to I-80. The road closes a mile ahead.

Construction: Stewart wants to get to I-80 quickly in order to make his flight. Stewart however is new in town and does not know his way to I-80. He stops to ask Bertha directions. Bertha tells Stewart that he should intend to take a right at the light, since she knows the quickest way is under construction.

 

It seems like knowing this information makes it the case that Bertha provides a further reason for Stewart that only she knows is a reason for intending to take a right at the light. It is within the realm of possibility that there are rational advice cases like Construction where the advisor does not point to a fact about one’s situation when providing their advice. Nor do they provide full disclosure of what they know about the situation, yet provides good advice with respect to the advisee’s attitudes.

One might question whether this is a genuine case of the ‘ought’ of rationality. I maintain that is is a case of an ‘ought’ of rationality. In Directions, Bertha delivers advice that is justified by Stewart’s antecedent attitude. Bertha is not pointing to a fact, or a feature of Stewart’s situation, differentiating the case from one of the ‘ought’ of reasons. Bertha points out that it would be irrational for Stewart to do anything other than to intend to take a right at the light given his antecedent attitudes.

One reason that these criticisms may hold is a false distinction between the ‘ought’ of reason and the ‘ought’ of rationality. The ‘ought’ of reason, by definition, intends to point out a fact about one’s situation, while the ‘ought’ of rationality points out one’s reasons views in abstraction from the reasons for them. In other words, the ‘ought’ of rationality points out one’s attitudes as a reason for adopting a further attitude.

It is not clear, however, why one’s mental states do not count as a fact that an advisor points to in order to provide advice. Nor is it at all clear why attitudes do not count as a feature of an agent’s situation. It seems painfully obvious that there is a fact regarding our attitudes and it is a fact in the world. Much like pointing out that one’s building is on fire, intending to go to the smoke shop, wanting to take a right at the light, or intending to be a vegetarian are all facts about the world, but they are facts of a different kind: mental facts.

Mental states are as much of a feature of a situation as a chair, table, and coffee mug. Mental states must count as facts. They get you to do the things that you do. To deny that they are facts is an absurd notion. This may not be what Kolodny has in mind, but it sure seems like an implication of his division of ‘ought’. If this is a consequence of his division of rational ‘ought’ from reasonable ‘ought’, then Kolodny makes a bad distinction that others do not make as non-reductionists, or all non-reductionists make the same mistaken distinction. If the former, then so much the worse for Kolodny; if the latter, so much the worse for the non-reductionist.