Thursday 27 December 2018

An Objection to Lord’s Reasons Responsiveness View

In his 2018 volume, The Importance of Being Rational, Errol Lord defends the thesis that “what it is to be rational is to correctly respond to possessed objective normative reasons” (p. 3).  Lord calls this thesis “Reasons Responsiveness”.  According to Lord, Reasons Responsiveness entails the following principle:

Necessarily, you are rational if and only if you correctly respond to reasons. (Lord 2018, p. 23)

In this blog post, I will briefly explain why I believe that Equivalence—and by extension, Reasons Responsiveness—is false.

To better appreciate Lord’s proposal, let us consider how it explains the irrationality incurred from having inconsistent beliefs or intentions.  Many (if not most) contemporary action theorists attempt to explain why it is irrational to have inconsistent beliefs or intentions in terms of something along the lines of the following wide scope norms.

Belief Consistency:
Rationality requires that [if one believes P, then one does not believe ¬P]

Intention Consistency:
Rationality requires that [if one intends to F, then one does not intend to ¬F]

Lord deems such wide scope norms unnecessary given that the same explanatory work may be done by Reasons Responsiveness. Simply put, in all cases in which an agent has inconsistent beliefs or intentions, they are also failing to respond appropriately to their reasons.  According to the present view, if one possesses decisive reasons to adopt a certain attitude or course of action, then one is rationally required to adopt that attitude or course of action (p. 11).  Moreover, if one possesses decisive reason to believe P, then one ipso facto possesses decisive reason not to believe ¬P.  Hence, we may explain why it is irrational for an agent to have inconsistent beliefs in terms of the requirement that they respond appropriately to their reasons.  The upshot is that we are free to dispense with Belief Consistency and Intention Consistency, and adopt Reasons Responsiveness in their place.

Pace Lord, I maintain that there are cases in which an agent violates Intention Consistency, but in which they do not violate Reasons Responsiveness. Hence, Reasons Responsiveness is unable to explain all cases in which an agent is irrational for having inconsistent intentions.

Consider an agent, let’s call her Edna, who goes to the supermarket to purchase some soup. She finds two identical cans of Campbell’s Tomato soup—can A and can B—on the shelf and forms the intention to take can A rather than can B.  Since both cans are indistinguishable in every salient respect, each reason Edna possesses to (intend to) take can A—e.g., she loves soup, Campbell’s Tomato soup is her favourite, the soup is on sale, etc.—is also a reason she possesses to intend to take can B.  What Edna lacks is a reason to take can A rather than can B or a reason to take can B rather than can A. Even so, Edna may intend to take can A rather than can B.  Moreover, she does not appear to be guilty of irrationality for so doing.

If the above analysis is right, then this poses a problem for Reasons Responsiveness.  This is because while it is rationally permissible for Edna to intend to take can A rather than can B, it is not rationally permissible for her to simultaneously intend to take can A rather than can B and intend to take can B rather than can A.  In sum, even in picking situations, one is not rationally permitted to have inconsistent intentions.  However, as we already noted, Edna has no reason to (intend to) take can A rather than can B. We therefore cannot explain why Edna is being irrational by saying she fails to respond appropriately to her reasons.  Ex hypothesi, she has none.  Hence, we appear to have a case in which an agent has inconsistent intentions but does not fail to respond correctly to the reasons she possesses.

Lord anticipates a similar kind of objection in chapter 2 of his volume.  However, the case Lord considers differs from the above example in a crucial respect.  Lord considers a case in which an agent possesses sufficient (but not decisive) reason to intend to F and sufficient (but not decisive) reason to not intend to F.  Such cases appear to pose a problem for Reasons Responsiveness since possessing merely sufficient reason to intend to F does not entail possessing decisive reason to not intend to ¬F.  The apparent upshot is that Reasons Responsiveness appears to allow that an agent with merely sufficient reasons may have inconsistent intentions.

Lord solves this problem by appealing to a principle he calls “Intentions Attenuate” (p. 31).  According to Intentions Attenuate, if one adopts the intention to F, the adopted intention attenuates the strength of all competing intentions, including the intention to ¬F.  Intentions Attenuate contrasts with the popular suggestion that forming an intention generates additional reasons to do what one intends.  Lord rejects this proposal due to bootstrapping concerns (and rightly so).  Intentions Attenuate avoids the boostrapping problems because it does not entail that merely forming an intention generates additional reasons.  Rather, it entails that an intention may either intensify or attenuate the weight of one’s pre-existing reasons.  However, it is the very feature of Intentions Attenuate that allows it to avoid boostrapping concerns that preclude it from solving the problem being limned in this post. Since Edna has no reason to intend to take can A rather than can B, there are no reasons to be either intensified or attenuated.  The upshot is that Intentions Attenuate is of no help to Lord in addressing the present worry.

One possible diagnosis of why Intentions Attenuate is ineffective against the Edna example is that it appears to presuppose that an intention is rationally permissible only if one has at least some reason in its favour.  (See p. 38 of Lord’s discussion.)  However, I believe intentions stand in contrast to belief in precisely this regard.  Plausibly, it is rationally permissible to adopt a certain belief only if one has some reason in its favour, at least when it comes to beliefs about external world.  For example, I cannot rationally believe that the number of stars in the Milky Way is even given that I lack any reason to believe that it is.  However, as the rational permissibility of picking illustrates, it may be rationally permissible to adopt an intention even if one has no reason in support of the intention.  Given this fact, we cannot hope to make sense of the consistency norms governing intentions in terms of an agent appropriately responding to her possessed reasons.

Lord, E. (2018) The Importance of Being Rational. Oxford University Press.

Friday 14 December 2018

Feldman & Conee Evidentialism and Incomplete Evidence

In a recent essay, “Between Belief and Disbelief”, R. Feldman and E. Conee (F&C) defend the following evidentialist thesis:
(EC)   Believing is the justified attitude when the person’s evidence on balance supports a proposition, disbelieving the justified attitude when the person’s evidence on balance supports the negation of a proposition, and suspension of judgement is the justified attitude when the person’s evidence on balance supports neither a proposition nor its negation. 
F&C observe that (EC) “implies that even slight support is sufficient to make belief justified”(p. 76). In this respect, (EC) stands in contrast to views according to which an agent may be justified in suspending judgement about P even if their evidence slightly supports P over ¬P.   F&C motivate (EC) with the following claim:

“Barely indicating truth is still indicating truth. Believing fits with this indication; withholding judgement does not. Thus, believing and not withholding is the attitude justified by minimal evidential support (p. 77).”

F&C go on to add the following two clarifications with respect to (EC):

First, they note that saying that having one’s evidence slightly support P over ¬P is enough to justify believing P is consistent with the idea that beliefs are accompanied by different strengths of conviction. Moreover, it is possible that beliefs themselves come in degrees, which correspond with the strength of conviction implicated by them. Given this fact, we may also hold that the strength of conviction with which an agent believes P (or their degree of belief in P) should mirror or be proportional to how much their evidence supports P. Hence, saying that having one’s evidence slightly support P over ¬P is sufficient for belief does not imply that it is sufficient for having full confidence in P or believing P with full conviction. 

Second, they emphasize that just because one believes P, it does not follow that one ought to act on one’s belief.  If one believes P with a very low level of conviction (or to a sufficiently minimal degree), it may be more appropriate to wait for additional evidence (assuming the option is available) rather than to act on one’s belief. These two points of clarification are meant to bolster the plausibility and palatability of (EC).

F&C do not go into great detail about what is entailed by an attitude being justified. They do note that “an attitude’s being epistemically justified does not imply having any sort of duty or obligation to take the attitude”(p. 79).  However, it is plausible (and F&C would most certainly agree) that an attitude is justified only if it is rationally permissible.  It also seems plausible that an attitude that is not justified is not rationally permissible.  (Both of these claims are consistent with saying that an attitude being justified does not entail any rational obligation to adopt that attitude.)  In sum, whether or not an attitude is justified appears to covary with whether or not it is rationally permissible, and not with whether or not it is rationally obligatory.  In order to make the notion of justification a bit more concrete (and hence more amenable to appraisal), I will frame my discussion of justification in terms of whether or not an attitude is rationally permissible.

I wish to grant F&C’s contention that having one’s evidence on balance slightly support P is enough for it to be rationally permissible to believe P, given the caveat about strength of conviction or degrees of belief adumbrated above. However, I resist the implication that it would be rationally impermissible to suspend P if one’s evidence very slightly supported P over ¬P.  This implication of (EC) seems particularly implausible in cases in which one is aware that the evidence currently available is known to be limited or incomplete.  For example, suppose that the evidence we currently have available slightly supports the proposition “String Theory is true” over its negation.  Suppose further that we know that the evidence bearing on whether or not String Theory is true is scarce or significantly incomplete. Under such circumstances, it seems as though it would be rationally permissible (i.e., justified) in suspending judgement about whether String Theory is true.  If this is right, then one may be justified in suspending P even if one’s available evidence slightly supports P over ¬P. 

To sum up, in cases in which we have reason to think that our currently available evidence is incomplete, we may be justified in suspending judgement even if on balance our evidence slightly favours P.  


Feldman, R.  and E. Conee (2018). “Between Belief and Disbelief” in Believing in Accordance with the Evidence: New Essays on Evidentialism, edited by K. McCain. Synthese Library: Studies in Epistemology, Logic, Methodology and Philosophy of Science 398: 71-89.