Kiesewetter, Benjamin: The Normativity of Rationality. Oxford: Oxford University Press 2017. 314 Seiten. [978-0-19-875428-2].

Reviewed by Eva Schmidt (Universität Zürich)

Gudrun intends to go to the Beyoncé concert, but she doesn’t intend to buy a ticket, although that’s the only way for her to attend the concert (and she’s aware of this fact). Gudrun’s attitudes are incoherent – she exhibits instrumental irrationality. We can criticize her for having this combination of attitudes: for, first, having a certain end, and second, not being willing to take the necessary means, despite the fact that, third, she knows that she can’t reach her end without taking the means. She rationally ought not to – has decisive reason not to – have this combination of attitudes. The same can be said about Gudrun if she accepts that she ought to go to the concert, but doesn’t intend to (akratic irrationality); if she believes that she has sufficient evidence that Beyoncé will be in Munich, but doesn’t believe that Beyoncé will be there (doxastic akratic irrationality); or if she believes that there will be a Beyoncé concert in Germany, believes that if there will be a Beyoncé concert in Germany, then it will be in Munich, but doesn’t believe that there will be a Beyoncé concert in Munich (modus ponens irrationality).

That rationality is normative in this way, that is, that we have decisive reason not to have incoherent attitudes, is hard to deny. Even so, over the last twenty years, there has been a lively debate over the normativity of rationality, couched in terms of structural requirements of rationality, i.e., requirements to avoid incoherent attitude combinations. While the debate’s participants agree that there are structural requirements – such as the instrumental requirement to intend to ψ if one intends to φ and believes that ψ-ing is a necessary means to φ-ing – there has been disagreement over whether these are normative, over whether subjects necessarily have normative reasons to comply with them.

Against this backdrop, Benjamin Kiesewetter proposes a novel way of accounting for the normativity of rationality in his book The Normativity of Rationality. According to Kiesewetter, (1) there are no structural requirements of rationality, neither normative nor non-normative. (2) Rationality is necessarily normative, but (3) the only requirements it provides are requirements for the subject to respond to her reasons. (4) Despite appearances, cases of incoherent attitudes then do not involve structural requirements, but rather rational requirements for the subject to respond to her reasons. To elaborate, in cases of incoherent attitudes, the subject is irrational; she has decisive reason to give up one of her attitudes.1 However, this is due not to structural requirements of rationality, but to reasons she has anyway and which bear on her attitudes individually. As part and parcel of this account, Kiesewetter defends (5) a perspectivist picture of reasons, according to which the normative reasons there are for a subject to φ are relative to her evidence. As this summary of the main theses of The Normativity of Rationality indicates, the book addresses a wide range of interrelated issues in metaethics and epistemology. It provides a clear, in-depth overview and discussion of the highly complex debates surrounding the normativity of rationality, all the while reconceiving what’s at issue in original ways that will shape these debates in the future. It is a must-read for anyone interested in themes at the intersection of reasons, rationality, and normativity.

Let me provide an overview over the book before turning to critical comments. Chapter 1 presents the four central cases of structural irrationality (akratic irrationality, instrumental irrationality, doxastic akratic irrationality, and modus ponens irrationality) and the corresponding (alleged) structural requirements. Chapter 2 defends the claim that rationality is normative in the sense that there is decisive reason to do what is rationally required. In chapter 3, Kiesewetter clarifies how to conceive of structural requirements and elaborates on the central distinction between narrow- and wide-scope structural requirements.

In the following three chapters, he shows that problems ensue on the assumption that there are structural requirements, no matter how one stands on the narrow-scope vs. wide-scope issue, e.g. with respect to the detachment of unacceptable claims regarding what the subject ought to do (chapter 4) or with respect to the question of what favors conforming to structural requirements in the first place (chapter 5). In chapter 6, he argues that problems along these lines arise even after one abandons the assumption that structural requirements are normative. Consequently, that there are structural requirements at all is “a myth” (p. 157) and we have to look elsewhere to account for the rationality at stake in cases of incoherent attitudes – which is Kiesewetter’s project in chapters 7 to 10.

In chapter 7, he defends the view that rationality consists in responding to one’s available reasons, i.e., the reasons that are part of one’s evidence. A subject has a certain body of evidence, which consists of external facts that she knows or that are internally accessible to her and of internal phenomenal occurrences that are also accessible. This body of evidence determines what reasons are available to her, and thereby provides decisive available reason concerning a certain response, i.e., it fixes how the subject is rationally required to respond. Kiesewetter faces the problem that, while evidence is partly mind-external on his picture and can thus vary independently of the subject’s mental properties, rationality is plausibly mind-internal: same mental properties, same rational requirements. But then a subject’s evidence, and so her available reasons, and so whether she has available decisive reason to φ, can vary independently of her mental properties, and so independently of whether she is rationally required to φ. This is unacceptable on a reasons-responsive view of rationality.

Kiesewetter responds by endorsing what he calls the “backup view” (p. 171). Here’s an illustration. Gudrun sees that her three-year-old son is underwater in the pool, flailing his arms. The fact that he is underwater in the pool, flailing his arms, is among her evidence. It provides her with decisive reason, first, to believe that her son is drowning, and, second, to intend to jump in the pool and save his life. Gudrun’s envatted mental duplicate Badrun merely hallucinates that her three-year-old son is underwater in the pool, flailing his arms. There is no such fact (after all, she is a brain in a vat), so the fact is not among her evidence and cannot provide her with any decisive reason. Apparently then, though Gudrun and Badrun – as internal duplicates – are both equally rationally required to intend to jump in the pool and save their sons’ lives, only Gudrun has decisive reason so to intend (and this is the problem for Kiesewetter). However, like Gudrun, Badrun has an appearance of her son’s being underwater in the pool, flailing his arms. And this appearance is evidence as well, which – according to the backup view – provides equally decisive available reason for Badrun to intend to jump in and save him. Generally, then, whenever a subject has decisive available reason for an attitude, so does her internal duplicate. For even if only the original subject has certain external facts as evidence, her deceived duplicate has backup evidence available that provides reasons of exactly the same strength, viz. her appearances. This precludes any mismatch between decisive reasons and rational requirements.

Chapter 8 contains Kiesewetter’s defense of a perspectivist account of normative reasons and ought. Only facts that are part of the subject’s evidence are normative reasons for her; only these facts determine how she ought to respond. Importantly, this is not a subjective, but an objective variant of perspectivism, as it doesn’t relativize a subject’s reasons (or what she ought to do) to her – possibly false – beliefs. Rather, it ties her reasons to her evidence, which consists of actual facts of her situation (including, but not exhausted by, facts about her experiences).

Further, Kiesewetter argues that objectivism about normative reasons and ought, according to which the subject’s epistemic situation makes no difference to her normative reasons or to how she ought to respond, cannot make sense of “known ignorance cases” (p. 202). In such cases, an agent has three options. For instance, Jill can choose between three medications for her patient’s skin complaint. She knows that medication A will relieve, but not cure the condition. Of medications B and C, she knows that one of them is the cure, while the other will kill the patient, but she is clueless as to which is which. Assume that medication C will de facto cure the patient. According to objectivism, then, Jill ought to administer medication C. Intuitively, however, she ought to administer medication A. Perspectivism does justice to this intuition: Given her evidence, there is a substantial probability for both medications B and C that they will kill the patient, whereas administering medication A carries no such risk. According to perspectivism, it is these facts about Jill’s evidential probabilities that are her reasons, not the fact that medication C is de facto best, of which she is ignorant.

In chapters 9 and 10, Kiesewetter applies his view to the four cases of structural incoherence, with the aim of showing that the irrationality here is due not to violations of structural requirements, but to a failure to respond to decisive reasons pertaining to the individual attitudes. Not just any failure to respond to one’s decisive reasons constitutes full-fledged irrationality, however; a subject is irrational when her attitudes guarantee a failure to respond to decisive reasons. Kiesewetter argues that this is true of practical akrasia, doxastic akrasia, and modus ponens incoherence. Here’s a simplified example. Gudrun exhibits doxastic akrasia – she believes that there is sufficient evidence that Beyoncé will be in Munich, but doesn’t believe that Beyoncé will be in Munich. Now she either has sufficient evidence that Beyoncé will be in Munich, or not. If she has sufficient evidence, she has decisive reason to believe that Beyoncé will be in Munich; since she doesn’t have this belief, Gudrun violates a reasons-responsive requirement. What if she lacks sufficient evidence that Beyoncé will be in Munich? Then she has decisive reason against believing that she has such evidence; her belief that she has sufficient evidence that Beyoncé will be in Munich is not responsive to this decisive reason, so she again violates a rational requirement. Overall, with this combination of attitudes, Gudrun simply couldn’t be fully responsive to her decisive reasons; so she is irrational.

Finally, in chapter 10, Kiesewetter extends his view to the case of instrumental incoherence, which poses the greatest difficulty. For an agent’s pursuit of an end is often merely rationally permissible, not required. In such cases, since the agent’s intention to pursue the end is not backed by decisive reasons, it cannot be argued (at least not in the same way as in the other cases of incoherence) that one of her attitudes is guaranteed to violate decisive reasons. Kiesewetter concedes that in some such “underdetermined cases” (p. 263), the subject is not irrational despite her incoherent attitudes. However, most of them involve an economic reason to decide between pursuing the end and not taking the means, viz. the reason that agents who exhibit instrumental incoherence have a higher risk of engaging in costly, pointless activity.

Turning now to criticisms, my biggest worry regarding Kiesewetter’s proposal is that the backup view, which he introduces to square his externalism about evidence and reasons with his internalism about rationality, only engenders further trouble. Focus on Gudrun and Badrun. Gudrun has the fact that her son is underwater in the pool and flailing his arms as evidence that provides decisive reason for her to intend to jump in and save him. Plausibly, one reason provided by this evidence is the fact that her son is drowning, which Gudrun knows. Badrun has only the appearance as of her son underwater in the pool and flailing his arms as evidence providing her reason to intend to jump in and save him. The reason provided by Badrun’s evidence is not the fact that her son is drowning (there is no such fact), but the fact that, probably, her son is drowning.2 Since Gudrun has the same appearances as Badrun, her evidence provides this ‘probability fact’ as a reason in addition to the ‘bare fact’ that her son is drowning.3

The first worry raised by this is that it seems that the bare fact is a stronger reason than the probability fact. Certainly, if it is evidentially probable that your son is drowning, you ought to intend to jump in the pool and save him; but intuitively, the requirement so to intend is even more pressing when it is simply the case that your son is drowning, and you know it. In support, notice that the known fact that your daughter is drowning in the pond next to the pool, as a reason to intend to jump in the pond to save her, outweighs the probability fact, but not the known bare fact that your son is drowning in the pool, as a reason to intend to jump in the pool to save him. But then the reasons provided by backup evidence are not as strong as the reasons provided by ordinary facts. This calls into question whether envatted duplicates are guaranteed to have decisive available reason for a certain response whenever their non-deceived counterparts do.

Second, assume that I am wrong and backup reasons are guaranteed to be as strong as ordinary reasons. If so, we can explain why Gudrun is rationally required to intend to jump in the pool and save her son both by appeal to the bare fact and by appeal to the probability fact – she has both of them as equally strong reasons. After all, she has the same appearances as Badrun; they have the same evidential weight as for Badrun; so they also provide her with decisive reason (viz. the fact that, probably, her son is drowning) to intend to jump in the pool and save her son. This raises a screening-off worry for the backup view: Both the bare fact and the probability fact explain why Gudrun ought to intend to jump in the pool. But only the probability fact explains additionally why Badrun ought so to intend, so it has greater explanatory power than the bare fact – it provides a better explanation. Consequently, an explanation of what Gudrun ought to intend had better appeal to the probability fact; yet it may leave out the bare fact. The bare fact that her son is drowning is then screened off, by the probability fact, from doing any explanatory work. Generalizing from this, the reasons provided by a subject’s backup evidence explain her rational requirements better than her reasons provided by the external world. The worry this raises for Kiesewetter is that his externalism about evidence and reasons now appears to be a gratuitous extra to his account – and this raises the worry whether the account is genuinely externalist.

My final worry is whether we can think of Badrun as rational with regard to how she bases her intention. Badrun, like Gudrun, intends to jump in the pool and save her son for the reason that (as she believes) he is drowning – she bases her intention on the merely apparent fact that he is drowning, which is not really a reason for her so to intend at all. The fact that, probably, he is drowning, though it is a reason provided by her evidence, is not utilized in her reasoning. Consequently, while she has decisive reason to so to intend, she does not intend for that reason; really, she intends to jump in the pool for no good reason. She is in that respect less rational than her internal duplicate Gudrun, who intends for a reason that is actually a normative reason. The problem is, in the end, that evidence and reasons externalism, even with the backup view, is not guaranteed to be compatible with internalism about rationality. For what is a rational basis for a subject’s intention in the good case, may be no basis at all (or an irrational basis) for the same intention of her envatted duplicate.

The discussed criticisms have hopefully shown that one of the deepest fault lines of Kiesewetter’s theoretical landscape lies in the combination of externalism about reasons and evidence with internalism about rationality presupposed by the reasons-responsive view of rationality. None of this should detract from the philosophical merit of his book, which offers not only a lucid as well as penetrating discussion of the status quo of the debate over the normativity of rationality, but also makes original and convincing contributions to that debate. Kiesewetter’s reasons-responsive account of rationality, as well as his brand of perspectivism, are certain to be essential points of reference for anyone who is interested in normativity, rationality, or reasons.


I thank Benjamin Kiesewetter for helpful feedback on an earlier version of this review.

  1. Exceptions to be discussed below.↩︎

  2. As in the known ignorance cases sketched above, the probability in question is evidential: It is a matter of how probable it is that Badrun’s son is drowning, given her evidence and applying objective Bayesian probability measures.↩︎

  3. In support of this claim, imagine that Gudrun is suddenly envatted without her knowledge. How does this affect the perceptual evidence she possesses? It blocks the bare facts that her perception would otherwise provide as evidence, while leaving untouched the probability facts provided as evidence by her perceptual appearances.↩︎

© 2019 Zeitschrift für philosophische Literatur, lizenziert unter CC-BY-ND-3.0-DE