Suspension of Judgement: Fittingness, Reasons, and Permissivism

Abstract

This paper defends three theses on the normativity of the suspension of judgment. First, even if beliefs have to fit the truth and disbelief the false, suspension can still have satisfiable fittingness conditions. Second, combining this view with specific theses on the link between fittingness and normative reasons in favour of attitudes commits one to the existence of reasons to suspend judgement, which are neither reasons to believe nor reasons to disbelieve. These independent reasons, in turn, generate a form of epistemic permissivism. Finally, I argue that there are different routes to derive this commitment to independent reasons for suspending judgement. Not only fittingness-centred approaches to epistemic normativity but also many analyses in terms of reasons are committed to this form of epistemic permissivism.

1. Introduction

In this paper, I bring two aspects of recent epistemological theorising together. On the one hand, there has been a very productive debate about the nature and norms of the suspension of judgement. I use ‘suspension’ as denoting a third, flat-out doxastic attitude aside from belief and disbelief.¹ On the other hand, there has been a more general interest in the notion of fittingness which exceeds the bounds of epistemology. An attitude is said to be fitting iff it gets something peculiar right about its object, or if the object merits the attitude.² This normative property of attitudes is taken to bear an important relation to the normative reasons for said attitudes. My aim is threefold. First, I defend the thesis that suspension has satisfiable fittingness conditions against a recent line of attack: I argue that satisfiable fittingness conditions for suspension are compatible with a common position on the fittingness of belief. Second, I show that this manoeuvre commits anybody who endorses a strong link between fittingness and normative reasons to a form of epistemic permissivism. Third, I investigate different routes to deriving this commitment to epistemic permissivism. As I demonstrate, this commitment extends beyond fittingness-centred theories.

I start by summarising a line of thought brought up against the very possibility of reasonable fittingness conditions for suspension (section 2). However, this argument rests on a dubious premise – it takes it for granted that suspension and (dis-)belief can never be fitting at the same time. In the third section, I point out why there is no reason to believe in said premise and, thus, there should be no problem assuming that suspension and (dis-)belief can be fitting at the same time. This thesis raises an immediate concern regarding whether the fact that the fittingness conditions of doxastic attitudes are co-satisfiable entails epistemic permissivism. The second half of the paper is occupied with exploring this question. First, I argue that co-satisfiable fittingness conditions for doxastic attitudes paired with a claim about the connection of fittingness conditions and normative reasons entail that at least some reasons to suspend exist that are neither reasons to believe nor reasons to disbelieve. These independent reasons to suspend, in turn, commit one to epistemic permissivism (section 4). Second, there are different ways to derive these independent reasons to suspend. A whole group of theories which analyse epistemic normativity in terms of reasons, of which fittingness-centred theories comprise only a fraction, are committed to sufficiently independent reasons to suspend and, thereby, to the same form of epistemic permissivism.

2. The case against fitting suspension

In this section, I consider an argument which was recently posited by Lee (2022) as part of a larger argument strategy against the, so called, fittingness-first programme, the family of theories according to which every normative notion can be reduced to the primitive notion of fittingness.³ According to Lee's reasoning, if these theories want to account for certain reasons to suspend judgement, they are committed to the following inconsistent set of propositions:

1. (1) TRUE: A belief that p is fitting iff p is true.

2. (2) SUSP: Suspending judgement as to whether p has to fit one's epistemic situation with regard to answering whether p.

3. (3) BWL (belief-withholding link): If it is fitting to believe p, then it is not fitting to withhold belief about p.

Since there can be cases where the epistemic situation merits suspension when the belief in question is true, these three propositions generate a contradiction. Thus, provided that beliefs have to fit the truth, suspension cannot have fittingness conditions and, as Lee continues in his reasoning, a fittingness-first theory along these lines should not appeal to reasons for suspension. In the remainder of this section, I closely examine this segment of Lee's resourceful paper.

First, I provide some further elucidations. TRUE is a widespread idea in the literature on fittingness. A fitting attitude gets something peculiar right about its object, and in the case of beliefs, the thing that the attitude is supposed to get right is the truth value of the believed proposition (see McHugh 2014). One might doubt this premise and opt for a different account of fitting beliefs. In that case, the above puzzle might not even emerge.⁴ However, identifying the fittingness conditions of beliefs with the truth of the believed proposition is close to orthodoxy, and most fittingness-first accounts of epistemic normativity in the literature seem to take it for granted. Furthermore, one might even argue that the relationship between truth and beliefs is at the very core of the meaning of ‘fittingness’ – at least as used in the more recent debate: ‘fittingness’ simply denotes whatever normative relation connects belief and truth (D'Arms and Jacobson 2000: 72). In this paper, I will follow this paradigmatic idea.

In the context of Lee's paper, SUSP is a necessary commitment for fittingness-firsters to explain the reasons for suspending judgement in terms of suspension's conditions of fit.⁵ However, it also has independent motivation. For instance, as Booth (2014) argued, if suspension has any fittingness conditions at all, its fittingness must differ quite drastically from that of other doxastic attitudes. Similarly, the only paper-length discussion of fitting suspension by Rosa (2020), endorses something along the lines of SUSP; furthermore, in their recent book, McHugh and Way (2022: chapter 7) allude to a similar idea. Moreover, if suspension is nothing but a particular higher-order belief about one's epistemic situation, TRUE entails SUSP.⁶

BWL constitutes the core premise of Lee's argument. It is also shared by Rosa (2020: 7ff.) in his work on suspension: believing p and suspending judgement as to whether p can never be fitting at the same time. Lee's defence of this idea is as follows:

The idea that belief is fitting if and only if it is true derives its plausibility from the idea that having a fitting attitude is a matter of getting things right. And whenever p is true, only believing p is a way of getting things right and neither withholding nor disbelieving gets things right. For a helpful analogy, imagine a horserace which horse A has won. It is only those who bet on A who can be said to have gotten things right: those who decided not to make a bet or those who bet on another horse have, each in their own way, failed to get things right. If so, Belief-Withholding Link is true. (Lee 2022: 3572)

BWL is justified by an appeal to the general nature of fittingness or getting things right. If there are two incompatible attitudes or actions, only one of these attitudes or actions can be said to get things right.

Second, I argue that Lee's reasoning can be extended to show that there will always be a contradiction as soon as the fittingness conditions of suspension are satisfied. To strengthen Lee's argument, I add three further premises:

1. (4) FALSE: A disbelief that p is fitting iff p is false.

2. (5) DWL (disbelief-withholding link): If it is fitting to disbelieve p, then it is not fitting to withhold belief about p.

3. (6) BIVALENCE: Any proposition is either true or false.

FALSE, like TRUE, is a commonly endorsed view in the literature on fittingness. If disbelieving p is equivalent to believing in the negation of p, FALSE follows from TRUE. DWL can be justified by the same line of thought that Lee used to motivate BWL. Finally, BIVALENCE can be taken for granted in these contexts.⁷ Together, these premises entail that there cannot be any satisfiable fittingness conditions for suspending judgement.

It must be noted that this problem is somewhat more general than I have indicated so far. First, even though Lee's argument is directed against the fittingness-first programme, the above conclusion might create trouble for a range of different theories. While satisfiable fittingness conditions for suspension are especially appealing for fittingness-firsters, they might also be endorsed by anybody who believes in a link between normative reasons for attitudes and their conditions of fit. If one believes, for instance, that the distinction between the right and wrong kinds of reasons should be analysed in terms of fittingness, one already encounters Lee's problem.⁸ Even reasons-first theories, i.e. accounts which posit that every normative notion can be ultimately analysed and reduced to talk of normative reasons, are not entirely safe from this issue.⁹ Second, the appeal to SUSP is not essential for the argument to work. As long as the truth view on fitting beliefs is combined with BWL and DWL, any theory of suspension that posits satisfiable fittingness conditions has to be rejected. For instance, the same point applies to Friedman's (2013, 2017) prominent account according to which suspension is a special inquisitive attitude. If any question with a true or false answer merits this attitude, the conjunction of BWL and DWL will create a contradiction.

To conclude, the addition of BWL and DWL to the orthodox idea that belief and disbelief have to fit the truth value of the (dis-)believed proposition makes satisfiable fittingness conditions for suspension impossible. Based on the somewhat plausible assumption that there are some reasons of the right kind to suspend judgement, the position is incompatible with several metanormative theories as well as certain accounts of the nature of suspension. Both of these upshots motivate closely reconsidering the justification of BWL and DWL. As I argue in the next section, the motivation for these claims, to which Lee alludes, is fundamentally flawed.

3. Fittingness can be permissive

The above justification of BWL and DWL referred to a general property of fittingness or of getting things right: fittingness is unique in that, for any two incompatible attitudes φ and ψ, the fittingness conditions of φ and ψ cannot both be satisfied at the same time. The most straightforward reading of ‘incompatible’ which is backed by Lee's text, is in terms of rational incompatibility: any subject which does both φ-ing and ψ-ing exemplifies a form of structural irrationality.¹⁰ Belief and suspension appear to be incompatible in this way. For instance, a subject who believes p and suspends judgement as to whether p passes Worsnip's (2021) inconsistency test for structural irrationality: if it were disclosed to some subject that they are both in a state of believing p and suspending with regard to whether p, they would be disposed to change their attitudes in due time.¹¹ Furthermore, in this interpretation, there are many true principles in the vicinity of BWL and DWL. A contradictory set of beliefs is structurally irrational, and, given the law of non-contradiction, the fittingness conditions of the contradictory beliefs cannot be satisfied at the same time. The same might be true for other, famous cases of structural irrationality, for instance, cyclic preferences and akratic states of mind.

Even though there are some true instances of the general principle Lee alludes to, the principle does not hold in its full generality. Fittingness can be permissive, i.e. there are cases in which two rationally incompatible attitudes are fitting at the same time. Although one cannot be in a rational state of mind by having these attitudes at the same time, each attitude might get something right about its object. This permissiveness claim is merely the negation of the universal uniqueness claim above. Hence, I do not want to say that any two incompatible attitudes might be fitting at the same time. Rather, whether or not a pair of incompatible attitudes can be fitting at the same time, must be decided on a case-by-case basis by checking whether the fittingness conditions in question are co-satisfiable. In this section, I present two examples of such pairs.

First, I address inter-level incoherence. I believe that it is sunny today, but I also have the higher-order belief that my evidence does not support such a first-order belief. Just as believing and suspending create an inconsistency, the combination of such first- and higher-order beliefs leads to a form of structural irrationality. They, too, pass Worsnip's inconsistency test. If it were disclosed to me that I have both the first- and the higher-order belief, I would be disposed to change my attitudes in due time. However, as seen in the previous section, both beliefs can be fitting at the same time. For example, the sun might be shining, yet all my evidence may speak against this proposition. Note that any proponent of a higher-order belief view of suspension should already take this observation as a decisive case against BWL and DWL.

Second, consider permissive choice situations: I contemplate which outfit to wear today. My choices include wearing (a) only a red rain coat, (b) only a blue rain coat, or (c) no rain coat at all. Furthermore, it will rain today, and I would rather not get wet. I have no preference as to whether to wear the red or the blue rain coat and, all things considered, it does not make any difference. In this case, I can intend either of the three actions (a), (b), or (c). Under any plausible account of the fittingness of intentions, both intending (a) and intending (b) turn out to be fitting. Presumably, an intention has to fit the practical value of the intended action (McHugh and Way 2022: chapter 3). Both (a) and (b) are on par concerning their practical value. Hence, they should be on par regarding their fittingness. If one of these intentions is fitting – which I assume it is – the other one will be fitting as well.¹² Yet, if it were disclosed to me that both options are incompatible, intending both (a) and (b) is an instance of structural irrationality.¹³ Their combination passes Worsnip's inconsistency test. Hence, to use Lee's idiom, in permissive situations such as the one envisaged here, there are different ways of getting things right. Intending (a) gets things right and intending (b) gets things right as well.

At this point one might object that I conflate fittingness with the deontic status of being permitted. Lee (2022: 3572f.) dismisses Rosa's (2020) account of the fittingness of suspension for the same reason.¹⁴ In short, while permissions do not agglomerate over conjunction, fittingness attributions do. I assume that this observation is correct. However, my take on fittingness complies with this demand. As my comments about structural irrationality already make apparent, even though both intending (a) and intending (b) are fitting at the same time, their fittingness does not imply that it is rationally permissible to intend (a) and intend (b) at the same time. Even though taking up the two intentions at the same time does not alter their fittingness concerning their object, their conjunction is not permissible. Thus, my appeal to the permissiveness of fittingness does not amount to conflating ‘fitting’ and ‘permitted’.

Another objection raised by several people is that the above examples do not refute BWL and DWL. It might be argued that doxastic attitudes are significantly different from other attitudes. Thus, even though fittingness might be permissive, the fittingness conditions of doxastic attitudes might not be co-satisfiable after all. However, this objection misses the point of my counterexamples. Here, I argue against the motivation or justification of BWL and DWL which I mentioned in the previous section. Pace Lee, I show that one cannot derive BWL and DWL from any general observation about the nature of fittingness or getting things right. Nevertheless, it might turn out that BWL and DWL are true.

The question remains whether there is anything else, besides Lee's general thesis, which lends some support to BWL and DWL. Here, one might point to Rosa's (2020: 8) perspective on BWL. According to him, the incompatibility of the fittingness of belief and suspension is justified because it explains the normative conflict between belief and suspension. This rational is indeed an abductive point in favour of BWL. However, it is also a fairly weak one. To explain the tension between certain attitudes in the above examples, it is not possible to rely solely on considerations of their fittingness. Therefore, a different theory of rationally incompatible attitudes must be introduced, and applying this theory to beliefs and suspension does not appear to yield a significant downside. Whether this task can be accomplished in a fittingness-first framework is a question of its own and answering it would exceed the bounds of this paper.¹⁵ However, as mentioned in the previous section, my concern here is not only with fittingness-first theories. Satisfiable fittingness conditions for suspension are needed for numerous different theories, only some of which come with such a commitment concerning the analysis of structural irrationality.

In conclusion, Lee's general motivation for BWL can be dismissed, and even Rosa's more concrete argument for BWL only goes so far. Hence, there is no good reason to endorse BWL and DWL before taking any stance on the fittingness of suspension.¹⁶ Therefore, I assume that the puzzle from the previous section is a good abductive reason to doubt BWL and DWL. In other words, any proponent of the truth view on fitting belief should not assume from the start that suspension cannot have satisfiable conditions of fit. In the next section, I explore some consequences of opting for co-satisfiable fittingness conditions for doxastic attitudes.¹⁷

4. On co-satisfiable fittingness conditions and epistemic permissivism

‘Epistemic permissivism’ denotes any position which claims that more than one doxastic attitude may constitute a rationally permissible response to a given set of evidence.¹⁸ Here, I am primarily concerned with intrapersonal rather than interpersonal permissivism, i.e. the idea that there are different rational responses open to one and the same subject at the same time. Furthermore, the discussion to follow will centre around a weak version of this thesis: I will only discuss the idea that suspension can be rational while either belief or disbelief would also constitute rational responses. In this section, I argue that co-satisfiable fittingness conditions combined with a fittingness-first take on normative reasons entail weak intrapersonal permissivism. I proceed in two steps: first, I show that the above claims force the acceptance of independent reasons to suspend judgement. Second, if there are reasons to suspend that are not reasons for either belief or disbelief, this version of epistemic permissivism follows.

As a preliminary remark, note that permissivism is not a thesis about fittingness, but about rational responses. Hence, not every situation in which two responses would be fitting is a permissive situation and vice versa. Consider some variations of the rain coat example from the previous section. Maybe, unbeknownst to me, there is some important practical difference between wearing the red rain coat and the blue one. Assume, for instance, that I promised someone to wear the red one, but completely forgot about it. In that case, this might be a permissive situation, but only one of the options would count as fitting. On the other hand, the two options might be on par concerning their fittingness, but I have misleading information to the contrary. For instance, somebody wrongfully tells me that I promised to wear the red rain coat that day. In that case, the situation might not be permissive after all. Therefore, even if the fittingness conditions of suspension and belief are co-satisfiable, this does not yet constitute a form of epistemic permissivism.

Over and above co-satisfiable fittingness conditions, a claim is needed about the connection between rationality and fittingness. First, I assume that substantive rationality boils down to a form of reason responsiveness: an attitude φ is rational iff it constitutes a proper response to the subject's normative reasons.¹⁹ Second, normative reasons might bear an important relationship to the fittingness conditions of the attitudes in question. I call this assumption the fittingness reasons link (FRL). In the current literature on the analysis of normative reasons, there are three such links. According to the explanation view, reasons are explanations of fittingness facts (Howard 2019). Hence, R is a reason for φ-ing iff R bears an explanatory relationship to the fact that φ's fittingness conditions obtain.²⁰ In contrast, the evidence view contends that reasons are evidence of fittingness facts (Whiting 2018, 2022).²¹ Thus, R is a reason for φ-ing iff R is evidence that φ's fittingness conditions obtain. Finally, according to the nomic view, R is a reason for φ-ing iff ceteris paribus R only obtains when φ's fittingness conditions are satisfied (McHugh and Way 2018, 2022).²²

I now combine these assumptions about rationality and FRL with the ideas from the previous section, namely the co-satisfiable fittingness conditions for doxastic attitudes. (Dis-)belief and suspension can be fitting at the same time. In that case, the fittingness conditions of belief and suspension appear to be independent in several respects. First, an explanation of why p is true (or false) is different from an explanation of why it is fitting to suspend with regard to p. Considerations that have a bearing on the former, might play a role in the latter. However, the explanation of why suspension is fitting should comprise further considerations. Second, evidence in favour of (or against) p is different from the evidence that the fittingness conditions of suspending with regard to p are satisfied. In this case, as before, evidence in support of p might be part of the evidence that suspension would be fitting, but further considerations might make the fittingness of suspension more likely. Finally, the kind of things bearing a nomic connection to the truth (or falsity) of p will be different from the kind of things bearing a nomic connection to the fittingness of suspending with regard to p. Thus, in all versions of FRL, the normative reasons in favour of beliefs and disbeliefs should come apart from the normative reasons in favour of suspension, i.e. there will be reasons in favour of suspension which are not reasons in favour of belief or disbelief.

Assume, for instance, that SUSP is correct, i.e. suspension is fitting iff one is in bad epistemic circumstances. First, the explanation of why the epistemic circumstances with regard to some p are bad will differ drastically from the explanation of why p is true or false. Second, there might be evidence that one is in bad epistemic circumstances with regard to p that is not evidence in favour of or against p, for instance, certain higher-order considerations. Finally, while evidence in favour of or against p might bear a nomic relationship to the truth of p, there will be other factors which have such a connection to the fittingness of suspension. On similar grounds, one might argue that certain higher-order considerations are nomically connected to the fact that one is in bad epistemic circumstances.²³

In the second part of my reasoning, I argue that if there are independent reasons to suspend judgement, then epistemic permissivism should be endorsed. Here is a straightforward argument which rests on four premises. First, which attitudes constitute rational options for some subject in a situation depends on the outcome of a weighing of the subject's normative reasons in favour of those competing attitudes. Second, reasons to suspend sometimes outweigh reasons to believe, and vice versa. To address the example from above, strong higher-order evidence in favour of suspension might outweigh one's first-order evidence in favour of belief. A subject S might receive reliable testimony that S is not in an epistemic situation to tell whether p. On the other hand, first-order evidence in favour of belief might be enough to outweigh less weighty higher-order evidence. For instance, if S receives the above testimony from an unreliable source, S's first-order evidence might be able to outweigh this less-weighty reason. Third, the weight of reasons is somewhat continuous and does not allow for gaps and jumps. Therefore, situations in which the normative reasons in favour of suspension and the normative reasons in favour of (dis-)belief are on par should at least be conceivable, i.e. it should be possible that both sets of reasons have an equal weight attached to them. Finally, if the weight of the reasons in favour of the two best-supported options is on par, then either of these options is rationally permissible.²⁴ Therefore, epistemic permissivism follows.

I close this section by elaborating on this commitment to permissivism. First, the preceding reasoning is in line with one major motivation behind the so-called uniqueness thesis, the negation of permissivism. According to the uniqueness thesis, there is only one doxastic attitude which is rationalised by a given set of evidence. A central line of reasoning behind this idea is that the evidential support relation between a set of evidence E and some proposition p can only take one of three forms: E can support p, support non-p, or be neutral in that regard. Hence, given that epistemic support is unique, there can only be one rational response to a body of evidence.²⁵ However, at no point in the above case for permissivism is the uniqueness of the evidential support relation cast into doubt. For instance, our premises did not relativise evidential support to the epistemic standards of different subjects (Schoenfield 2014) or the relative weight one assigns to the goals of gaining true beliefs and avoiding falsehoods (Kelly 2013). It is only stated that aside from the set of reasons which comprise the evidence of p's truth or falsity, there are additional normative reasons which rationalise suspending on whether p. Thus, one can accommodate a basic motivation of the uniqueness thesis while still proposing a form of permissivism. Hence, the above case for permissivism from the reasons to suspend is, in a sense, less committal than other routes to establish this position. However, this characteristic does not preclude the above form of permissivism from any criticism. Most standard objections which have been levelled against permissivism still apply. Thus, one might question whether an arbitrary choice between two permitted doxastic options can be called ‘rational’ (Stapleford 2019; White 2005), or whether permissivism deters the value of rationality (Horowitz 2014, 2019) or that of rationality ascriptions.²⁶

To conclude, endorsing the view that suspension and (dis-)belief can be fitting at the same time, and maintaining any version of FRL entails weak intrapersonal epistemic permissivism. Thus, even though fittingness-firsters can evade the puzzle envisaged in the second section, they must buy into epistemic permissivism; therefore, they are susceptible to its problems. Yet, such a commitment to permissivism is merely contentious. It is not a detrimental objection and might even be motivated on other grounds.²⁷ The culprit that generates this commitment is the fact that FRL leads to independent reasons to suspend. As I argue in the next section, one can motivate these independent reasons on quite different grounds, i.e. the commitment to independent reasons is not peculiar to the thesis of co-satisfiable fittingness conditions for doxastic attitudes.

5. Another route to independent reasons in favour of suspension

As argued in the previous section, normative reasons to suspend on whether p that are neither reasons to believe nor disbelieve p entail epistemic permissivism. A proponent of co-satisfiable fittingness conditions for suspension is committed to these independent reasons. However, other arguments in the literature also come with such a commitment. In this section, I elaborate on these ideas. As I demonstrate, any proponent of FRL should endorse independent reasons to suspend anyway. Yet, this commitment generalises to many other positions which analyse epistemic normativity in terms of reasons.

First, a group of defenders of pragmatic encroachment appeals to independent reasons to suspend. According to pragmatic encroachment, the epistemic status of one's belief formation might depend on the practical circumstances of one's situation, such as the potential costs of getting things wrong. Since practical reasons in favour of belief and disbelief are highly contentious, some scholars have identified these practical factors as reasons in favour of the third doxastic stance of suspension.²⁸ Because these practical reasons to suspend are supposed to be neither reasons to believe nor reasons to disbelieve, this take on pragmatic encroachment is committed to independent reasons to suspend.

Second, one might argue that higher-order evidence is a reason to suspend judgement, but deny that it has any direct bearing on belief or disbelief. This view has been defended by Lord and Sylvan (2021) who regard it as a plausible middle ground in the debate about the impact of higher-order evidence.²⁹ Furthermore, as Lee (2022) argues, proponents of FRL, TRUTH and FALSE cannot view higher-order evidence as a reason to believe or disbelieve any first-order proposition. Hence, if they maintain that higher-order evidence has an impact on the rationality of first-order belief formation, they might view higher-order evidence as a reason to suspend.³⁰ Thus, another instance of independent reasons to suspend arises.

Finally, Lord and Sylvan (2022) propose that suspension is sensitive to zetetic or inquiry-related factors. If one accepts Friedman's (2017, 2019) account of suspension as the most general inquisitive attitude, these zetetic reasons in favour of suspension should not come as a surprise. For instance, one might argue that the fact that a question is not central to one's research agenda is a reason for suspending judgement on it rather than trying to form a correct belief about it and, subsequently, wasting precious cognitive resources on an irrelevant issue. Since neither belief nor disbelief is supposed to be sensitive to these zetetic factors, Lord and Sylvan are, again, defending independent reasons to suspend.

However, all of these points are contentious. One might argue, for instance, that neither of these reasons has a bearing on the normativity of the third doxastic attitude, but on something else.³¹ As McGrath (2021a, 2021b) contends, philosophers commonly conflate the third doxastic attitude with certain prior mental actions such as waiting or postponing one's judgement (cf. Crawford 2022). In a similar vein, one might point to Wagner's (2022) position on suspension, which also does not view it as the third doxastic attitude. In her account, suspension is the mental action of committing to one's indecision. However, if there are independent reasons to do these non-doxastic things, they do not lead to the envisaged independent reasons to suspend, i.e. reasons to suspend which compete with the reasons to believe and disbelieve in determining which doxastic attitude is permitted. Therefore, upon paying closer attention to the different stages of one's deliberation, neither of the above positions leads to independent reasons to suspend.³²

There is a good case to be made in favour of accounting for many of the above points in terms of such a two-stage model. However, as Schroeder (2021: 137–49) contends, certain considerations escape this model. The most crucial one is that believing that p appears to be irrational as soon as the evidence with regard to p is tied, or even close to being tied. This phenomenon has come to be known as prohibitive balancing: if the evidence with regard to p is close to being tied, neither belief nor disbelief should be permissible options.³³ Whatever normative reasons explain these prohibitions, those reasons must pertain to the doxastic stage of the two-stage model. Therefore, they cannot be explained in terms of reasons to wait or to postpone one's judgement.³⁴ In Schroeder's view, these reasons are primitive reasons against believing or disbelieving. However, if one does not endorse such primitive reasons against φ-ing that do not reduce to reasons in favour of some alternative option to φ-ing, one could easily construe his proposal in terms of reasons to suspend. Although there might be good reasons to endorse such a primitivist position on reasons against (Snedegar 2018), any proponent of FRL will have its problems with allowing for them. This is because the reasons in question lack the required link to the fittingness of belief or disbelief (Lee 2022; Roeber 2016). For instance, the fact that the evidence with regard to p is almost balanced is neither evidence that p nor evidence that not p. Thus, to account for prohibitive balancing, there is a good case to be made for reasons to suspend that are neither reasons to believe nor disbelieve.

One might object that these reasons to suspend are not independent in the sense required for the argument of the previous section to work. After all, the reasons to suspend judgement are facts about the reasons to believe or disbelieve. Hence, according to the current suspicion, I cannot use them in the argument of the preceding section. However, according to FRL, facts about the evidence can only amount to reasons to suspend if they exemplify the required link to the fittingness conditions of suspension. If something along the lines of SUSP is correct, suspension is fitting iff one is in dire epistemic circumstances. The fact that the evidence is close to being tied bears the required link to this condition of fit.³⁵ As I have argued in the previous section, however, the reasons to suspend will proliferate. Anything that bears such a link to the fact that one is not in a position to know will count as an independent reason to suspend. Even if a very narrow view on the fittingness of suspension is adopted, it amounts to the same result. For instance, assume that suspending on whether p is fitting iff the evidence in favour of or against p is close to being tied. In that case, any higher-order evidence that the first-order evidence is tied may, again, be an independent reason to suspend.³⁶

At this juncture, a proponent of FRL might reasonably complain that the commitment to independent reasons to suspend is not unique to their position. To round off this section, I demonstrate that facts about the evidence generate sufficiently independent reasons to suspend for a host of different positions which do not subscribe to FRL. There are two premises which my argument requires. First, I am targeting any analysis of prohibitive balancing in terms of the weighing of reasons which endorses a minimally evidentialist position in that, at least in some situations in which prohibitive balancing is at work, evidence that p is the only reason in favour of believing p and evidence that not-p is the only reason in favour of disbelieving p. Proponents of FRL usually endorse such a form of minimal evidentialism. However, independent of any talk about fittingness, this account of reasons to believe is quite popular among epistemologists. Second, the reasons to suspend which are supposed to account for the phenomenon of prohibitive balancing are quite weak when the evidence is decisive, and extremely strong when the evidence is balanced.³⁷ Let X be the weight of reasons to suspend. X is the value of a function f(E) which takes the whole set of evidence E as an input. The closer E approximates a balance of evidence, the more X increases. Thus, if the evidence is close to being tied, the reasons to suspend will outweigh the reasons to believe. A similar result will be achieved concerning the reasons to disbelieve. Therefore, suspension is the only permitted option; consequently, prohibitive balancing can be accommodated in this weighing model. However, as long as f(E) does not exhibit any jump discontinuities, my argument from the previous section still applies because there must be a point at which the reasons in favour of belief are as weighty as X.³⁸

To conclude, while some of the proposals for independent reasons to suspend might look doubtful, I have argued that the most paradigmatic reasons to suspend judgement might be sufficient for the previous section's argument to work. If one believes that one has a normative reason to suspend if the evidence is close to being tied, one must combine an appeal to FRL with an endorsement of epistemic permissivism. However, this commitment extends to many other positions that account for prohibitive balancing in terms of the weighing of reasons.

6. Conclusion

The goal of this paper was to propose three theses on the normativity of suspended judgement. First, I defended the idea that, even if one endorses the most prevalent take on the fittingness of belief, suspension can have satisfiable fittingness conditions. Fittingness is permissive; thus, there is no additional problem in allowing for situations in which (dis-)belief and suspension are both fitting at the same time. Second, co-satisfiable fittingness conditions for doxastic attitudes commit any proponent of FRL to the existence of some reasons to suspend which are neither reasons to believe nor reasons to suspend. These independent reasons to suspend, in turn, establish a form of epistemic permissivism. Finally, one can show that this commitment to independent reasons to suspend and to permissivsm is not peculiar to the proposal of co-satisfiable fittingness conditions for doxastic attitudes. On the one hand, defenders of both the truth view on fitting beliefs and FRL, insofar as they think that there are reasons to suspend at all, should be epistemic permissivists. On the other hand, there is a good case to be made that this commitment extends to many analyses of epistemic normativity in terms of reasons.

During the preceding discussion I left several things unquestioned and open for further inquiry. First, I assumed from the very beginning, both, that suspension is an attitude and that it might have fittingness conditions and normative reasons that speak in favour of it. Even though these are prominent assumptions in the literature, one might question them. Second, I left fittingness conditions for suspension completely unspecified. I merely defended the idea that suspension and belief could have co-satisfiable conditions of fit. Therefore, depending on which view on suspension one deems correct, one would need to specify the corresponding conditions of fit. Finally, the commitment to epistemic permissivism merits further scrutiny. If it turns out that this position is unbearable, one would be forced to concede that there are no independent reasons to suspend. This concession might, in turn, put many fittingness-first proposals and also a large number of reasons-centred theories in jeopardy.³⁹