Justification, Conformity, and the Norm of Belief

Abstract

According to a popular view in contemporary epistemology, a belief is justified if, and only if, it amounts to knowledge. Upholders of this view also hold that knowledge is the fundamental norm governing belief and that conforming to this norm is both necessary and sufficient for justification. I argue against the claim that mere norm conformity is sufficient for justification. Rather, justification requires norm conformity for sufficient undefeated reasons. An important consequence is the rejection of the claim that conformity to the norm of belief is necessary and sufficient for epistemic justification. I illustrate some interesting consequences of this result.

Selon une thèse populaire en épistémologie contemporaine, une croyance est justifiée si, et seulement si, elle est une connaissance. Les défenseurs de cette thèse soutiennent également que la connaissance est la norme fondamentale de la croyance et que la conformité à cette norme est à la fois nécessaire et suffisante pour la justification. Je conteste l’affirmation selon laquelle la simple conformité à une norme suffit à justifier une croyance. La justification exige la conformité pour des raisons suffisantes et «invaincues» (undefeated). Une conséquence de ceci est le rejet de la thèse selon laquelle la conformité à la norme de la croyance est nécessaire et suffisante pour la justification épistémique.

According to a popular view in contemporary epistemology, a belief is justified if, and only if, it amounts to knowledge.¹ This claim fits within a general externalist framework of epistemic normativity in which the notion of knowledge plays a central role. In this framework, a justified belief is a belief conforming to the fundamental norm governing this attitude. For a belief to be justified, conformity to this norm is both necessary and sufficient. Any beliefs that do not conform to this norm are not fully defensible from rational criticisms, and thus not fully justified—at best they can be blameless and excusable if they manifest a general disposition to comply with the norm. If only knowledge can justify belief, and norm conformity is necessary and sufficient for justification, then only knowledge can be the norm of belief. The resulting framework is one in which knowledge is the necessary and sufficient condition for a series of properties including epistemic justification, full epistemic defensibility, and conformity to the norm of belief.

This broadly externalist knowledge-based model of epistemic normativity has several virtues. For example, the model is particularly well placed to address traditional problems affecting externalist accounts of justification, such as the New Evil Demon Problem.² A main rationale for this model relies on parallels with the use of central normative notions in other normative disciplines, in particular in the practical and legal domains, where the notion of justification is most often employed and discussed. The suggestion is to look at how notions such as ‘justification,’ ‘reason,’ and ‘excuse’ are used in these domains, and theorize by analogy about the corresponding notions in the epistemic domain.³ This methodology relies on the assumption that normative frameworks in different domains share some deep similarities.⁴ This approach has the advantage of bypassing intricate epistemological debates muddled by technical terminology and marred by an excessive appeal to intuitions about specific cases,⁵ and allows epistemologists to avoid neglecting important and subtle normative distinctions, such as those between ‘justified,’ ‘rational,’ and ‘excused.’ The hope is that once these distinctions are appreciated in their full generality, real progress in the debate will be possible.

In this paper, I shall grant many of the assumptions of those who endorse this model: first, I will assume the same general framework and methodology; second, I will assume that the arguments advanced by these philosophers succeed in showing that justification cannot require anything less than norm conformity, that justification requires full defensibility from rational criticisms, and that only knowledge can grant full justification to belief. My contention is with the claims that mere conformity to a norm suffices for epistemic justification and that only knowledge can be the norm of belief. This is where, in my view, the equation

$$knowledge\, = \,full\,defensiblity\, = \,justification\, = \,norm\,conformity$$

fails, at the level of the last equivalence. I will show that the claim that norm conformity is sufficient for justification is not supported by analogical considerations about the use of the corresponding notion in the practical and legal domains. I will argue that there are very good grounds to reject this claim and endorse a view in which justification requires more than norm conformity. More precisely, I will suggest that justification requires conformity to a norm for sufficient undefeated reasons. This claim has received substantial support from many authors in other normative domains. While this conclusion concerns primarily practical and legal justification, one can mount an argument by analogy for the conclusion that a justified belief requires conforming to the norm of this attitude for sufficient undefeated reasons. Besides being of interest in itself, this result has very important consequences for the general model of epistemic normativity discussed by these philosophers. In particular, assuming that a justified belief requires knowledge, if justification can require more than conformity to the norm of belief, it follows that the norm of belief can be some property falling short of knowledge, such as truth.

Here is the plan of this article. In §1, I introduce the general externalist normative framework assumed by upholders of the knowledge-based model. I stress in particular a disagreement amongst philosophers endorsing this framework on whether justification requires mere conformity to a norm or something more than that. In §2, I present four arguments against the view that mere norm conformity is sufficient for justification. Some of these arguments also indicate that justification requires conformity for sufficient undefeated reasons. I also consider and address two possible objections to these arguments based on the alleged distinctions between personal and act justification and between act-type and act-token justification. In §3, I argue that, once we abandon the assumption that justification requires mere conformity to a norm, a familiar line of argument in support of the knowledge norm of belief fails. I also show that if we accept the claim that justification requires more than mere norm conformity, a similar line of argument turns out to support a norm of belief requiring a condition weaker than knowledge, such as truth. In §4, I sketch how a model of epistemic normativity can be developed from a truth norm of belief within the present externalist normative framework. I also stress some important differences between this model and the views of other epistemologists endorsing a truth norm of belief.

Two preliminary remarks are in order here. First, as I said above, the present discussion will assume that upholders of the knowledge-based model are right on several issues. I am aware that some of these assumptions are rather contentious. Philosophers disagreeing with such assumptions are free to read my conclusions as merely conditional. Second, for ease of exposition, my discussion in this paper will mainly follow the framework and terminology developed by Clayton Littlejohn, who has spent more effort than others in developing and defending the present model and methodology.⁶ With minor terminological modifications, the same arguments apply to other similar views.

1. The Normative Framework

The general normative framework assumed by upholders of the knowledge-based model is broadly inspired by works in contemporary practical philosophy and philosophy of law (e.g., Maria Alvarez, John Gardner, Joseph Raz, Judith Jarvis Thompson), as well as by more traditional works on justification and excuses (e.g., John Austin, Peter Strawson). This framework develops around the notions of norm, reason, justification and excuses.

Norm. The central notion in this model is that of a norm, where this notion is here used in a general sense not restricted to the moral domain. Norms in this broad sense include all kind of obligations, permissions, and prohibitions.⁷ Norms governing specific normative subdomains are issued by independent authoritative sources and characterize those subdomains: for example, norms governing morality specify moral normativity, laws specify legal normativity, and so on. Besides moral norms—such as the duty to keep one’s promises—typical examples of norms considered by these philosophers are rules of games and laws, norms whose conditions of satisfaction and application are fully external and objective.

An important distinction in this framework is between conformity and compliance to a norm. An agent conforms to a norm when she does what the norm requires and violates the norm when she doesn’t. An agent complies with a norm when, in addition to conforming to it, she does it because she is guided by that norm. Normative guidance here shouldn’t be conceived as involving full awareness of the norm. An implicit form of sensitivity to normative reasons is sufficient.⁸ We can conceive compliance with a norm as equivalent to conformity on the basis of reasons that the norm provides. One can conform to a norm without being guided by it, or even without intending to comply with it. If Cody promises Lara that he’ll go to the party tonight, he will conform to the norm of promise-keeping if and only if he does go to the party. No matter whether Cody intends to keep his promise and tries his hardest to keep it, or whether he reasonably believes he has kept it, if Cody doesn’t go to the party, he will break his promise. And, if Cody does go to the party, he will conform to the promise-keeping norm no matter the reason for which he went to the party.⁹

Reason. Reasons are facts or true propositions.¹⁰ Normative reasons are facts counting in favour of performing certain actions or holding certain attitudes and contributing to making these actions and attitudes right or appropriate. The ‘favouring’ relation is understood as objective and independent of the access of agents to reasons. That today is Sunday is a normative reason for Jack not to go to work, even though Jack doesn’t ‘have’ this reason, i.e., he doesn’t know that it is Sunday, and consequently cannot be motivated by that not to go to work. In the present framework, norms determine which facts count as reasons. For example, the moral duty to keep one’s promise makes the fact that Cody promised to Lara to go to the party tonight a reason to do what he promised. Motivating reasons are a subset of normative reasons. They are the reasons for which a subject performs an action or holds an attitude, and highlight the features that the agent took to make her actions or attitudes right, appropriate, or fitting.¹¹

Justification. While philosophers endorsing this framework agree on the notions of norm and reason, for justification the agreement is only partial. On the one hand, there is a general agreement that a necessary condition for justification is doing what is all things considered right or permissible: in order to be justified to φ, there must be undefeated sufficient normative reasons to φ. An agent who does what she ought not to do cannot be justified, no matter her perspective on the facts.¹² Justification to φ is compatible with there being pro tanto reasons not to φ. For example, someone killing in legitimate self-defence is justified to the extent that she did what she ought to do in her particular circumstance, even though she also had pro tanto reasons not to kill. Such reasons are not cancelled or undermined, but merely defeated, outweighed by other reasons that make it all things considered right to kill in these specific circumstances. The thought that justification requires doing what is supported by sufficient undefeated reasons can also be put in terms of norm conformity: when the norms governing a certain φ-ing do not generate conflicting demands, if S justifiably φ-s, S conforms to these norms.

On the other hand, these philosophers disagree on whether doing what is all things considered right or permitted (thereby acting in conformity to the relevant norms) is also sufficient for justification. Littlejohn explicitly identifies the conditions for justification and those for norm conformity. For example, he claims that “[s]omething is justified when there’s sufficient reason for it and we determine whether there’s sufficient reason for something by considering whether it conforms to the relevant norms”¹³ and “[t]o determine whether S’s φ-ing was justified, we have to ask whether S conformed to the norms that applied to her in φ-ing.”¹⁴ For Littlejohn, we are justified to φ simply by the presence of sufficient normative reasons to φ, no matter what our motivating reasons are.¹⁵ Here, Littlejohn’s view diverges from that of other philosophers in the practical and legal domain often quoted in his works. For example, Gardner explicitly denies that justification is mere norm conformity. After criticizing the idea that there are two notions of justification, one subjective and related to motivating reasons and the other objective and related to normative reasons, he focuses on the ‘fundamental conceptual’ question of whether justification depends upon normative (guiding) reasons or motivating (explanatory) reasons:

The answer, irritating but unavoidable, is that it depends upon both. No action or belief is justified unless it is true both that there was an applicable (guiding) reason for so acting or so believing and that this corresponded with the (explanatory) reason why the action was performed or the belief held. … [F]rom whatever point of view one claims justification for one’s actions or beliefs, one claims justification only if one claims both that there were, from that point of view, reasons for one to act or believe as one did and that one’s reasons for performing the act or holding the belief were among these reasons.¹⁶

According to Gardner, for being justified to φ, it is not sufficient that there are some undefeated normative reasons to φ. In addition, some of these reasons must also be the one(s) for which the agent φ-ed. When an agent φ-s justifiably, a normative reason to φ becomes also the reason that motivates the agent to φ and explains why she φ-ed. The present disagreement on the notion of justification reflects a more general dispute in criminal law between those considering justification to be a matter of mere conformity to norms, sometimes called ‘deed’ theorists, and those who think that justification requires more than that, such as conformity for sufficient undefeated reasons.¹⁷ In the next section, I will provide arguments against the former view and in favour of the latter.

Excuse. Justifications should not be confused with excuses. Excuses are appropriate only when one lacks justification. They are defences appealing to how one reasonably saw the world rather than how it really was, pointing to some non-culpable lack of information about the circumstances (blameless mistakes, misevaluation of the situation …).¹⁸ An agent is excusable for φ-ing if, even though there are not sufficient undefeated reasons for the agent to φ, she φ-ed on what she wrongly but reasonably took to be sufficient undefeated reasons for φ-ing.¹⁹

2. Is Justification Mere Conformity to a Norm?

In this section, I consider four arguments against the view that mere norm conformity is sufficient for justification. Some of these arguments also positively support a notion of justification as doing the right thing for sufficient undefeated reasons. The first three arguments are original; the fourth is Gardner’s. I also consider and address a couple of possible objections to the view that justification requires more than norm conformity.

Argument 1: Evidence from Legal Practice

Some preliminary support for the idea that justification requires more than mere norm conformity comes from one of the domains in which the notion of justification is more often used, namely, criminal law. As Gardner observes, “So far as I can see, our judges persist in using the word ‘justification’ to refer mainly to legal justifications proper, i.e., to legally recognized reasons for acting which were also the relevant agent’s reasons for acting in the case under consideration.”²⁰ Gardner mentions, as a paradigmatic example, the so-called Dadson doctrine, named after a real case involving a prisoner’s ‘unknown self-defence.’ The sentence says: “The prosecutor not having committed a felony known to the prisoner at the time when he fired, the latter was not justified in firing at the prosecutor.”²¹ As Gardner observes, this sentence implies a notion of justification according to which, in order to be justified, the prisoner firing upon the prosecutor must have at least known the prosecutor’s intentions at the time when he fired. It must be stressed here that even the critics of the view that justification is more than conformity, such as Paul H. Robinson, recognize that a dual requirement for justification (that the actor both performs the right deed and acts for the right reason) would get support from current law, which denies a defence to the unknowingly right actor who uses force against an attacker for the wrong reason.²²

Argument 2: Full Defensibility Requires More Than Conformity

A point on which many agree is that justification requires full defensibility from possible rational criticisms. According to Gardner, justification is a type of defence that can fully exculpate an agent for φ-ing (even if it can admit pro tanto reasons not to φ). An agent is fully justified to φ only when her φ-ing can be fully defended from possible rational objections. Only the unobjectionable is in no need of justification.²³ Littlejohn seems to agree on this point. For example, he claims that proper blame excludes justification. The facts in light of which someone can be properly blamed threaten the justificatory status of her attitudes or actions.²⁴ Some of the arguments of both Littlejohn and Timothy Williamson for the claim that the norm of belief cannot be anything less than knowledge rely more or less explicitly on this assumption.²⁵

This notion of ‘full defensibility’ doesn’t square well with the idea that justification is mere norm conformity. Rather, it fits well with the idea that full justification requires, in addition to the presence of sufficient reasons for the subject to φ, that the subject also φ-s for some of these reasons, for only in the latter circumstance could her φ-ing be fully defensible from possible objections. Mere norm conformity seems insufficient to alone make the agent fully defensible and rationally unobjectionable. If, for example, there are sufficient reasons for an agent to φ, but she takes there not to be sufficient reasons to φ, her φ-ing will be open to criticisms and thus not fully defensible, even though it was right for her to φ in the circumstances. Reflection about specific cases can help us see this point more clearly:²⁶

Reckless Driver. Ramon drives his car through a crossroad completely ignoring the presence of traffic lights on his way. As it happens, lights are green. Ramon’s action is permissible: the presence of green lights is a sufficient reason to cross the road. Even though Ramon did a permissible thing, his driving through the crossroad is not fully defensible. In particular, his action was reckless and not exempt from criticisms for not showing the due respect to street regulations.

Lucky Chess Player. Levon is playing a chess match in a tournament. It is his turn to move and he is seriously undecided between three possible moves. He is running out of time and must make a move in the next few seconds. He randomly chooses one move from amongst the three candidates. As it happens, that is the best move. However, he didn’t make that move because he took it to be the best. Levon’s move is not fully defensible and beyond criticism. Conformity to a standard of good moves doesn’t provide a full defence for Levon’s choice in the situation. His random move, though right, needs some excuse (e.g., ‘well, I didn’t choose on the basis of a well pondered choice, but I was in time trouble and had to make a move …’).

Accidental Promise Keeper. Elena has promised Mike to be in London on Monday. While she doesn’t make any effort to keep the promise, by mere accident, she happens to be in London on Monday. Although Elena did what she promised, one may still complain about her conduct, for she didn’t keep the promise for the right reason. Intuitively, Elena’s action is not fully defensible and exempt from criticisms. Apologies to Mike would make perfect sense in the circumstance.²⁷

The mere conformity view of justification entails that, since the agents in the above cases do what there is reason for them to do, they are justified, fully defensible, and beyond criticisms. In particular, this view delivers the verdict that, as long as the agents’ actions conform to the norms relevant in their circumstances, these agents and their actions can be as defensible as someone who in their place would have acted for good reasons. This is a very counterintuitive consequence of the view. As the above examples show, it seems obvious that the subjects in the above cases cannot claim full defensibility for their actions, which, although right, are clearly not exempt from rational criticisms. It seems also obvious that subjects in counterpart cases conforming to norms for good reasons could claim more defensibility than those in the present cases.

A related problem well illustrated by the above cases is the following. Many philosophers in the practical and legal domain consider justification a type of defence of which a rational subject could avail herself, along with other defences, such as excuses. Full defensibility presupposes that in normal circumstances a fully rational agent aware of her own motivations is in a position to fully defend herself from possible challenges. However, the agents in the above cases do not have this ability: Ramon, in Reckless Driver, completely ignores that there was a reason to drive his car through the crossroad, namely, that lights were green. He is unable to avail himself of this reason if someone challenged him asking why he passed through the crossroad. Similarly, both Levon and Elena, in Lucky Chess Player and Accidental Promise Keeper, cannot properly avail themselves of the reasons there were for doing the things they did. Indeed, it seems appropriate for them to defend their actions by putting forward excuses rather than justifications, and excuses are appropriate only in the absence of justification.²⁸

A possible reply in defence of the conformity view is that, while the agents in the above examples act in conformity to a norm, they are violating some other norm sanctioning, not the act, but the attempt or the intentions of the agent. These agents would be at the same time justified in virtue of conforming to the act-norm, but unjustified for violating the attempt-norm. This strategy was first discussed by Robinson, the main proponent of the conformity view. Robinson considers cases of agents doing the right things but for the wrong reasons. Here it is a variant on his well-known attacker-thought-to-be-a-jogger example (which is significantly similar to the Dadson doctrine considered above):

Attacker-Thought-To-Be-A-Jogger. Paul sees that what looks to him like a jogger is approaching. Actually, this person is not a jogger but a robber planning an attack on him. However, Paul is completely ignorant of this fact. To him the person is just a common jogger running there by chance. When the person is sufficiently close, Paul gives him a strong punch on the face, causing the person’s loss of consciousness, and thus preventing the aggression. However, Paul completely ignores the bad intentions of the person and the threat. His motives to punch the person are malicious and completely unreasonable (say, to test the strength of his punch on the first person he meets).

Robinson’s diagnosis of this type of case is that the use of force against the attacker-thought-to-be-a-jogger is fully justified, because Paul acts in conformity to the law permitting the use of force when this is necessary to prevent an imminent unlawful harm. However, Paul may be liable for an attempt to unjustifiably assault another person.²⁹

This diagnosis of the case is implausible for several reasons. The main problem is that it is implausible that someone can be punished for an attempted aggression when she actually succeeded in performing that aggression. In this circumstance, the attempt is inseparable from the performed action. There simply are not two things to evaluate: an action and an attempt, but a unique thing, an action, which constitutively involves a successful attempt. We cannot punish the attempt without punishing the action. At most, we can punish the action for the attempt it involves rather than for its consequences—i.e., we can explain why we punish the action by pointing to a specific aspect of it. But, contra Robinson, we should judge the action to be unjustified and we should punish it. The very same considerations apply if we substitute ‘attempt’ with ‘intention.’ Since in the above case Paul’s intention is an inseparable constituent of his action, we cannot punish the intention without punishing the action. Paul doesn’t merely try to use force on the passerby; he actually uses it. Hence, he is liable for assault, not merely for attempted or intended assault.³⁰ The existence of a real threat for Paul can at most provide reasons to mitigate the punishment on the ground that the action is partially defensible under a specific aspect of it, namely, for its consequences. Similar considerations are valid for the above examples: what is deemed not fully defensible (and thus unjustified) in all these cases is not the attempt or the intention to perform a forbidden action, but the very performance of an action which, though right, is not done for sufficient reasons.³¹

Argument 3: Justification Is Comparable, Norm Conformity Isn’t

Justification admits of comparisons. It is natural to compare the justifications that one has for doing different things or believing distinct propositions. For example, it is natural to say things like ‘I am more justified to believe that 2+2=4 than that Kathmandu is the capital of Nepal,’ or ‘when it comes to helping others, altruism is a better justification than egoistic reasons such as receiving credit for it.’³² In general, we tend to ascribe more or better justification to actions or attitudes done or held for better reasons. This is problematic for the conformity view, for conformity to a norm doesn’t admit of degrees or comparisons: if a norm requires one to φ, an agent conforms to it if she φ-s, and doesn’t if she fails to φ. One can be more or less blameworthy, rational, or excusable, but one cannot be more or less right, permitted, or forbidden. It is a well-known and recognized fact that permission, obligation, and forbearance do not admit of degrees or comparisons (either an action is obligatory or it is not),³³ and norm conformity is just a matter of doing what is obligatory or permitted and avoiding doing what it is forbidden. A conformity view equating justification and conformity doesn’t have a simple explanation of comparative justification.

It is helpful here to compare this to the doing-the-right-thing-for-sufficient-undefeated-reasons view of justification (hereafter, DRTSUR). The latter view can easily explain different strengths of justification, for according to this view the reasons for which one acts matter for justification: one can φ for sufficient reasons R, which can be better or worse compared to other sufficient reasons R* for φ-ing. In this perspective, an action can be more justified than another, for there may be two sets of reasons for which one may φ which are equally sufficient to φ, but one set of reasons is better than the other. For example, while altruism and receiving credit are both sufficient reasons to help others, and both can justify the action, helping others for the former reason is better than for the latter. In contrast, the conformity view cannot explain why doing something for certain reasons is more justified than doing it for others. According to this view, the reasons for which one performs the action do not play any justificatory role at all.

An upholder of the conformity view may reply that her view can account for justification comparisons in the following way. This view admits that, within a set of permitted actions, there can be more reasons to perform certain actions than others. Justification comparisons could then depend on the comparative reasons there are—reasons determining what it is all things considered right to do—rather than those for which one acted. The problem with this reply is that strength of justification doesn’t depend on the reasons there are, but on the reasons for which one acts. Compare two agents, A and B, for which there are all things considered sufficient reasons to use force against aggressors. However, while B uses force against her aggressor for sufficient reasons (she knows about the imminent threat), A completely ignores the threat and uses the force on the attacker for vicious motives (say, because she likes causing harm to people). If we had to compare justification based on the reasons there are, A’s use of force would be as justified as B’s. But it seems clear that, while both actions are right, B’s action is more justified. The obvious explanation is that B’s action is performed for sufficient reasons, while A’s action is not. This, again, favours the DRTSUR view of justification over a mere conformity one.³⁴

A related advantage of the DRTSUR view is that it is able to provide an account of the variable liability to punishment, criticism, and blame of the subject in different situations.³⁵ In particular, the account explains the grades of liability of an unjustified agent. For example, it can account for why doing the right thing for the wrong reason makes a subject liable to criticism, but to a lesser degree than one who does the wrong thing for the wrong reason. Since the account distinguishes wrong from unjustified deeds, it allows for associating different liabilities to each condition. Furthermore, the view can admit different specific liabilities, criticisms, and appraisals (e.g., punishments, fees, duty to show regret and apologize, propriety of feeling guilty and repent, praise …) depending on the specific features making the action unjustified (lack of conformity, consequences, badness of the motives, or strength of the reasons on which the agent acted). The alternative view collapsing justification on conformity cannot provide an equally simple explanation of the variability of liabilities and assessments. The latter view seems forced to accept an implausible collapse of assessments. For example, it would deliver full exemption from liability for the unknowingly right actor in Attacker-Thought-To-Be-A-Jogger, which would conflict with both ordinary intuitive assessments and law courts’ judgements.

Argument 4: Gardner’s Argument

Gardner provides the following argument against the conformity view and for the DRTSUR view. Gardner observes that “it is quite pointless to cite, by way of justification, an undefeated reason for which one did not act, even though it would have been alright for one to act upon it if one had been minded to do so.”³⁶ Gardner also observes that citing such a reason would be insufficient to justify one’s action. One would have to provide some further reason on which one acted in order to clinch the justification. However, once one provides such further reason, any other reason for which one didn’t act becomes irrelevant for the justificatory status. This shows that acting for sufficient undefeated reasons is both necessary and sufficient to justify one’s action.

Two Possible Objections and Replies

One may argue that the notion of justification is ambiguous between two senses: on the one hand, there is act justification, which is an appraisal of the act or attitude and bears on the mere conformity to normative demands; on the other hand, there is personal justification, which is an appraisal of the agent and her virtuous conduct and concerns also the motives and reasons for which the agent acts. This distinction would reduce the contrast between the two considered views about justification to a terminological dispute. Upholders of the conformity view could then argue that the notion relevant for epistemic justification is that of act justification, concerning belief rather than the believer subject. My reply to this objection is that the contrast between personal and act justification may be useful in specific technical contexts, but it doesn’t correspond to any important real distinction. As a matter of fact, in our ordinary talk, we don’t distinguish between an agent being justified to φ and the agent’s φ-ing being justified. In particular, there are no circumstances whatsoever in which we attribute lack of justification to the agent but not to her action. If Paul is not justified in claiming a reward, then also Paul’s claiming the reward is unjustified.³⁷

Someone may suggest alternative versions of the present objection.³⁸ One may hold that ‘justification’ is ambiguous between two senses, but the ambiguity does not concern the distinction between act (attitude) and agent’s justification. Rather, it concerns some other notions of justification expressing different specific kinds of evaluation of an act or attitude. For example, it may be suggested that a certain act is justified* in virtue of displaying conformity to the norm, or avoiding bad consequences, but the same act is not justified** in virtue of manifesting bad will, lack of responsiveness to reasons, and so on. My response to this objection relies on the observation that, as it has been argued in literature discussed in this and the previous sections, justification is that type of defence that requires full defensibility from possible criticisms. An act or attitude φ is fully justified only when it can be fully defended from any rational objections.³⁹ As Gardner observed, only the unobjectionable is in no need of justification.⁴⁰ Undoubtedly, our actions and attitudes can be assessed according to several standards such as correctness, rationality, blameworthiness, safety, and so on. However, full justification is different from any such specific standard in the following respect: while an action can conform to specific standards and still be the object of criticism for failing to conform to others, an action that is fully justified is fully defensible from any kind of criticisms, unobjectionable under any normative respect. Thus, a justified action cannot fall short of conforming to any particular normative standard.⁴¹

More plausible is the appeal to a distinction between the justification of a specific action and that of that type of action performed in relevantly similar circumstances. I think that this distinction tracks a real difference in our ordinary assessments of actions, attitudes, and agents. However, I don’t think it can be of any help for the upholder of the conformity view. The distinction is one between, on the one hand, the justification of an agent and her act in a particular circumstance, given the reasons for which she actually acted, and on the other hand, the justification of an action and a rational agent performing it in relevantly similar idealized circumstances. While a particular action φ could be unjustified if not performed for sufficient reasons, φ-ing would indeed be justified provided that a rational agent in similar circumstances φ-ed for sufficient reasons. So, for example, when we say that the use of force is justified when one is under serious threat, we mean that an agent’s using the force for defending herself from a serious threat would be justified. This is compatible with some agent’s use of force not being justified in similar circumstances, as in the Attacker-Thought-To-Be-A-Jogger case. In this respect, justification assessments when attributed to an act-type seem clearly derivative from justification assessments applied to particular actions. Obviously, this distinction doesn’t constitute any real threat for the DRTSUR view since, in both type and token assessments, justification depends on both doing the right thing and doing it for sufficient reasons.⁴²

It is also worth stressing that justification attributions to act-types are less common than to particular acts. While we say that the driver’s stopping at the red lights was justified, we usually don’t say that to stop at the red lights, qua type of act, is justified. More precisely, while it is appropriate to attribute permissibility and rightness to a type of action in abstraction from the reasons for performing it, it sounds odd to attribute justification to an action viewed without any connection to the agent’s reasons for performing that action. A simple explanation is that justification depends on the reasons for which an action or attitude is performed, and when assessing an act-type sounds particularly odd, this is because it is left indeterminate whether that act-type is conceived as performed for sufficient reasons. In the rare cases in which we attribute justification to a type of action, we tend to conceive that action as performed by a rational agent who would perform the action for sufficient reasons.⁴³

3. Justification, Norm Conformity, and the Knowledge Norm

In §2, I argued that justification requires more than norm conformity. This conclusion blocks a familiar line of argument for the knowledge norm briefly sketched in the introduction. We can reconstruct the argument as follows:

1. 1) A belief is justified if and only if it is fully defensible from rational criticisms (assumption)

2. 2) Knowledge is necessary and sufficient to make a belief fully defensible (assumption)

3. 3) Knowledge is necessary and sufficient to justify a belief (from 1 and 2)

4. 4) For any subject S and action or attitude φ, it is justified for S to φ if and only if S conforms to the relevant norm governing φ-ing (assumption)

5. C) S’s belief that p conforms to the norm of belief if and only if S knows p (from 3 and 4)

Let’s assume, for the sake of argument, that premises 1 and 2 are true. We have already discussed motivations for premise 1 in §2. The main motivation for premise 2 adduced by upholders of the knowledge-based model is that, on the one hand, any epistemic condition falling short of knowledge is somewhat defective and open to criticisms, and on the other hand, nothing more than knowledge seems to be required for a belief being fully defensible (at least from an epistemic perspective).⁴⁴ Let’s also grant that 3 follows from 1 and 2. My objection to the argument concerns 4. Alleged support for this premise would come from parallels with the use of the notion of justification in other normative disciplines. However, as argued in §2, such parallels indicate that conformity to the relevant norms governing an action or attitude is not sufficient to justify that action or attitude. Something more than that is needed, such as conforming for sufficient undefeated reasons. We should thus substitute 4 with the weaker 4*:

1. 4*) For any subject S and action or attitude φ, it is justified for S to φ only if S conforms to the relevant norm governing φ-ing (assumption)

From 4* and the other premises, we can conclude that knowledge is a sufficient condition to conform to the norm of belief, but not a necessary one. This conclusion is compatible with the norm of belief being some condition weaker than knowledge. The above argument for the knowledge norm of belief fails.

Furthermore, a similar line of argument turns out to support a norm of belief requiring a condition weaker than knowledge. In §2, I have argued that justified φ-ing requires more than mere conformity to the relevant norm governing φ-ing—it requires conformity for sufficient undefeated reasons.⁴⁵ If so, mere conformity to the norm requires a condition weaker than justification: the condition for justification implies the condition for norm conformity, but not vice versa. Accordingly, we should expect that the condition required by the norm of belief would be weaker than the one necessary and sufficient for epistemic justification. A belief can conform to this norm without being justified. If we keep the claim that knowledge is necessary and sufficient to justify a belief (3, derived from 1 and 2 in the previous argument), we reach the conclusion that the norm of belief requires some condition weaker than knowledge, such as truth. The argument can be reconstructed as follows:

1. 3) Knowledge is necessary and sufficient to justify a belief (from 1 and 2)

2. 4⁺) For any action or attitude φ, condition C required for conformity to the relevant norm governing φ-ing is weaker than condition J required for justified φ-ing (J implies C but not vice versa) (assumption)

3. 5) The condition required by the norm of belief is weaker than the condition required for justified belief (from 4⁺, instantiation)

4. C⁺) The norm of belief requires a condition weaker than knowledge (from 3 and 5)

4. Toward a Truth-Based Model of Epistemic Normativity

Until now, I have argued that it is wrong to identify justification with mere norm conformity, as upholders of the knowledge-based model do. I have also shown that this conclusion is problematic for a familiar line of argument supposed to support the knowledge norm, and that it rather lends support to a norm of belief requiring a condition weaker than knowledge, such as truth. In this section, I consider an alternative model of epistemic normativity, which develops within the general externalist normative framework introduced in §1, but which assumes a truth norm of belief. A truth norm requires that one believe what is true, or only what is true.⁴⁶ The aim is not to provide a full defence for this truth-based model. My more modest goal is to illustrate how an alternative model of epistemic normativity can be built within this framework, compatible with some central assumptions of knowledge-based models and preserving their externalist core and many of their insights and advantages, but leading to different conclusions about a number of epistemic properties.

The interesting aspect of a truth-based model is that this model doesn’t lead to a collapse of the conditions for several normative properties on a unique condition, as the knowledge-based model does. In particular, within this model, it is possible to distinguish between, on the one hand, the condition for norm conformity (truth) and, on the other hand, the condition for conforming for sufficient undefeated reasons, full defensibility and epistemic justification (knowledge). The truth-based model can leave open the possibility of conforming to the truth norm without doing it for sufficient reasons. An example is when a subject believes some true proposition p but her belief isn’t based on sufficient evidence. In this case, the belief that p is permissible but not fully defensible, and thus not justified.

If we take seriously the idea that epistemic reasons are all and only the reasons that the norm of belief provides, the adoption of truth as the norm of belief leads to a form of reason infallibilism.⁴⁷ If, for example, the norm requires believing only the truth, it provides sufficient (undefeated) reasons to believe a proposition p only if p is true. It is impossible that S believes p on the basis of sufficient reasons R, and yet S’s belief that p is not true. If the belief had been false, there would have not been sufficient reasons to believe it—in exactly the same way in which if someone thinks that she made a promise but she didn’t, there would be no reasons to do what she believes she has promised (at least none issued by a promise-keeping norm). Thus, according to the truth-based model, the only reasons (not) to believe p are the truth (falsity) of p or any truth q guaranteeing the truth of p.⁴⁸

A consequence of the adoption of such a truth-based model concerns justification. It is plausible that, within this framework, if a subject S believes a proposition p for sufficient undefeated reasons, she also knows p. This is because, as I said above, if there are sufficient undefeated reasons to believe p, these reasons guarantee the truth of p, and if S believes p on the basis of these reasons, then she also knows p. If this is true, then within this model we can maintain, with upholders of the knowledge-based model, that a belief that p is fully defensible and justified only if the subject knows p. This allows for preserving a neat distinction between assessments in ‘good cases,’ in which the subject knows and is justified, and ‘bad cases,’ in which the subject’s belief falls short of knowledge and is not justified—at most, it can be blameless and excusable if it is held reasonably. However, the truth-based model admits a richer spectrum of assessments compared to the knowledge-based one: assessments of permissibility, rightness, and correctness are associated with norm conformity (viz., true beliefs or the absence of false beliefs), assessments of excusability are associated with rational beliefs falling short of knowledge, and assessments of full defensibility and justification are associated with beliefs that conform to the norm for sufficient undefeated reasons, which in the picture I’ve just sketched amount to knowledge.⁴⁹

The present model has the resources to address a popular line of argument against the truth norm of belief. Roughly, according to this argument, only a knowledge norm can account for both objective and subjective, outward and inward looking assessments of belief and explain why a belief that falls short of knowledge is epistemically defective.⁵⁰ This argument presupposes that (i) nothing less than knowledge can grant a belief’s full defensibility from epistemic criticisms and (ii) that norm conformity is sufficient to grant full defensibility. From these assumptions, the argument concludes that no epistemic condition weaker than knowledge can be the norm of belief. As argued in §2, premise (ii) is false: full defensibility, as justification, requires more than norm conformity. We can thus maintain that the norm of belief is some weaker condition than knowledge, such as truth, while also maintaining that only knowledge can grant full epistemic defensibility and justification.

Before concluding, I would like to stress that the truth-based model of epistemic normativity sketched in this section doesn’t differ only from the knowledge-based model, but also from traditional models identifying the norm of belief with truth.⁵¹ As a matter of fact, the respects under which the present model differs from other traditional truth-based ones are much more significant than the differences between this model and the knowledge-based one. I will mention here only what I take to be the most striking dissimilarities. Traditional truth-based views often do not distinguish between justification and rationality, and systematically take these properties as compatible with beliefs falling short of norm conformity. In contrast, in the present truth-based model justification and rationality are radically different assessments playing distinct roles in our evaluative practices: while rationality concerns the responsiveness of the subject to apparent reasons (what look like reasons from the subject’s perspective)⁵² and is compatible with the existence of rational beliefs that do not conform to the norm of the attitude, a justified belief requires full defensibility and conformity to the norm for sufficient undefeated reasons. A related difference is that the present truth-based model, like the knowledge-based one, preserves a disjunctive account of epistemic assessments in good cases (in which the subject knows and is justified) and bad cases (in which a belief falls short of knowledge, is not justified, and can at most be excused). On the contrary, traditional views do not distinguish between justification and excuses and admit identical normative assessments in good and bad cases.

In the present section, I have provided a rough and incomplete picture of what a general model of epistemic normativity would look like if we adopt the general externalist framework considered in §1 and take truth to be the norm of belief. Many arguments deployed for a knowledge-based model can be used to support the present model. In particular, parallels with the use of corresponding normative notions in other normative disciplines provide a strong case for this model. No doubt the present account is very sketchy and imprecise in several respects, and several problems not considered in this paper need to be addressed. My aim in this section was just to illustrate a possible alternative to the knowledge-based model of epistemic normativity, based on the same externalist normative framework and enjoying most of the same virtues, but preserving a difference between conditions for justification and for norm conformity. Investigating further the merits and shortcomings of this model is a task for future occasions.