Why should we care about following epistemic norms – such as the norm that we should base our beliefs on adequate evidence?Footnote 1 One answer is that we care about epistemic norms because we care about having true beliefs. But what makes truth so special for us? Here is where the controversy begins. Instrumentalists hold, very roughly, that following epistemic norms will lead to other things we place value in – such as that we will survive or be able to engage in practical reasoning.Footnote 2 Since, as human beings, we should care about these things, we should also care about following epistemic norms. Another view aims to ground epistemic normativity in the constitutive features of belief.Footnote 3 Proponents of this view hold that beliefs are by their very nature directed toward the goal of truth. If an agent fails to base her beliefs on factors having to do with the truth, she will either fail to possess the concept of belief or cease to be a genuine believer. On still another view, our concern for truth is grounded in the constitutive conditions of agency. According to one such proposal, which was developed by Veli Mitova (Reference Mitova, Grajner and Schmechtig2016), the hallmark of agency is the drive to sense-making: we cannot act unless the world makes sense to us. But in order for the world to make sense for us, we need to respect epistemic norms. Therefore, epistemic norms have authority over us. There are even more proposals on offer.Footnote 4
In this paper, I explore an alternative view about the source of epistemic normativity. The view takes inspiration from Christine Korsgaard’s (Reference Korsgaard1996, Reference Korsgaard2009) account of the source of moral normativity. As Korsgaard (Reference Korsgaard1996) has stressed with respect to moral reasons, reasons gain their authority from being grounded in the self-identities of agents. If I take a reason for some course of action as authoritative, then this reason conforms to those things I take to be valuable and that form part of my identity. For these identities, Korsgaard (Reference Korsgaard1996: 83) has coined the term “practical identities,” which she conceives of as “description(s) under which you value yourself, under which you find your life worth living and your actions worth undertaking.” The value of an agent’s self-image is, on Korsgaard’s view, ultimately grounded in the value of humanity: a presupposition of valuing one’s self-identity is that one values humanity in general. Korsgaard claims that only by appealing to the value of humanity, which is unconditionally valuable, might we account for the normative status of a given practical reason. She argues that appealing to something that only possesses conditional or instrumental value will open the door to moral skepticism. In case an agent possesses a conditional reason to follow moral norms, this agent might still ask why moral norms have authority over her. By contrast, Korsgaard’s account allows us to escape this problem because it provides the agent with an unconditional reason to comply with the demands of morality.
In this paper, I explore the prospects of a Korsgaardian account of the normative authority of reasons for beliefs. This view has not been considered either in Korsgaard’s writings or in the recent philosophical literature. I will argue that this account is a serious competitor to the alternative accounts on offer. I proceed as follows. In the first section, I will clarify some key concepts. In the second section, I will first introduce Korsgaard’s account of the source of moral normativity and then consider how her proposal might transfer to the epistemic domain. Along the way, I will also address some concerns that have been presented against Korsgaard’s position. In the third section of the paper, I will take a closer look at two objections that might be leveled against the account developed herein. In the final section of this paper, I critically assess competing Kant-inspired accounts of epistemic normativity. Despite their merits, I argue that none offers a genuine alternative to the Korsgaardian view defended in this paper – mainly for the reason that they all fail to ground epistemic normativity in something of intrinsic value.
1. Preliminaries: Setting the stage
What are epistemic norms? To a first approximation, epistemic norms are norms prescribing epistemically good processes of belief formation, maintenance, and revision. Epistemic norms are norms that also tacitly guide agents in the way they form beliefs. One example of an epistemic norm is what might be called “the evidence norm,” which claims that you should proportion your beliefs to the evidence available to you.
In addition, believers might be guided by epistemic norms, which demand not to neglect counterevidence and to be consistent, respectively.
If agents and their beliefs are subject to norms, the question arises as to how to specify the exact form and content of these norms.Footnote 5 Moreover, alongside traditional epistemic norms governing beliefs, some have argued that there are zetetic norms that govern inquiry.Footnote 6 Recent work on zetetic epistemology has raised questions about the relationship between epistemic and zetetic norms, such as whether some epistemic norms are also zetetic norms or whether traditional epistemic norms might be in tension with zetetic norms.Footnote 7 I will not go into these issues here, since they are not relevant for the main purposes of this paper.Footnote 8
But what are we looking for when we ask for the source of epistemic normativity? A first interpretation of this question might be that we are simply seeking an answer as to why agents should follow epistemic norms.
However, some have opted for a stronger reading of the question of what grounds epistemic normativity. Hilary Kornblith (Reference Kornblith1993: 363) writes:
If you tell me that a belief of mine is unjustified, this gives me reason to give up that belief. The epistemic claim is something about which I should care, and an account of the source of epistemic norms must explain why it is that I should care about such things.
The quote from Kornblith suggests that an account of epistemic normativity should offer an explanation of why an agent should care about following epistemic norms.
Should we opt for the weaker or the stronger reading? I think we should address the stronger reading (NORM-SC). The main reason is that only this reading will do justice to the characteristic of epistemic norms that they are reason-giving for an agent—that is, that an agent who complies with or contravenes epistemic norms in forming a belief thereby has a reason for holding or abandoning this belief. And an account of the authority of epistemic norms should explain this feature of epistemic norms.
If we adopt the view that epistemic norms possess reason-giving force, an account of the authority of epistemic norms should satisfy at least two constraints, which are widely accepted in metaethics and which equally translate to the epistemic domain.Footnote 9 The two constraints claim that if x is reason-giving for an agent, then (i) x motivates an agent and (ii) x has authority over agents.
I find both constraints plausible. Yet, for those who remain unconvinced, let me briefly explain why I think we should accept them. Both constraints flow directly from what it means to possess a reason.Footnote 10 As Bernard Williams (Reference Williams and Williams1981) has argued, a reason for an action must be able to motivate an agent to engage in that particular action. Williams claims that the motivational force of reasons results from the fact that reasons are always reasons for someone.Footnote 11 If an agent’s reason could not motivate an agent, then this reason would not be able to explain an agent’s action on the basis of her particular reason. This also applies to the case of beliefs. Think of cases of inquiry, which are equally governed (at least in part) by epistemic norms like (Evidence). If I fail to have sufficient evidence for a particular hypothesis, I will not place belief in that particular hypothesis. Part of the reason as to why I will believe that hypothesis is that epistemic norms motivate me to comply with them. By contrast, in case I have sufficient evidence for a particular hypothesis, this will license belief in the hypothesis in question. As with the case of actions, epistemic norms possess motivational force.
I have chosen cases of inquiry to illustrate the idea that epistemic norms motivate agents. I wish to note, though, that some might consider controversial the view that norms pertaining to beliefs possess motivational force. One might argue that in order for this view to be intelligible, we need to presuppose the possibility of genuine cognitive agency – that agents can exercise control over their beliefs, just as they can exercise control over their actions. However, only a very narrow class of beliefs is susceptible to direct voluntary control. There is some discussion in the recent literature of how to conceive of motivation in the epistemic domain and the corresponding phenomenon of cognitive agency.Footnote 12 However, I do not have space to go into these issues here.
Turning to the second constraint, there are examples that illustrate the close tie between reasons and normative authority. Consider the following case. Suppose you intend to drink a glass of gin but have good reason to believe that the glass in front of you contains petrol instead of gin. You thereby have a good reason to refrain from taking a sip from the glass. Your belief that the liquid you see in that glass in front of you is petrol has authority over you.Footnote 13 Likewise, if epistemic norms are reason-giving, they will possess authority over agents. Thus, if we wish to explain the reason-giving force of x, we need to account for the authority of x as well.
2. A Korsgaardian account of the source of epistemic normativity
2.1. Korsgaard on the source of moral normativity
In The Sources of Normativity, Korsgaard aims to answer the basic question of morality, “Why be moral?” As Korsgaard observes, this question threatens morality with a regress. On Korsgaard’s view, we need an endpoint, “an unconditional reason,” where our justifications to follow moral norms come to an end. The demand for an unconditional reason is particularly pressing when the object of morality requires a difficult sacrifice on the part of the agent. In such cases, an agent might refrain from fulfilling the demands of morality. An answer to the moral question, on Korsgaard’s view, should be able to dispel any doubts an agent might have about acting upon a moral reason.
Korsgaard’s answer to the moral question proceeds in three steps, two of which are relevant to the topic of the present paper. First, she observes that human consciousness is essentially reflective. When an agent is confronted with a mental state, such as an impulse, this agent might step back and consider whether to take this particular mental state as a reason for some action. As Korsgaard observes, in comparison to other animals, humans have the ability to distance themselves from their impulses and reflect on them.Footnote 14 By reflecting on her mental states, an agent arrives at a conclusion as to whether she should endorse a given mental state as a reason for an action. From the perspective of a naturalistic worldview, some might be inclined to deny that reasons are a genuine feature of the way we humans engage with the world. However, Korsgaard asserts that reasons are an essential part of practical life; to deny reasons, she claims, would be like “denying the reality of color.”
But which principles guide agents in conscious reflection? Which moral reasons an agent will endorse are, according to Korsgaard, determined by a principle that has its roots in the autonomy of the will. For Korsgaard, a reason (or maxim) qualifies as moral if it conforms to “the Kingdom of Ends,” that is, if all rational beings would agree to act upon this reason (or maxim) in a worldwide cooperative system. The principle Korsgaard has in mind corresponds roughly to one of the formulations of the categorical imperative we find in Kant.
In the second step of her argument, Korsgaard argues that moral reasons, which are determined by the principle of the Kingdom of Ends, gain their authority from being grounded in the agent’s self-identity. Korsgaard points out that in the case of reasons for actions, we tend to reject certain courses of action and the corresponding reasons by saying to ourselves, “I couldn’t live with myself if I did that” (Reference Korsgaard1996: 86). In her view, these reactions stem from the authority or value we assign to our self-identities. As Korsgaard (Reference Korsgaard1996: 101) puts it, “Your reasons express your identity, your nature; your obligations spring from what that identity forbids.” For reasons for actions, Korsgaard calls the identities in question “practical identities,” which she considers value-laden ways an agent conceives of herself. As Korsgaard (Reference Korsgaard1996: 83) explains:
The conception of one’s identity in question here is not a theoretical one, a view about what as a matter of inescapable scientific fact you are. It is better understood as a description under which you value yourself, a description under which you find your life to be worth living and your actions to be worth undertaking. So I will call this a conception of your practical identity.
She goes on to argue that the practical identity of an agent comprises several such identities, such as being a human being, being a man or a woman, or being a member of a certain ethnic or religious group. All these identities will give rise to reasons and obligations. But Korsgaard claims that “it is the conceptions of ourselves that are most important to us that give rise to unconditional obligations. For to violate them is to lose your integrity and so your identity, and no longer to be who you are” (Reference Korsgaard1996: 84). She identifies as the most basic self-conception of ours, our identity as human beings. In order to assign value to any conception we have of ourselves, we need to value ourselves as human beings first. She further claims that if we value ourselves as humans, we thereby value humanity in general.Footnote 15
If broken down into individual steps, her argument might be put as follows:
-
(1) Reasons for actions gain their authority by appealing to our practical identities.
-
(2) Our practical identities need to be justified themselves in order to confer a normative status on a particular reason.
-
(3) Practical identities are anchored in the value of humanity: an agent is committed to valuing humanity in general in order to value her practical identity.
-
(4) Valuing humanity constitutes an unconditional reason for an agent.
-
(5) Therefore, our practical identities are justified.
Armed with this argument, Korsgaard claims to be in a position to provide an answer to the moral skeptic: the value of humanity serves to halt the regress of reasons because valuing humanity constitutes an unconditional reason.
However, there are reasons to be critical of Korsgaard’s argument.Footnote 16 What I take to be the Achilles’ heel of Korsgaard’s argument is premise (3). It is a live possibility for an agent to value her practical identity without placing any value in humanity. Such an agent might simply not take the unconditional value of humanity for granted and might ask, “Why should I value humanity?” Korsgaard is aware of this possibility – and the resulting consequence that it will jeopardize her argument. She offers two replies. First, she claims that we cannot justify the value of humanity by philosophical argument alone. She urges that humanity only possesses value for us if we find life worth living. But whether we do find our life worth living will depend on how our life looks to us. Consider, as an example, a manically depressed person. This person might fail to assign any value to humanity, while most of us, who do not suffer from chronic depression, will value humanity. Korsgaard says:
The price of denying that humanity is of value is complete practical normative scepticism … Is there an argument against that kind of scepticism, a reason not to commit suicide? In one sense, the right reply is that there can’t be, since life itself is the source of reasons. In another sense, the right reply is that this is not an issue to be settled by philosophical argument alone … [V]alue only exists if life is worth living, and that depends on what we do. (Reference Korsgaard1996: 163)
But will this argument convince someone who doubts that humanity possesses unconditional value? Some critics object that Korsgaard’s argument is incompatible with her starting point: if valuing our lives would be beyond the need for justification and could halt the reflective regress, then apparently the same would be true of our practical identities since they seem to encode everything we value about our lives.Footnote 17 As a result, this might be a reason to doubt that Korsgaard’s first argument for the value of humanity will establish its conclusion. Therefore, it seems we need another justification for the value of humanity.
Korsgaard presents a second, somewhat stronger argument for the value of humanity. Her second argument is a transcendental argument that aims to demonstrate that agents need to acknowledge the value of humanity in order to act upon a reason. As Korsgaard argues, rejecting the idea that humanity possesses unconditional value would lead to a form of practical skepticism. She writes:
Since you are human, you must take something to be normative, that is, some conception of practical identity must be normative for you. If you had no normative conception of your identity, you could have no reasons for action, and because your consciousness is reflective, you could then not act at all. Since you cannot act without reasons and your humanity is the source of your reasons, you must endorse your own humanity if you are to act at all. (Reference Korsgaard1996: 92–3)
Let us call this argument the “Transcendental Argument for the Value of Humanity”:
-
(1) An agent cannot act without having normative reasons.
-
(2) The normativity of a reason is grounded in the agent’s practical identity.
-
(3) The value of an agent’s practical identity is dependent on the value of humanity: in order for an agent to value her practical identity, she needs to value humanity.
-
(4) Therefore, the value of humanity is the source of the normativity of practical reasons.
-
(5) Therefore, an agent cannot act unless she values humanity.
However, in this dialectical setting, premise (3) appears to be question-begging: from the fact that we value our practical identity, it does not follow that we necessarily need to value humanity.Footnote 18 It might be the case that an agent just takes her own identity as authoritative on no basis at all. Unless we have an independent argument for why we should value our identities only on the basis of the value of humanity, Korsgaard’s argument does not seem to provide an answer to the challenge posed by the reflective regress. I will address these concerns in the next section. I will outline a different view from Korsgaard’s of how to conceive of the relation between self-identities and the value of humanity. Roughly, my idea will be to state the relation between the value of humanity and the value of our identities in terms of a relation of normative grounding. The resulting view still preserves the main idea of Korsgaard’s proposal but avoids the objections just considered.
2.2. A Korsgaardian account of the source of epistemic normativity
Do Korsgaard’s ideas transfer to the epistemic domain? I will now explore how Korsgaard’s ideas might provide an answer to the question of what grounds the authority of epistemic norms. I will stick to the two main steps of her argument as sketched in the previous section.
To come to the first step of Korsgaard’s argument, notice that what she calls “reflective consciousness” is a phenomenon we also witness in the epistemic domain.Footnote 19 We also reflect on our beliefs and consider whether they are held on the basis of reasons or evidence. Admittedly, few beliefs we hold are backed up by reasons that also survive processes of conscious reflection. Perhaps the main portion of our beliefs results involuntarily from engaging with the world. It seems, therefore, that deliberating on the epistemic credentials of beliefs is limited to a very narrow class of cases: we reflect consciously, for example, on those beliefs for which we encounter counterevidence or perhaps on those that greatly matter to us practically. Nevertheless, there are close parallels between the practical and the epistemic domains in that we also witness cases of conscious reflection regarding those reasons that speak in favor of our beliefs.
In comparison to the moral domain, however, reflection in the epistemic domain is guided by very different principles. It is plausible to suppose that there is a set of epistemic norms that underlie our epistemic assessments and also guide agents in processes of conscious reflection. Examples of epistemic norms might be those given above – that you should only believe upon sufficient evidence or that you should not neglect counterevidence. Epistemic norms, therefore, contrast with Korsgaard’s Kantian principle of the “Kingdom of Ends.” While Korsgaard claims that a Kantian account of the content of moral norms will imply a form of “constructivism” about morality – which avoids commitment to realism about moral facts – it is less clear whether it makes sense to understand epistemic facts as ultimately the products of the human will and hence as so constructed.Footnote 20 Note that the account of epistemic norms needed in the present context is not committed to a particular conception of the nature of epistemic facts.
The second step of Korsgaard’s argument from above claims that reasons gain their authority from being grounded in the agent’s self-identity. Now, one might ask whether, with respect to epistemic reasons, the self-identities of agents do play a similar role as in the moral case. Note first that, as with the moral case, we tend to reject certain beliefs and the corresponding reasons by saying to ourselves, “I couldn’t live with myself if I believed that.” Consider, as an example, belief in the non-existence of climate change. If I deliberate on whether I should believe that there is no man-made climate change, my initial reaction would be to flat-out reject this particular belief and consider adopting this belief as being inconsistent with who I am. It is part of my identity as a rational being to base my beliefs on adequate evidence, and I conform to this identity if I don’t disregard evidence from trustworthy sources, such as expert testimony. Believing upon insufficient evidence or shallow reasons will be in conflict with this part of my identity.
It is natural to conceive of the identity that underlies our reactions in cases like these as our epistemic identities or identities as rational believers. While practical identities, as Korsgaard conceives of them, concern how we act, epistemic identities might be understood as regulating which evidential states we should take at face value or what we should believe. In comparison to our practical self-identities, which also comprise idiosyncratic identities, such as that we are members of certain ethnic or religious groups, epistemic identities do not seem to display an equal amount of variation among individuals or groups of individuals. Yet, factors such as expertise or socio-economic status arguably impact epistemic identities and will, therefore, also cause a variation in epistemic identities among individuals.
But the crucial question is, whence do epistemic identities gain their authority? I can think of the following options:
If we consider each of these options in light of the constraints introduced above, I think the most promising option is (Humanity). That is, the authority of epistemic identities flows from being part of our identity as human beings and, therefore, by being ultimately grounded in the value of humanity. Let me now briefly explain why I think we should opt for (Humanity).
Consider first (Brute). Recall the two constraints that I introduced above. The first constraint claims that an account of the authority of epistemic norms should be able to explain the motivational force of epistemic norms, while the second constraint has it that epistemic norms have authority over agents. An account of the source of epistemic normativity should therefore be able to answer the question of why agents should care about following epistemic norms. It remains questionable whether (Brute) satisfies both of these constraints. One might concede that epistemic identities have authority over agents for the simple reason that they are self-identities and that they might, for this matter, possess motivational force for an agent. However, the idea is much harder to sustain that epistemic identities merit caring about on their own or in an intrinsic or unconditional sense. It seems that this view faces Smith’s fetishism charge, in that we will have the wrong attitude toward the object of merit, just as in cases of sexual fetishism. Therefore, (Brute) does not seem to fare well as an answer to the question as to why agents should care about following epistemic norms.
Now, how about (Pragmatic)? The idea of this proposal is that in order to engage in practical reasoning, I need to follow epistemic norms. Suppose I have the desire to grab a cup of coffee right now at a café close to my office. However, I remember that the café closes in about half an hour. I conclude that I should leave my office now in order to grab a coffee. I take action and leave my office. But in order to do so, I have to rely on epistemic norms – in particular, the norm of believing upon sufficient evidence. Does (Pragmatic) fare any better than (Brute) as an answer to the question as to why an agent should care about following epistemic norms? I think it does. The simple reason is that agents should care about agency and, for that matter, should care about following epistemic norms. Therefore, (Pragmatic) seems to satisfy both constraints that I introduced at the beginning of this paper. Note, however, that (Pragmatic) will only provide an agent with an instrumental or conditional reason to follow epistemic norms: agents should care about agency because agency is instrumental to other goods. Agency itself does not possess any unconditional or intrinsic value. Yet, if we adopt Korsgaard’s demand that an account of the normative authority of a reason should be able to end the reflective regress, then (Pragmatic) does not seem to be the solution we are looking for.
How about (Humanity)? (Humanity) claims that agents should follow epistemic norms because their epistemic identities are part of their identity as human beings. And their identity as human beings will possess authority over them because this identity is grounded in the value of humanity. In the spirit of Korsgaard, one might argue that since humanity possesses unconditional or intrinsic value, the regress of justification for taking a reason to be normative thereby comes to an end. Note that (Humanity) provides a straightforward answer to the question of why agents should care about epistemic norms: since agents should care about their identity as human beings and about humanity, they should also care about epistemic norms. (Humanity) thus seems to satisfy the demands of an explanation for why agents should follow epistemic norms.
But recall my discussion of Korsgaard’s argument for the value of humanity from above. There appears to be a lacuna in her argument in that it might be an open question for some agents as to why they should place any value in humanity at all, even though they might still value their practical identity. Consider Korsgaard’s response. She first claims that we will value humanity if we find life worth living. Her second response is centered around the idea that valuing humanity is a precondition for valuing anything at all. However, as we have seen above, both responses fail to be conclusive.
However, if we conceive of the relation between the value of humanity and the value of our identity as human beings in an alternative way, we might escape the objection that Korsgaard’s defense of the value of humanity faces. Think of the relation between the value of humanity and the value of our identities as being a case of normative grounding: on this proposal, our epistemic identities, which are part of our identity as human beings, will gain their authority by being grounded in the value of humanity. Humanity can, on this possibility, still be unconditionally or intrinsically valuable and transmit its value to our identities. Yet, there is no requirement that an agent who takes on her identity as authoritative need to place value in humanity. As a result, even in the case an agent might question whether she should place value in humanity at all, this will not imply that the corresponding grounding claim is false. This proposal can be developed within Korsgaard’s overall framework, since she wishes to avoid committing to a form of realism about value. Let me now briefly elaborate on this proposal.
First, take the notion of grounding. Most of its proponents conceive of grounding as a non-causal relation of determination, which is asymmetric, irreflexive, and transitive and distinct from relations such as supervenience, identity, or conceptual analysis.Footnote 21 Proponents of grounding claim that it is a non-epistemic relation that obtains independently of whether we think or know that it obtains. It is in this respect much like causation.Footnote 22 Paradigmatic examples of grounding include the groundedness of the existence of non-empty sets in their members, the groundedness of dispositions in categorical properties of an object, or the groundedness of mental states in physical states.
There is plausibility to the idea that we should also conceive of the determination of normative or evaluative properties in terms of grounding. Likewise, it makes sense to conceive of the transmission of normative properties from one class of facts to another class of facts in terms of grounding. The reason is that other candidate relations such as supervenience, identity, or conceptual analysis possess the wrong features to account for what we have in mind when we claim that normative properties are determined by natural or any other types of properties.Footnote 23 The arguments should be familiar to those who are acquainted with the grounding literature. Supervenience and identity both fail to possess the right formal features: supervenience and identity are both symmetric, while grounding is an asymmetric relation of determination. Grounding is therefore in a position to account for the metaphysical priority of the properties that do the grounding in comparison to those that are so grounded, while supervenience and identity are not relations that might explain this fact. Conceptual analysis, in contrast to supervenience and identity, is a relation between concepts. Yet, when we state that normative properties are grounded in non-normative properties, what we wish to express is that non-normative properties determine, in a metaphysical sense, normative properties. However, stating this relationship in terms of conceptual relations would miss the mark in terms of what we have in mind when we claim that normative properties are determined by non-normative properties.
A second component of the present proposal is that humanity possesses unconditional or intrinsic value. Note that this assumption is independent of any account of how humanity might come to possess its value. Even if we assume that values are ultimately determined by our attitudes, this is still compatible with the idea that the value of humanity might ground the value of our identities. Note that Korsgaard endorses a judgment-dependent view about value. On her proposal, what is valuable is ultimately determined by what people value: something is valuable if our valuing this object is able to “withstand reflective scrutiny” (Korsgaard Reference Korsgaard1996: 93). Korsgaard claims that the value of humanity (or any other valuable item) is nothing over and above our valuing it “in the full light of reflection,” just as there is nothing more to an entity being green than our seeing it as green under appropriate circumstances (Korsgaard Reference Korsgaard1996: 266). In the present context, I cannot discuss the merits of this proposal. As I said above, for the purposes of my argument, we need only assume that humanity is unconditionally or intrinsically valuable.
We can now state an argument of how the value of humanity might ground the authority of epistemic norms:
-
(1) The normativity of an epistemic reason is grounded in an agent’s epistemic identity.
-
(2) Epistemic identities are part of our identity as human beings.
-
(3) The value of an agent’s identity as a human being is grounded in the value of humanity.
-
(4) Humanity is unconditionally or intrinsically valuable.
-
(5) Therefore, humanity is the source of the authority of epistemic norms.
Now, we are in a position to counter the objection that it might still be an open question for an agent as to why she should value humanity. If our identities get their authority by being grounded in the value of humanity – that is, if premise (3) is true – then agents who value their identity and at the same time question whether they should place value in humanity commit a normative mistake. Denying the value of humanity while at the same time endorsing one’s practical identity would be much like denying the existence of Socrates while affirming the claim that {Socrates} exists. In contrast to Korsgaard’s argument, which requires an agent to value humanity in order to value her self-conception, the present proposal does not make such an assumption. It is, therefore, not prone to the objection sketched above.
How does the present account compare to rival accounts of the source of epistemic normativity? The present account will imply that there is a close connection between practical and epistemic normativity. If we accept the present proposal about the source of epistemic normativity, epistemic and moral normativity will – within the Korsgaardian framework – be two sides of the same coin. Each form of normativity will be governed by a different set of norms, which are not reducible to one another. In addition, both forms of normativity will concern distinct domains, in that epistemic normativity concerns the evaluation of beliefs, while moral normativity pertains to the evaluation of actions. But both forms of normativity will ultimately have the same source in that they are grounded in the value of humanity. The present account thus implies that there is a unity to both forms of normativity.
I think that this is an attractive feature of the present account and separates it from its competitors. Recall the positions I introduced at the beginning of this paper. Instrumentalist accounts of the authority of epistemic norms claim that epistemic norms gain their authority because following them is conducive to other things we place value in, such as survival or practical reasoning. Even if the versions of these accounts should satisfy the main desiderata introduced above, they will not imply that there is a unity to both forms of normativity.
The same verdict will apply to accounts that locate the source of epistemic normativity in the constitutive features of belief.Footnote 24 The constitutivist view comes in two varieties. On the normativist version, epistemic norms gain their authority from being constitutively tied to the concept of belief: agents should care about following epistemic norms because such caring is constitutive of applying the belief concept. On the second version, the teleological account, epistemic norms have authority over agents because beliefs aim at the truth. And agents should care about following epistemic norms because basing beliefs on truth-related factors is constitutive for being a good or genuine believer. However, if we consider these accounts in light of the idea that there is a unity to moral and epistemic normativity, it will become apparent that both accounts will not imply that both forms of normativity bear a connection to each other. Take normative constitutivism first. The fact that the concept of belief seems to exhibit a constitutive connection to the concept of truth and thereby to epistemic norms does not shed any light on the question as to where moral norms gain their authority from. If proponents of this view would in addition endorse the thesis that the concept of action bears a constitutive connection to, say, the concept of well-being, then there might be structural parallels between both forms of normativity. Nevertheless, this would not show that moral and epistemic normativity have a common root. A similar verdict will apply to the teleological version of constitutivism, on which beliefs are directed toward the goal of truth. If proponents of this view would further claim that it is constitutive for actions to be directed toward a particular goal, say, well-being, then the accounts of moral and practical normativity the constitutivist gives would also exhibit structural parallels. But, again, this is a much weaker similarity between both sorts of normativity than in the present suggestion.
Finally, consider Mitova’s (Reference Mitova, Grajner and Schmechtig2016) suggestion, which holds that following epistemic norms is a necessary precondition of agency. Her proposal is that in order for the world to make sense to an agent, this agent needs to follow epistemic norms. However, as with the previous proposals, this account does not shed any light on why there is a unity between moral and epistemic normativity. I therefore conclude that the present proposal will account for the unity of both realms of normativity in a superior way than its rivals.
3. Two objections to the present account
3.1. What about deviant epistemic identities?
How does the present account deal with agents who have deviant epistemic identities, such as adherents of conspiracy theories? Adherents of conspiracy theories hold beliefs that are not confirmed by genuine or scientific evidence. They believe, for example, that the earth is flat, that there is no man-made climate change, or that the 2020 US presidential elections were rigged. An important feature of those who adhere to conspiracy theories is that they do not experience any conflict between their identities and the beliefs they hold. They feel perfectly fine with being at odds with the truth and what the sciences or experts teach us about their conspiratorial beliefs. Let’s call adherents of conspiratorial beliefs “conspiracy theorists.” Conspiracy theorists are counterexamples to the present account because it seems constitutive of their identities to discard certain genuine pieces of evidence and therefore to not assign authority to traditional epistemic norms, at least as they apply to their conspiratorial beliefs.
The conspiracy theorist parallels very closely a character that causes trouble for Korsgaard’s account of moral normativity. Gerald Cohen (Reference Cohen and O’Neill1996: 167–88) has considered the case of Mafiosi, who fail to take moral norms as authoritative and adhere to their own code of honor. Nevertheless, they endorse their identities upon reflection and resemble morally decent agents in surprising ways. Mafiosi take care of their community, show loyalty toward other members of their clan, cooperate with their peers, and display moral emotions. The case of the Mafia mobster shows that endorsement of reasons for actions on the basis of a particular identity is not a litmus test to discriminate between morally good or bad reasons. Cohen insists that reflective endorsement of reasons on the basis of self-identities merely shows that something matters to an agent in that the principles constitutive of the agent’s identity will define the bounds of her actions. But reflective endorsement fails to establish the authority of a set of genuine moral principles for this agent. As a result, Cohen claims, we should look elsewhere for the source of the authority of moral norms.
Let me briefly consider Korsgaard’s response to this objection and whether it might serve to exclude the possibility of persons who hold conspiratorial beliefs on the basis of their identity. Korsgaard (Reference Korsgaard1996: 256) admits in response to the Mafioso case that mobsters seem to be bound by genuine obligations. Yet she claims that “there is no coherent point of view from which [the mobster’s project] can be endorsed in the full light of reflection” (Korsgaard Reference Korsgaard1996: 256). As Korsgaard explains, the Mafioso’s obligations are merely “shallow and incoherent.” The Mafioso has, for example, obligations of kinship and friendship toward his peers. But because he does not embark on “the search for the unconditional,” as Korsgaard puts it, his obligations are not grounded in the intrinsic or unconditional value of humanity. Therefore, Korsgaard concludes, the mobster’s obligations are not genuine obligations.
Korsgaard (Reference Korsgaard2009) provides a closely related type of response to the case of the Mafioso. She insists that the Mafioso does not engage in autonomous agency. As Korsgaard argues, the possibility of genuine agency requires “respect for humanity in your own person and that of every other [person]” (Reference Korsgaard2009: 25–6). Immoralism involves a loss of agency because valuing humanity is for Korsgaard a precondition for valuing anything at all, and any reason for a particular action receives its normative force from being ultimately grounded in the value of humanity. If an agent does not place genuine value in humanity, this agent could still be considered to value something instrumentally or conditionally. But, as Korsgaard claims, this agent would merely react favorably or unfavorably to some aspect of reality, rather than engage in the activity of valuing something in a genuine way. By lacking respect for humanity, the agent, on Korsgaard’s view, will not be in a position to constitute herself as an agent because the principles of practical reason are, in her view, also the constitutive principles of agency.
It might be evident from my presentation of Korsgaard’s response to the Mafioso case that her diagnosis will not be of help to deal with the conspiracy theorist. Recall that Korsgaard first argues that the Mafioso has shallow or incoherent obligations because they are not grounded in the value of humanity. Even though it might be argued that the conspiracy theorist has incoherent epistemic obligations – for the simple reason that she in general follows epistemic norms regarding virtually all of her beliefs, yet refrains from following them for a particular subset of them – the conspiracy theorist might nonetheless still value humanity intrinsically or unconditionally. It is even plausible to assume that the conspiracy theorist values humanity in the exact same way as those who do not adhere to conspiracy theories. Therefore, Korsgaard’s response to the Mafioso case will not apply to the conspiracy theorist.
Does her second response fare any better? Korsgaard’s (Reference Korsgaard2009) response claims that the Mafioso will fail to engage in genuine autonomous agency because she fails to value humanity. But again, this diagnosis will not be of help in explaining why the conspiracy theorist is not fully bound by epistemic norms. It is perfectly compatible for the conspiracy theorist to value humanity in the exact same fashion as someone who does not hold conspiratorial beliefs. Therefore, Korsgaard’s response will not be of help in dealing with the present case.
Now, what might be said in defense of the present account from the possibility of persons who have identities allowing them to disrespect epistemic norms in particular instances? A simple possibility to deal with these epistemically deviant identities might be to claim that they fail to be coherent epistemic agents upon unbiased reflection. On this diagnosis, upon unbiased reflection, conspiracy theorists will realize that they are incoherent in taking the authority of epistemic norms at face value. In particular, regarding beliefs about other closely related subjects and their conspiratorial beliefs, say, beliefs about how macroscopic objects behave under normal conditions or beliefs about what climate science teaches us about topics other than climate change, they seem to endorse the evidence the sciences provide. Nevertheless, for their conspiratorial beliefs, they rely on spurious evidence that deviates from contemporary science. If the conspiracy theorist were to reflect in an unbiased way, she would realize that there is a conflict in her acceptance of the sources of evidence that she deems reliable. For a particular class of beliefs, the conspiracy theorist will rely on respectable evidence, while for a narrow subclass of beliefs, she will fail to adhere to scientific evidence. Therefore, upon unbiased reflection, the conspiracy theorist will discover a tension in her acceptance of epistemic norms.
3.2. What happened to truth?
How does the present account accommodate the stock-in-trade idea that there is a close connection between epistemic norms and truth? On the face of it, there is no kinship between the value of humanity and the notion of truth. If the authority of epistemic norms is ultimately grounded in the value of humanity, one might ask whether this will secure that those who follow epistemic norms will likely end up with true beliefs. Therefore, one might question how the present proposal accommodates the connection between epistemic norms and truth.
The present account does not incorporate the notion of truth. Nevertheless, the present account is still able to accommodate the idea that there is a close connection between following epistemic norms and arriving at true beliefs. The reason is that our epistemic identities are sensitive to truth-tracking factors and therefore will only comprise norms that are truth-conducive. As I mentioned at the beginning of the paper, I do not wish to commit myself to a particular conception of epistemic norms. Yet, the norms mentioned above exhibit a close connection to truth in that they make it likely that a believer who follows them will end up with true beliefs. I therefore think that the present proposal will account for the connection between epistemic norms and truth.
4. Assessing alternative Kantian accounts of epistemic normativity
In recent literature, several accounts of epistemic normativity that develop Kantian ideas have been proposed.Footnote 25 How do these proposals compare to the account developed in this paper? And – perhaps more importantly – do they constitute genuine alternatives to the view suggested here? In what follows, I discuss four such accounts, developed by Catherine Elgin (Reference Elgin2013, Reference Elgin2017, Reference Elgin, Matheson and Lougheed2021), Kurt Sylvan (Reference Sylvan2020), James Andow (Reference Andow2023), and Alix Cohen (Reference Cohen2014, Reference Cohen, Himmelmann and Serck-Hanssen2021, Reference Cohen2024), respectively. A central problem shared by these accounts is their failure to anchor epistemic normativity in anything of intrinsic value. As a result, they are not in a position to halt the reflective regress. Despite their merits, I conclude that the Korsgaardian account sketched in this paper is superior to existing Kantian accounts of epistemic normativity. I set aside work on Kant, which deals exclusively with questions regarding how we should interpret Kant’s ideas on epistemic normativity, or how Kant’s account of epistemic normativity relates to his account of moral normativity.Footnote 26 Engaging with those questions would take me too far afield from the central concern of this paper.Footnote 27
Elgin (Reference Elgin2013, Reference Elgin2017, Reference Elgin, Matheson and Lougheed2021) develops a Kant-inspired account of epistemic normativity. Elgin’s account draws on Kant’s famous distinction between autonomy and heteronomy.Footnote 28 A heteronomous epistemic subject is bound by constraints that she “neither makes nor endorses” (Elgin Reference Elgin2017: 92). A heteronomous agent simply forms beliefs, without being able to endorse or dismiss those beliefs on the basis of reflection. An autonomous epistemic agent, by contrast, reflectively endorses her beliefs and her methods of belief formation. But what is reflective endorsement? On Elgin’s proposal, reflective endorsement involves adopting and maintaining methods of belief-formation that serve an agent’s epistemic ends. Elgin extends one formulation of Kant’s Categorical Imperative into the epistemic domain: the “Epistemic Imperative” demands that an epistemic agent should “accept only considerations that she could advocate and endorse as a legislating member of a realm of epistemic ends” (Elgin Reference Elgin2017: 105). One key feature of Elgin’s proposal is that epistemic normativity is inherently intersubjective: epistemically respectable methods of belief-formation are ones that other agents within one’s community are eager to accept on the basis of reflective endorsement. She claims that “since the realm of epistemic ends is supposed to be the arena within which epistemic agents live their cognitive lives, those commitments must mesh” (Elgin Reference Elgin2017: 144).
Elgin’s proposal differs fundamentally from the account developed here. Her aim is not to explain why an agent ought to follow epistemic norms. Rather, her account aims to uncover the normative basis on which agents select the right methods of belief-formation. Still, one might infer from her Kantian framework that the epistemic imperative has authority over all rational agents, just as Kant’s moral categorical imperative binds agents simply in virtue of their rationality. Yet this proposal raises a further question: why does rationality have authority over agents? Why does rationality constitute an end in itself? Elgin does not address these questions in her paper, nor does she provide any indication of how to answer them. Without an account of the normative authority of rationality, Elgin’s Kant-inspired account of epistemic agency cannot address the guiding question of this paper and thereby halt the reflective regress. The view developed here, by contrast, claims that epistemic normativity is ultimately grounded in the value of humanity. Since humanity is intrinsically valuable, epistemic norms are grounded in something that is of intrinsic value, which helps to halt the reflective regress.
Sylvan (Reference Sylvan2020) develops a view that he calls “Epistemic Kantianism.” Sylvan’s proposal is tailored to give us an account of epistemic justification – that is, to specify the conditions that constitutively characterize epistemic justification. His central thesis is that epistemic justification is to be understood in terms of the notion of epistemic respect. On Sylvan’s proposal, the fundamental epistemic value is accuracy, and a justified belief is a belief that manifests the right way of valuing accuracy – namely, by respecting it. For Sylvan, respect is not a mere feeling toward accuracy nor mere esteem for it. Rather, it is a norm-governed mode of responsiveness to the value of accuracy. To respect accuracy is for an agent to let accuracy structure what she is permitted to believe. As Sylvan argues, justified belief constitutively requires evidential sensitivity: an agent shows respect for accuracy if she is responsive to her evidence. On Sylvan’s proposal, epistemic justification is fundamentally deontological rather than consequentialist – a feature that allows epistemic Kantianism to avoid the problems that consequentialist accounts face.Footnote 29
It should be evident from the foregoing discussion that Sylvan addresses a different question from the one pursued here. His focus is on developing an account of epistemic justification, not on explaining what grounds the binding force of epistemic norms. Even though Sylvan’s proposal is not designed to address the guiding question of this paper, his account might nonetheless be understood as entailing an answer to the question of why an agent ought to follow epistemic norms. Sylvan might claim that epistemic norms possess authority over all agents because agents ought to display respect for accuracy. Even if one interprets his account as implying that agents ought to follow epistemic norms because they ought to respect accuracy, a further question remains: why ought one to respect accuracy? Respect for accuracy does not by itself halt the reflective regress, since it remains intelligible for an agent to ask why accuracy should command her allegiance. Sylvan might reply that respect for accuracy is a constitutive condition of epistemic agency. Yet, this constitutivist strategy faces a familiar difficulty: what is normatively significant about being an epistemic agent or a genuine believer? Unless epistemic agency itself is shown to possess unconditional value, the reflective regress persists. The same difficulty arises for any view that grounds epistemic normativity in constitutive features of epistemic agency.
Another Kantian account of epistemic normativity is developed by Andow (Reference Andow2023). Andow argues that epistemic consequentialism – understood as the view that what is epistemically permissible is determined by conduciveness to epistemic value (truth) – fails because it licenses problematic epistemic trade-off cases.Footnote 30 Andow contends that this creates an opportunity for a deontological account of epistemic normativity, modeled after Kant’s Formula of Humanity. On Andow’s suggestion, one’s own (and others’) rational agency is an end in itself. Andow claims that the epistemic version of Kant’s Formula of Humanity allows us to derive perfect as well as imperfect epistemic duties not to engage in certain types of belief-forming behavior – where the former duties allow us to exclude cases that give rise to epistemic trade-offs. He concludes that the Formula of Humanity offers a promising framework for an account of epistemic deontology, which avoids the problems of consequentialist accounts. As Andow concedes, however, this Kantian approach remains conditional on the promise of grounding epistemic normativity in the Formula of Humanity.
Andow’s account is centered on Kant’s Formula of Humanity: an agent ought to respect her own and others’ rationality as an end in itself. Yet, Andow’s view also leaves the guiding question of this paper unanswered. Why ought one to respect rational agency as an end in itself? Why does rationality possess unconditional value? As with the preceding accounts, the possibility of raising these questions indicates that the reflective regress has not been blocked.
Similar options are available to Andow as to the other proposals discussed above for addressing this worry, though, again, these responses are not considered in Andow’s paper. He might claim that respect for rationality is a constitutive feature of epistemic agency: without valuing one’s own (or others’) rationality, an agent cannot engage in epistemic agency. As I argued for above, this view faces the problem that we need an account of the unconditional value of epistemic agency. Or Andow might appeal to Kant’s idea that the categorical imperative has authority over agents simply in virtue of their rationality. But, again, without an account of the unconditional value of rationality, his proposal is not in a position to halt the reflective regress. Therefore, Andow’s account does not fare any better than the accounts discussed above regarding the guiding question of this paper.
Finally, Cohen has explored across a series of papers the possibility of developing a Kantian account of epistemic normativity.Footnote 31 In Cohen (Reference Cohen2014), she sketches a Kantian account that is “capable of contributing to current debates about the ethics of belief” (318). In contrast to Andow’s proposal, Cohen’s (Reference Cohen2014) central claim is that Kant’s Formula of Universal Law can be extended to the epistemic realm. Kant’s Formula of Universal Law, as applied to maxims for belief, claims that “I ought never to think except in such a way that I can will that my maxim should become a universal law” (Cohen Reference Cohen2014: 322). On Cohen’s suggestion, Kant’s Formula of Universal Law serves as a universalizability test for maxims of belief. In case a maxim fails to be universalizable (either by generating a contradiction in conception or in will), agents ought to discard that maxim from a normative perspective. She argues that Kant’s Formula of Universal Law rules out paradigmatic cases of unjustified or irrational beliefs, e.g., those based on ignoring evidence. Moreover, Kant’s Formula of Universal Law grounds a duty to rely on the (undefeated) testimony of others, since, on Cohen’s view, a universal maxim of incredulity would both undermine the very practice of testimony and conflict with our essential cognitive dependence on others.
But, again, her Kantian proposal is not designed to address the guiding question of this paper – namely, of why an agent in general ought to follow epistemic norms. Rather, Cohen’s proposal focuses on the question of which maxims of belief-formation are epistemically permissible.Footnote 32 Still, there are options available to Cohen to address the guiding question of this paper. If Cohen’s view is understood as claiming that agents ought to follow epistemic norms because they ought to respect Kant’s Formula of Universal Law, we will not arrive at a satisfactory answer to the guiding question of this paper. As already discussed above in connection with Sylvan’s proposal, the reason is that it remains an open question why an agent ought to display respect for Kant’s Formula of Universal Law. Cohen might perhaps respond that Kant’s principle plays a constitutive function in guiding agents in their belief-forming behavior: which epistemic maxims are permissible will be determined by the universalizability test. Therefore, agents have no other option but to rely on Kant’s Formula of Universal Law in the sense that the Formula of Universal Law serves as a necessary precondition for epistemic agency.
I have two objections to this proposal. First, in order for Kant’s Formula of Universal Law to play this role, it needs to deliver the right verdicts with respect to all possible cases. But there appear to be maxims that are epistemically pernicious but still pass the universalizability test. One such class of maxims is discussed in Andow (Reference Andow2023: 14–9). Suppose an omnipotent being offers to bestow me with the capacity to form many true beliefs (and very few false ones) in the future if I now believe some random proposition p (for which I have neither reason to believe that it is true nor that it is false).Footnote 33 The maxim underlying my belief would claim, “Believe p in order to secure many true beliefs about other subject matters in the future.” Suppose it were a universal law of belief formation that agents always accepted offers from an omnipotent being that resulted in them acquiring the ability to form many true beliefs (and very few false ones) in return. As should be obvious, this law passes the universalizability test. Suppose our world were populated by epistemic agents who always seized opportunities to maximize the amount of true beliefs they hold. That would have a positive impact on epistemic communities as a whole: if the world were populated by better sources of testimony, everyone in this community would be better off from an epistemic perspective. Still, there is something amiss with the permissibility of epistemic trade-offs (and a maxim that licenses them as permissible), primarily because epistemic trade-offs appear intuitively wrong.Footnote 34
But even if we set aside concerns about the adequacy of Kant’s Formula of Universal Law, the proposal under consideration faces a second problem. It presupposes that rational agency constitutes an end in itself. Without an account of the unconditional value of epistemic agency, this proposal fails to halt the reflective regress.
I conclude that existing Kant-inspired accounts of epistemic normativity fail to be serious competitors to the account sketched in this paper. None succeeds in halting the reflective regress. By grounding epistemic normativity in the intrinsic value of humanity, the account defended here avoids this difficulty and offers a more promising foundation of epistemic normativity.
5. Conclusion
In this paper, I have sketched a Korsgaardian solution to the question of what grounds epistemic normativity. On the account proposed here, epistemic norms gain their authority from being grounded in the self-image of agents, paralleling the way moral norms gain their authority on Korsgaard’s proposal. I have further proposed to trace the normativity of epistemic self-identities back to their being part of our identity as humans and therefore gaining authority by being grounded in the value of humanity. I have argued that the resulting view has advantages over existing proposals, in particular those that also develop Kantian ideas. For all their strengths, existing Kantian accounts of epistemic normativity fall short of providing a credible alternative to the Korsgaardian framework, as they fail to anchor epistemic normativity in something of intrinsic value. I take this to speak in favor of the account sketched in this paper.
Acknowledgements
For helpful comments and suggestions on an earlier version of this paper, I would like to thank Christos Kyriakou and an anonymous reviewer for this journal, as well as audiences at the Unity of Normativity workshop (Vienna), the 2022 meeting of the Swiss Society of Analytic Philosophy (Neuchâtel), and the Theoretical Philosophy Colloquium (Dresden). Research for this paper was supported by the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG), project number 516156675.
Open access
