1. INTRODUCTION
The contemporary debate over responsibility for belief is divided over the issue of whether such responsibility requires doxastic control, and whether this control must be voluntary in nature.Footnote 1 It has recently become popular to hold that responsibility for belief does not require voluntary doxastic control, or perhaps even any form of doxastic “control” at all.Footnote 2 We can be responsible for our doxastic attitudes, some have argued, even though such attitudes are in some sense non-agential,Footnote 3 or at least, are not subject to any form of direct or voluntary control. In this respect, responsibility for belief seems (to some) importantly different from responsibility for action, the latter of which appears to essentially involve direct, voluntary control.
However, Miriam McCormick (Reference McCormick2011, Reference McCormick2015) has recently argued that doxastic responsibility does in fact require (quasi-voluntary) doxastic control, and that our beliefs are the products of our agency, in much the same way that our actions are. She argues that the doxastic control we exercise, which grounds responsibility for belief, is a form of “guidance control.” Guidance control is a compatibilist form of control that does not entail that one could have done (or believed) otherwise. It is an essentially diachronic, temporally-extended form of control that is bound up with how one sees oneself and how one is seen by others. Understanding doxastic responsibility as grounded in guidance control, McCormick argues, avoids a number of issues encountered by other accounts of doxastic responsibility, such as accounts that see responsibility for belief as grounded (only) in reasons-responsiveness or in one's character.
In this paper, I pursue a negative and a positive task. In the first part of the paper, I will argue that McCormick's account of doxastic responsibility in terms of guidance control faces serious difficulties. In particular, I will argue that grounding doxastic responsibility in guidance control requires too much for agents to be the proper targets for attributions of doxastic responsibility. I will focus my criticisms on three cases in which McCormick's account gives the intuitively wrong verdict. I will suggest that agents can be appropriately held responsible for their beliefs while failing to meet the conditions required for guidance control. If so, doxastic guidance control does not ground responsibility for belief.
After criticizing McCormick's account of doxastic responsibility, I will move on to the second part of the paper, in which I offer my own positive account. In particular, I develop a modified conception of McCormick's notion of “ownership of belief,” which I call Weak Doxastic Ownership. I employ this conception to argue that responsibility for belief is possible even in the absence of guidance control. In doing so, I argue that the notion of doxastic ownership can do important normative work in grounding responsibility for belief without being subsumed under or analyzed in terms of the notion of doxastic control.
2. BACKGROUND: GUIDANCE CONTROL AND DOXASTIC RESPONSIBILITY
McCormick borrows the general notion and central parts of the theory of “guidance control” from Fischer and Ravizza (Reference Fischer and Ravizza1998), and extends their account into the doxastic realm. I will be concerned in this paper with only McCormick's own presentation and use of the notion, and thus will not concern myself with how (if at all) it diverges from Fischer and Ravizza's. As McCormick employs it, there are two components of guidance control: (a) reasons-responsiveness, and (b) ownership. An agent exhibits guidance control over an attitude, such as a belief, when that attitude is the product (the “upshot”) of a historically reasons-responsive process or mechanism, which the agent correctly recognizes as her own, and thereby “takes ownership” of the mechanism and its products, e.g., her beliefs.
A mechanism or process is reasons-responsive, in the relevant sense, if it involves the capacity to respond to various considerations as reasons, e.g., to respond to evidential considerations as epistemic reasons. An agent is reasons-responsive in the manner required for guidance control, on McCormick's account, if she employs a mechanism or process that is itself reasons-responsive, and is able to “guide” her belief-formation and revision in response to considerations that she understands as reasons.Footnote 4 This “guidance” centrally involves keeping one's beliefs in line with one's own higher-order judgments about what one ought to believe.Footnote 5
The notion of “ownership,” being the second condition for guidance control, is somewhat more difficult. Taking ownership of a belief involves regarding it as ‘one's own’, which in turn involves taking responsibility for it, and for the mechanism or process that produced it. The notion of “taking responsibility” here is understood developmentally and historically. McCormick (Reference McCormick2015: 112) describes it in the following way:
Taking responsibility is understood historically. As one comes to view oneself as an agent – as having an effect on the world as a consequence of one's intentions, decisions, etc. – one comes to view oneself as a fair target for the reactive attitudes, such as being worthy of blame or praise. By viewing oneself as an appropriate target for the consequence of a particular mechanism (say, ordinary practical reasoning), one thereby takes responsibility for it and the behavior resulting from it. Once one takes responsibility for a particular mechanism, then this ownership extends to future operations of the mechanism. It is a process that occurs over time where we develop a concept of ourselves as engaged in a kind of conversation.
It is important to note, as McCormick herself emphasizes elsewhere, that taking responsibility for one's beliefs need not be or involve an explicit, conscious act. Rather, the fact that one has taken responsibility can be revealed by one's general behavior and one's other attitudes. In particular, McCormick says, “Even if we never consciously endorse a mechanism, we can still have ownership of it. … I have taken responsibility if my practices reveal that I have accepted the expectation that I keep my beliefs of this kind in line with my higher order judgments of how I ought to believe.”Footnote 6
McCormick appears to be working with something like what some have called the answerability model of responsibility, which attempts to combine core features of the accountability and attributability models.Footnote 7 The features of the latter two models that answerability attempts to synthesize, and which McCormick appears to think are both important, are the second-personal structure involved in holding someone accountable to a demand or expectation, as well as the notion that things for which we are responsible must reflect our evaluative judgments or be products of our agency. McCormick is concerned with attributability via the notion of “ownership” of belief, while she appears concerned with accountability via her focus on the reactive attitudes, which are often understood as ways of holding others accountable to demands or expectations.Footnote 8 So, like the answerability model, her account attempts to combine these two features with respect to doxastic responsibility. Thus, for the purposes of this paper, I will follow McCormick in thinking of responsibility in terms of answerability in this sense, where being answerable is ultimately supposed to involve being the proper target of a demand for reasons or justification for one's belief (or other attitude).Footnote 9
So McCormick's general picture is this: doxastic responsibility is grounded in a form of doxastic agency that is centrally defined by the capacity for guidance control. Guidance control over one's doxastic attitudes necessarily entails (a) that the attitudes are the products of a reasons-responsive mechanism or process, and (b) that one recognizes the mechanism or process as one's own, takes ownership of it, and take responsibility for its products. Whether or not the reactive attitudes are appropriately directed at an agent in virtue of her doxastic attitudes will be determined by the extent to which that agent exercised or had the capacity for guidance control over the attitude in question.
3. PROBLEMS FOR GUIDANCE CONTROL AS A REQUIREMENT ON DOXASTIC RESPONSIBILITY
I will now argue that McCormick's account requires too much for agents to be the proper targets for attributions of doxastic responsibility. I will do so via consideration of three cases in which an agent fails to satisfy the ownership condition on guidance control, and yet in principle still appears to be an appropriate subject of doxastic responsibility. Each case will represent a structurally different way in which one might fail to satisfy the ownership condition.
3.1 Case 1: Replicated Beliefs
The first way that one can fail to satisfy the ownership condition on guidance control and nevertheless still be the appropriate subject of doxastic responsibility is by having one's beliefs produced by a mechanism other than that which one has (historically) taken responsibility for and recognizes as one's own. Remember that for guidance control, one must not only take responsibility for one's beliefs, but also for the mechanism or process of which they are “upshots.” But consider the following case:
Replicated Beliefs: Jane has a degenerative brain disease that causes her to gradually lose her beliefs, in the same way that other conditions cause one to lose one's memories. The disease does not, however, affect her other cognitive or executive capacities. Jane's brother John, a brilliant computer scientist, designs a computer program that makes exact virtual copies of her beliefs each morning, and replaces any beliefs that she has lost at night while she sleeps, though Jane is unaware of this. Eventually, all of Jane's beliefs have been replaced with copies generated by the computer program. Nevertheless, Jane still experiences those beliefs as her own, endorses them, and they still reflect and are grounded in her judgments and values.Footnote 10
Once enough time has passed, none of Jane's beliefs are products of a mechanism or process that she has taken responsibility for or identifies as her own. So she fails to satisfy the ownership condition on guidance control, and thus McCormick's account would tell us that she cannot be appropriately held responsible for any of her beliefs. But if we grant that Jane's case is possible, then it seems plausible to suppose that Jane could be appropriately held responsible for her beliefs even though she lacks guidance control. So it is possible in principle for one's beliefs to be produced by a mechanism other than ‘one's own’ and for one to still be appropriately held responsible for them.Footnote 11
3.2 Case 2: Doxastic Swampman
The second way that one can fail to satisfy the ownership condition on guidance control is by lacking the right kind of socio-causal history. Consider the following case:
Doxastic Swampman: Imagine that a being, Swampman, spontaneously comes into being out of the churning sludge of a swamp. Swampman has a full complement of cognitive abilities, i.e., is capable of reasoning, language, belief-formation and revision, etc. And imagine that Swampman immediately goes ahead and starts forming and professing beliefs about the world, most true, but some false. And imagine that a group of regular old human agents are there to witness all this. They start conversing with Swampman, who is fully able to interact linguistically with them, and appears to understand how his perceptual faculties are related to his belief-forming and linguistic mechanisms.Footnote 12
Once the shock of what they have witnessed wears off, and the humans seem confident that Swampman is a doxastic agent, would they be in a position to properly hold him responsible, i.e., answerable? One might have worries here about whether Swampman's could even have beliefs at all. For example, given that he has no causal history, one might doubt that his mental states could have intelligible content, or could bear reference to singular entities or even natural kinds, since Swampman has never been in causal contact with those entities or kinds.
One way to address this worry is to imagine that Swampman could at least have beliefs with intelligible a priori (e.g., moral) content. For beliefs with a priori (or perhaps purely conceptual) content, the lack of a casual history is less clearly a problem. McCormick's account tells us that, because Swampman has no developmental history whereby he came to see himself as the proper target of the reactive attitudes, he cannot possibly be appropriately held responsible for any of his beliefs. But imagine that Swampman quickly forms and professes the belief that the humans are inferior to him and should serve him as their master. He is adamant about this belief and insists he is correct, though he takes no objectionable or violent action against them. Would he not be the appropriate target of blame, resentment, indignation, etc. in virtue of this belief? Could he not be appropriately held answerable for this belief, at least in principle?
This is a rather fantastic case, and I do not mean to suggest that what should be said about it is extremely clear. One might have no clear intuitions at all. But even if Swampman is not appropriately held responsible, the fact that he has no social-developmental history of the relevant kind might seem to be rather far down on our list of worries about him with respect to whether or not he is responsible for his beliefs. Presumably our most serious worry is simply whether Swampman really understands what he's saying, doing, and thinking. However, if we were somehow convinced that Swampman really did understand what he was saying, doing, and thinking, it doesn't seem beyond the pale to think that he would still be the appropriate target for attributions of doxastic responsibility. His lack of a developmental history, in particular a history of taking responsibility for his beliefs, does not appear to exempt him in principle from being appropriately held responsible for his beliefs – though of course other things might.
3.3 Case 3: Isolated Society
The third way that one can fail to satisfy the ownership condition on guidance control is by not belonging to a community that engages in the relevant practices of epistemic evaluation. Consider the following case:
Isolated Society: Imagine that we discover and make contact with a previously unknown and isolated human society, and are able to engage in linguistic communication with them. This society has no discernable social practice of taking ownership or responsibility for their beliefs, and no practice of doxastic appraisal, though they do certainly seem to have beliefs. While they do appear to employ certain evaluative practices, none of these practices seem to involve taking or attributing responsibility for belief.
If such a society is conceivable, which it certainly seems to be, McCormick's account would tell us that no members of that society would ever be appropriate targets for attributions of doxastic responsibility, unless they somehow came to adopt practices of taking ownership and responsibility for belief. Otherwise, all members of such a society would in principle be systematically exempted from doxastic responsibility, since they would categorically fail the ownership condition.
But this is the wrong result. It also threatens to make doxastic responsibility culturally relativistic. If, after sustained observation and interaction, we became confident that members of such a society indeed had beliefs, and that they understood themselves as having belief-like mental states, it's far from obvious that we would not be entitled to hold them responsible for their beliefs. There might be a sense in which holding members of this society responsible for their beliefs would seem ‘unfair’, given that they would likely lack an understanding of the concept of doxastic responsibility. But if they understood the general concept of normative responsibility (e.g., as applied to actions), attributing doxastic responsibility to them would not be inappropriate, at least certainly not in principle.
3.4 McCormick's reply
The above three cases strongly suggest that McCormick's ownership condition on guidance control is too strong. However, McCormick anticipates and responds to something like this objection. I will quote her at length so as not to obscure any of the details of her response:
Another worry with the ownership account is that on the one hand, it can be too easy to duck responsibility by refusing to take responsibility and, on the other hand, one can be held responsible when one ought not to be because one has mistakenly taken responsibility. Should responsibility really depend on attitudes of the fallible agent? Fischer and Ravizza discuss this worry at length and I am satisfied with their response. First, we must remember the dialogical and historical aspects of their account. As I emphasized [above], taking responsibility is not a single act that one chooses to do or fails to choose to do. The price of failing to take responsibility is high and not one that many people would be willing to incur. In viewing oneself as an agent and as an “appropriate participant in the family of reactive attitudes,” one thereby takes responsibility. If one does not see oneself in such a way, one would be cut off from most meaningful human relationships; it requires one “to relinquish autonomy and to remain a fragmented self that is constantly in danger of ‘slipping away.’” There is, indeed, a “subjectivist” component to the ownership account in that an agent has to have a certain view of himself to be responsible.Footnote 13
I think this response falls short of really addressing the most persistent form of the worry. McCormick suggests that the “price” of refusing to take responsibility for one's beliefs is so high and thus “not one that many” people would accept. But if “not many” would pay this price, does she mean to allow that some would? Or even at least that some could choose to pay the price? If so, then McCormick has granted the existence of counterexamples to her account. If some agents could and/or would be willing to pay the price associated with refusing to taking responsibility for their beliefs, then those agents would effectively be able to exempt themselves from doxastic responsibility. But as I've suggested above, this appears implausible. It appears that agents can be proper targets for attributions of doxastic responsibility, in principle, regardless of whether they regard themselves as such, or whether they exercise guidance control.Footnote 14 The fact that there is a “high price” associated with “ducking” responsibility by refusing to take ownership is ultimately neither here nor there in terms of addressing the worry at its most general level.
It is important to note that McCormick does allow that there might be exceptional cases, like the three I considered above, that may not appear to be fully captured by her account. However, if there are exceptional cases or potential counterexamples to a general account of a phenomenon, we want something principled to say about them. McCormick does have something to say, which is that in many such cases, failing to participate in the “family of reactive attitudes” would cut one off from most meaningful human relationships. And further, “it requires one ‘to relinquish autonomy and to remain a fragmented self that is constantly in danger of ‘slipping away.’”Footnote 15 Relinquishing autonomy would likely mean that one does not qualify as enough of a rational agent to be appropriately held doxastically responsible, and so the cases then would be captured by her account. This response will capture some such cases, perhaps including Doxastic Swampman. But it's far from clear that this response will work for Replicated Belief and Isolated Society. I think we can plausibly imagine that the citizens of the isolated society are autonomous, non-fragmented agents with various meaningful human relationships. And even though Jane, in Replicated Belief, can't technically take ownership of her beliefs (given their origin), she nevertheless remains an autonomous agent capable of meaningful relationships. So while McCormick may allow that there are likely exceptional cases like these, more needs to be said about them and what they imply about the nature of doxastic responsibility.
4. DOXASTIC RESPONSIBILITY WITHOUT GUIDANCE CONTROL: ON WEAK DOXASTIC OWNERSHIP
I have now offered various criticisms and three counterexamples to McCormick's account. The main upshot has been that agents can lack guidance control by failing the ownership condition, and yet still appear to be appropriate subjects of doxastic responsibility. However, I nevertheless think that some version of the notion of ownership of belief is normatively central to doxastic responsibility. I thus now want to argue that a modified conception of ownership of belief (hereafter: doxastic ownership) can do important normative work in grounding responsibility for belief without being subsumed under or analyzed in terms of the notion of doxastic control, as McCormick does. I will call this modified conception Weak Doxastic Ownership. I call it “weak” only in a contrastive sense, in order to highlight that it is less demanding than McCormick's conception of doxastic ownership.
On the account I will develop in this section, the most fundamental form of doxastic responsibility – answerability – follows from the nature of belief itself, or rather, from what it is to hold a belief. So what makes one answerable (i.e., responsible) for one's beliefs is just that one holds them. What this requires in the first instance is that one satisfy an attributability condition with respect to the beliefs – namely, that the beliefs are properly understood as one's own in the normatively robust sense relevant to responsibility. This sense of a belief being ‘one's own’ is where the notion of doxastic ownership will play a central role: what it is for a belief to be properly attributable to an agent will be for her to satisfy the condition of weak doxastic ownership. The sense of doxastic ownership that I am interested in is the following:
Weak Doxastic Ownership (WDO): An agent takes weak doxastic ownership of a doxastic attitude just in case she holds it for reasons she takes or is disposed to take herself to possess, and the attitude reflects an evaluative judgment that she regards or is disposed to regard as her own, i.e., she is disposed to reflectively endorse it as expressing her values.
The most obvious and salient kind of evaluative judgment that a belief might reflect is a judgment about what a body of evidence supports, or more generally, what a body of reasons supports. An agent's belief that p will presumably often reflect her (dispositional) evaluative judgment that the available evidence (or some body of evidence E) sufficiently supports believing that p. Alternatively, if one holds that we can believe for non-evidential reasons, as McCormick does, the relevant judgment might just be my available normative reasons sufficiently support believing that p, where some or all of these reasons might be non-evidential ones.Footnote 16
However one may object that my characterization of WDO has the following problem: it will fail to capture cases in which agents believe for no reason.Footnote 17 If WDO is a condition on doxastic responsibility, the objection goes, then it will tell us that agents who hold beliefs not based on reasons are exempted from responsibility, but this is plainly incorrect. This is indeed a worry for WDO, but it can be defused. I think it is in fact very unusual for agents not to believe for reasons that they are not disposed to take themselves to possess. This is compatible with thinking that, e.g., perceptual beliefs are not initially formed and held for reasons. I think when we talk about agents believing ‘for no reason’, what we typically mean is no good or salient reason, or perhaps no immediately identifiable reason. But those kinds of cases are capturable by WDO. On the other hand, if there are in fact cases where an agent actually holds a belief not based in any way on what she is disposed to see as her reasons, I suspect one of two things will be true: the attitude in question will in fact not be a belief at all, or the agent will be alienated from the belief in such a way that it in fact will be appropriate to exempt her from responsibility for it.
To return, I propose to understand WDO as a kind of attributability condition: satisfying WDO with respect to a doxastic attitude makes that attitude one's own in the normatively robust sense relevant to responsibility. Call this normative attributability. WDO is partly supposed to account for cases where a belief doesn't seem to be attributable to an agent in the way that would legitimate attributions of responsibility. For example, beliefs that are ‘implanted’ via posthypnotic suggestion or brainwashing will likely not be connected with an agent's judgments, values, and dispositions in the way that would satisfy WDO.Footnote 18 In such cases, though these beliefs might ‘belong’ to the agent in the mere sense that they inhere in her psychology – call this descriptive attributability – they are not normatively attributable to her.
I next want to return to my remark above that doxastic responsibility follows from the nature of belief itself. My central suggestion here is that belief itself implies a norm of answerability, where this is the norm that effectively makes one the apt target of a demand for reasons or justification. That is, the holding of a belief by default makes one answerable for that belief. The intuitive justification for this claim is that if someone e.g., professes to believe that p, it always is apt (if not appropriate) to ask them why.Footnote 19 Remember that if an agent satisfies WDO with respect to a belief – the belief is normatively attributable to her – then she sees or is disposed to understand that belief as held for or supported by her reasons. So the suggestion here is that belief is characterized by a norm of answerability that makes one answerable for one's belief in relation to the reasons by which one (partly) satisfies WDO.
The claim that belief intrinsically implies a norm of answerability is likely a contentious one. However, one might take the claim to follow from some version of normativism about belief, namely, the view that belief is intrinsically subject to certain norms and/or evaluative standards, or that certain norms are ‘built into’ belief.Footnote 20 For example, Pamela Hieronymi (Reference Hieronymi2008) appears to take a norm of answerability to follow from something like normativism. She says, “believing brings with it its own distinctive form of answerability. In believing, you are answerable for reasons that you take to show the belief true.”Footnote 21 For Hieronymi, this is in large part because beliefs are uniquely justified or rationalized by “constitutive reasons,” i.e., reasons bearing on the truth of their content. However, one could adopt an answerability norm even if one is an epistemic pragmatist (like McCormick), where the norm would be that one is answerable for reasons that one takes to support holding or adopting one's belief, not merely reasons that show the belief to be true. Thus, an answerability norm is in principle neutral between pragmatism and, e.g., evidentialism or ‘truth-essentialism’ about belief.
The view I'm developing here may sound similar enough to Hieronymi's that one might wonder whether I'm really offering anything new. However, her view is in fact quite different from my own. While Hieronymi does think that a norm of answerability is intrinsic to belief, what really does most of the work of making an agent the proper subject of doxastic responsibility for her is the notion of evaluative control.Footnote 22 Evaluative control is a non-voluntary, reasons-responsive form of doxastic control by which we ‘control’ our beliefs by evaluating the truth of their content. Evaluative control is effectively a form of doxastic deliberation by which we settle on our beliefs as the answers to questions about what is the case. By exercising evaluative control over our beliefs, Hieronymi holds, we make them constitutive parts of our “moral personality,” and they thereby reveal something deep about our characters, our rational selves, or the “quality of our will.” So, on her view, doxastic responsibility ends up being a species of, or at least grounded in, responsibility for self.Footnote 23 So Hieronymi's account is still fundamentally based on the notion of doxastic control, albeit a rather idiosyncratic one. The account I am offering here, on the other hand, is concerned with neither doxastic control nor responsibility for character or self, and so is importantly distinct.Footnote 24
At this point, it may seem as though I've partly lost sight of the important normative role that doxastic ownership was supposed to play in grounding doxastic responsibility, and that the proposed answerability norm of belief has taken center stage. However, while answerability is indeed essential to my account, doxastic ownership still does the most fundamental normative work of making our beliefs ours in the way that allows for the answerability norm to apply. We can only be answerable for beliefs that are normatively attributable to us, and this is why, I suggest, the notion of doxastic ownership is essential. The nature of weak doxastic ownership as characterized above also helps make sense of why being responsible for a belief is a matter of being answerable to a demand from others for reasons in support of that belief.
To summarize and conclude this section: I've now argued that the notion of doxastic ownership can do significant normative work in an account of doxastic responsibility without appeal to the notion of doxastic control. Belief essentially implies a norm of answerability, but it is doxastic ownership that makes us proper subjects of this norm. The final upshot of the account I've offered is that we are responsible for our beliefs not because (and when) we exercise control over them, but rather because (and when) we own them in the way discussed above. This helps us see why the agents in Replicated Belief, Doxastic Swampman, and Isolated Society might lack guidance control but nevertheless be proper subjects of doxastic responsibility. Insofar as the agents in those cases are capable of believing for reasons they see as evidence, and of making and endorsing evaluative judgments, it appears that they all can satisfy the weak ownership condition I've articulated despite lacking guidance control.
5. OBJECTIONS AND REPLIES
5.1 The objection from control, take 1: ownership requires control
I now want to consider some objections to the account I've offered. The first salient objection one may raise is that, even if we grant that a form of doxastic responsibility is implied by the nature of belief itself, doxastic ownership requires or involves a form of control. So one does not count as owning a belief unless one (can) exercise a form of control over it. Thus, doxastic responsibility cannot be divorced from doxastic control, the objection goes, even if such responsibility is grounded in doxastic ownership and the normativity of belief.
The plausibility of this objection will depend heavily on what form of doxastic ‘control’ one thinks is essential to ownership. Conceptions of doxastic control vary so widely that some versions of the objection will look significantly different from others.Footnote 25 This is made more difficult by the fact that some things that get called ‘control’ don't seem to involve much more than reasons-responsiveness (or, e.g., for Angela Smith, judgment-sensitivity).Footnote 26 Thus, there is the danger of getting caught up in a potentially merely verbal debate here regarding what does or doesn't count as a form of doxastic ‘control.’ However, if one thinks that the relevant form of control is something like indirect voluntary control, whereby we can voluntarily alter our beliefs by e.g., gathering further evidence, it is implausible to think that this is required for ownership. This is because it is clear that a belief can be normatively attributable to us even when we are unable to exercise indirect voluntary control over the attitude. An agent can satisfy weak doxastic ownership even when there is simply no further evidence to gather regarding whether p, or when she is otherwise unable to exercise “managerial” or “manipulative” indirect doxastic control.Footnote 27 For example, Kate might believe that extraterrestrials have visited Earth, but lack indirect voluntary control over her belief because she is totally unsure of what evidence, if any, would or might change her mind.Footnote 28 Similarly, a lack of indirect voluntary control does not keep us from believing on the basis of reasons that we take to support our belief.
It is beyond the scope of this paper to argue that no version of doxastic ‘control’ is required for doxastic responsibility, given the aforementioned diversity and plurality of the various conceptions. For forms of ‘doxastic control’ that involve only some form of reasons-responsiveness, it may be that such a capacity is required by my conception of doxastic ownership, but I think it is far from clear that such capacities really count as forms of ‘control’. However, this rather quickly leads into what looks like a merely verbal dispute, and so I will not linger on the issue here. My point here has just been that doxastic ownership does not plausibly require or essentially involve the capacity to intentionally alter our beliefs.
5.2 The objection from control, take 2: responsibility requires agency
One may next have the following worry: one can only be held responsible for exercises of one's agency, or the results of such exercises. But if answerability is a form of responsibility, how are we to understand the way in which it is connected to one's agency? Typically, exercises of agency are understood as or in terms of exercises of control; thus, the motivation for grounding doxastic responsibility in some form of doxastic control. But if it is not ‘control’ that allows for the connection with agency, then what?
In response, I want to suggest that believing itself can be understood as agential. To believe in the way that human animals do implies the employment of rational capacities that make possible believing for reasons. This is arguably why answerability follows from the nature of (human) belief itself: it represents an exercise of cognitive agency. Here I am inspired by Matthew Boyle's (Reference Boyle2009, Reference Boyle2011) view of doxastic agency, according to which belief implies “the activity of reason” and involves the actualization of rational capacities.Footnote 29 While belief is a state of an agent, Boyle suggests it is an active state: it reflects a kind of continuous rational activity on the part of the doxastic agent.Footnote 30 Boyle's suggestion is that belief is not best understood merely as a passive state towards which we bear a non-agential relation unless it is through the extrinsic activities of deliberation and judgment. Rather than being exercised only in events or processes of coming to believe, our doxastic agency might also be exercised in “‘energetic’ activities of holding rationally-grounded attitudes toward particular propositions.”Footnote 31
However, Boyle's view faces the following problems. Boyle suggests that belief is active in the sense that it represents something like the continuous assent to or acceptance of a proposition as something to-be-believed in light of one's reasons, or the “enduring actualization of [an agent's] capacity to hold a proposition true for a reason she deems adequate.”Footnote 32 But “continuous assent” or “continuous acceptance” only seem to be notably active when they are understood as occurrent mental acts. Presumably acceptance, and perhaps also assent, can be non-occurrent mental states, but it is not clear that they make belief active in a robust sense when they are non-occurrent.Footnote 33 Furthermore, it's not clear that all forms of belief involve anything like continuous assent or acceptance. This means that many of our beliefs, especially our dormant, tacit, and dispositional beliefs, will come out as non-agential even for Boyle. So the sense in which belief is supposed to be a ‘continuous rational activity’ is not clearly captured by continuous assent or acceptance.
Nevertheless, we can still capture the way in which belief is unique and interesting qua state without having to say that it involves a continuous activity in Boyle's sense. I think Boyle is right to emphasize the way in which belief seems to involve a kind of continuous rational feature or a form of rationally important continuity. Belief is unlike other states in that it essentially involves a kind of temporal or synchronic stability. One can go from the state of being seated to that of being standing and back again in seconds. But belief isn't like this: part of being in a state of belief is that that state extends over time in a certain way. This is what, e.g., makes it a state of belief and not merely a state of one's brain at a particular time. But what is this sense in which belief is essentially diachronic? I want to suggest that the state of belief is sustained over time by our doxastic agency, e.g., our counterfactual sensitivity to various considerations as normative reasons. This need not be understood in terms of continuous assent or acceptance, but rather only our persistent counterfactual sensitivity to considerations bearing on the truth or falsity of our beliefs.Footnote 34 This explains why belief states can't simply go in and out of existence: their continuous and synchronic stability reflects the continuity of the perceived normative force of various reasons by the agent. Finally, the idea would be that states that represent our enduring rational sensitivity to reasons, like belief, and so exhibit this kind of stability and continuity are subject to certain norms, e.g., a norm of answerability.
The view that belief itself is agential is highly contentious. I am unable to defend it fully here, but I do think it helps us to understand three things. First, it helps us see why the locus of doxastic agency, and thus also doxastic responsibility, could be belief or believing itself, and not merely things extrinsic to belief, such as belief-system maintenance (see Chrisman Reference Chrisman2018) or practices of indirect doxastic influence (see Peels Reference Peels2017).Footnote 35 Second, the view helps us see why some version of normativism might be true: belief is intrinsically subject to certain norms in part because believing is an exercise of rational agency. And finally, for the same reasons, it can help us see why one of these norms intrinsic to belief might be a norm of answerability: namely, because we are answerable for the exercise of our agency.
5.3 Objection 3: the present account is too liberal
The third objection I'd like to consider is that the account I've offered is too liberal, and so will imply that agents are responsible for their beliefs in a range of cases in which it is implausible to ascribe such responsibility, such as for young children, or cases where an agent has been brainwashed into holding a belief, or where the belief has otherwise been ‘implanted’ in some unusual manner. In such cases, the objection goes, my account will tell us that the agents are still responsible for the implanted beliefs, since this follows simply from the fact that their beliefs are attributable to them – and this is the wrong result. The correct thing to say is that such agents are excused from responsibility, since their beliefs are the products of force, manipulation, or – in the case of young children – insufficiently rational processes.Footnote 36
This objection points to an important worry, but it can be accommodated by my account. It is true that any theory of responsibility, doxastic or otherwise, will need to have something to say about how excuses and exemptions from responsibility function. But the account I've offered can allow that agents in the above kinds of cases may be excused from responsibility in one of two different senses. The initial question here is whether implanted beliefs, or a young child's beliefs, would really be normatively attributable to the relevant agent. While, e.g., a brainwashed agent's implanted beliefs may be descriptively attributable to her, if the beliefs are not normatively attributable to her, then she would not be subject to the answerability norm of belief, and thus would not be responsible. So some of the potential problem cases identified by the objection will be accommodated by the fact that the relevant beliefs will not be normatively attributable to the agents.
However, it is perhaps possible that some brainwashed agents, or young children, will satisfy doxastic ownership with respect to some of their beliefs. But here we can separate responsibility (i.e., answerability) simpliciter from praise and blame, or from negative and positive epistemic appraisal.Footnote 37 One can be excused from praise or blame without being excused from answerability itself. If an agent's beliefs are the result of brainwashing or implantation, we might excuse her from, e.g., blame for failing to satisfactorily respond to a demand for reasons. That is, we might excuse her from being negatively evaluated for failing to have good reasons for their beliefs. The same might be said about the case of young children. But this is compatible with such agents nonetheless remaining responsible qua answerable. So, while default doxastic responsibility, in the form of answerability, will apply as long as one's beliefs are normatively attributable to one, this does not imply that one will always be blameworthy for, e.g., irrational or false beliefs. Holding that one might be excused from epistemic blame or praise by contingent factors is perfectly compatible with the claim that doxastic responsibility is implied by the nature of belief itself and by doxastic ownership.
So, on my account, excuses from responsibility will function not at the level of answerability or ownership itself, but at the level at which one's rational performance in response to the demands associated with answerability is evaluated. Thus, if the brainwashed agent, or the child, can properly be said to own their beliefs in the sense of normative attributability, they will thereby be subject to the answerability norm of belief. So they are not excused from responsibility simpliciter. Rather, they are potentially excused from blame for failing to meet the rational demands of answerability, assuming they indeed fail.Footnote 38 Much more would need to be said here, ideally, about the nature of doxastic praise and blame. My account in this paper has not been of praise and blame, but rather of the basic form, structure, and ground of doxastic responsibility.Footnote 39
6. CONCLUSION
I've done two things in this paper. First, I've offered three cases about which McCormick's view that doxastic responsibility entails guidance control appears to give the wrong result. The cases have shown three different ways that agents can fail to satisfy McCormick's ownership condition of guidance control, but still seem to be proper subjects of doxastic responsibility. Thus, the sense of ‘ownership of belief’ required by McCormick's account of guidance control is, I've argued, too strong.
Second, I've developed a modified conception of doxastic ownership, which I've called Weak Doxastic Ownership. I've proposed understanding this form of doxastic ownership as a condition of normative attributability, the satisfaction of which makes an agent's belief(s) ‘their own’ in the sense relevant to responsibility. I've also suggested that belief is intrinsically subject to a norm of answerability, such that if one satisfies the doxastic ownership condition, one is thereby subject to this norm, and so is answerable for the relevant belief(s). The final upshot of the account I've offered is that doxastic responsibility is not grounded in, nor does it require, doxastic control, but rather follows partly from the nature of belief itself, and from the way in which we typically own our beliefs in a normatively robust fashion.
I would like to conclude by briefly considering some of the advantages of my account, at least as compared to McCormick's. The first is that, as already noted, it allows us to see why the agents in Replicated Belief, Doxastic Swampman, and Isolated Society could be proper subjects of doxastic responsibility despite lacking guidance control. This points to a second, broader advantage, namely that my account allows us to make sense of a range of unusual kinds of cases without having to say that the agents in such cases suffer from fragmented identity or diminished agency. This is due in part to a third advantage, which is that my account allows us to separate responsibility from blame and praise: agents can be excused from negative or positive epistemic evaluation without thereby being excused from doxastic responsibility itself. Finally, my account, if correct, allows us to reorient the debate over doxastic responsibility away from the issue of doxastic control – and thus potentially away from worries surrounding doxastic involuntarism – and towards issues of doxastic ownership and the normativity of belief.Footnote 40
