Is Practical Deliberation Bound by a Coherency Requirement? Foundational Normative States, Volitional Conflict, and Autonomy

Abstract

Harry G. Frankfurt has put the problem of volitional conflict at the center of philosophical attention. If you care fundamentally about your career and your family, but these cares conflict, this conflict undermines the coherency of your decision standard and thereby your ability to choose and act autonomously. The standard response to this problem is to argue that you can overcome volitional conflict by unifying your foundational motivational states. As Frankfurt puts it, the ‘totality of things that an agent cares about’ plus his ‘ordering of how important to him they are effectively specifies his answer to the question of how to live’ (The Reasons of Love (Princeton: Princeton University Press, 2004), p. 23). In this paper, I critically assess the three main reasons given for such a coherency requirement: 1) we can do only one action at a time; 2) our motivational states come with normative pressure towards coherency; and 3) conflicting motivational states provide us with an incoherent decision-making framework. I conclude that these reasons do not ground a coherency requirement for practical deliberation and argue that we can autonomously express ourselves as volitionally conflicted by acting on our conflicting motivational states over the course of multiple actions.

1. Introduction

Harry G. Frankfurt (1988) has put volitional conflict at the center of philosophical attention. Volitional conflict is the phenomenon whereby two conflicting motivations for action cause us ‘not to know how to act’. The conflict between two things a person truly cares about illustrates this phenomenon: e.g., do I, as a physician, take on an extra shift in the wards or do I, as a parent, stay home to care for my child and give her the attention she needs? Most philosophers working within moral psychology see the provision of a convincing answer to the problem of volitional conflict as a hallmark of a theory of (moral) agency. They do so because they assume that volitional conflict pulls an agent in two incompatible directions, making it impossible for her to decide what to do and therefore to act autonomously. Especially in cases where the agent has two deeply held but conflicting cares resolve about what to do is especially difficult. Frankfurt's reflections (1988) on ambivalence provide a good example.

The ambivalent agent, from a commonsense standpoint, feels torn because she does and does not want to perform the same action at the same time. This is unlike the depressed agent who finds nothing to be motivated for or the doubting agent who is undecided regarding two alternatives. The ambivalent agent knows what she wants but cannot convert this into a coherent perspective on what to do. It is therefore unhelpful to advise her to ‘pull yourself together’: there is nothing to pull together as the ambivalent agent is whole. The whole itself is conflicted. Neither is it helpful to tell her to ‘make up your mind’: the agent's ambivalence is grounded in deep motivations that define her mind. Thus, the ‘made up mind’ itself is conflicted. Frankfurt also explores the idea that the agent should give up on one side of the conflict – the advice to ‘undo your mind’ – but this is as ineffective. If the ambivalent agent could give up one side of the conflict, she wouldn't be ambivalent in the first place. Frankfurt therefore concludes that the motivational disunity that defines ambivalence ‘either tends to paralyze [the agent's] will and to keep him from acting at all, or it tends to remove him from his will so that his will operates without his participation’ (Frankfurt, 1988, p. 21). This makes his ultimate advice to ‘be sure to hang on to your sense of humor’ if the ambivalence doesn't resolve on its own (Frankfurt, 2004, p. 100).

Philosophers working within moral psychology share Frankfurt's assessment that being pulled in two incompatible directions by one's deeply held cares makes it impossible to act autonomously. This has given rise to a widely shared commitment to the claim that autonomous action is possible only if an agent has a coherent volitional make-up: i.e., a coherent set of motivational states.¹ In other words, philosophers within moral psychology assume that if an agent's motivations direct her to undertake different, incompatible courses of action, then they make it impossible for practical deliberation to come to a conclusion regarding what to do.

In contrast to this claim, I will argue that practical deliberation can settle on a course of action even if the agent is volitionally conflicted. Furthermore, I will conclude that practical deliberation does not only do this despite the volitional conflict, but also by taking the conflict as its ground: i.e., as the autonomous self-expression of the agent's most deeply held cares (Section 6).

To pave the way for this conclusion, I will first explicate (Section 2) three of the central reasons why most philosophers working within moral psychology posit the coherence of our motivational states as a condition for practical deliberation and autonomous agency: 1) we only can undertake one action at a time; 2) our motivational states come with normative pressure towards coherency; and 3) our motivational states constitute the standpoint from which we deliberate, and therefore a conflict between our motivational states undermines the activity of deliberation before it can start.

In Sections 3 to 5, I argue that the three reasons ground a coherency requirement for practical deliberation only hold on the assumption that we choose each action in isolation. I conclude that there is no reason for this assumption, arguing practical deliberation governs over multiple actions at a time, i.e., that we can also consider the rationality of our current choice in light of possible future ones. This makes it possible for an agent to deliberate and decide vis-à-vis multiple actions; she can thus decide to express her volitional conflict, undermining the need for a coherency requirement.

2. Volitional Conflict, Foundational Normative States, and Practical Deliberation

Imagine a country-loving pacifist whose country goes to war.² As parliament decides to revoke the suspension of military service for everyone under 45 years, she receives a call for duty. She discovers that her country allows for conscientious objections based on pacifist beliefs. Assuming that both her pacifist beliefs and her love for her country are fundamental to who she is, this situation creates a volitional conflict in her. Her pacifist convictions cause her to assess the war and the prospect of joining the military negatively, but the love she feels for her country gives her a strong reason to sign up for military service and defend her country's honor. Reflecting on possible higher-order compromises, she considers joining as a medic. Even this, however, is tacit support of warfare, something she disapproves of as a pacifist. She is at a loss regarding what to do: does she have more reason to join the army (as a medic) or to submit a conscientious objection?

2.1. The Generic Model of Agency: Foundational Normative States & Practical Deliberation

Within moral psychology, the standard response to such a conflict is to ask what the agent's deepest and ‘most true’ desires, cares, projects, or values tell her to do. As Frankfurt formulates this response, the ‘totality of things that an agent cares about’ plus his ‘ordering of how important to him they are effectively specifies his answer to the question of how to live’ (2004, p. 23). This standard response is grounded in a generic model of agency that is shared by philosophers working within moral psychology.

On this generic model, an agent acts autonomously if and only if she governs her choices and actions through non-arbitrary control, i.e., without a force external to the agent motivating the action. The agent gains this non-arbitrary control over her motivations for action through a process of identification, affirmation, or endorsement based on the agent's deepest cares. If an agent truly cares about her career, for example, choosing to make extra hours at the office instead of spending this time with friends and family is autonomous. In this way, the agent is ‘fully behind’ the action (Watson, 1987, p. 153).³ Such a process of affirmation grounds the agent's choices and actions in motivational states that are expressive of who she is and who she wants to be. I will call these states that grounds an agent's choices and actions her foundational normative states.

Within the confines of this generic model of agency, philosophers have posited or argued for different types of foundational normative states. For Christine M. Korsgaard (1996b) they are practical identities, for Bernard Williams (2006) they are an agent's ground project, for Marya Schechtman (2014), Alasdair MacIntyre (2016), and J. David Velleman (2006) they are the central tenets of an agent's autobiographical narrative, for Frankfurt (1988) and John J. Davenport (2012) they are an agent's cares (her volitional necessities), and for Michael Bratman (2007) they are the agent's self-governing policies.

The generic model of agency assumes a tight connection between an agent's foundational normative states and her practical deliberations. For Bratman, as we will see in the next subsection, self-governing policies indicate which desires can be taken up as subjectively justifying consideration in deliberation: i.e., which desires come with ‘subjective normative authority’ (Bratman, 2007, p. 91). Frankfurt points out that our ‘deliberations […] must be guided by procedures and standards in which it is appropriate for us to have mature confidence’ (Frankfurt, 2006, p. 29). This confidence stems from care as ‘the creator of inherent or terminal value and of importance’ (Frankfurt, 2004, p. 56). For Korsgaard, practical identities are ‘standing sources of incentives, as well as principles in terms of which’ an agent accepts or rejects ‘proposed actions’ in deliberation (Korsgaard, 2009, p. 22). Lastly, Velleman claims that an agent is bound in her deliberations to her autobiographical narrative to maintain ‘the internal coherence of the story itself’ (Velleman, 2006, p. 216).

Throughout this paper, I accept this assumption regarding a tight connection between foundational normative states and practical deliberation. What I take issue with is the additional thesis that non-arbitrary control is secured only if the agent's foundational normative states are coherent, thus implying that an agent's deliberation is bound by a coherency requirement. The underlying idea of this requirement seems to be that if an agent is volitionally conflicted, then ‘movement [within his practical deliberations] in any direction is truncated and turned back. However an agent starts out to decide or to think, he finds that he is getting in his own way’ (Frankfurt, 1999, p. 99). Coherency is thus deemed essential for deliberation to come to an end. In other words, an agent's normative states provide her with a decision-making framework that guides her deliberations. If there is conflict within this framework, the agent cannot reach an autonomous conclusion.

Let us illustrate this generic model of agency with Bratman's theory of planning agency.

2.2. An Illustration: Bratman's Self-Governing Policies

Bratman's choice of foundational normative states are self-governing policies (2007). Such policies differ from intentions in that they coordinate an agent's behavior over an extended period and do not have specific actions as their object. Policies also differ from plans. Plans constitute a blueprint for obtaining the concrete desired ends an agent sets for herself, such as writing a paper or having a meal. The function of self-governing policies is more general: they coordinate and organize the execution of an agent's plans, thus ranging over recurring events (Bratman, 2007, p. 42). Examples of self-governing policies are the policy of developing and upholding a ‘strong concern for honesty in writing’ or of letting ‘a desire for revenge or a desire to demean’ never be ‘a legitimate consideration in one's deliberation’ (ibid., p. 33).

For Bratman, self-governing policies define which desires in an agent's volitional make-up are considered to provide ‘a justifying reason in motivationally efficacious practical reasoning’ (ibid., p. 39). In other words, self-governing policies define how desires, intentions, and plans are to be weighted in practical deliberation: a self-governing policy establishes the ‘relevant weights to be given to desired ends in [practical] deliberation’ (ibid., p. 40).

The familiar question can be raised of how an agent can know that her self-governing policies are truly expressive of who she is. According to Bratman, only a specific subset of self-governing policies has this quality: those with which an agent is satisfied (ibid., pp. 34–36). Similarly to Frankfurt's notion of wholeheartedness (1999, Ch. 8), satisfaction depends on a lack of conflict between an agent's self-governing policies. In other words, if self-governing policy A undermines the role of self-governing policy B (and vice versa), then an agent cannot be satisfied with policy A. It follows that any actions based on this policy are not undertaken autonomously (Bratman, 2007, p. 35). For example, policy B may exclude the desire to eat sweets as setting a justifying consideration in deliberation, whereas policy A may set eating sweets as a justificatory end.

Not only does policy B block ‘the central organizing and coordinating’ role of policy A, it also constitutes a motivation to exclude policy A from the agent's volitional make-up (ibid., p. 35). Volitional conflict between an agent's self-governing policies makes it the case that policy A cannot speak for the agent, because the fact that this policy undermines policy B indicates that the agent is not satisfied with it. As such, coherence is not a rational requirement that is external to the functioning of self-governing policies. Rather, coherence is part and parcel of their functioning, it is built into their nature.

Furthermore, according to Bratman, incoherence among an agent's foundational normative states also makes it impossible for her volitional make-up to guide her practical deliberations. This claim comes to the fore in a recent collection of essays (Bratman, 2018). In this collection, Bratman clarifies that an agent's self-governance depends on the ‘guidance of thought and action by where the agent stands’: i.e., by the standpoint constituted by her plans and ‘norms of plan rationality’ (ibid., p. 202). This standpoint needs ‘to be sufficiently coherent to constitute a clear place where the agent stands on relevant practical issues’. Otherwise, it will be unable ‘to guide choice’ (ibid., p. 211). In other words, the norms of plan rationality that constitute the agent's standpoint ‘track forms of [synchronic and diachronic] coherence that are essential to a planning agent's self-governance’ (ibid., p. 210), as without such norms of coherence the planning agent finds no way to ‘weigh pros and cons with respect to specific decisions’ (ibid., p. 203).

2.3. The Reasons for a Coherency Requirement that Guides Practical Deliberation

In the example of Bratman's theory of planning agency, we have encountered two of the three reasons for a coherency requirement. I discuss the first reason – that the nature of foundational normative states themselves provide normative pressure towards coherence – in Section 4. I do so in light of Davenport's contention (2012) that foundational normative states come with a ‘self-perpetuating principle’ that provides a reason to exclude competing or conflicting cares from one's volitional make-up. The second reason is that an agent's foundational normative states provide the agent with the standpoint from which she deliberates about what to do. I discuss this reason in Section 5, considering Agnes Callard's claim (2018) that conflicting values provide conflicting guiding standpoints for deliberation, thereby making it impossible to engage productively in practical deliberation from the start.

The third reason for a coherency requirement is that an agent has only one body to act with and therefore can undertake only one action at a time.⁴ As Korsgaard claims, ‘[y]ou are a unified person at any given time because you must act, and you have only one body with which to act’ (Korsgaard, 1989, p. 111). Therefore, ‘on any given occasion, we can only do one thing’. This makes it necessary to ‘[m]ake up your mind, or even better, Pull yourself together’ if you are volitionally conflicted (Korsgaard, 2009, p. 134). Yonatan Shemmer (2012) expresses a variant of this ground for a coherency requirement.⁵ In contrast to the local pressure for coherence expressed by Korsgaard, Shemmer introduces the idea of a global pressure. He argues that although a degree of volitional conflict does not undermine autonomous action, too much conflict does: a person with too much volitional conflict does not have enough time to give expression to all her cares. I will turn to both variants in Section 3.

3. Does ‘One Action at a Time’ Limit Practical Deliberation?

Does the fact that we can do only one action at a time pose a limit on practical deliberation? Korsgaard makes use of split-brain cases in one of her early papers to argue ‘yes’.

A person whose corpus callosum is severed can get by in daily life just as well as any of us, but under experimental conditions notable things happen. A person given a stimulus to the visual systems of her brain's right hemisphere (and thus presented in her left visual field) may still be able to pick out what she perceives with her left hand while being unable to name or describe the object. She might even claim to see nothing. The reason for this is that the left hemisphere is in control of speech. With the corpus callosum severed, the right hemisphere cannot communicate its visual input to the left hemisphere. The left hemisphere thus ‘speaks the truth’ as it lacks the visual input.⁶

In trying to make sense of how such a person can function without any observable abnormalities outside of experimental settings, empirical research suggests that either the brain hemispheres communicate via external means (called cross-cues) or the person makes sure that stimuli are presented to both eyes (Volz et al., 2018). Based on these empirical findings, Korsgaard argues that both hemispheres can be understood as functioning autonomously ‘on their own’: there are two minds or two ‘consciousnesses’, each with their own goals, in one body.⁷ However, as they have only one body with which to act, they need to ‘come to an agreement’ regarding what to do to in order for ‘both hemispheres’ to be effectual (Korsgaard, 1989, p. 111). If they do not come to an agreement, then they undermine each other in reaching their separate goals.

According to Korsgaard, a volitionally conflicted (non-split brain) agent is confronted with the same task: you ‘are a unified person at any given time because you must act, and you have only one body with which to act’ (ibid.). This ‘logic of action’ binds an agent to (synchronically) unifying her foundational normative states in practical deliberation – via prioritization, for example. Based on the need for ‘an agreement’, Korsgaard even concludes that there is ‘the raw necessity of eliminating conflict among our various motives’ (ibid.). However, it is questionable whether this agreement necessarily implies the elimination of volitional conflict.

Let us explicate this considering a volitionally conflicted (non-split brain) agent. The generic model of agency states that an agent's foundational normative states ground the autonomy of her choices and actions. Turning this around, we can say that an agent's choices and actions are the expressions of her foundational normative states. With this reformulation, we can see that actions are a time-local expression of foundational normative states, which are themselves time-extended and persistent. The agent at the beginning of Section 2 may therefore express her care for being a pacifist (one of her foundational normative states) by joining a war protest. However, in saying that she truly cares about this, she will also have expressed this care in other actions that promote nonviolence, independent of this specific issue.

The insight that foundational normative states are time-extended and persistent supports the following claim: an agent can be said to care about pacifism only if she gives expression to this care over the course of multiple actions.⁸ She cannot claim to have this care (foundational normative state) if she acts on it only once by joining a war protest, without showing concern for expressing pacifist values on other occasions. Frankfurt stresses this: ‘the notion of guidance, and hence the notion of caring, implies a degree of persistence. An agent who cared about something just for a single moment would be indistinguishable from someone who was being moved by impulse’ (Frankfurt, 1988, p. 84).

Against this background, the question of whether our ability to do only one action at a time places a coherency requirement on practical deliberation should be answered with ‘no’. As actions are time-local expressions of time-extended foundational normative states, the volitionally conflicted agent can extend her decision-making over multiple time-local actions. The conflicted agent can give expression to her conflicting volitions over multiple actions. The conflicted agent is thus not restricted by the logic of action (‘one body, one action’) in her deliberations. She can choose and act autonomously by spreading the conflict over multiple actions.⁹

3.1. Choosing, Picking, and the Worry of Global Pressure for Coherence

The provided argument gives rise to two concerns. The first concern comes from the thought that even if an agent can express her volitional conflict over multiple actions, she still needs to make time-local decisions at each time-local occasion. The worry is that she will be able to make these time-local decisions only if her foundational normative states provide a unified decision-making framework in which her priorities are coherent. Bratman reflects this concern when he claims that without norms of coherence, an agent cannot ‘weigh pros and cons with respect to specific decisions’ (Bratman, 2018, p. 203). I contend that this conclusion can be resisted by denying the need on such occasions for choice under a decision standard (that is ultimately grounded in one's foundational normative states).

To make this case, I must introduce the distinction between picking and choosing. This distinction is defended by Edna Ullmann-Margalit and Sidney Morgenbesser (1977). According to them, picking happens in situations in which ‘preferences are completely symmetrical’ (Ullmann-Margalit & Morgenbesser, 1977, p. 757). Although the agent has no reason to pick either option A or option B if her preferences are symmetrical, she has reason to pick one of the options. Take the example of an agent who picks butter from the supermarket shelf. According to Ullmann-Margalit and Morgenbesser, this person does not need a decision standard, as she is without any reason to prefer one option over another. However, as she does prefer to have a butter, she does want to resolve the situation. The idea is that she can do so by merely picking one, without a decision standard. This situation therefore falls outside the domain of decision theory, strictly speaking. Ullmann-Margalit and Morgenbesser formalize picking-situations by using three criteria: first, the agent cannot select both options; second, she has no preference for one over the other; third, the agent prefers to select either of the options over none of them (ibid., p. 777).

I contend that these three criteria also hold for giving expression to volitional conflict at specific time-local occasions.¹⁰ First, the conflicted agent cannot choose to express both conflicting foundational normative states. Second, the agent wants to give expression to both and has no preference for one over the other. Third, the agent prefers to give expression to something that is foundational to whom she is over not acting at all. The situation the agent finds herself in thus qualifies as a ‘picking-situation’ and she can proceed without a decision standard: nothing depends on the choice of action on this occasion as long as the agent guarantees she will express both (conflicting) normative states over the course of multiple actions.¹¹ Authors in the debate thus seem to make a mistake by assuming that every choice situation requires a decision standard (grounded in an agent's foundational normative states).

If it is granted that picking allows an agent to be relieved from the local pressure for coherency, it could still be argued that there is a global pressure for coherency. This is the second concern, articulated by Shemmer (2012). Shemmer introduces a broad principle of coherence, the PCB, which states that agents should limit their number of goals as well as prefer goals that share resources for their satisfaction (Shemmer, 2012, p. 164). This principle allows for the idea that an agent can accommodate volitional conflict locally, but only to a certain extent. According to Shemmer, the agent has a global coherency requirement if she does not want to become arbitrary in her actions: she cannot stretch herself too thin over too many goals to still claim to pursue goals in a meaningful way. This global pressure thus comes into view when deliberation spans over multiple choice-situations.

First, it should be noted that Shemmer himself states that ‘the broad principle of coherence (PCB) is not a necessary constraint’: ‘many of us, maybe most of us, have reasons to be coherent, integrated, and unified, but not all of us do and certainly not all of us do to the same degree’ (ibid., p. 167). Furthermore, the problem Shemmer answers to with the PCB does not seem to be a problem of coherence per se. Rather, it seems a problem of non-commitment and doubt. To see this, it is helpful to introduce Korsgaard's idea (1996a, p. 181) of the ‘backward determination’ of our foundational normative states. Korsgaard means by this that if an agent asserts, for example, that she cares for her career, she needs to give expression to this care in and through her choices and actions. Only in this ‘backward fashion’ does she make her care into a foundational normative state. Korsgaard's idea becomes insightful if it is combined with that for volitional conflict to be possible – the threat to autonomous agency under discussion in this paper – the agent's foundational normative states should be in conflict.

Yet, I contend that the agent who stretches herself too thin cannot be said to have foundational normative states, as she cannot have ‘backward determined’ her cares as foundational to her. As such, the problem this agent is confronted with is not so much incoherence, but rather a situation like the doubting agent who is undecided regarding the alternatives she has for action. This agent, that is, is undecided regarded her commitments to any foundational normative states that ground her choices and actions. The PCB, in my reading, does therewith not point us at a problem with coherency, but rather at the danger of a lack of commitment to any foundational normative states, however incoherent these states as a set may be. As such, if there is the fact that we should limit the number of goals, it does not come from a requirement of coherence, as Shemmer proposes, but rather from the fact that a lack of commitment to specific goals undermines our capacity for autonomous agency.

This does not imply that we need to pursue all our cares one hundred percent of the time. As Shemmer (2012, p. 169) points out, our goals are open-ended and often not well-defined. This also means that how much time an agent spends on giving expression to a certain care (such as being an attentive parent) depends on how much time she allocates to this care considering her other cares. Consequently, an agent with many conflicting cares ‘is likely to try and satisfy each one of them at different times and not to satisfy any of them to the degree that someone with fewer [cares] could do’ (Shemmer, 2012, p. 167).¹²

4. Are we Bound in Deliberation by a Self-Perpetuating Principle Inherent to Foundational Normative States?

John J. Davenport (2012) argues in Narrative Identity, Autonomy, and Mortality that foundational normative states have a self-perpetuating quality. Based on this quality, Davenport argues that it is rational for an agent to remove any contradiction among her foundational normative states. I follow him in explicating the argument in terms of cares, which he sees as foundational to whom an agent is.¹³ Davenport (2012, p. 13) points out that an agent's cares include ‘precommitments to future motivation’. These are observable in that caring about upholding a friendship (for instance) just involves ‘remaining loyal to’ one's friend (Davenport, 2012, p. 13). This is what Davenport means by cares being self-perpetuating: caring about X includes the commitment to stay motivated to care about X in the future. In giving expression to the care by acting on it, an agent affirms her commitment and strengthens her care. Let us call this self-perpetuating aspect of cares the self-perpetuating principle: SPP in short.

Against the background of the SPP, Davenport argues that it is irrational for an agent to maintain and give expression to conflicting cares: if the pacifist above is truly committed to continuing to care about her pacifism, it does seem irrational to uphold her care for her country that is at war. This latter care would be in direct conflict with and undermine her pacifism care. As Davenport formulates it, it is irrational to ‘try to accommodate’ projects that ‘would contradict my current highest priorities’, as ‘I should want to hinder that future’ in which I stop caring about my highest priorities through the pursuit of contradictory projects (ibid.). Based on the SPP, Davenport thus claims that practical deliberation is bound by a principle of coherency. The question posed by this section is whether such a coherency principle follows from the SPP.

I resist this move based on what I indicated at the end of the previous section: an agent makes a foundational normative state her own by giving expression to it on multiple time-local occasions. This also means that if an agent periodically acts against a (time-extended) foundational normative state, she does not undermine her commitment to it. She would do so only if she does so too often and thus displays a lack of care over an extended period. Once an agent has ‘backward-determined’ her foundational normative states, these states gain a certain stability. Korsgaard ascribes this quality of stability to our practical identities (her choice of foundational normative states). In Sources of Normativity, Korsgaard discusses an agent who acts against a practical identity she is committed to. She argues that this possibility ‘does not come from the fragility of identity, but rather from its stability. It can take a few knocks, and we know it’ (Korsgaard, 1996b, p. 103). I take it that we are actually quite familiar with this phenomenon in our own lives: we cannot be said to have given up on our commitment to our career by occasionally choosing to procrastinate.

The mistake Davenport makes, therefore, is to think that the self-perpetuating quality of cares – foundational normative states – requires a coherent volitional make-up because the agent will otherwise undermine her own autonomy. Davenport implicitly holds that the self-perpetuating nature of cares needs to be affirmed on every possible occasion, but this is not the case. The stability of the agent's cares as diachronic, time-extended states, I argue, allows her to choose not to give expression to a care on a particular occasion without her losing her commitment to it.

It thus follows that the SPP can be satisfied without a coherent set of cares (foundational normative states). A bitter spouse can give expression both to her bitterness and to the love she feels towards her partner on separate occasions. This does not undermine her commitment to either the bitterness or the love; nor does she lose her autonomy because she is conflicted. By acting on both her bitterness and her love, she actually accommodates the SPP of both motivational states. In this way, she merely affirms her volitional conflict by giving expression to herself as conflicted. The stability of the agent's cares makes this possible.

A coherency principle binding practical deliberation thus does not follow from the SPP.¹⁴ I have therefore defended the possibility that an agent may act against her care on occasion and still be understood as being committed to it, albeit in a conflicted manner. Insofar the agent is conflicted and chooses to give expression to this conflict, however, this conclusion gets it exactly right.¹⁵

5. Do Foundational Normative States Individually Provide the Standpoint for Practical Deliberation?

The last ground for why philosophers take a coherency requirement to be necessary is the following. Insofar as an agent's foundational normative states shape their standpoint for deliberation, volitional conflict provides an incoherent deliberative standpoint. This incoherent deliberative standpoint makes it impossible for the agent to engage in deliberation. Callard has recently articulated why she takes this to be the case, so I will assess this last ground by considering her theory.

Central to Callard's conception of moral agency is the notion of valuing. She uses Samuel Scheffler's hybrid account (2012) in which ‘to value something’ is understood as a complex psychological state in which an agent desires to act in favor of the object valued, believes the object is valuable, is emotionally vulnerable regarding what happens to the value-object, and monitors herself concerning whether her affective, conative, and motivational responses align with her value.¹⁶ This self-monitoring aspect is relevant for our purposes. For Callard, our values (foundational normative states) provide the standpoint for deliberation from which we determine whether our own affective, conative, and motivational responses are appropriate considering these values. Callard explains that ‘when we value something we react to our own responses to the valued object, experiencing affective, conative, and motivational responses […] as merited or appropriate’ (Callard, 2018, p. 119). As such, our values have a function in Callard's theory of moral agency that can be compared to Bratman's self-governing policies. Our self-governing policies determine whether a desire provides ‘a justifying reason in motivationally efficacious practical reasoning’ (Bratman, 2007, 39). Our values determine whether not only our motivational states, but also our affective and conative states, are appropriate and thus also subjectively justified.

For Callard, the need for a coherency requirement comes from the fact that if an agent's values are conflicted, then she is unable to coherently monitor her affective, conative, and motivational responses. As her values provide the standpoints from which the agent monitors her responses, she has at least two standpoints that produce conflicting evaluations. It won't help to suggest that the agent takes a step back from both conflicting values in order to adjudicate between them: since her values constitute her deliberative outlook, she has no further values to retreat to. For our purposes, the consequences of being unable to retreat further are especially relevant considering the agent's motivational responses. As the agent cannot take up a coherent standpoint regarding her motivations, her conflicting values leave her without a standpoint from which to engage in deliberation (Callard, 2018, pp. 127–132).

Take the pacifist agent who loves her country. She is motivated to support her country (which is at war) by joining the army; this aligns with the love she feels for her country. However, the part of her self-monitoring activity that is grounded in her valuing pacifism evaluates this support negatively. Vice versa, staying true to her pacifism is appraised negatively by her self-monitoring activity grounded in her valuing her country. The pacifist's problem is that she has two values that each provide a standpoint from which the courses of action open to her are evaluated in opposite ways. As values are taken to be foundational, this agent seems to have no further standpoint she can step back to. Callard thus concludes that it is impossible for the volitionally conflicted agent to take up a standpoint from which she can engage in practical deliberation.

As practical deliberation becomes impossible because the agent cannot choose between her two conflicting values, Callard proposes a non-deliberative route for escaping the volitional conflict. I will demonstrate two things by discussing this non-deliberative route. First, I will show that it is implausible for this route to be fully non-deliberative; second, I will show that the agent can form an alternative standpoint from which to start deliberation if she takes her set of (conflicting) values to constitute a standpoint for practical deliberation.

For Callard, the non-deliberative route is possible, as values never have ‘equal strength’ within an agent's volitional make-up over an extended period. This means that quite ‘often we simply find that we have, for a while now, been aspiring to become more loving and less spiteful in some relationships’ (ibid., p. 146). Based on her awareness of the relative strength of her values, the agent can mark out ‘one attitude as a kind of target toward which she orients herself, and another (or others) as a danger from which she must turn herself away’ (ibid., p. 146). The conflicted agent should thus cultivate the stronger value and neglect the weaker. By orienting herself towards the stronger value, the agent gives shape to a coherent evaluative outlook (ibid., p. 143).

Callard's non-deliberative route still includes a deliberative aspect, however. Insofar as the agent chooses and acts on her awareness of the stronger value, Callard introduces a deliberative moment in which the agent chooses to orient herself towards the stronger value. The problematic aspect here is that Callard implicitly introduces a decision standard that is not based in any single value the agent has. This aspect can be seen if we ask why an agent should care from the standpoint of an individual value about the (relative) strength of her other values. As we have seen in the foregoing section, each value (foundational normative state) comes with a SPP: taken by itself, it wants to be reproduced, and the strength it has relative to other values seems irrelevant for this. By introducing a decision standard based on the relative strengths of the agent's values, Callard invites the agent to take up a standpoint that is grounded in the agent's set of values insofar as she uses their relative strength to ground her choices for action.

Callard's non-deliberative route appears to be driven by the implicit assumption, shared more broadly in this debate, that each individual value (foundational normative state) constitutes a deliberative standpoint by itself. This is evident in Callard's contention that an agent cannot engage in deliberation because of her conflicting values. It is unclear why this assumption must be accepted, however. It seems plausible, and even likely, that an agent can take the set of her values to constitute her deliberative standpoint.¹⁷

The plausibility of this possibility can be strengthened by discussing the example of the country-loving pacifist. We presented the country-loving pacifist as encountering the problem of having two values that each provide a standpoint from which the courses of action open to her are evaluated in opposite ways. Even a possible higher-order compromise such as joining the war effort as a medic is caught up in the crosshairs of her volitional conflict: as a pacifist, she simply cannot give expression to herself as a person who supports war, even tacitly. However, if the country-loving pacifist takes up a deliberative standpoint constituted by both her values, then she might get a different perspective on her choice situation.¹⁸ For example, she may come to see that there are several courses of action that could express her volitional conflict: she may, for example, choose to join the war as a medic while also opposing the war effort via social media – even by showing the horrible injuries that war causes.

6. A Final Worry: Can Volitional Conflict Ground Autonomy?

If correct, the arguments of the previous three sections establish that there is no principled reason available in the current debate to favor a coherency requirement for practical deliberation. First, the fact that we have one body does not limit practical deliberation: we can deliberate by considering acting on multiple occasions during an extended period of time. Second, the SPP does not restrict practical deliberation, because time-extended foundational states come with a kind of stability that allows us occasionally to act against them. Third, volitional conflict does not limit practical deliberation by offering conflicting standpoints for deliberation, as the agent can take the (conflicting) set of her foundational normative states as a standpoint. In short, I have made a preliminary case for the claim that an agent can autonomously express herself as conflicted if she gives expression to both sides of her volitional conflict on multiple occasions over an extended period of time.

In this last section, I wish to address a final worry. This worry seems to arise universally once an agent is implicated who creates her core self (her foundational normative states) ex nihilo. Susan Wolf formulates the concern as follows: the idea of self-creation introduces a ‘prime mover unmoved, whose deepest self is itself neither randomly nor externally determined, but is rather determined by itself – which is, in other words, self-created’ (Wolf, 1988, p. 52). This worry may be seen as especially pertinent to the conclusion argued for here: insofar an agent's foundational normative states are conflicted, it can be asked from what deliberative standpoint the agent decides to express herself as conflicted or to favor one side of the conflict over the other. In other words, does my argument implicitly rest on the idea of an ‘outside perspective’ over and above the agent's foundational normative states from which she chooses and acts?

To respond to this worry, I first want to point out that our foundational normative states have only a limited plasticity, because we have only limited control over their existence. A claim that is widely shared. Velleman, for example, argues that the agent does not have full control over the substantive self-concept by which she represents ‘which person and what kind of person’ she is (Velleman, 2006, p. 3). Her self-concept has some degree of plasticity, as the agent's self-understanding influences who she is: thinking of herself as an introvert causes her to behave as an introvert. At the same time, there are limits to this plasticity: wanting to be interested in sport (because your partner is, say) may or may not succeed in making yourself interested in sport (ibid., p. 6). You can put yourself in a position in which such an interest can grow and develop, but you cannot guarantee that the interest will follow.¹⁹ This idea finds ready support from philosophers such as Frankfurt (1988) and Williams (2006). However, Davenport (2012, Ch. 2) too says that our cares are based on the preconscious narrative elements that constitute our lives (he calls these ‘narravives’), over which conscious control is limited. Even Korsgaard, who defends a voluntaristic image of agency, writes that ‘we sure stumble into some of our deepest concerns, perhaps most obviously, the ties associated with family, ethnicity, and nationality, but also sometimes and to some extent our religions, friendships, and careers’ (Korsgaard, 1996b, p. 257).

Thus, we often do not choose our deepest concerns, we encounter them. Once encountered, we can try to indirectly change our foundational normative states, but we cannot change them at will. The implication of this is that if an agent is volitionally conflicted, she cannot will the conflict away. Her attempts to overcome the conflict might work but might also fail. As her control is limited, her deliberative standpoint may thus be fundamentally conflicted.

Having established that volitional conflict can be difficult if not impossible to overcome, due to the limited plasticity of our foundational normative states, we can turn to the expressed worry. From what perspective can an agent autonomously decide whether to express the volitional conflict or to side herself with one of the two conflicting foundational normative states? In other words, can an autonomous agent be autonomous if she acts based on conflicted foundational states?

To answer this question, I use Velleman's conception of agency to show that disunity on the level of foundational normative states is compatible with acting autonomously. According to Velleman (2006), an agent acts autonomously if her action is intelligible in light of her autobiographical narrative. Velleman starts with the claim that an agent's actions are autonomous ‘due to the agent's causal understanding’ (Velleman, 2006, p. 7).²⁰ This causal understanding means that agents understand the world, including their own actions, in folk psychological causal terms. As choosing and acting for reasons means to do what makes sense, actions become autonomous if the agent is intelligible to herself. If we combine both ideas, Velleman's theory of agency can be stated as follows: an autonomous agent takes her own causal history as the basis for her actions. As Velleman argues that an agent's causal history is his autobiographical narrative, an agent's ‘capacity for causal understanding gets redescribed as his capacity for coherent narration’ (ibid., p. 8). Summarized, an agent becomes intelligible to herself by developing a self-concept that pictures her as the protagonist in the story of her life – her autobiographical narrative.

For Velleman, an agent maximizes her self-intelligibility if she acts in line with her past, her life-story: If ‘there was a more intelligible story for [the person] to enact, by choosing to do something else, there was a better rationale for doing that thing instead’ (Velleman, 2000, p. 29). This explains why volitional conflict becomes a serious volitional problem for him: if an agent loses ‘a coherent conception of who he is […] he may feel that he has lost his sense of self or sense of identity’ (Velleman, 2006, p. 4). (In Bratman's terms, this agent is dissatisfied with her self-governing policy.) The underlying reason for Velleman's conclusion seems to be that the loss of a sense of self occurs because the agent stops being intelligible to herself under circumstances of volitional conflict. Considering the agent's conflicted life-story, an action is both intelligible and unintelligible, thus causing a loss in her sense of self. If this is correct, then we are committed to a tragic conclusion: incoherency makes an agent less autonomous, but she may simultaneously fail in her efforts to become more autonomous by resolving the conflict, because her self is not fully plastic.

This tragic conclusion can be resisted, though, if we acknowledge and accept that foundational normative states are partially non-plasticity. Consequently, agents are not in full control of their foundational normative states. This makes it intelligible for an agent to express her volitional conflict as an expression of her desire to act in line with her autobiographical narrative, accepting the conflict part of her life-story. Furthermore, due to the non-plasticity this might be the only way for the agent to give (authentic) expression to her foundational normative states as they are.

In other words, it is psychologically plausible to claim, pace Velleman and others, that the conflicted agent may experience a (greater) loss in her sense of self if she gives up expressing either of the volitions that constitutes the conflict. Agents can thus find themselves in a bind whereby getting rid of one of the opposing forces in their volitional make-up is worse than being conflicted.

We should resist the temptation to see this as a higher-order compromise that assumes coherency between the agent's foundational normative states or, in Bratman's and Frankfurt's terminology, with which the agent is satisfied. The country-loving pacifist may begrudgingly perform actions expressive of both conflicting foundational states: she may see her work as a medic as a betrayal of her pacifist ideals, and also see her social media activity as a betrayal of her love for her country. However, she may realize that expressing her conflict is the best course of action open to her and therefore be satisfied with her own dissatisfaction. She may realize this if she takes both conflicting values as her deliberative standpoint (as constitutive of her autobiographical narrative). In this way, she explicitly rejects coherence as the ideal state of her set of foundational normative states and lets the volitional conflict constitute her deliberative standpoint, choosing and acting autonomously. She does not take a step back to a standpoint beyond her foundational normative states.

Moreover, I take the experience of being satisfied with our dissatisfaction as a more often encountered phenomenon, especially for people who aim at a high level of accomplishment regarding foundational normative states in more than one domain (e.g., career and parenthood; friendship and elite sport). In such cases, sacrifices must be made. These sacrifices can easily lead to situations in which the agent is dissatisfied considering each value taken in its own right; she may even be dissatisfied with the balance she has to strike. However, she might be satisfied with the fact that she expresses herself as volitionally conflicted, as this is the best expression of her foundational normative states in light of their limited plasticity. In other words, the agent can make the volitional conflict her own, make it the basis of her life-story, and act autonomously based on it.

María Lugones (1987) beautifully illustrates this phenomenon in her writing. From Lugones’ texts, we learn that she struggles with accommodating two conflicting aspects that are each constitutive of who she is. In struggling, she strives to give (authentic) expression to her foundational normative states. She describes herself as someone with strong roots in Latino culture while also being a lesbian; this makes her move in two different worlds that create a volitional conflict in her. Within Hispanic culture, lesbianism is rejected as an abomination; within lesbian culture, there is no space to integrate Hispanic values. The expectations she has of herself are reinforced by how both her Latino and lesbian social environments allow certain ways of expressing herself in these respective roles and foreclose others (Lugones, 1992). She therefore experiences volitional conflict, always falling short of one ideal. Through an explorative process, Lugones comes to embrace the conflict as constitutive of her identity. She writes, ‘I do not know whether the two possibilities can ever be integrated so that I can become, at least in these respects, a unitary being. I don't even know whether that would be desirable’ (Lugones, 1990). In giving expression to being a divided being, Lugones seems to have found a way to give autonomous expression to her foundational normative states.²¹

In summary, even if we accept that a normative pressure towards coherency stems from our foundational normative states taken individually (the SPP), it is a psychologically real possibility that a normative pressure against coherency can arise from our foundational normative states taken as a set. In a conflicted agent's search for how she can give autonomous (authentic) expression to her foundational normative states, the agent explores where she stands vis-à-vis the volitional conflict. Whereas sometimes she may find that giving in to the pressure towards coherency from an individual foundational state is expressive of her (most authentic) self, it seems equally plausible that the normative pressure against the coherency of her foundational normative states is most expressive of who she is.

Importantly, this conflicted agent may still feel essentially torn and on one level dissatisfied with the volitional conflict. Despite this being the case, I have argued that this does not undermine her capacity for autonomous deliberation, choice, and action.²²