Repetti's Buddhism, Meditation, and Free Will (2019) is composed of two major arguments. The first is a comprehensive defense of soft compatibilism. Repetti's argument against incompatibilism is summarized by others and is the focus of Gregg Caruso's comments, and I will leave the evaluation of this multifaceted argument to them. My comments are related to Caruso's, however, in that I aim to more fully specify the semicompatibilist account of free agency that is supposed to be the basis of moral responsibility. Second, although many of Repetti's arguments against libertarian and incompatibilist positions do not rely on any distinctly Buddhist assumptions, and thus can stand alone, Repetti argues (1) that Buddhist Meditation enhances mental freedom and autonomy, and (2) that his soft compatibilism coheres best with Buddhist core ideals and aspirations. Repetti emphasizes the accounts of the unlimited mental freedom of the Ārya meditation virtuoso, and he argues that the Buddhist conception of the Ārya presupposes robust (near unlimited) mental freedom, and thus metacognitive and volitional control (Repetti 2019, 20). Karin Meyers discusses his specifically textual argument in her comments and I also leave it aside. My focus is on Repetti's nontextual, philosophical argument for free will and specifically on the conception of autonomy that it presupposes.

Repetti argues that meditation enhances mental freedom, increases volitional control, and thus free will. This argument is largely based on the first‐person phenomenology of meditation. Repetti claims, relying on his own experience, that in intense meditative practice one has direct evidence for free will. In particular, he argues that when meditating, one can experience oneself as distinct from the five aggregates, which are supposed to constitute and exhaust the self. He concludes that Buddhist meditation thus provides evidence for agency and perhaps an agent. Repetti thus cautions us against “the increasingly viral meme in the exchange between Buddhism and Western … culture's embrace of mindfulness—no self; thus no autonomous self” and argues that this inference “is misleading and dangerous” (Repetti 2019, 5; italics added).

As a result, Repetti is at best ambivalent when it comes to the Buddhist doctrine of no‐self (Repetti 2019, 19–20). Recently, there has been a wave of interest in the possibility of “agentless agency” (Siderits, Thompson, and Zahavi 2011). Repetti is skeptical, and at best agnostic, about the thesis that human agency is agentless. Instead throughout his book, Repetti flirts with a “Personalist” or Pudgalavadin view of the self. Pudgalavadins argue, roughly, that the Buddhist doctrine of no‐self is compatible with a conception of a unified person that is the subject of karmic responsibility (Carpenter 2014, 2015). Pudgalavadin conceptions of the self were common in early Indian Buddhist debates about the nature of the self and karma. Although all such data is contested, it is possible that “twenty‐five percent of Buddhist monks in India in the seventh century CE were Personalists [Pudgalavadins]” (Strong 2015, 132–34). ¹ Although Repetti is clearly attracted to the Buddhist Pudgalavadin view, he does not develop or defend it at all (Repetti 2019, 10, 28–30, 94–97, 100, 127–28, 148).

In this comment, I agree with Repetti that the concept of the self or person seems to be necessary for agency and responsibility. But unlike Repetti, I have no doubt that the doctrine of dependent origination and no‐self are the foundational doctrines of Buddhist metaphysics—and crucially Buddhist virtue ethics. Nonetheless, I agree that the Buddhist path presupposes robust agency. I will say more about the difference between our two views in a moment, but to put my cards on the table, my view embraces the viral meme: no self; thus no autonomous self. On the other hand, I hope to contribute to his project by sketching a conception of agency that fits nicely with his defense of semicompatibilism.

In the end, I defend a Buddhist‐inspired conception of ego‐less agency, and argue that Kantian conceptions of autonomy are incompatible with the doctrine of no‐self. I argue that Repetti's defense of free will relies on what seems to be a Kantian conception of autonomy. But I also argue that Kantian autonomy is incompatible with the doctrine of no‐self. If I am right, then he needs to either defend free will without autonomy or more fully develop a non‐Kantian conception of autonomy. The literature on autonomy is vast but Repetti focuses on the arguments of Harry Frankfurt and John Martin Fischer in his defense of free will, and so I too will start with their distinct contributions. After discussing Frankfurt and Fischer, I will first argue that Kantian autonomy, including Korsgaard's recent development of Kant's approach, is incompatible with the doctrine of no‐self. Second, I consider Frankfurt's more recent non‐Kantian conception of autonomy and also reject it. Here, I build on David Velleman's critique of Frankfurt and defend Velleman's alternative minimalist conception of the self, which I suggest provides a plausible contemporary interpretation of the Buddhist doctrine of no‐self.

FRANKFURT ON META‐VOLITIONAL CONTROL AND FISCHER ON REASON‐RESPONSIVENESS

Repetti's compatibilism relies on both Frankfurt's hierarchical conception of the will and Fischer's conception of reason‐responsiveness. Although both of these conceptions of the will are useful for fleshing out a compatibilist approach, they are also distinct. Indeed, the emphasis on reason‐responsiveness was introduced in part to overcome perceived problems with simple hierarchical accounts of free will. Repetti focuses on hierarchical structure and control and seems to assume this includes reason‐responsiveness, but these are distinct capacities. He argues that meditation increases mental control, but he does not explain how it increases reason or Dharma‐responsiveness. The obvious answer is that meditation increases one's awareness of dependent‐origination and no‐self, and the dissolution of the self in turn increases lovingkindness and compassion—but Repetti never makes this argument. He instead claims that his practice of meditation increases his agential control over his self, and argues that agential control is best captured by the concept of “source autonomy.” In short, he argues that the practice of meditation provides evidence of an autonomous self.

To set the stage, here is a quick sketch of the Frankfurt's view, the problem with it, and Fischer's response (Frankfurt 1971; Fischer 1987; Ekstrom 2000). Frankfurt points out that in addition to first‐order desires and inclination, humans also have second‐order desires that take other desires as objects. For example, I am tired and so I'm inclined to go to sleep, but I also want to stay awake, and thus develop a (second‐order) desire not to want to fall asleep. Second‐order desires, when they are effective in motivation, add an internal element of control over first‐order desires. Frankfurt's early account identifies this higher order control with freedom of the will. The obvious question to ask is why having second‐order desires in itself increases freedom. If an organism acts simply on the strength of first‐order desire, then the desires are in control, and not a person. Frankfurt calls this sort of being a “Wanton” because they are controlled by desire. The initial Frankfurtian assumption is that the second‐order desires reflect the perspective of the person and their evaluation of their own first‐order desires. An effective second‐order desire is thus a candidate for an act of will, a volition. But this conclusion is hasty. That a desire is higher order may be necessary, but it does not seem to be sufficient for a full account of the will. Willing an end is a reflective process, and presupposes a hierarchical conception of motivation. Nonetheless, second‐order desires can also be alien to the person's perspective and thus also not a reflection of the person's will. Gary Watson has argued that we need to distinguish valuing and wanting and a free will presupposes an additional valuation system of preferences, which is distinctive from the hierarchical structure of desires (Watson 1975). On Watson's account it is the match or “mesh” between values and motivations that matters for free agency and responsibility. This of course raises new questions about the distinction between value judgments and mere preferences. If I experience what I take to be irrational guilt (e.g., guilt that results from internalized values that I reject), my guilt may take my desires as its object but still not reflect my will.

In response to this type of objection to his simple hierarchical model of the will, Frankfurt (1987) has argued that willing also includes, and requires, a person's identification with their motivating desires. Identification, he argues, is an additional necessary condition of freedom of the will. Is “identification” alone sufficient for freedom of the will? I think not. In the penultimate section, I will return to Frankfurt's account of “identification” and his more recent account of autonomy as grounded in “contingent volitional necessities.” For our purposes now, simply note that we can distinguish brute unreflective identification with preferences and a more reflective endorsement of preferences (or character traits) that is based on reasoning. When I act based on my reasoning, my actions are more distinctly my own. Impulsive or unreflective action does not involve deliberative agency at all, even if it is hierarchically coherent. As we will see later, Frankfurt realizes that he must modify his account of identification to respond to this concern (and other problems). As I will argue, my own view is that we should switch tracks and follow John Martin Fischer approach instead. He has argued that freedom of the will and moral responsibility presuppose that a person is reason‐responsive.

Although Repetti does not emphasize its distinctive role and contribution to the idea of meta‐volitional control, he also incorporates Fischer's conception of reason‐responsiveness into his defense of compatibilism. Fischer's central idea is that Frankfurt's simple hierarchical conception (or Watson's valuational model) is “purely structural and ahistorical,” whereas “moral responsibility is a historical phenomenon; it is a matter of the kind of mechanism that issues in action.” A purely structural theory is completely insensitive to the type of mechanism that causes the elements of the self to cohere. As incompatibilists argue, “The mesh between the elements of different preference systems may be induced by electronic stimulation, hypnosis, brainwashing, and so on” (Fischer 1987, 104–05). Free agency in contrast presupposes that the person's motives (and character) are reason‐responsive (Fischer 1987, 99). Fischer also argues that as long as we are reason‐responsive it doesn't matter if determinism is true or if indeterminism is true. This view is called semicompatibilism and this is also Repetti's view. ²

This brings me back to Repetti's argument. Repetti rightly emphasizes the meta‐cognitive control that is a common result of the practice of meditation. As Fischer emphasizes, however, it is not enough that the “mesh” between volitions is induced by meditation, as it might also be induced by hypnosis or electronic stimulation. Volitional control, even when enhanced by mediation and mindfulness, is a mere means and can be directed toward any set of ends. This is one of the main criticism of corporate McMindfulness. ³ As Fischer argues, for responsibility and free will, the mechanism in question must also be reason‐responsive. Meditation as a mechanism of meta‐cognitive control must also be reason‐responsive, and this is why the Buddhist path also includes morality and wisdom. Mental freedom, as Repetti argues, thus requires Dharma‐responsiveness. Indeed, in addition to meta‐volitional control, meditation is also supposed to provide insight into the Dharma; specifically an awakening to impermanence, dependent‐origination, and no‐self. And awareness of dependent origination and no‐self is supposed to undermine egocentrism and increase lovingkindness and compassion. My main reservation about Repetti's defense of free will is that he does not incorporate the interdependent relationship among meditation, insight, and virtue into his Buddhist conception of free will. If he did, I argue that the doctrine of no‐self would play a more central role.

AGENCY, DEPENDENT ORIGINATION, AND NO‐SELF

As I mentioned already, Repetti is clearly inclined to think that agency requires an agent and thus a self, and he argues that we should not treat the Buddhist no‐self doctrine as sacrosanct. He writes,

My intuition is that regardless of how much the no‐self doctrine is repeatedly asserted to be the central doctrine—if not the sine qua non—of Buddhism, it is an open question to what extent elements of Buddhism imply that there is agency or even an agent. There are no sacred cows in philosophy … the only mistake the Pudgalavadins (followers of the person‐affirming doctrine) made was to fade into extinction. (Repetti 2019, 147–48)

My own view is that the doctrines of impermanence, dependent‐origination, and no‐self are the foundational metaphysical doctrines of Buddhism because they are essential to the Four Noble Truths and the Buddhist path. The essence of the “Four Truths” is that Dukkha (understood as psychic suffering, unease, and/or dissatisfaction) is rooted in a fundamentally mistaken, even delusional, self‐conception. The source of suffering and discontent is ignorance of impermanence, of the nature of suffering, and of the truth of no‐self.

In the Buddha's first teaching, he presents the 12‐linked Chain of Dependent Origination that constructs the self and leads to suffering (and is also the cause of rebirth). The 12‐linked chain also emphasizes the embodied, emergent, and cyclical and recursive nature of self‐constitution. Here is a quick interpretive summary. ⁴

The chain starts with (1) Ignorance of the nature of reality, specifically ignorance of impermanence, the nature of suffering, and the truth of no‐self. (2) Volitions and Intentions form in ignorance, and they determine (3) Consciousness in the next life. In the Buddhist scheme of rebirth and karma, we are born into each new life based on the volitions and attachments formed in our previous lives. Reinterpreted within a life (without rebirth), my current self, my narrative and practical identity, is the product of the volitions and attachments that have formed over the course of my life; but this identity is formed in ignorance of the nature of reality, including the real nature of myself; and this is the source of unease, clinging, stress, anger, and thus dissatisfaction.

Steps 4–9 elaborate 1–3 and emphasize the embodied nature of consciousness and the dependence of our perceptions, sensations, feelings, and desire on our particular perceptual capacities. At (re)birth, we are born into (4) a Sentient Body with (5) Six Senses (the regular five plus inner awareness), and our senses determine our (6) Sensory Stimulations, which determine (7) Hedonic Feelings of pleasure, pain, or indifference. From this base, (8) Desire or Attachment and Aversion forms, and extends itself into (9) Egocentric Appropriation, which involves thinking of things as “me and mine.” And all of our egocentric appropriation provides (10) the Roots of Becoming and (11) a Renewed Consciousness (i.e., rebirth). In short, egocentric attachment and appropriation constructs our sense of ourselves; our narrative and practical identities. The last step in the chain (12) is Old Age and Death, which marks and starts the repeat of the cycle.

On my reading, the 12‐linked chain is a description of the cyclical process that constructs the minimal self out of the aggregates, including its practical and narrative identity. The Buddha's original chain of dependent origination does not include the centrality of a practical and narrative identity, which is a focus of my discussion below. The process of “egocentric appropriation,” however, is easily understood as a process of forming and identifying with a practical identity and at least weakly narrative self‐conception. In addition, the chain of dependent origination pinpoints the centrality and distorting effects of ignorance, egocentric appropriation, and metaphysical essentialism. The 12‐linked chain starts with volitions and attachments rooted in ignorance, and our practical identity is thus formed in ignorance of three characteristics of existence, which are impermanence, suffering, and no‐self. Ignorance is a central causal condition of the egocentric identifications that are the source of unhappiness. Liberation from the chain of dependent origination thus requires an awakening to the fundamental interdependent nature of the self and reality.

Understanding the doctrine of no‐self and changing our self‐conception is the prescription for transcending suffering. Enlightened action requires re‐conceiving and re‐constituting oneself so as to reflect the thoroughly impermanent and interdependent nature of all things, and thereby eliminate ego‐centrism. On my view, Buddhist insight meditation is aimed at removing ignorance, and the foundation of this ignorance is belief in an autonomous self. Internalizing a new conception of the self as impermanent, dependently originated, and interconnected is a cognitive component of the four virtues (the Brahma‐viharas) of lovingkindness, compassion, sympathetic joy, and equanimity. In this way, the Buddhist path increases reason‐responsiveness. I elaborate on this process in the final section and also see Karin Myers comment on Repetti for a fuller account.

On the other hand, intentional action does require situating oneself in relation to others and the world. Following the Buddhist eightfold path, which includes developing the Brahma‐viharas (the four immeasurables), requires robust agency. The self is dependently originated, but it nonetheless has the capacity for robust agency. It seems clear that we need a dependently originated conception of personhood that is a sufficient basis for robust agency. So interpreted, the doctrine of no‐self is compatible with personhood in the sense of a robust agent. This conception of “no‐self,” which will include a minimalist conception of the self, is rightly characterized as a contemporary version of the lost Pudgalavadin conception of the person.

Although we are both sympathetic to Pudgalavadin conception, Repetti also argues that meditation and mindfulness provide a self‐reflective perspective and executive control that is over and above the five aggregates (which are physical bodily form; sense perception; sensations of pleasure and pain or indifference, volitions and emotions; and consciousness of physical and mental states). In particular, he describes his own meditation experience as embodying a perspective distinct from, “a space between,” the five aggregates that are supposed to constitute the self (Repetti 2019, 20). In discussing meditative distance from the five skandhas, or aggregates of the self, he writes that:

[It is] this space in which all thoughts, interpretations, evaluations, judgments, intentions, emotions, and potential actions are optional, are not constitutive of me—that informs my belief that I am free to choose among them … The fact that I am not these things—that I am not any of the aggregates—does not entail that I lack this feature of mental autonomy, however small its range. Nor does the fact that I am able to experience these phenomena as ‘not me’ entail that nothing is me, or that I am not, for I may well be this executive awareness itself, this agency itself, or some more subtle combination of all of these phenomena. We can leave open the question of what I am for now. (Repetti 2019, 20–21; original italics)

The rest of my comment takes up and focuses on this open question. First, it should not go unnoticed that the suggestion that “I” might be “executive awareness,” “agency itself” sounds remarkably Kantian. Although, as far as I can tell he never fully commits himself, Repetti's conception of the self and source autonomy seems to be more Kantian than Buddhist. In particular, this passage is reminiscent of Korsgaard's Kantian conception of autonomy and of the self as “over and above” the aggregates. Repetti's line of reasoning, however unintended, suggests an implicit commitment to a conception of the locus of agency as somehow ‘over and above the aggregates,’ a conception that appears to run counter to standard Buddhist interpretations. Since Repetti explicitly includes executive control in the space between the aggregates, I follow Korsgaard and often refer to this as conception of the autonomous self as over and above the aggregates. In the rest of my discussion, I will first develop the parallel with the Kantian approach, and then sketch an alternative non‐Kantian view of agency, that I believe better fits Repetti's Buddhist compatibilism. I will argue that, yes, Buddhists need a robust conception of agency, but nonetheless should reject a Kantian conception of autonomy because it is inconsistent with the doctrine of dependent origination. On the one hand, I argue that Repetti should reject the idea that the self is agency itself, but on the other hand, the view I develop also fits nicely with his main argument for compatibilism.

SOURCE AUTONOMY AND KANTIAN AUTONOMY OF THE WILL

Repetti contrasts “leeway autonomy” with “source autonomy.” Leeway autonomy is the leeway to do otherwise than what one actually chose. I think that it is natural to relate “leeway autonomy” with meta‐cognitive and volitional control. If I were to choose otherwise, then I would act otherwise. ⁵ My actions reflect my considered preferences. On my view, leeway “autonomy” or meta‐volitional control is not sufficient for free will. As explained above, this is the point of Fischer's objection to Frankfurt and Watson. Free will requires self‐regulation, which Repetti calls “source” autonomy. Although source autonomy is not subjected to a careful analysis by Repetti, he does develop the contrast with leeway autonomy, and identifies “source autonomy” with higher order, reason‐responsive, self control. Repetti explains that it is “the ability to have it be the case that what one chooses and does is ultimately ‘up to’ oneself, such that the agent is the ultimate causal origin of its choice or action”—and this ability is first equated with “self‐regulation” and then fleshed out with a counterfactual analysis of self‐regulation.

First, I have a problem with the idea that anything has an “ultimate causal origin” because everything's origin is embedded in a multifaceted scheme of dependent origination (Garfield 2015; Cummiskey and Hamilton 2017; Meyers 2018). Nonetheless, as I interpret Repetti, the agent is the “ultimate causal origin of its choice” in the requisite sense (for self‐determination and responsibility) when the agent is reason‐responsive and the agent's reasons determine choice and actions.

Since Repetti does connect source autonomy with reason‐responsiveness (and thus Dharma‐responsiveness), how is this connection related to his account of the phenomenology of meditation? He says that through meditation he experiences increased source autonomy. The focus on executive control and autonomy, includes volitional control, which goes beyond mere awareness or focused attention. It seems like what would be available to the meditator, or what is experienced, is a subject who can arbitrarily pick a mental focus, rather than a self that chooses for reasons—in the way I might arbitrarily select one can of soup from a shelf of 20, as opposed to the way I might choose a graduate school to attend. Repetti owes us an account of the connection here between what he experiences when meditating and reason‐responsiveness. ⁶

What exactly is the connection between being a meditation virtuoso and reason‐responsive free will, which presupposes wisdom and virtue. Even if we grant that meditation increases leeway autonomy (understood as basic volitional control), we still need an analysis of the relation between the practice of meditation and source autonomy (understood as reason‐responsiveness). How does meditation make us more reason‐responsive? These are the questions that I think need significantly more exploration by Repetti. Since Repetti focuses on autonomy and suggests that the self might be “executive awareness” or “agency itself” (21), Kant's conception of the autonomy of the will is an obvious starting point for additional exploration and comparison (for a discussion of Buddhist Modernism and Kant, also see Cummiskey 2018).

To summarize, as Repetti emphasizes, source autonomy requires reason‐responsiveness. And the idea of reason‐responsiveness essentially presupposes an account of justifying reasons. Repetti centers his account of free will on the Kantian idea of autonomy and to be autonomous is to be self‐legislating. Autonomy contrasts with heteronomy, which refers to actions motivated by, or unduly influenced by, factors external to the agent. Source autonomy presupposes that my motivating reasons are not external or “alien” causes. But as we saw above when discussing Frankfurt, in cases of addiction, compulsion, or impulsive anger, my own desires can overcome me and thus be alien to me. For action to be an act of will, we must recognize reasons and identify with the desires that motivate actions (or inaction).

Rousseau famously made this point, which inspired Kant's account of autonomy as pure practical reason. Rousseau writes, “For to be driven by appetite alone is slavery, and obedience to the law one has prescribed oneself is freedom” (1762, Bk I ch. 8). Kant picks up on this, in his distinctive manner, and argues,

The will is a kind of causality belonging to living beings in so far as they are rational; freedom would be the property of this causality that makes it effective independent of any determination by alien causes … [this] explanation of freedom is negative and is therefore unfruitful for attaining insight into its essence; but there arises from it a positive conception of freedom, which is richer and more fruitful. (Kant 1785, 446)

As I understand Repetti, leeway autonomy is the same as Kant's negative freedom. Negative freedom, however, is not sufficient for self‐regulation. To paraphrase, Kant argues that since the will is a kind of causality, and causality is not lawless, freedom of the will also is not lawless. An uncaused action is also not caused by me, and thus it is not a product of my will. If we agree that a free will must be independent of “alien causes,” then the will also cannot be caused by mere natural laws that determine my actions without regard to my will. It follows, Kantians argue, that my rational nature itself must be the source of the principle (or law) that motivates my action.

If we focus on the first person deliberative perspective, the distinction at work here is clearer. There is a difference between trying to predict an action (even one's own action) and deliberating and deciding which action one should take. We see that deliberation assumes that my will is free in the sense that it is governed by my deliberation, and not by other external causes. As Repetti emphasizes, when deliberation determines my actions that does not undermine my free agency. In this sense, the deliberative perspective presupposes practical freedom. From this Kant famously concludes,

[W]hat else can freedom of the will be but autonomy, that is, the property that the will has of being a law unto itself? The proposition that the will is in every action a law unto itself expresses, however, nothing but the principle of acting according to no other maxim than that which can have at the same time have itself as a universal law for its object. Now this is precisely the formula of the categorical imperative and the principle of morality. Thus, a free will and a will subject to moral laws are one and the same. (Kant 1785, 447; italics added)

To be a law unto oneself is to give oneself laws; laws have the unconditional form of categorical imperatives and are necessarily universalizable maxims; and universalizable maxims also must treat rational nature as an end‐in‐itself, and not a means only.

Much ink has been spilled on evaluating the soundness of Kant's argument, and this is obviously not my focus here. My point is that since Repetti identifies a free will with source “autonomy,” his argument is incomplete without a fuller, more substantive, account of autonomy of the will. I assume that Repetti does not accept the Kantian account of autonomy of the will, but instead wants to (un‐controversially) link source autonomy with reason responsiveness. But the idea of source autonomy calls out for more specification; and in particular a specific account of the source of reason. We have already seen that Repetti also suggests that “I may well be this executive awareness itself, this agency itself” (Repetti 2019, 21), which certainly sounds like a classic Kantian view. In short, if Repetti does not accept the Kantian conception of pure practical reason, with its companion conception of morality (as a system of categorical imperatives and universalizable maxims), what is his alternative (Buddhist) account of the source of normativity? In the next section, I sketch Christine Korsgaard's contemporary Kantian conception of practical reason, and then in “Velleman's ‘Buddhist’ Critique of Frankfurt's Contingent but Essential Self” section, return to Frankfurt alternative, essentialist, non‐Kantian conception of autonomy.

KORSGAARD ON PRACTICAL IDENTITIES AND SELF‐CONSTITUTION

Buddhist ethics focuses as much on character as it does on actions. The Buddhist eightfold path involves a total reorientation of the self. As explained above in our discussion of the Buddha's first teaching (“Agency, Dependent Origination, and No‐Self” section), our self is formed, constructed, and continually reconstructed in a state of fundamental ignorance, and it must thus be reformed and re‐constituted in light of the wisdom of the Dharma. The Buddhist path is simply a path of increased Dharma‐responsiveness that optimally leads to full awakening. I also argued, and Repetti agrees, that reason/Dharma‐responsiveness is a necessary condition for free will (“Frankfurt on Meta‐Volitional Control and Fischer on Reason‐Responsiveness” section), and it thus provides the normative basis for self‐constitution. In the last section, I argued that Repetti's account of source autonomy at least appears to be a Kantian account.

Korsgaard's Kantian account of the source of our reasons, and of self‐constitution, provides a more contemporary starting point for thinking about a Buddhist account of reason‐responsiveness (1996, 2009). She begins by arguing that our practical identities are the source of our reasons.

Practical identity is a complex matter for the average person there will be a jumble of such conceptions. You are a human being, a woman or a man, an adherent of a certain religion, a member of an ethnic group, a member of a certain profession, someone's lover or friend, and so on. All of these give rise to reasons and obligations. Your reasons express your identity; your nature; your obligations spring from what your identity forbids. (1996, 101)

Korsgaard's conception of practical identity is a minimally narrative conception. She argues that our intentional actions both constitute and reflect our practical identities. Particular actions (that is, acts done for a purpose) are required by particular contingent identities, and in acting from that practical identity we actually constitute ourselves as persons (2009, 20–21). The normativity of our actions is rooted in our practical identities and we affirm our identity through our actions. As she explains it, “valuing yourself under a certain description consists in endorsing the reasons and obligation to which that way of identifying yourself gives rise” (2009, 24).

It is important to note that Korsgaard does not embrace an unduly voluntarist, disembodied, or unencumbered conception of practical identity. Our embodied, finite animal nature is also an essential part of our human nature. She also acknowledges that we are born into families and cultures and that our practical identities are initially largely unchosen. Of course, other identities, like one's profession, are reflectively chosen and cultivated. Whether initially chosen or unchosen, all of our particular practical identities are subject to self‐reflection and revision. The important point for Korsgaard is that our practical identities are normative and constitutive of us, and we endorse and reaffirm them through our particular choices and actions.

Following Kant, however, she also argues that we must endorse our identities from an autonomous perspective that is distanced from, and over and above, our contingent identities (also see Okrent 2001). In the end, she concludes that our rational nature is both an unchosen and an inescapable practical identity, and as such it is the (unconditional) source of normativity. This unconditional (first‐person) perspective is the perspective of a rational agent as such, and not based on any of our particular substantive identities. If we abstract away all of our contingent identities, all that is left, she argues, are formal principles of practical reason. At this point her argument follows Kant, and she argues that the unity of persons and the overall normativity of actions are determined by the coherence between (1) our diverse contingent identities and (2) the higher order constraints provided by formal principles of practical reason: Kant's Hypothetical Imperative and the Categorical Imperative.

This is just a sketch of Korsgaard's argument, which I develop more fully in another article (2011). ⁷ For our purposes here I would like to focus on the centrality of the detached autonomous perspective in Korsgaard's argument. Like Repetti, she argues that human beings experience themselves as an executive agency that is in some sense distinct from and “over and above” their contingent identities. From this premise, she eventually concludes that the unconditioned source of normativity is pure practical reason, which must itself be the purely formal principle captured by Kant's Categorical Imperative. If one takes up Repetti's suggestion that I might be “executive awareness itself” (2019, 21), then, as Kant and Korsgaard argue, it seems to follow that purely formal principles of reason must be the foundational principles of practical reason (and thus of reason‐responsiveness). Although I will not reproduce the objections here, like many others I am skeptical that one can derive substantive moral principles from the ultra‐thin notion of a finite rational being. The bare notion of a rational being is too thin to provide a basis for any substantive reasons for action. So, even if we have this one practical identity necessarily, it hard to see how one can use it as an Archimedean point for reflective endorsement. If we distance ourselves from all of our contingent identities, by definition nothing of substance is left. Of course, Korsgaard and Korsgaardians have much to say in response, and Repetti might choose to develop and defend her approach. What is the alternative, however, if we cannot derive substantive principles from Kantian pure practical reason?

Let me sum up my take away from the argument so far. Korsgaard's account of the normativity of our practical identities seems right to me. And it is true that, at least in principle, we can reflectively evaluate any of our particular identities and endorse or reject it based on reasons. The reflective, hierarchical structure of motives, values and reasons, allows us to “stand back” from even our dearest practical identities. In this sense, it is true that I can endorse or reject any practical identity. But it does not follow that I can distance myself from all of them at once. In evaluating any particular identity, the evaluative reasons are always themselves dependent on other values and principles that we provisionally accept. There is simply is no evaluative “view from nowhere.” ⁸ In addition, agency and subjectivity is always dependently originated. There is no perspective that is not causally/historically, perceptually and conceptual conditioned. There is no perspective of pure reason, or pure ego, or of an unencumbered and unconstructed self. This brings us back around to Frankfurt's conception of autonomy.

VELLEMAN'S “BUDDHIST” CRITIQUE OF FRANKFURT'S CONTINGENT BUT ESSENTIAL SELF

As discussed in opening section, Frankfurt first attempts to capture our self‐conception as autonomous by focusing on wholehearted identification with a particular conception of our selves. The first take on this idea was that agents are free and autonomous when they act on preferences that they reflectively endorse. As we have already seen this interpretation of autonomy leaves out the source of the reflective endorsement, and it thus could be based on factors clearly “external” to the person; for example, hypnosis or brain‐washing. Autonomous motivation must be internal so that the person is the source of the action.

Frankfurt thus amends his account and argues that “a person acts autonomously only when his volitions derive from the essential character of his will” (Frankfurt 1999, 132). ⁹ Of course this sounds too Kantian for Frankfurt, and he thus emphasizes and develops a non‐Kantian conception of autonomy of the will. In reference to Kant's account of the will as pure practical reason, Frankfurt writes,

[T]his pure will is a very peculiar and unlikely place in which to locate an indispensable condition of individual autonomy. After all, its purity consists precisely in the fact that it is wholly untouched by any of the contingent personal features that make people distinctive and that characterize their specific identities. (1999, 132)

By focusing on the contingent, and dependently originated, nature of persons, Frankfurt moves his account closer to one that might be compatible with a Buddhist approach. Unlike Kant and Korsgaard, Frankfurt argues that autonomy must reflect the contingent features of the person, but he also realizes that the concept of autonomy, of setting oneself laws, involves necessitation. To give oneself a law is to be bound by it; our principles bind us in a way that mere preferences do not. For Korsgaard our rational and finite animal nature is our essential nature and it is thus both essentially who we are and the source of the law that bind us. (And it is also the basis of our fundamental equality.) For Frankfurt, autonomy must both necessitate and be contingent, and he thus argues (1) that we each have a distinctive essential, inescapable self, and (2) that this essential self is volitionally constructed, and thus contingent. The past process of self‐construction is contingent, nonetheless (he argues) the future self is essentially constrained by a volitional core of the present composition of the self. Although the self is contingently constructed, the self that is constructed is essentially who we are and its elements are thus inescapable. Autonomy of the will is thus analyzed as based on the “contingent volitional necessities” that constitute our essential self. In short, our essential self is thus dependently originated in that it has a contingent causal origin, but according to Frankfurt our contingent self becomes an essential self by a process of volitional necessitation.

Our present self is constructed by our past self, which was itself contingently constructed. There is nothing controversial in that. There is a trivial sense in which my future self depends on my current self. But what does the concept of “volitional necessity” add? Why also think my current self is now essentially who I am in such a way that it constrains the identity of my future self in some nontrivial sense? Why think the core of the self is unrevisable? If we focus on the ability to “stand back” from any of our practical identities, and evaluate them in light of one's reasons and values, there is no “volitional necessity.” At best it might be true that some values are so central and well‐grounded that the person cannot imagine ever having good reason to change them. Frankfurt's focus on essential identification, however, implies that they could not be rejected even if the person believed that they had sound reasons for doing so. Although a person's reasons will always be based on their current (contingent) practical identities and self‐conception, this simply does not make any particular identity essential or necessary.

As David Velleman points out, the really interesting thing about this essentialist self‐conception is that it is clearly false but also so attractive (2006). Velleman writes:

In light of how implausible the notion of personal essences can be when applied to particular cases, we have to wonder why it remains so attractive. Frankfurt never offers us a convincing example of someone who ceases to be the same person because of abandoning a project, betraying a commitment, or undergoing some other change of heart; nor does he offer any argument for thinking that motivational changes can have such momentous results. He simply asserts that our projects and commitments are sometimes essential to who we are. We welcome his assertion, but not as something of which he has convinced us; we welcome it as something that we, too, want to say about ourselves. The question is why we want to say it. In the absence of examples or arguments to show that we have motivational essences, what moves us to apply this self‐description? (2006, 341)

From a Buddhist perspective, Frankfurt has clearly captured our subjective experience of Egocentric Appropriation, which is the ninth‐link in the twelve‐linked chain of dependent origination (see “Agency, Dependent Origination, and No‐Self” section). Egocentric appropriation is the process by which we segregate both our passions and the world into that which is viewed as me or mine, and that which is other and not‐mine; and this process of appropriation is itself central to our (deluded) narrative self‐conception. Velleman also argues that the fiction of an essential inner‐self is an unhealthy delusion. Velleman focuses on Frankfurt's conception of “wholeheartedness.”

[W]holeheartedness is an object of wishes that do not necessarily represent a healthy trend in our thought. Our attraction to the idea of being wholehearted is one manifestation of the fears that move us to defend ourselves against our own emotions. Hence our affinity for Frankfurt's ideal may not indicate that he's right about the constitution of the self; it may indicate no more than our own defensiveness … When we defend ourselves against unwelcome emotions, we would like to think that we are expelling, excluding, or (in Frankfurt's term) “extruding” them from ourselves. We are trying to neutralize troublesome elements of our psyches, and one way to neutralize troublemakers is to banish them beyond some enforceable boundary. When we picture the inner sanctuary of the self, we are picturing a defensible territory—which is precisely what's needed for successful defenses. Given our wish for this safe haven, however, our belief in its existence may be another case of wishful thinking. (Velleman 2006, 346–47)

Velleman concludes:

When Frankfurt describes us as identifying with some of our motives and alienating others, his description rings true, I suspect, because it accurately describes this common defensive fantasy. We do indeed identify with some of our motives, but we thereby engage not in self‐definition but self‐deception. We identify with some of our motives by imagining ourselves as being those motives, to the exclusion of whatever might complicate or conflict with them. (2006, 353)

There are no actual “contingent volitional necessities” that could constitute an essential self, and the belief in an essential self is an act of self‐deception. We want to think of ourselves (and others so as to blame or praise them) as having fundamental and unchanging essences, but there simply is no such thing. Velleman follows Sartre and characterizes this type of self‐essentialization as “bad faith.” Buddhists argue that this self‐delusion actually is the source of unhappiness, unease, and discontent.

PRIMAL CONFUSION, BUDDHIST PERFECTIONS, AND EGO‐LESS AGENCY

The Four Noble Truths are focused on the nature and source of suffering and the path that transcends suffering. Suffering is the result of a mistaken conception of ourselves and our orientation to the world. As Jay Garfield emphasizes, suffering results from ignorance, and this ignorance is the result of a “primal confusion” best captured in the idea of a “twofold self‐grasping.” The “first grasping” is to grasp oneself as a privileged subject and assign special ontological and moral importance to the referent of the ‘I’. “The second, which follows from the first, is to see everything else as existing in relation to the self, as “mine” [or not mine] … to experience reality through this structure is irresistible, perhaps an essential character of human phenomenology. It is the view that takes us each to live at the center of the universe … and is the view reflected in the indexical system of every natural language” (Garfield 2015, 9–11).

The Kantian picture of source autonomy, with an independent autonomous self and pure practical reason, is itself a consequence of the primal confusion that is the source of suffering. Repetti is correct that “Meditative practice increases mental freedom and free will” (2019, 21; original italics), but he needs to also emphasize that mediation increases freedom because it helps us overcome our primal confusion and thereby increases Dharma‐responsiveness.

Returning to Korsgaard's conception of practical identities, the problem is that our practical identities are shaped by our primal confusion and twofold self‐grasping. The soteriological point of Buddhist meditation and philosophy is to reorient our practical identities, so that we “awaken” and see the world as it really is, and thereby transcend suffering (Garfield 2015, 11–14). On the view I am defending, we still need a narrative self and practical identity to engage and act in the world, but we must resolve and work to reconstitute our conception of our self. Our core practical identities must be reconstituted so that they have a fundamental nonegocentric focus (Mackenzie 2013, 2015, 2016, 2017). ¹⁰

In addition to cultivating mental freedom through meditation and philosophical reflection, we reorient ourselves, and become Dharma‐responsive by cultivating virtue—which is also accomplished through meditative practice and increased mindfulness. In particular, we can reorient ourselves by developing the Brahma Viharas, the Four Immeasurable Virtues of lovingkindness, compassion, sympathetic joy, and equanimity. Karin Myers develops this more fully in her comments on Repetti, but the basic idea here is simple enough. By developing these dispositions, we undermine and counter‐act our egocentric dispositions. Each of the Four Immeasurables has an opposite afflictive emotion that must be overcome if we are to reorient ourselves (Mackenzie 2016; McRae 2016; Heim 2017). The afflictive emotions are afflictions; anger and jealousy, for example, disrupt one's inner peace and bring no real satisfaction. Cultivating the virtues counteracts and cures us of these afflictions; and these virtues are constitutive of a nonegocentric re‐constitution of the self. Developing loving‐kindness counteracts ill‐will, anger, and hatred. Developing compassion counteracts malice and cruelty. Developing sympathetic joy (in the good fortune of others) counteracts envy, resentment, and jealousy. Developing equanimity counteracts indifference and egoism. Equanimity is more than impartiality; it involves expanding one's circle of care and concern to encompass all people and indeed all sentient beings. The perfection of virtue is boundless love and compassion. Of course, this is a lofty and distant ideal—the perfection of lovingkindness, compassion, sympathetic joy, and equanimity is an immeasurable and boundless state that (perhaps) no one can achieve. Nonetheless, and crucially, mere progress toward the ideal diminishes egocentrism, mitigates existential suffering, and increases inner peace.

The Buddhist eightfold path, which includes developing the Four Immeasurables, presupposes a robust conception of agency and self‐constitution. And self‐constitution presupposes first‐person agency. First‐person agency, however, does not require the autonomy and independence of the Kantian self. Although the minimal self and narrative self are useful explanatory concepts, accepting their nonreducible status does not imply that there is a metaphysically substantial, or otherwise deep or essential self.

Instead, the narrative self is thoroughly bound up in the chain of dependent origination. The goal is thus to reconstitute ourselves so that our practical orientation accurately reflects its interdependent and nonegological nature. To be clear, on my minimalist conception of the self, we do experience the world from a particular point of view and our subjectivity does have a “zero‐point” that orients us in relation to the world of others and objects. The zero‐point, however, is not ontologically or normatively significant. We also have a thin and “episodic” conception of ourselves as the subject of past experiences and a thicker but distinct conception of ourselves that is constituted by identification with our personality traits (Nichols 2014; Hogan and Nichols 2015). It is also the case that humans have higher order recursive capacities that provide the basis for self‐reflection and normative evaluation. None of this, however, requires an essential self or unconditioned autonomous self, and it certainly does not justify an egocentric normative orientation.

Following his critique of Frankfurt's volitional essentialism, David Velleman defends what I believe is a useful Buddhist minimalist conception of the self. I quote Velleman at length in order to let him speak for himself rather than attribute to him what I take to be Buddhist minimalism in a nutshell (compare MacKenzie 2015, 2016, 2017).

In my view, “self” is just a word used to express reflexivity—that is, the coincidence of object and subject, either of a verb or of the activity that it represents. (“She accidentally cut herself.”) … As a word expressing reflexivity, “self” has various uses in various contexts, including several contexts that are of interest to philosophy. “Self” can express the reflexivity of the control that an autonomous [sic] agent exerts over his own behavior; the reflexivity of the memories and anticipations that link a temporally extended person to his past and future; or the reflexivity of any first‐personal attitudes that he may hold. Although “self” expresses reflexivity in each of these contexts, there is no single entity to which it refers in all of them. We shouldn't assume, in other words, that there is something called The Self that governs a person's behavior when it is self‐governed, persists so long as the person remains himself, and is the object of his self‐concept or self‐image … There is no kernel or core whose presence in past or future persons makes themselves of his; there are only the psychological connections that mediate his reflexive references to them, thus enabling him to think of them first‐personally… If a person's relation to past and future selves doesn't depend on a shared subset of attributes and attitudes, then it doesn't depend on anything that might be the object of his self‐regard. (Velleman 2006, 354–55)

The pervasive reflexivity of human thought leads us to think of ourselves as an essential core. This thin first‐personal and reflexive perspective, however, is not a suitable object of self‐regard. Indeed, our evaluative identification with our first‐person orientation is the source of the primal confusion and twofold self‐grasping that leads to suffering. Although this is a longer story, from a Buddhist perspective, our basic egocentric orientation also leads us to not appreciate the pervasive conceptual dependence of all experience, and the deeper nature of impermanence and dependent origination. In short, on my view, subjective experience and agency do require a minimal self, but our practical identity is a changing, shifting, and contingent construction. Agency is never truly unconditioned, and thus human freedom, as significant as it is for self‐transcendence, is never truly autonomous.

CONCLUSION

Although I have developed a conception of agency that is distinct from Repetti's focus on autonomy, I believe that the position sketched here fits nicely with his semicompatibilist defense of free will. Repetti emphasizes the centrality of agency and reason‐responsiveness, and my approach tries to link these elements of his view with a particular conception of the doctrine of dependent‐origination and no‐self. On my view, our mistaken self‐conception is indeed the core philosophical doctrine of Buddhism, which must be developed, defended, and incorporated into an adequate account of the source of normativity, reason‐responsiveness, and human agency. If Buddhist meditation actually supports the idea an autonomous self, over and above the aggregates, as Repetti suggests, this fact would undermine the primary Buddhist approach to normativity and reason‐responsiveness. In addition, the doctrine of no‐self also plays a central role in most (perhaps all?) Buddhist accounts of the virtues of equanimity and boundless compassion. If meditation actually reveals an autonomous self over and above the aggregates, then the centrality of the agent is not a primal confusion; which would mean that we need a new argument for the virtue of compassion. To avoid this result, we need a non‐Kantian account of agency that is consistent with the soteriological point of the doctrine of no‐self. My view is that agency does presuppose minimal subjectivity and personhood, but not a substantive self or Kantian autonomous agent; and that free agency, freedom of the will, is realized through ego‐less agency. On this interpretation of no‐self, the point is not that there is no subject of agency; the point is that the subject does not center their agency on themselves.

ACKNOWLEDGMENTS

I would like to thank Paul Schofield, Karin Meyers, Mark Siderits, and Rick Repetti for their comments on drafts of this article. I would also like to thank and acknowledge two of my students who wrote their senior theses on Buddhism and free will: John Ricatto (“Buddhist Agency and Emotional Resistance,” Fall 2017), and Paul Phillips (“Buddhist Perspectives on Western Reason‐Responsive Compatibilism: Transcending Reactive Moral Attitudes,” Winter 2018).