1  Introduction

Meditation practice allegedly enables practitioners to distance themselves from their own mental states, thereby increasing self-regulation over their mental states, on the one hand. However, as practitioners advance through the practice, they allegedly experience an increasingly diminishing sense of self, on the other hand, what Rick Repetti (2019) has described as ‘agentless agency'. But the very concept of ‘agentless agency' seems prima facie oxymoronic. In this chapter I focus primarily on how Thomas Metzinger (2015), Asaf Federman and Oren Ergas (2018), and Rick Repetti (2019) have addressed this dilemma. Then I develop a proposal that I think works to forge a path through the horns of this dilemma. However, this proposal results in a new dilemma, but one only facing Repetti's account.

2  The context

Buddhist meditative practice and the by-now widely popular derivative mindfulness-based interventions (MBIs) in psychotherapy (“mindfulness meditation” in general) are of philosophical interest largely due to their having seemingly paradoxical attributes related to self-consciousness and self-control. These meditative techniques, on the one hand, appear to enhance self-awareness, self-regulation, and therefore agency, with the ultimate result of becoming a meditation virtuoso (in Buddhist parlance, an ārya), a being who is able to “think or not think whatever thought he wants to, have the intention he wants to, etc.” (Repetti 2019, p. 157; emphasis in original). Metzinger (2015) calls the capacity for second-order control of one's first-order mental life “M-autonomy” (mental autonomy). By contributing to M-autonomy, meditation seems to bolster the self, turning an individual person into a more autonomous agent, less subject to external forces or internal impulses, and more self-directed. On the other hand, meditation seems to simultaneously diminish the experience of effortful control in the practitioner and undermine the sense of self that is in Buddhism negatively associated with ego-clinging and attachment, yielding a kind of “choiceless awareness” of one's mental life, without effort or judgment.

Recent work by Federman and Ergas (2018) and Repetti (2019) has investigated these phenomena specifically in relation to agency and free will, seeking to in one way or another resolve the paradox, but without making any metaphysical claims about the status of the self. Repetti, in particular, has argued that nothing in Buddhist philosophy and the soteriological aims of Buddhist practice entails that there cannot be such a thing as human agency, and in fact that meditative practice contributes to agency rather than diminishing it or demonstrating its illusoriness. He holds that the Buddhist description of the ārya allows for agency, free will, and therefore moral responsibility, in a way that evades the most powerful arguments against such capacities. Furthermore, he suggests that there is empirical evidence demonstrating that meditative practice does have the agency-enhancing effects that Buddhists claim it does, thereby demonstrating the real existence of agentive capacities. Furthermore, he argues that in addition to the reality of agency, it may be the case that there are genuine agents.

Even if one accepts the arguments of (most) Buddhists and naturalistic philosophers against the existence of a “metaphysically superlative” self, i.e., one that is permanent, unchanging, and wholly distinct from the flow of mental and physical events that comprise the life of a person, Repetti argues, there may exist some other kind of self. He holds that claiming that a “self” exists is a weaker claim than claiming that an “agent” exists, for selves are necessary for agents, but not vice-versa. Still, he sees no reason why there may not also be agents. And if agency requires an agent, then it seems there must be agents or agent-selves, because the reported experience and observed behavior of expert meditators suggest that, at least in some cases, genuine agency exists.

I argue in what follows that the only thing revealed and enhanced by meditative practices that might reasonably be called an “agent-self” is generic, devoid of distinctive personal qualities. This generic “agent-self” is not the sort of thing that can ground any distinction between individual persons. The “agent-self” in this case merely refers to the attentional and discriminative processes that allow the meditator, to the degree that they have become virtuoso meditators, to choose the thoughts they want, i.e., to choose the thoughts that yield the actions that contribute to greater M-autonomy and further detachment from first-order subpersonal mental states and experiences. The self of an ārya, insofar as she is an agent, is merely an agent of the Dharma, the set of principles and/or ritual behaviors adhered to by Buddhist practitioners with the ultimate goal of attaining ‘enlightenment', a complete awareness of the true nature of reality as emptiness, and an escape from the karmic cycle of rebirth with its inherent experience of suffering. Or in more modestly, and in more culturally, metaphysically, and religiously neutral terminology, the meditation virtuoso, insofar as she is an agent, is an agent of ‘mindfulness', which denotes the principles and behaviors themselves that engender greater awareness of and detachment from one's own mental states and awareness of their nature as insubstantial and fleeting, and subsequently engender the agentive mental capacity of M-autonomy.

The paradox of self and agency in meditation may thereby be resolved, for while meditation entails a rejection of any kind of self-identification, it is compatible with and perhaps constructive of a different conception of selfhood and agency. The virtuoso meditator may be an agent in some sense, but ceases to regard any distinctive personal qualities as essential to their personal identity, including any motivation besides that of increased mindfulness.

3  The cognitive phenomenology of meditation

In a typical beginner-level mindfulness exercise, the meditator is instructed first of all to attend to their own breathing. They are then told to notice any time they find their mind wandering, note that they have become distracted, and return their attention to the breath. The idea is that over time this practice will enhance their ability to hold attention fixed on breathing, and eventually on some other physiological or mental object without mind-wandering. According to Metzinger, intentionally moving one's attention from one inner state to another is a mental action, an example of M-autonomy, whereas mind-wandering is merely a mental behavior which is subpersonal and non-autonomous (2015). He claims that for most people, nearly two-thirds of their mental life is of the latter type.

Metzinger defines M-autonomy as a meta-mental ability or set of abilities, the key component of which is a veto ability over other occurrent mental contents,

the ability to control the conscious contents of one's mind in a goal-directed way, by means of attentional or cognitive agency. This ability can be a form of rational self-control, which is based on reasons, beliefs, and conceptual thought, but it does not have to be. What is crucial is the “veto component”: being mentally autonomous means that all currently ongoing processes can in principle be suspended or terminated.

(2015, p. 279)

Metzinger further describes M-autonomy as follows, saying that it:

is the capacity for causal self-determination at the mental level. It is based on a complex and graded functional property, which comes in three major degrees: the phenomenally represented knowledge that oneself currently possesses this specific ability, executed attentional self-control, and cognitive self-control.

(Id., p. 280)

Mindfulness meditation appears to enhance M-autonomy, which includes various types of self-control. Metzinger goes on to argue that M-autonomy and the related “epistemic agent model” (EAM), whereby an individual knows they have the capacity for M-autonomy, are necessary for having a first-person perspective and therefore for being a person. He writes that,

to possess a first-person perspective means to operate under a specific kind of conscious self-representation, a PSM [phenomenal self model] that portrays the system as an epistemic agent, as an entity that is actively searching for and optimizing its knowledge, for example by controlling its own high-level, quasi-symbolic processing as a cognitive agent … or by actively sustaining and controlling the focus of attention.

(Id., p. 281)

If Metzinger's account is correct, then mindfulness meditation enhances the cognitive capacities that are constitutive of having a certain kind of self and of being a rational agent. However, as more advanced steps in an MBI reveal, the eventual goal of the practice is not to exert intentional control over one's mental life, but in a sense to release that control. As Federman and Ergas put it:

The cognitive theory at the basis of these courses suggests that it is the thinking process itself that causes harm, not merely the contents of thoughts. Hence, an altogether different approach to thinking is necessary: instead of trying to control or change thoughts, it is better to accept their reality and “decenter” from them. As participants learn that even controlling attention is difficult and cannot completely counteract thought processes, they are encouraged to retreat into an even more passive mode of acceptance and letting be.

(2018, p. 730)

Metzinger recognizes this second level of mindfulness practice as well:

In advanced stages of so-called “open monitoring” meditation, however, the aperture of attention has gradually widened, typically resulting in an effortless and choiceless awareness of the present moment as a whole. Whereas in beginning stages of object-orientated mindfulness practice the meditator identifies with an internal model of a mental agent directed at a certain goal-state (“the meditative self”), meta-awareness of the second kind is typically described as having an effortless and non-agentive quality.

(2015, p. 289)

Further differentiating the two stages, Metzinger adds:

In the first case an EAM [epistemic agent model] is present, leading to a process that would still count as personal-level, whereas in the second case we have meta-awareness without an EAM. It is important to understand that these are distinct phenomenological state-classes. Interestingly, even the neural correlates pertaining to this difference between “trying to meditate” and “meditation effortlessly taking place” are already beginning to emerge (Garrison et al., 2013).

(Metzinger 2015, p. 289)

Hence, here we see another indication of the paradoxical nature of the mindfulness enterprise. Mindfulness practice begins by enhancing the processes constitutive of the agent-self, but then goes on to apparently jettison them in favor of a non-agentive kind of awareness. In the earlier stage one learns to intentionally choose which first-order states to attend to and retain and which to allow to pass away, in a sense curating the self. However, in the later stage, the meditator becomes a non-judgmental voyeur of their mental life, as if it did not belong to them.

By recognizing that their first-order mental states arise independently of conscious control, due to factors that are unconscious, the meditator learns that these aspects of their mental lives are, in a sense, not their own. By viewing their mental states as arising independently of their conscious will, the meditator learns not to identify with those mental states and so ceases to view them as essential to their self or identity. As Repetti puts it, “A common, liberating, meditation-based insight about thoughts may be contrasted with the Cartesian ‘I think; therefore, I am': the Buddhist's ‘I think; but I am not my thoughts'” (2019, p. 155).

The events that make up one's mental life now seem entirely contingent, such that the meditator can allow them to flow away from attention so that they seem to disappear entirely, having no further downstream effects on the meditator's mental life and subsequent behavior. As the quote just above from Repetti implies, this appears consistent with one aspect of the Buddhist philosophical doctrine of anatman or no-self, as presented in the classical Buddhist story of the monk Nagasena and King Milinda, according to which it is as much a cognitive error to identify oneself with any particular aspect of one's mental life or even of all the elements of one's mental life taken together as a whole, as it is to identify a chariot with any one of its particular physical parts or of the particular set of parts at one time (as any one of them may be swapped out without entailing the destruction of the chariot) (Menander and Nagasena 2015).

One way to think about the paradox is to describe the experience of mindfulness in terms of Strawson's (2009) distinction between self-as-subject and self-as-object. As an object, the self is merely the stream of mental events, viewed by the self-as-subject, whatever that may be. However, since these events are recognized as not under the volitional control of the self-as-subject, they begin to be regarded as not-self.

Not identifying with one's first-order mental states is liberating because it breaks the causal connection between individual mental states and between mental states and behavior. If a thought is something that just happens to arise in one's mind and is not something intentionally chosen or essential to one's identity, then one can allow it to pass away and out of one's mental life, which is helpful when a thought would otherwise have destructive consequences, leading to depression, distraction, or even self-harm. ¹

4  Repetti's Buddhist soft compatibilism

Mindfulness meditation plays a key role in the theory of free will that Repetti develops and argues for, and which he sees as consistent with the core of Buddhist philosophy. He follows Buddhist philosophers such as Santideva and Tsongkhapa in holding that “meditative mind training” is a means of attaining greater mental freedom, that:

practice taming the scattered mind, training emotions, noticing thoughts, feelings and intentions when, as, and how they arise, and cultivating the ability to observe them dispassionately, with detached equanimity, leads to greater insight and greater mental freedom.

(Repetti 2019, p. 152)

By cultivating the meditative abilities of “mindfulness” and “one-pointedness”, the practitioner is able to develop the kind of self that is understood by many compatibilists as necessary for free will, one that is maximally reasons-responsive, and hierarchically harmonious, the latter being Frankfurt's criterion (1971). The expert meditator, captured in the Buddhist figure of the ārya, possesses these abilities to an incredible, “titanic”, degree. The ārya can:

think or not think whatever thought he wants to, have the intention he wants to, etc. Those leeway abilities (to do X or not X) extend throughout all factors of the Eightfold Path, and illustrate having the sort of mind one wants to have – freedom of the mind, mental freedom. To the extent the ārya approximates mental freedom, she possesses mental autonomy and volitional self-regulation or free will – both leeway autonomy and source autonomy. She possesses leeway autonomy because if she wanted to think or intend otherwise, she could, … and she possesses source autonomy because it is her powerful will and control that govern what she even entertains, chooses, and does, as opposed to the power of unregulated desires pushing or pulling her into actions.

(Repetti 2019, p. 157)

So, mastering mindfulness meditation gives the ārya a kind of free will, one that is consistent with causal determinism, but one that, Repetti argues, can also ground moral responsibility, for it evades the most powerful arguments against compatibilist accounts of free will and responsibility. For instance, Strawson's (1994) ‘Impossibility Argument' holds that a person cannot be ultimately responsible for their actions because they can't be the ultimate cause of them. If one's actions are determined by the character one possesses at some time t, then they must be responsible for having that character, and therefore must have brought it about that they are that way, which means they must be responsible for actions performed at t–1 which caused them to have the character they have at t, but to be responsible for the actions performed at t–1, they must be responsible for the character they had at t–1, and so on, generating a regress that continues to times either when the person was too young to be reasonably held responsible for anything, or if there is a sense in which babies or even fetuses can be held responsible for their actions, then times before they ever existed. To be the ultimate cause of their actions, a person would have to have created themselves ex nihilo, which is logically impossible. However, according to Repetti, the ārya evades this argument because their actions are not determined by anything that precedes their enlightenment, i.e., their development of complete M-autonomy. As Repetti puts it,

without being a logically impossible causa sui ab initio (cause of oneself from the beginning), she is a “relative” causa sui post facto, because she has conquered the self that emerged before she was able to self-regulate, and completely equalized its conditioning, transformed its tendencies, eliminated those undermining self-regulation, and maximized those promoting self-regulation, to the titanic level.

(2019, p. 158)

Repetti believes that mindfulness meditation allows a person to develop a kind of free will such that they can be morally responsible for their actions, because mindfulness practice allows one to sever the power of the thoughts, sensations, and desires that have arisen in the past and continue in the present to arise unbidden out of unconscious subpersonal processes of which one has no control, to generate further thoughts and motivate actions. Instead, the ārya has only the thoughts, sensations, and desires they want to have, and therefore āryas only undertake the actions they want to undertake.

5  What kind of self does an ārya have?

According to Repetti, the self of an ārya is entirely self-formed and self-directed. Their first-order mental states are fully determined by their second-order volitions. But what second-order volitions would such a person have? If they're not determined by the ārya's personal history, then where do they come from? If, as Repetti claims, the ārya can “dis-identify with, detach from, and not act on any mental contents whatsoever” (p. 157), then on what basis does she decide to have the contents that she in fact does, or any contents whatsoever? Repetti has an easy answer for that, derived from the Buddhist tradition itself, from which mindfulness meditation originates.

The Buddhist meditator is guided from their first forays into mindfulness practice, all the way to their mastery of the discipline, by the prescriptions of the Buddha, summarized by the ‘Eightfold Path' to enlightenment: Right View, Right Intention, Right Speech, Right Action, Right Livelihood, Right Effort, Right Concentration, and Right Mindfulness. Presumably, even if this is not the explicit intention of the meditator, mindfulness practice will still engender the kinds of positive qualities achieved by following this path. And if it's true that a meditation virtuoso's second-order volitions have no necessary connection to anything in their pre-meditative life, then any second-order volitions they end up with must be derived from the meditation itself.

In this way, the self of the ārya is constituted by discrimination and selection processes that comprise a perfect decision procedure for generating Right thought, action, etc. Assuming that there is such a thing as the Right View, the Right Action, etc., to have in any situation, then this decision procedure should be generic, i.e., the same for everyone. The self of the ārya, since it is the only truly self-chosen self, is the only true self. But this has the strange result that all selves are really qualitatively indistinct. Everyone who has established a genuine, autonomous, self through meditation has the very same self as anyone else who has done so.

Put more precisely: It is the case that since people find themselves in different situations with different contingent features, each ārya may have different specific first-order mental contents in what we may refer to as their selves. However, their second-order volitions, i.e., to do and have the sort of mind that the Eightfold Path commends, will all be the same, though they may have cultivated them in various idiosyncratically diverse manners. So, the kind of self that is produced by successful Buddhist meditation (assuming, for the sake of argument, that there is only one kind of Buddhist meditation, or that the different kinds are nonetheless species of a greater genus the members of which all share the relevant properties) is radically different from how we ordinarily think of ourselves, i.e., as what makes each of us unique, what distinguishes one person from another.

This result should be of no surprise to Buddhists, and in fact makes sense of the connection between the no-self doctrine and the Buddhist idea that suffering stems from attachment. Perhaps the greatest source of attachment, and therefore of suffering, is one's attachment to the specifics of one's individual personality, preferences, emotional dispositions, etc., not to mention one's attachment to the uniqueness of one's self such that one fears its inevitable obliteration in death. This can all be described as ‘ego-clinging' and is perhaps the most fundamental form of attachment recognized by Buddhism. So, āryas, by detaching from anything that marks them out as particular individuals, become liberated, but in so doing cease to be individuals at all at least in this regard: They are type-identical, though they remain numerically distinct, i.e., āryas 1, 2, 3, etc. may be counted, however otherwise identical they are in the type or category sense, just as identical blank DVDs remain numerically distinct prior to their programming; but even if – and here is the real challenge to our ordinary conception of agent-selves – all such DVDs are given the exact same programming, but are used by different users inputting different data, they remain arguably analogously identical to otherwise identical āryas encountering different external circumstances. Another way to put the point is that while they may differ in terms of narrative identity (they have different hagiographies), they do not differ in terms of characterological identity (they have identical virtue-epistemic profiles).

Another way of construing the same idea is that their ‘self' is something akin to Aristotle's “active intellect”, which is considered universal though it operates differently within different individuals, though without the metaphysically superlative character of Aristotle's principle. Aristotle distinguishes between two elements of cognition:

the one [passive] sort is intellect (nous) by becoming all things, the other sort by forming all things, in the way an active condition (hexis) like light too makes the colors that are in potency be at work as colors.

(2001, On the Soul, Book III, ch. 5, 430a10–25)

For Aristotle, the active intellect “is deathless and everlasting” (by which he may be referring to the timelessness of abstract logical or mathematical truths or reasoning about them), which way of describing them is generally inconsistent with Buddhist doctrine. However, the idea that cognition requires both passive and active elements, with the active element universal to all thinkers, and only the passive element specific to individuals, seems to somewhat capture the dynamic between a potentially autonomous, but generic, self as an agent and a non-autonomous, but specific, self as an object of awareness, as I have outlined it in the foregoing.

6  Conclusion

I have focused in this chapter on the paradoxical nature of the self-controlling and self-transcending identity, or the lack thereof, of the ārya, the meditation virtuoso, which category of remarkable individuals includes putatively enlightened beings like the Buddha. I have brought together primarily the ideas of Metzinger (2015) and of Federman and Ergas (2018) to pose an objection to Repetti's (2019) account of ārya autonomy. Repetti argues that the ārya has maximal autonomy, but on my analysis, consistent both with Metzinger's and Federman and Ergas's analyses, Repetti seems to be attributing what these other accounts describe as the sort of control that āryas exhibit at the first stage of successful practice to the sort of control-transcending flow-type states that āryas experience at the second stage of practice.

I have argued, further, that if Repetti were correct in this conflation, his account would face another puzzle, namely, the counter-intuitive idea that all āryas have not only perfectly identical views, desires, and the like, but also, and much more problematically, an identical identity, which latter is a prima facie oxymoron. While most Buddhists would not have a problem with the identity of enlightened beings, given that they accept the widespread interpretation of the no-self doctrine, according to which enlightenment consists precisely in seeing through the illusion of self, ² in which case what they all share is a non-identity, and non-identities are all the same, any account that asserts that āryas have complete autonomy, in the first, self-controlling, sense, cannot easily mount this problem by claiming that what is identical across all enlightened beings is that they have no identity at all.

I have focused in this chapter on a specific set of philosophical problems facing a specific account of the successful stages of meditative practice, as compared with similar but importantly differing accounts of the same later stages of practice, in a metaphorical sense focusing on the trees and not on the broader view of the forest. However, the broader, forest-level point implicit in this display of philosophical analysis, and perhaps most important for this Handbook, is that meditation practice promises to lead practitioners to greater autonomy in its first stage of attainment and to greater self-transcendence in its second, culminating stage, and that philosophical analysis of the differences between them is both philosophically important for understanding the validity of meditation as a philosophical activity and for understanding the logically possible outcomes of following that practice to its culminating conclusions.