4.1 A pragmatic diagnosis

One conclusion one might draw from the performative fallacy is that no Sane has any meaning because Sane cannot have truth conditions. So either they are nonsense or they share their meaning with all other contradictions.

This is the wrong conclusion to draw. Yes, if we accept all five assumptions, what is said in a Sane leads to a contradiction, but this does not entail that they are meaningless. Remember that the contradiction follows not from what is said, but from saying it. Therefore, the proposition expressed by what is said can be the case. The utterance is perfectly intelligible – it simply is necessarily false when I say it.

The question is then: Why should someone say something that is immediately falsified by saying it? Maybe what is said is not what is meant. If people assert that they don’t exist, they violate conversational principles (Grice, 1991), namely the maxim of quality: “Try to make your contribution one that is true.” If a Sane cannot be true, it triggers conversational implicatures: the Sane must mean something other than what the speaker says. Just as obviously false statements like “I wasn’t myself that night”, “There is no time like the present!”, or “We’re busy doing nothing” can be uttered in a meaningful way, so can Sane.

What might a speaker mean with a Sane? They may intend to convey things like “I don’t feel well” or “Nobody is noticing me”, or express desires like “I wish I weren’t here”, or boast that “I got totally lost in what I was doing, paying no attention to myself”, or “Look, I have acquired enlightenment, admire me!”, and so on. None of this requires any form of actual ego-dissolution.

Instead of being a truthful report, what appears to be a report about a non-egoic state performs a different conversational function. It therefore does not require anybody to make the ontological commitment that non-egoic experiences exist. In fact, the hearer can decode the message only because such reports are openly self-contradictory, similar to a case where a speaker excuses her previous erratic behaviour by saying “I wasn’t myself”. Some variants of Sane might then be a complex way to communicate one’s unease with oneself. This is a good fit for cases of apparent ego-dissolution in depression.

However, this pragmatic diagnosis has two drawbacks. First, it does not explain the details with which such reports are usually embellished. Reconsider Saks talking about herself dissolving like “a sand castle with all the sand sliding away in the receding surf”. These metaphors fulfil no further pragmatic function. Rather, they fit our understanding of Sane as detailed reports. Second, this diagnosis hardly explains cases of written and kept-hidden reports of non-egoic states, e.g. in diaries. Given that these writings exist and may be revealed after an author’s death, we should be open to the possibility that some of them are intended as earnest reports, not as pragmatic play.

If we can reasonably rule out the intention of triggering drastic pragmatic implicatures and deviations of what is meant from what is said, we need another diagnosis for what underlies the expression of a Sane for cases that do not lend themselves to pragmatic explanations. This leads us to cognitive diagnoses.

4.2 Cognitive diagnoses

If all the signs suggest that a speaker actually means what she says in a Sane, despite the apparently self-defeating nature of Sane, then we have to ask what brought the speaker to make such an assertion. The possibilities include: (a) the speaker is confused and does not actually believe what she says; (b) if the speaker shows signs of believing what she says, she must be irrational; or (c) if the speaker shows signs of believing what she says and signs of being rational, she intends to deceive us.

5.1 Rejecting the Sane paradox

The pragmatic and the cognitive diagnoses are all available to explain why a Sane may be given by a person. Each of these diagnoses allows us to accept the presuppositions of the Sane paradox and remain agnostic as to whether phenomenal ego-dissolutions exist at all.

Even though we should accept these diagnoses as parts of a toolbox, they do not lend themselves to building a research programme focused on the phenomenality of ego-dissolution. For this, we have to reject the paradox.

Let me re-address the presuppositions behind the Sane paradox. Non-egoic experiences are supposedly marked by the following features:

(Cease)

A feeling of self is missing in some experiences.

(Access)

The absence of a feeling of self is cognitively accessible.

(Difference)

There is a felt difference in the feeling of self between the time before/after and during non-egoic episodes.

Two additional assumptions give rise to the Sane paradox.

(Unique)

For each individual, there is one and only one self that is the referent of self-reference and fulfils the functions of self-reference, or only one mechanism facilitating all kinds of self-reference.

(PhenSelf)

A self (or any mechanism facilitating self-reference) necessarily comes with a phenomenal feeling of self.

Only together do they preclude the possibility of reporting on non-egoic experiences.

But we might reject Unique, Cease, or PhenSelf and thereby preempt the Sane paradox. If not the feeling of self, but merely a feeling of self dissolves, then another mechanism facilitating self-reference might bring about relevant de se beliefs. If only one feeling of self exists but it doesn’t dissolve, then de se beliefs can occur. If self-referencing mechanisms need not show themselves in experience, then the relevant de se beliefs can occur.

But before I go into these non-paradoxical accounts, is there maybe a way to account for all the features of Sane while accepting the assumptions leading to the paradox? Someone may defend the idea that one can provide Sane from episodes of full ego-dissolution by relying on specific forms of memory. I do not think that this works well. Let me explain why.

5.2 Can we defend total ego-dissolution against the Sane paradox?

Total ego-dissolution would be the case if Unique, Cease, and PhenSelf were all accepted. So there is only one singular self or mechanism facilitating self-reference and self-consciousness which necessarily manifests itself in a specific feeling of self and which is inactive or nonexistent during such episodes. If this were the case, then we should not get any Sane when a feeling of self ceases. In this framework, if we ascribe a non-egoic experience to ourselves, then there must have been a self or a mechanism facilitating self-reference in order to bring about this de se ascription.

Some may accept these theses, but still find the Sane paradox unconvincing, specifically when it comes to memory. It might be easy to see why one cannot report on oneself as currently undergoing an episode of total ego-dissolution. One would need to speak of oneself in the third-person singular or use a definite description that happened to refer to oneself (e.g. “the speaker of this sentence”). For the type of de se reference necessary for a self-ascription, however, we would need some way of modelling ourselves in order to refer to ourselves as ourselves – and not merely to ourselves by accident.

But one might think this: Even though one could not report on such experiences while one is undergoing them, one might remember them and report on them afterwards as having happened to oneself. Metzinger (2004) apparently rejects this possibility, because “How can you coherently report about a selfless state of consciousness from your own, autobiographical memory?” Certainly, autobiographical memory relies on self-reference across time, which relies on a model of the progression of a self. But why shouldn’t some other form of memory suffice? Who else could this experience have happened to, when all the experiences that I can remember are necessarily my own? Could I not validly infer later on that this experience was my own? If so, we need not reject any of the premises. Instead the Sane paradox is restricted: Only present-tense Sane are impossible, but we may still report veridically about non-egoic experiences, if only retrospectively.

This might seem like a reasonable reply, but only because it relies on a misunderstanding: Semantic memory is not the same as episodic and autobiographic memory. That is, some memories of facts will not be de se memories (Conway, 2005; Conway & Pleydell-Pearce, 2000; Klein & Loftus, 2014; Klein & Nichols, 2012). I remember a lot of things that are facts, but not about me, e.g. when the Second World War ended, that Paris is the capital of France, that Joaquin Phoenix had a brother called River. Even if a fact concerns ourselves, a memory of that fact is not automatically a de se memory (Conway, 2005; Klein & Loftus, 2014). For example, you might remember that someone knocked over the lamp during the party last night, but you might fail to remember that it was you who knocked it over.¹⁷ As David Lewis (1979) argues: No amount of knowledge of facts, even up to divine omniscience, can replace knowledge de se. And it is only knowledge de se that allows us to locate ourselves in the world. God would fail to know where and who she was if she had only factual omniscience. If one wants to defend the possibility of veridical reports from total ego-dissolution, one has to show how one can form de se memories of one’s own episode of total ego-dissolution. How could that be?

Two stages are the most promising candidates during which memories may be imbued with a de se aspect: the formation and the retrieval (or recall) of the memory.¹⁸ In cases of total ego-dissolution, the de se aspect of memory cannot arise during formation, because at that point a self (or a mechanism facilitating self-reference) and therefore the basis for self-reference is missing. This is Metzinger’s, Gennaro’s, and Foster’s point. Therefore, the de se aspect must be inserted during reconstruction, injected during retrieval, added on during recall.

Unfortunately, there is no specific noticeable trace of what exactly is “added on”: A false memory of me knocking over a lamp feels just as real as a veridical memory of me knocking over a lamp. Sometimes, it does happen that we “add ourselves” to memories of affairs that did not involve us (Hyman Jr, Husband, & Billings, 1995; Loftus & Pickrell, 1995). Someone might reply that these are therefore not memories at all, but pseudo-memories that are simply phenomenologically indistinguishable from real ones; what makes something a memory is not how something feels, but whether what is remembered happened as it was remembered. Yes, there can be false memories with a de se component added later; but there can also be true memories with a de se component added later. This is obviously right. As an example: I cannot remember things from my early childhood. Most of what I know from that time, I know from pictures and hearsay. Say I see myself on one of these photos in the arm of my uncle Karl smoking a pipe. I know that it happened to me, but I know this not by remembering the experience itself. However, after a while, I might forget how I formed this knowledge that my uncle Karl held me while smoking a pipe. I might form a de se memory about this fact that I take to be one formed in my early childhood. Earlier, it was a fact about myself that I knew, stored in semantic memory; now I remember it as a de se fact with a phenomenological de se component that was added on later – and thus it has been turned into an episodic memory.

The problem is not that there cannot be true or accurate memories of facts with a de se component added on later. The problem arises from the “adding on” itself. For if this “adding ourselves in” is part of the reconstruction, then such memories fail to show that there actually was an episode of ego-dissolution – for it is indistinguishable from the outside or the inside whether the de se aspect was added on later or accurately remembered as being present during the experience. If the de se aspect was previously present, then the experience was egoic. If one agrees that it was a later mis-reconstruction, addition, or embellishment, then one should be doubtful that the rest of the experience was remembered accurately. We cannot be sure that we are learning something about the deep structure of non-egoic experiences if the reports are edited. But one cannot distinguish between these cases.

But what if the de se addition is justified by another memory system? What if one does not remember a proposition about someone and “add on” the de se component; what if instead one remembers imagistically?¹⁹ That is, one remembers the sensations felt during that episode as sensations, as one conjures up the scene of what it was like to sit in one’s mother’s kitchen as a toddler. In this remembered stream of sensations, a felt self could be missing. But because sensations always have to be mine (I don’t feel anybody else’s), I can reasonably infer that these sensory impressions with a lack of felt self must have been mine. What is added is a possessive de se component, not a phenomenal one: The experiences were mine, but I was not in these experiences. But again: What ensures that the de se component was not there in the remembered experience?

What if someone holds that a self is a necessary requirement for experiences in general? Any form of consciousness, we are often told, involves a first-person perspective. And what is a first person other than a self? Maybe it is the perspectival nature of experience which must point at someone, but not at anyone: at a self as the owner of the experience, as that whose experience it is, as the centre of the perspective. The feeling of self is not something extra, but is somehow ingrained in the structural composition of phenomenal sensations because of their perspectival nature. Maybe this justifies the inference that those were my experiences if I become aware of the perspectival nature of my imagistically remembered sensations.

I agree that our sensations, especially our auditory and visual experiences, are perspectival. But I do not think that this forces us to accept a self outside experience, a mechanism facilitating self-reference, or even a feeling of self as part of our experiences. Our experiences are perspectival and thereby indicate a centre in experience. But the perspectival nature of experiences does not necessarily indicate oneself as the centre; it just indicates someone. Consider, as an analogy, Being John Malkovich.²⁰ In this movie, people can enter the mind of John Malkovich through a door in the 7½th floor of a Manhattan building. When a person enters through that door, we see the succeeding scene unfold from John Malkovich’s first-person perspective. Obviously, the visual imagery suggests that these experiences are someone’s, but they need not be mine. Neither set-up nor framing specify whether these visual experiences are those of the person entering the mind of John Malkovich or of John Malkovich himself. And when seeing the scene as an observer, I need not feel like I am John Malkovich. The perspectival nature of the visual sensations is preserved, but they do not necessarily therefore feel like my own perspective, when I see the scene. The same goes for the people undergoing the experience in the movie: They do not confuse themselves with John Malkovich, but simply partake in his stream of consciousness – they experience this stream as John Malkovich’s. Therefore, conceptually, the perspectival nature of sensory imagery is not sufficient to indicate that these sensations are one’s own, that the vanishing point of this perspective coincides with my self. The difference between experiencing a perspective and experiencing a perspective as being my own is then not one of sensory impressions at all. Selfhood is not a sensory quale and not part of the imagistic array I remember or experience. This does not mean that there is no feeling of self at all, but simply that there is no sensation in any sensory modality that is that feeling of self.

If the perspectivity of our sensory experiences does not suffice for a feeling of self inside experiences or necessarily indicate a self outside our experiences, then attributing a remembered sensory experience to myself is something problematic for allegedly non-egoic episodes. In order to remember this stream of consciousness as being my own, a de se component is again “added on” to the perspective in recall. And again, we – whether as memorisers or as external observers – cannot determine whether the de se component was present in the first place or added on later.

So if there are non-egoic experiences, then we cannot have first-person reports of them, for this requires reference to oneself as oneself. If there are such experiences, then we cannot have autobiographical memories of them for the same reason. If they exist, then no purported autobiographical memory would prove their existence, for these would be indistinguishable from mis-reconstructions or mis-categorisations of egoic episodes as non-egoic ones. While this does not prove the nonexistence of episodes of total ego-dissolution, there is no way to demonstrate their existence by providing “reports”. This puts them in the same category as unicorns, whose existence one also cannot prove by pointing to “reports” and drawings of them.

We should therefore consider alternatives to total ego-dissolution where we do not need to posit something being “added on” to a memory. Then, it might be less controversial how we can see Sane as indicators, in some sense, of non-egoic episodes.

5.3 Multiple feelings of self: plurality or modularity

Let us accept Cease and PhenSelf but reject Unique: A feeling of self can temporarily cease to exist but there could be many differentiable phenomenal “selves” in our experiences. These might include a feeling of ourselves as a body, another as a thinker of thoughts, another as a person in social relations, and so on – each likely associated with a dissociable cognitive mechanism that facilitates this form of self-consciousness. Obviously, these distinct feelings of selves in each of us may influence one another, but they could still be dissociable. If they are dissociable, one feeling of self can cease to exist while others persist. Then, one can reasonably say that one of these – a, but not the, phenomenal self – dissolves while a system retains other mechanisms sufficient for de se beliefs or cognitive self-reference. Even though he does not endorse it, this position may be developed from work by Raphaël Millière (2017), who, in order to give an account of what happens in ego-dissolution, helpfully distinguishes between different forms of self-consciousness, e.g. the disruption of self-referential thoughts, of narratives about oneself, of body ownership, of bodily self-awareness, and of bodily self-location (and extension).

We can construe the thesis of “multiple selves” (i.e. multiple feelings of self at one time in one system) in two ways: as modularity or plurality. In the case of modularity, there is no single homogeneous mechanism bringing about a feeling of self; different modules specialise in different tasks, among them those mentioned by Millière (2017). Each module can bring about a kind of self-reference and self-relatedness independently from others, and each of these forms may, if we accept PhenSelf, manifest itself in phenomenal consciousness in a distinct way. But just as the knife, saw, and corkscrew still contribute to one Swiss Army knife, each self-module contributes to a larger, integrated, and unique feeling of self, which allows for mediation between these modules and a coherence between these feelings as indicating one and the same self. A disturbance in the modules affects the character of a feeling of self, but there is still one feeling of self. Because each self-module can fail independently and because such disturbances are noticeable phenomenologically, there can be reports of such partially non-egoic experiences – but only if some of the mechanisms necessary for self-reference are intact, for they enable the formation of first-person reports (e.g. the modules for cognitive or linguistic self-reference) or autobiographical memories.

In the case of plurality, these modules are not part of a larger, integrated feeling of self. Each contributes to forms of self-reference, but each produces its own distinct feeling of self – without any overarching unique feeling of self that they are part of. Some of these can be disturbed while others remain unaffected and therefore enable first-person reports and autobiographical memories of (partially) non-egoic experiences. Simply, there is no unique feeling of self, but multiple feelings of self, each associated with its own distinct mechanism facilitating some form of self-reference.

In both cases, the expression “I” in first-person reports will be ambiguous: It can refer to the bodily self, the social self, the cognitive self, or the reflexive self. If so, then the performative self-contradiction could be prevented by disambiguating. Consider Sartre’s statement again:

(Sartre)

When I run after a streetcar, when I look at the time, when I am absorbed in contemplating a portrait, there is no I.

In this example, the first two instances of “I” obviously refer to a physical body that runs and looks, the third “I” refers to a cognitive self, while the fourth “I” refers to the reflexive self – some being present in experience, some not. If we disambiguate in this way, the paradox ceases to arise:

(Sartre^* *)

When Body-I run after a streetcar, when Body-I look at the time, when Cognitive-I am absorbed in contemplating a portrait, there is no Reflexive-I.

This partial breakdown in the chorus of the selves could be what is expressed in Sane. Non-egoic experiences then need not be total ego-dissolutions; they are only partial disruptions of feelings of self and their associated self-reference-enabling modules.

But I see some drawbacks. First, this explanation does not seem to fit Saks’s report, where she claims that the centre (singular) does not hold. The definite article “the” here suggests a uniqueness: one and only one centre. Apparently, there is one coherent feeling of self for her, and it has gotten lost. Nor did only some of the feelings of self (as in the case of plurality) get lost; nor did the feeling of self change its character (as in the case of modularity). If the uniqueness expressed by Saks is not just loose talk, we should search for alternative explanations that do not rely on a multiplicity of selves or feelings of self.

The second drawback I see is: The account needs to explain how different feelings of self interact in order to give rise to the appearance of one integrated self, a feeling of self as inhabiting this body rather than that, and as the thinker of thoughts, and also as the experiencer of sensations, plus as the person addressed by others, and so on. If such an explanation is not provided, this approach remains unconvincing because this high degree of integration is the most prominent features of selves: their individuality – their lack of partition. A feeling of self should share this undividedness in order to be a feeling of one self as an individual. If such an explanation for overarching integration of feelings of self into one is provided, the multiplicity thesis loses some of its drive. Because now, there is apparently one overarching process that forms a coherent feeling of self out of several. The multiplicity thesis also becomes ambiguous: It is unclear what is disturbed in non-egoic states – is it the integrating process or the parts being integrated?

A third drawback I see is that because this explanation relies on multitudes, either of modules contributing to a larger feeling of self or of unintegrated feelings of self, it is not very parsimonious. With Occam’s razor, we should prefer an explanation that covers the same territory with fewer entities postulated.

We can also consider an evolutionary take: The amount of sugar one’s body burns is a constraint on one’s evolutionary success. Finding the right proportion, the sweet spot, between energy conservation on the one hand and access to resources (like food, status, or mates) on the other hand is an evolutionary imperative. So we have to show what an organism gains something by spending energy. If each of the multiple mechanisms for self-reference bringing about a feeling of self demands energy to be active, then the multiple-selves account looks like a very costly model in comparison to one that claims only one unique mechanism facilitating all feelings of self. An explanation without such pluralism might then be preferable, if not for theoretical reasons then for reasons of energy conservation.²¹

Both of these reasons for rejecting the multiplicity account are comparative: Of two or more accounts, they tell us which to prefer. So we should look at further alternatives. If multiplicity is the only game in town, it wins by default. But wouldn’t it be cool if there could be more to ego-dissolution than just the disturbance of some-but-not-all feelings of self?

5.4 Ego-expansion

Let us accept Unique and PhenSelf but reject Cease: There is only one feeling of self but it does not dissolve. It is the dissolving of a unique mechanism for self-reference or a unique self-model, indicated by the loss of a feeling of self, that leads to the Sane paradox, so rejecting Cease disarms the paradox. So what happens in such states if it is not a feeling of self dissolving?

Here is one option:²² Rather than contracting into nothingness, the feeling of self expands until everything in consciousness becomes part of it or attached to it. Instead of being nothing, in these states the feeling of self is associated and involved with everything. In this way, our engine for self-reference is still active – indeed, it is hyperactive. If the feeling of self is still there and an indicator of the activity of a mechanism facilitating self-reference, then we can get first-person reports and self-ascriptions of experiences of ego-expansion.

But why would they be about ego-dissolution if the feeling of self is not dissolving? The problem might arise from an ambiguity in how we conceptualise such experiences: In both the case of ego-dissolution and ego-expansion, the self/other distinction loses its unique meaning. Why? An empirical concept is uniquely meaningful only if there are some things that fall under its extension and some that do not. If everything falls under the extension of “feeling of self”, it is not the feeling of self that dissolves, but the boundary between a feeling of self and the feeling of anything else in consciousness. It is the concept of “self” that becomes meaningless, because it lacks any boundary and allows for no distinctions in one’s mind during full ego-expansion. The categories self or belongs to me become meaningless if everything is experienced as belonging to me. In a common but naïve stance, where one projects what is in consciousness onto the world, one is everything and one with the universe. One’s concept of “self” would be extensionally indistinguishable from the concept of things being identical to themselves. The same holds in the case of ego-dissolution: The category of other or not belonging to oneself becomes meaningless because it encompasses everything in one’s mind. So the self/other distinction loses its meaning under both circumstances. “Self” needs “other” to make sense because the two are mutually exclusive opposites. In cases where everything falls under self or everything falls under other, it is just as meaningful to claim “The self ceased to exist” as it is to claim “The other ceased to exist”. In both total ego-dissolution as well as full ego-expansion, the term self becomes empty just as the term other becomes empty. Thus, first-person reports of “ego-dissolution” might be adequate descriptions of a conceptual dichotomy dissolving.

So in cases where there is no empirical distinction between the concepts “self” and “other” available to oneself, it might make just as much sense to say that everything is other (report of ego-dissolution) or to say that everything is me (report of ego-expansion). But which is the experience underlying this report? Ego-dissolution or ego-expansion? Total ego-dissolution is disqualified because it does not easily square with Sane. (Or, at least, it makes assumptions one need not share.) Therefore, if Sane are indicators of a feeling of self-alteration, and if there is only one feeling of self, we are left with ego-expansion as a suitable alternative.

In the extreme case of ego-expansion, everything going on in consciousness would become part of the self-model, such that nothing could be meaningfully referred to as “other”. Only unconscious processes would be exempt from such a hypertrophying self. We would feel as involved in pushing the clouds across the sky as we would in moving our legs. Reports collected by Berkovich-Ohana et al. (2013, pp. 5–6) reflect this tendency for ego-expansion when meditators speak of the bodily space being larger, bodily sensations being wider, a “sense of expansion”, a “center of space [becoming] endless”, that there were “little bodily boundaries compared to the usual feeling”, and so on. Data by Preller et al. (2018) are suggestive of an explanation: They found that “LSD induces hyper-connectivity predominately in sensory and somatomotor areas” (Preller et al., 2018, p. 3). Together with an increased hyperconnectivity in the sensory thalamus, a classic neural hotspot for consciousness and its unified nature, this can be interpreted as: more and more active neural correlates of sensory conscious states are incorporated into and associated with one’s own bodily self.²³

This oceanic explanation of non-egoic states clearly marks such episodes as egoic: A feeling of self would be present! And it explains the misconceptualisation that happens in Sane as benign. It might also explain why certain people who claim to have achieved self-annihilation do not simply stop speaking but instead use a plural form. Consider Foster (2016) talking about Baker, the author who wanted to capture what it is like to dissolve and become one with a flock of peregrines: “If Baker is to be believed, it worked [...] the pronouns changed from ‘I’ to ‘we’.”²⁴ This change to the first-person plural may indicate that a felt self has expanded and now encompasses more than what the speaker usually associates with himself. Ego-expansion might thus be a suitable alternative to the multiplicity account and to total ego-dissolution.

An obvious problem for the ego-expansion account is that it does not straightforwardly explain differences in describing these experiences. If we believe that the feeling of self is unitary, then only two options remain for attaining experiences without a self/other distinction, namely total ego-dissolution and ego-expansion. Total ego-dissolution, if we buy into the Sane paradox, is undetectable or ineffable. But still, we have to explain why subjects differ in their replies to questionnaires or how they describe their experiences: Some answers suggest an “anxious ego-dissolution”, some an “ego-inflation” or an “oceanic boundlessness”. In this account, we have to explain all in terms of ego-expansion. How might this work?

I am sympathetic to ego-expansion, so I propose the following empirical hypothesis: Subjects conceptualise their experience of ego-expansion as “no self” or “all self” depending on where they are focusing when the expansion happens. Consider this: Our attention can be on our thoughts, our body, our breathing, our digestion, or other aspects of our body and mind. But it can also attach itself to external events, like the sky, water flowing, sounds of neighbours arguing, and so on. It can be on self-related thoughts (I still have to feed the cat) or thoughts where the content is unrelated to us (There is always a prime number between n and 2n). We know that under the influence of LSD, our attention is altered. Subjects who have taken LSD often focus with extreme intensity on small details, often at the cost of ignoring everything else. Say that while the feeling of self is slowly expanding, one focuses not on one’s own hand or breathing but on a feature of the environment (e.g. the veins of a leaf) or a mathematical truth. When attention is shifted again, as the person tries to get back to herself, she realises that she does not find herself where she expected herself to be; she has lost the familiar feeling of her usual, expected self-boundaries because they slowly expanded while she was distracted. The mismatch of predicted self-boundaries in experience and the lack of felt self-boundaries is then interpreted as ego-dissolution – because if it is not where it always is, it must be gone! Compare this to a case where people who have taken LSD focus on their own body or on their body in relation to the environment. The process of a feeling of self expanding is felt as a process: We can attend to and appreciate the dynamics and characteristics of its unfolding. As the endpoint of maximal ego-expansion is reached, we (not I) know how we got there. The mismatch of prior felt self-boundaries and the lack of felt self-boundaries is interpreted as an expansion of self because we witnessed our feeling of self expanding. What I suggest is that whether we categorise such an episode as ego-dissolution or oceanic self-boundlessness depends solely on how aware we were of the process of expansion. This suggests that people who are more introspective or sensitive, focusing on their body and mind, or more trained in keeping their attention partially on themselves might experience more episodes of oceanic boundlessness than people who lack such training or are more externally focused or do more rumination (i.e. thinking non-self-related thoughts and getting lost in them).²⁵

Coming back to the Sane paradox: We can expect first-person reports of such episodes of ego-expansion. And we can understand why some of them might rather be phrased in terms of ego-dissolution. So this interpretation can explain the specific structure and existence of Sane if ego-expansion underlies them. Ego-expansion is comparatively more parsimonious than the multiplicity account. However, there is an even more parsimonious solution available.

5.5 No-ego revelation

For total ego-dissolution, we accepted both that there is only one feeling of self (Unique) and that it dissolves (Cease), which got us the Sane paradox on the condition that mechanisms for self-reference must show themselves in phenomenal consciousness (PhenSelf). For the multiple-selves hypothesis, we rejected Unique and accepted Cease, which led to a non-parsimonious but acceptable diagnosis. For ego-expansion, we accepted Unique and rejected Cease, leading to a more parsimonious explanation in comparison to the multiplicity account. Logically, we could reject both Unique and Cease. However, there are two ways in which we can reject Cease: First, there are many feelings of self and everything stays the same; or, second, there is no feeling of self to begin with, so no feeling of self can dissolve. This second interpretation denies, to some degree, PhenSelf: A mechanism facilitating self-reference does not need to show itself in phenomenal consciousness at all. This, the no-ego account, is the most parsimonious attempt at a diagnosis.

So, what if there never was a phenomenal self? Here, we deny the basic presupposition of phenomenal ego-dissolution, because something has to exist before it can cease to exist. But I argued in the beginning, against some interpretations of Hume and Sartre, that this account fails to capture the difference between being inside and outside a non-egoic episode. So, how can we reply to this challenge?

At its core, the no-ego account explains this apparently felt difference as a difference in beliefs about one’s experience. Certain beliefs, e.g. about the proper grammatical construction of sentences, may coerce us into believing that there is a self in consciousness. Because how else could we make sense of statements like this:

(Dream)

I dreamt I was Napoleon.

The second instance of “I” in Dream appears to pick something out in the dream, i.e. it is about what I dreamt of, not who dreamt it. This suggests that there is a feeling of self in experiences like dreaming. But this leads to paradoxes similar to the Sane paradox for cases like “I dreamt I was dead and gone.” or “I sometimes imagine myself as never having been born.” How do we explain these away? I suggest: Grammar bewitches us.

As an analogy, think of Mark Twain complaining about “The Awful German Language” (1880). In German, articles indicate a gender: masculine (der), feminine (die), or neuter (das). According to Twain, these grammatical genders do not make any sense: “In German, a young lady [das Mädchen] has no sex, while a turnip [die Rübe] has. Think what overwrought reverence that shows for the turnip, and what callous disrespect for the girl.” We can imagine someone so confused by this grammatical artefact that they really think that any young lady is genderless and acquires a gender only after puberty, when she turns into a woman, or that turnips are actually all female. This odd fellow has been misled by grammar.

Just like our odd fellow who superimposed grammatical gender onto external reality, it might be that we naturally force our conceptions about proper grammar onto experiences themselves. It is not that we find a self in experience; rather, a self is introduced in the way our language forces us to report on experiences.

But, as I said before, such a theory fails to explain a reported change: Things appear to be different before and after in comparison to during the non-egoic episode. If there never was a self in experience, what is it that happens in such episodes of falsely labelled “ego-dissolution”?

I suggest an ironic solution: The fundamental irony of spiritual enlightenment is that it is a basic and general fact that phenomenal experiences are always selfless but we hardly ever notice this. It takes special circumstances for one to cognitively register that one never felt like anybody. Nothing changes in phenomenality if we attain these states. We just recognise that something has always been lacking: a feeling of self. This realisation that the idea of being a self is only a side-effect of grammar and not something we actually experience would be deeply ironic: Becoming enlightened by attaining a non-egoic state of experience is nothing special or extraordinary – it simply is our basic mode of being.

But one does change, if only doxastically: Before and after the episodes, one has false beliefs about the structure of one’s experiences; during these episodes, one gains true beliefs about the structure of one’s experiences. The difference is a difference in our belief system. And because all we claim is that there is no self-reference in phenomenal consciousness, we can still hold that there are unconscious processes that facilitate self-reference. We can therefore have purely cognitive de se beliefs without any phenomenal component indicating a lack of self-reference in phenomenal consciousness.

Despite claiming that there is no feeling of self, the no-ego account still belongs to the phenomenological diagnoses because it makes a claim about phenomenality: There is no feeling of self and there never was. In these episodes, we realise that we confused our way of speaking with our ways of experiencing. This is the revelation one attains in non-egoic experiences.

Revelations can be short-lived or persistent, frightening or enlightening. But they need not come with a phenomenal difference. People may react differently to them: Some may feel as if they have lost something because a welcomed illusion has been revealed as just that – an illusion. Language still has them under a spell, and they may therefore think that something must have been there before.²⁶ Others might truly feel the glow of epistemic progress and accept that language is confusing. They either stop speaking or start using a form of language that allows them to avoid a first-person pronoun, like speaking Japanese, or in English using only plural forms, third-person pronouns, or definite descriptions (e.g. “the speaker”) to avoid being bewitched again by language. This doxastic difference between before/after and during the episode can have widespread cognitive effects, like most doxastic differences.

Obviously, we can have reports here because this account of non-egoic experiencing does not exclude the possibility of any form of cognitive self-reference. It just claims that there never was a phenomenal self, not that there never was any form of self-reference. All it rejects is that forms of self-reference ever show themselves in experience. Non-egoic states aren’t anything special.

The no-ego account is obviously highly counterintuitive. For many, it is an obvious fact not just that they are someone, but that it feels like something to be someone. That this belief is so widespread may merely underline how deep the confusion runs. We need a convincing error theory for why this delusion that there is a feeling of self is so common. Pointing to grammar may get us only so far. But a proponent of the no-ego account may offer the following explanation, which relies on two theses.

First: In any language, we track cognitive differences in speaker and audience. According to a pragmatics-first approach (see e.g. Bar-on, 2013; Moore, 2016, 2017, 2018; Scott-Phillips, 2015), language was developed for communication and coordination, not necessarily for cognition. In communication, we have an embodied speaker and audience. Both are co-represented during an act of communication. However, communication happens in the real world, not just in our minds – and in the real world, speaker and audience differ. It is good communicative practice to keep track of the intentions and beliefs of the different individuals involved in a communicative act. Only if we are aware that someone differs from us in their beliefs may we be able to deceive or teach. In the framework of Gricean communication, tracking audience and speaker separately is in fact basic to any communicative act. And because language is one of the most important tools we have for categorising and capturing the structure of our experiences, we transpose into descriptions of experience this difference between speaker (i.e., us) and audience (the others).

Second, we are by nature naïve realists: In our natural stance towards our experiences, we take ourselves to be in contact with the real world, not with mental representations or phenomenal experiences standing in for the real world, unless we reflect on or are made aware of specific illusions. For this reason, we often pick out experiences by what we take them to stand in for in the world. For example, we may say that we had an experience as of a lamb on the horizon – although we only experience a hazy fleck of white. Could be a lamb, could be a Hungarian sheepdog (a Komondor), or even a white car. So we picked out this experience based not on how we experienced it, but on what we think it relates to in the world outside our minds. In such cases, there is a “crossing over” from what we believe about the world onto how we categorise our experiences themselves. In this process, we may come to believe that there are certain structures in our experiences even though these structures pertain only to the world outside.

Here is an illustrative example adapted from Palmer (1999, p. 209): We know that each of us has only one left and one right hand. And we think that our visual experiences present us with only one left and one right hand when we look at either one. But because of the set-up of our eyes, anything outside of where we focus produces non-aligned images on our retinas. This non-alignment shows itself as “seeing double” outside of where we fixate (including an area called “Panum’s fusional area”), which is sometimes interpreted as depth. The effect is most striking if we stretch an arm out and focus on our thumb or on the background. If we fixate on the thumb, we see double in the background; if we fixate on the background, we visually experience two thumbs. But this is true for all visual experiences, thanks to the architecture of our visual system. So we do not actually experience one left hand if we fixate on the right; we experience two. Yet in forming our beliefs about experience, we superimpose how we think the world is – where we actually only have one left and one right hand – onto what we think our experiences are like.

This forcing-onto-experience might also happen with “fundamental” structures: structures that stem from a basic feature of our dominant form of communication, namely tracking which intentions and beliefs belong to us as speakers and which ones belong to our audience. It might be that there is no felt self in phenomenal consciousness, but merely cognitive tracking of which mental states are under our control and which ones are outside of it. Obviously, today we exploit language for cognition, not merely for communication. Still, tracking which mental states belong to the speaker and which to others remains important. And this tracking of our own mental states as our own could be superimposed onto consciousness even though it is not in consciousness.

In this view, there is no feeling of self. There is, however, a self superimposed on an experience, necessary for registering this experience as one being had by the speaker, and not necessarily shared by the audience. In most cases, the speaker will fail to notice this superimposition as a purely cognitive add-on and will instead take it to be an accurate apprehension of the experience itself. I am uncertain whether I buy into this sketch of an error theory, but it certainly is a plausible one which a proponent of a non-egoic account might offer. This error theory explains why the ego-delusion – the delusion that there is a felt self – is so widespread: It exploits a mechanism at the core of every Gricean communicator.

5.6 Summary

How should we proceed if we captured a Sane in the wild? If we have reason to believe that (concerning this specific Sane) what is said reflects what is meant, we reject the pragmatic diagnoses. Then, if we have reason to believe that the utterer of this Sane is neither obviously deluded, nor irrational, nor deceiving us, we exclude the cognitive diagnoses. That leads us to the phenomenological diagnoses, where four options are available. In each, the speaker uses this Sane to express something concerning her experiences. Of these, only one (total ego-dissolution) leads to the Sane paradox. This leaves three live options.

Must we favour one of these phenomenological diagnoses over the others for all cases? Or should we expect that each may apply to some but not all cases of “non-egoic” ego-alteration? Should we therefore seek one general phenomenological diagnosis, or be pluralists who proceed on a case-by-case basis? This is what I focus on in the last section.