In the earliest texts of the Upaniṣads, dating from the seventh century B.C.E., dreamless sleep is singled out as one of the principal states of the self, along with the waking state and the dream state. Various characterizations of dreamless sleep are given. Some texts characterize it as a state of oblivion, while other texts describe it as a mode of unknowing or non-cognitive consciousness that lacks either the outer sensory objects of the waking state or the inner mental images of the dream state (Raveh 2008). It is this second characterization that we find in the later texts of the Yoga and Vedānta schools. These texts also present a basic form of philosophical argument for dreamless sleep being a mode of consciousness. The argument runs as follows: When you wake up from a dreamless sleep, you are aware of having had a peaceful sleep. You know this directly from memory, so the argument asserts, not from inference. In other words, you do not need to reason, “I feel well rested now, so I must have had a peaceful sleep.” Rather, you are immediately aware of having been happily asleep. Memory, however, presupposes the existence of traces that are themselves caused by previous experiences, so in remembering that you slept peacefully, a peaceful feeling must have been experienced. To put the thought another way, the memory report, “I slept peacefully,” would not be possible if awareness were altogether absent from deep sleep; but to say that awareness is present in deep sleep is to say that deep sleep is a mode of consciousness.
To my knowledge, the earliest version of this argument comes from Vyāsa’s third or fourth century C.E. commentary on Patañjali’s Yoga Sūtras.[1] Patañjali defines yoga as the stilling or restraining of the “fluctuations” of consciousness (Yoga Sūtras I:2). When this stilling is accomplished, the “seer” or “witness” can abide in its true form, namely, pure awareness; otherwise the “seer” identifies with the fluctuations of consciousness—with the movements of thought and emotion (I:3–4). Patañjali identifies five kinds of fluctuations or changing states of consciousness: correct cognition, error, imagining or conceptual construction, sleep, and memory (I:5–6), and he defines sleep as a state of consciousness that is based on an “absence” (I:10).
As the traditional commentaries indicate, “absence” does not mean absence of consciousness; it means absence of an object presented to consciousness.[2] Deep and dreamless sleep is a kind of consciousness without an object. When we are awake we cognize outer objects, and when we dream we cognize mental images. When we are deeply asleep, however, we do not cognize anything—there is no object being cognized and no awareness of oneself as knower. Nevertheless, according to Yoga, we feel this peculiar absence while we sleep and we remember it upon awakening, as evidenced by our saying, “I slept peacefully and I did not know anything.”
Before we examine the debate arising from this argument, let me mention an obvious objection that would occur to us today, especially given what we know from sleep science. The objection is that retrospective subjective evaluations of sleep may be unreliable (Baker et al. 1999), so we cannot assume that the subjective feeling upon awakening of having slept peacefully is based on a veridical memory of a peaceful sleep. An extreme case of the unreliability of self-reports about sleep comes from insomnia patients (Perlis et al. 1997; Rosa & Bonnet 2000; Zhang & Zhao 2007). These patients frequently display sleep-state misperception; that is, their subjective assessments of the quantity and quality of their sleep deviate strongly from the objective, polysomnographic measures. For example, they often identify themselves as having been awake when they are woken up from polysomnographically-defined sleep, they tend to overestimate sleep-onset latency (the length of time it takes to go from full wakefulness to sleep), and to underestimate total sleep time as compared with polysomnographic measures (Perlis et al. 1997). Even in healthy individuals, the feeling of having slept well could sometimes deviate from objective measures. One could feel refreshed upon awakening, yet the objective measures might show that one’s sleep was physiologically restless or intermittent; or one could feel fatigued upon awakening, yet the objective measures might show that one’s sleep was physiologically deep and undisturbed. In short, although it is conceptually true that a veridical episodic memory implies having undergone an experience whose content corresponds, to some degree, to that of the memory, it is an empirical matter whether or to what degree any given waking memory impression of sleep is veridical. It is also an empirical question whether episodes of peaceful sleep typically lead to the awareness of having slept peacefully and whether this feeling can occur even when sleep is disturbed.
This line of thought, however, is not decisive against the Yoga argument. Strictly speaking, all this argument needs is the possibility of there being veridical waking memories of having been deeply and dreamlessly asleep in order logically to establish that awareness can be present in at least certain phases or types of dreamless sleep. The argument does not need to establish that waking memory impressions are typically veridical, only that they can be. Indeed, as we will see later, the Yoga viewpoint can allow that ordinary sleep-state perception and retrospective subjective sleep-state evaluations may be unreliable. I will come back to this point at the end of the paper.
A more direct objection to the argument, however, is to challenge the premise that waking retrospective reports of sleep are ever memory reports. The philosophers of the Nyāya school (Naiyāyikas) make this challenge. They maintain that the statement, “I slept peacefully and I did not know anything,” expresses an inferential cognition, not a memory report, and that consciousness is entirely absent in dreamless sleep. Given how one feels upon awakening, one infers one had a peaceful sleep and no memory of any dreamless sleep awareness is involved.
Advaita Vedānta, in turn, argues against the Nyāyan viewpoint. The debate between them focuses in particular on the ignorance occurring in dreamless sleep, and specifically on how we know or establish the waking report, “I knew nothing.” While we are asleep we know nothing of this ignorance; we come to know it only upon waking up. Yet given that we do not remain ignorant of our own ignorance, how is this knowing of not-knowing possible? The Naiyāyikas claim that we infer we were ignorant because we do not remember anything, but the Advaitins argue that retrospective oblivion is no proof of a prior lack of consciousness. Moreover, when we wake up we have the feeling of having been asleep and having not known anything. This feeling, the Advaitins claim, is better regarded as a kind of memory brought about by the traces of previous experience. So, in some sense, we must experience our ignorance—the unknowing stillness of our mind—in dreamless sleep.
In reply, the Naiyāyikas claim that we have no consciousness in dreamless sleep, and that when we wake up we make an inference by reasoning in the following way: “While I was in deep sleep, I knew nothing, because I was in a special state (I was not awake) and I lacked the necessary means for knowledge (my senses and mental faculties were shut down).” Of course, the Naiyāyikas are not saying that we explicitly make this inference when we wake up. What they are saying is that what looks like memory is really a case of implicit reasoning taking this inferential form.[3]
In order to understand the kind of inference that the Naiyāyikas think we make, as well as why the Advaitins reject the Nyāyan position, it will be helpful to state the inference in the form of the standard Nyāyan syllogism, which forms an important part of the Nyāyan theory of inferential knowledge.
Suppose we are looking at a hill and you say to me, “There is fire on the hill.” I doubt what you say, however, so you need to convince me. You point to the hill and say, “There is smoke on the hill.” I see the smoke and I am convinced. According to the Nyāya, if we want to unpack how perception and inference have worked together to convince me that you are right, we need to formulate the inferential cognition in the following five steps:
1. There is fire on the hill.
[This is the proposition to be proven. It is what you think when you look at the hill, and it is what you want to convince me is the case.]
2. Because there is smoke on the hill.
[This is the reason you give to support what you say.]
3. Wherever there is smoke there is fire.
[This step states the universal concomitance between the presence of smoke and the presence of fire.]
4. As in the case of the kitchen.
[This step provides an example or actual case of the concomitance, to which we both agree.]
5. There is fire on the hill.
[This step states the conclusion, which is the proposition with which we began, but now stated as established and generated by the preceding inferential process.]
Let us now take this five-step syllogism and apply it to the case of dreamless sleep.[4] The Nyāya view is that our knowledge that we knew nothing in dreamless sleep is based on the following sort of inference:
1. While I was in dreamless sleep, I knew nothing (there was an absence of knowledge in my self).
2. This is because (i) I (my self) was in a special state (that is, not awake) or (ii) I (my self) lacked the necessary means for knowledge (that is, my senses and mental faculties were shut down).
3. Whenever (i) I (my self) am in a special state (whenever I am not awake) or (ii) I (my self) lack the necessary means for knowledge (whenever my senses and mental faculties are shut down), I know nothing (there is an absence of knowledge in my self).
4. As in the case of fainting or a blow to the head.
5. While I was in dreamless sleep, I knew nothing (there was an absence of knowledge in my self).
Notice the parallel between the previous inference concerning fire and the present inference concerning dreamless sleep. In the previous case, our concern is to establish the presence of fire on the hill. In the present case, our concern is to establish the absence of knowledge in the self during dreamless sleep. Nevertheless, the form of reasoning is the same.
Again, the Naiyāyikas are not saying that we explicitly go through this inference step by step when we wake up. What they are saying is that we know by inference that we were ignorant during dreamless sleep, and that our inference can be shown to be correct when we make explicit all the steps that it contains. So there is no need to suppose that there is any kind of consciousness during dreamless sleep.
The Advaitins respond by arguing that this inference is faulty and cannot be how we know that there is an absence of knowledge during sleep. The problem is that I need some way to know or establish the reasons for inferring that I knew nothing—namely, that I was in a special state and that I lacked the means for knowledge—and there seems to be no way for me to do this without my relying on the kind of memory these reasons were supposed to obviate.
The first reason the Naiyāyikas give for me to infer that I knew nothing is that I was in a special state, that is, a state different from the waking state. But how do I know that I was in a special state? If I say, “Because I knew nothing in this state,” then I am reasoning in a circle.
The second reason the Naiyāyikas give for me to infer that I knew nothing is that the means for knowledge were lacking—that is, that my senses and mental faculties were shut down. But here too we need to ask, how do I know that these means were lacking? How do I know my senses and mental faculties were inactive?
Suppose I say, “I infer my senses were shut down because they feel refreshed when I wake up.” But here the same basic problem repeats itself. How do I know or establish that there is a relationship between my senses feeling refreshed and their previously having been inactive? Would I not need to have some experience of knowing that my senses were inoperative together with an experience of knowing I feel refreshed in order to establish a relationship between the two? But while I am asleep I do not have any experience of knowing my senses are inactive; I know this only upon awakening. So how do I establish this relationship? If I appeal to yet another inference, then it looks like I am headed off on an infinite regress.
More generally, the only way I can know that the means for knowledge were absent in deep sleep is by knowing that there was no knowledge present in this state. Only by knowing the effect—my not knowing anything—can I infer the cause—the absence of the means for knowledge. So unless I already know what the inference is trying to establish—that I knew nothing—I cannot establish the reason on which the inference relies.
The Advaita Vedānta conclusion is that I know on the basis of memory, not inference, that I knew nothing in deep sleep. In other words, I remember having not known anything. But a memory is of something previously experienced, so the not-knowing must be experiential.
It is important to highlight the larger metaphysical disputes about the self and cognition that drive this debate. For the Naiyāyikas, the self is a non-physical substance. Unlike Descartes, however, who held that consciousness is the essence of the non-physical mind, the Naiyāyikas maintain that the self is the substratum of consciousness and that consciousness is an adventitious quality of this substratum that is present only given the appropriate causal conditions, namely when the sensory and mental faculties are functioning to cognize objects. In addition, cognition consists in the taking of a separate object as content and never in taking itself as its own content.[5] (In the case of introspection, a second-order cognition takes a separate first-order cognition as its object.) For the Advaitins, however, the self is pure consciousness, that is, sheer witnessing awareness distinct from any changing cognitive state. Thus, unlike the Naiyāyikas, the Advaitins cannot allow that consciousness disappears in dreamless sleep, since they think (as do the Naiyāyikas) that it is one and the same self who goes to sleep, wakes up, and remembers having gone to sleep. In addition, for the Advaitins, cognition consists in a reflexive awareness of its own occurrence as an independent prerequisite for the cognition of objects (Ram-Prasad 2007). In other words, the defining feature of cognition is reflexivity or self-luminosity, not intentionality (object-directedness), which is adventitious. Thus, during dreamless sleep, although object-directed cognition is absent, consciousness as reflexive and objectless awareness remains present.
It may help to use the modal notions of necessity and possibility to describe the difference between these views. For the Naiyāyikas, to be in a conscious state is to be in an object-directed state. Given that dreamless sleep is not an object-directed state, it is necessarily the case that consciousness is absent from this state. Nevertheless, if it could be shown that object-directed cognition can occur in dreamless sleep, then the Nyāya could allow for the possibility of consciousness during dreamless sleep. Such consciousness, however, would have to be intermittent or episodic, since object-directed cognitions come and go. What the Nyāya cannot allow is that consciousness is intrinsically reflexive or self-revealing (self-luminous), or that it can occur without an object. Furthermore, for the Nyāya, consciousness requires a substratum, since consciousness is a mental quality, and mental qualities require the substratum of the self. Therefore, although the self continues to be present during dreamless sleep, consciousness is absent. The Advaitins agree with the Naiyāyikas that the self remains continuously present during dreamless sleep, but they maintain that the self is pure consciousness—consciousness as intrinsically reflexive and self-revealing, not as contingently and adventitiously object-directed. So, for the Advaitins, consciousness cannot ever be absent from dreamless sleep, which is to say that it is necessarily the case that consciousness is present throughout dreamless sleep.
Given these differences, the Nyāya might be thought to be more flexible than Advaita Vedānta with regard to the specific issue about dreamless sleep, since the Nyāya can allow for the possibility of intermittent consciousness during dreamless sleep, whereas Advaita Vedānta cannot allow for any absence of consciousness in this state.
Despite this limitation of the Advaita Vedāntan view, it is possible to extract a key phenomenological idea from its metaphysical commitments. This idea is that when I wake up from a dreamless sleep, it seems that I can sometimes knowingly say I have just emerged from a dreamless sleep, and this saying seems to be a reporting of my awareness, not the product of having to reason things out (Kesarcordi-Watson 1981). It is this thought that provides a premise of the Advaita Vedāntan argument for consciousness continuing in dreamless sleep, and this thought is logically distinct from the Vedāntan belief that the self is essentially pure consciousness.
This phenomenological thought, however, is open to the objection that, given an apparent memory, it does not follow that the state apparently remembered was consciously experienced. For example, we may have apparent memories of childhood events, yet their presence does not imply that these events were consciously experienced, for the memory impressions may have been acquired from other sources of information, such as things our parents told us or family photographs. Similarly, during dreamless sleep, information may accumulate non-consciously from a variety of interoceptive and exteroceptive sources, and upon awakening we may realize that something was going in our mind while we were asleep, though at the time we had no experience of it.
At one level—the level of the empirical psychology of memory—we can make the same reply here that we made above to the objection to the Yoga argument, namely that all the argument requires is the possibility of there being genuine veridical episodic memories upon awakening of having been peacefully asleep; the argument does not need to establish that every apparent waking memory is such a memory. Unlike remote memory (of the sort we have for childhood events) or semantic memory (memory for learned facts or words), episodic memory is standardly taken to require that the events “encoded” in memory are experienced at the time of encoding. So, if there are possible cases upon awakening in which there is any kind of genuine episodic memory “retrieval” of the dreamless-sleep state, it follows that in such cases something about the state of being dreamlessly asleep must have been experientially encoded.
At another level—the level of cross-cultural philosophy of mind—we can see in the Vedāntan phenomenology the basis of a transcendental argument. Transcendental arguments aim to deduce what must be the case in order for some aspect of our experience to be possible. In the present case, the aspect of experience with which we are concerned is not simply that we sleep but that we know that we sleep. What are the necessary conditions of possibility for this kind of self-knowledge? To put the question in a more phenomenological way, how is it possible for you as a conscious subject to experience yourself as one and the same being who falls asleep, who does not actively know anything in being asleep, and who emerges from sleep into waking life? The Vedāntan view is that a retrospective inference across the gap of a complete absence of consciousness will not suffice to make this kind of unified self-experience possible. Rather, you must have some kind of experiential acquaintance with dreamless sleep as a mode of your conscious being.
We can take a further step and think about the Vedāntan argument not just from a Kantian transcendental perspective but also from a Husserlian transcendental phenomenological perspective. From this perspective, the core of the Vedāntan argument concerns not so much episodic memory in the sense of the distinct mental act of recollection but rather what Husserl calls “retention”—the holding onto the just-past as an intentional content belonging to our consciousness of the passage of time, including our own mental lives as flowing in time. The Advaita Vedāntan thought is that, at the moment of waking up, I can experience by retentional awareness my having just been asleep and my having not known anything. What Nyāya fails to see, according to Vedānta, is that I need this kind of retentional awareness in order to have the first-person knowledge that I slept and to ground any retrospective inference I may subsequently make.
Of course, even if we suppose that there is or can be such a direct memory in the form of a retentional awareness of the deep sleep state, the presence of such a memory would not suffice to prove the continuous presence of consciousness throughout the entirety of dreamless sleep. After all, the presence of such a memory seems compatible with there having been moments or periods during which consciousness vanishes completely, with the sleeper remembering only the later smoothed-out and mentally-merged, conscious parts of sleep. Nevertheless, if dreamless sleep allows for or includes phases in which awareness is present, then this state cannot be defined as one in which consciousness is absent.
Another important Advaita Vedāntan thought is that when I say I just woke up from a dreamless sleep, the first-person pronoun does not refer to my autobiographical self—my self as I represent it in personal memory. Rather, it picks out my consciousness or subjectivity itself. To use a phenomenological idiom, it picks out the “ipseity” or minimal selfhood of consciousness in contrast to the ego as a mentally represented object of memory or reflection. But whereas the Advaitin takes this minimal selfhood to be a transcendental “witness consciousness” (Gupta 1998), it is open to us today to maintain that it is my embodied self or bodily subjectivity, or what phenomenologists would call my “pre-personal lived body.” In this way, we may be able to remove the Advaita Vedāntan conception of dreamless sleep from its native metaphysical framework and graft it onto a naturalist conception of the embodied mind—a conception that should also appeal to the Cārvāka or naturalist school of Indian philosophy (see Ganeri 2012, pp. 69–97), besides being tractable for cognitive science.
Cognitive science is also relevant to an interesting disagreement between Yoga and Advaita Vedānta concerning cognitive activity during dreamless sleep. Advaita Vedānta maintains that cognitive activity ceases during dreamless sleep and only consciousness remains, whereas Yoga maintains that cognitive activity continues during dreamless sleep (see Dasgupta 1922, pp. 460–61). To understand this difference it is important to note that both traditions distinguish between consciousness, which is the self-luminous (reflexive) and passive witnessing awareness, and the mind, which is the intentional or object-grasping cognitive system. Moreover, in the Yoga view, the mind is material, and so is not different from the body (see Schweizer 1993). According to Yoga, deep sleep is a subtle or reduced state of the mind, specifically of the “inner sense” (antaḥkaraṇa), which includes both mental cognition (manas, which processes and integrates sensory material, and buddhi, which intellectually discriminates and judges) and the sense of ego (ahaṃkāra, the feeling, “I am”). Thus, for Yoga, cognitive activity, particularly the formation of memories, continues subliminally in deep sleep, and this process is physical or physiological. According to Advaita Vedānta, however, the mind, specifically the inner mental sense, shuts down entirely in deep sleep, leaving only the passive “witness consciousness” and the life processes of the body. If we set aside the question of consciousness and ask whether cognitive activity, specifically memory formation, occurs during deep sleep, the answer from cognitive science is unequivoval, for evidence from psychology and neuroscience indicates that memory processes are strongly present in deep sleep (Diekelmann & Born 2010; Walker 2009). These processes include both passive and active forms of memory consolidation (the strengthening of newly-acquired memories and the integration of them with older ones). Of course, this kind of memory consolidation is thought to occur in the absence of consciousness, so this evidence does not support the Yoga and Vedāntan view that consciousness continues in dreamless sleep. Nevertheless, the evidence does support the Yoga view that physiologically-instantiated cognitive processes continue in dreamless sleep, contrary to both Advaita Vedānta and Nyāya, which believe the mind shuts down in dreamless sleep.
The claim that mental activity ceases in dreamless sleep while consciousness remains creates another difficulty for the Advaita Vedāntan view. If the inner sense stops functioning in dreamless sleep, then how is the waking memory, “I slept peacefully and I did not know anything,” formed? Episodic memory requires the encoding of experience, so if there is no experience of “I” in dreamless sleep, then how can I remember that I slept well?
The Advaita Vedānta answer is clever (see Dasgupta 1922, pp. 460–461). In deep and dreamless sleep, ignorance completely envelops the mind. Since the ego sense is inoperative, it doesn’t appropriate this ignorance to itself, so there is no feeling of the ignorance belonging to an “I.” At the moment of awakening, however, the ego sense, grounded on the felt presence of the body, reactivates, and the mind starts up its cognitive workings. Immediately, the ego sense appropriates the lingering impression or retention of not-knowing and associates this retention with itself, thereby generating the retrospective thought, “I did not know anything.”
From the Vedānta. perspective, this “I” is not the true self; it consists in a mistaken superimposition of the self onto the mind-body complex. The true self is the egoless “witness consciousness” (egoless, because it is not a function of the ego sense). The Advaitin take this “witness consciousness” to be transcendental and not essentially embodied. It is open to us today, however, to suppose that if there is some kind of egoless and basal consciousness that can continue to be present in dreamless sleep, then it is a fundamentally embodied consciousness, perhaps a minimal mode of sentience consisting in the feeling of being alive. This thought provides another example of how it may be possible to separate the Advaita Vedāntan conception of consciousness in dreamless sleep from its original metaphysical framework and graft it onto a contemporary naturalist conception of the embodied mind.
If we project some terminology from contemporary philosophy of mind onto Yoga and Advaita Vedānta, then we can say that dreamless sleep counts for these Indian philosophers as a “phenomenal state” or a state of “phenomenal consciousness”—a state that has a phenomenal character or for which there is something it is like to be in that state. What is it like? Yoga and Vedānta describe deep and dreamless sleep as peaceful, as one undifferentiated awareness not divided up into a sense of being a distinct subject aware of a distinct object, and as blissfully unknowing. From a contemporary naturalist perspective, this conception could be taken as a description of a quiescent and tranquil form of sentience or the feeling of being alive. Under this description, dreamless sleep would not count as a state of “access consciousness”—a state whose phenomenal content or character we can cognitively access, hold in working memory, and use to guide our attention and thinking. We seem to have no cognitive access to being asleep during sleep; rather, we gain access retrospectively in the waking state. On this conception, in dreamless sleep we are phenomenally aware but we have no cognitive access to that awareness at the time.
Ultimately, however, this way of conceptually parsing the Yoga and Vedāntan view will not work. A central commitment of Yoga and Vedānta, as well as Indo-Tibetan Buddhism, is that we can gain access to the state of dreamless sleep through meditative mental training. I will come back to this idea at the end of this paper. But first we need to consider the default view of consciousness and dreamless sleep in cognitive neuroscience.