“Only the dead can know what it means to be dead.”—Ananda Coomaraswamy
Philip Larkin’s poem “Aubade” articulates his fear of death in chilling terms. It describes a man who hates his job and gets drunk every night. Then, before dawn, he wakes, and with the gathering light, he fixates on the certainty of his own death and what it will mean for him. Larkin is clear that it means complete cessation of the self, that there is no possibility of an afterlife, and that this absence of the self is the most terrifying thing in the world. He describes his terror of:
the total emptiness forever,
The sure extinction that we travel to
And shall be lost in always. Not to be here,
Not to be anywhere,
And soon; nothing more terrible, nothing more true.
Larkin is contemptuous of those who try to palliate the emptiness of death through a belief in a heaven, or through philosophical arguments that claim death should not be feared. Religion is:
That vast, moth-eaten musical brocade,
Created to pretend we never die.
Stoicism in the face of the certainty of death is an impotent weapon, bringing no comfort:
Courage is no good:
It means not scaring others. Being brave
Lets no one off the grave.
Death is no different whined at than withstood.
And this clear-sighted terror of the inevitability of death, and the consequent loss of the self forever, always lurks in the margins of our waking consciousness, “just on the edge of vision.” In everyday life it is something that is not openly considered, and so it can be pushed to one side. Only in moments of quiet lucidity, “when we are caught without/People or drink,” does the full sense of the imminent cessation of our being come into focus. And the breaking light of dawn, which gives the poem its ironic title, provides just such a moment for morbid self-introspection.
“Aubade” is a brilliant and terrifying poem. It expresses perfectly the modern attitude towards death: the belief that the self will end, that religion is a palliative, and that death is to be feared. It does so with a cold and precise use of language and rhyme that seduces the ear and convinces the mind. It is a poem which conveys the terror of “the emptiness forever,” but whose author is not able to believe in anything better. His intellectual integrity prevents him from performing the sleight of mind necessary to believe in an afterlife which modern science has forbidden to us.
As far as the modern attitude towards death goes, there is no such thing as a metaphysics of death. There is only the physical process of the termination of the organism. This, as Larkin correctly points out, is because we no longer believe in religion, and so have abandoned the belief that there resides in the individual an immortal soul that will endure after death. We tend to concur with most scientists that a living organism is a material construct whose sense of self and whose consciousness are phenomena arising from purely material processes. If human consciousness is a merely material construction then it would logically follow that it would disappear with the death of the body. It should then be worth considering whether or not the belief that consciousness emerges from purely material processes is correct.
If the materialist paradigm is correct, and there is no ”immortal soul” energizing our being, then it means that it is, in principle, possible to map the entire neurological network so that it could be replicated. As long as the materialist paradigm holds, it will allow for the enormously complicated task of identifying and measuring the neural connections in the brain and reconstructing them. If we, like Larkin, disallow the existence of an eternal spirit, or some such other ethereal essence, then there is nothing in principle that should prevent us from creating an artificial intelligence from which a sense of consciousness could arise. All that is required is that the unique material composition of the brain can be simulated in a computer program and then, as a consequence of the complex interactions of the neural network, consciousness will arise. If consciousness is some sort of illusion produced by material interactions in the brain, then the building of a brain in an artificial intelligence program should give rise to the spontaneous creation of an individual consciousness within a virtual world. Success in this task would demonstrate that it is possible for consciousness to arise from purely material foundations, without the need to evoke an ”immortal soul,” and we would then agree with Larkin that death is ”emptiness forever.”
The attempt to create an intelligent machine has been ongoing since the dawn of the computer age. The belief that consciousness emerges from material processes in the brain leads to the inevitable conclusion that it is therefore, in principle, possible to recreate consciousness outside of the human brain; in a computer, for example. The challenge for researchers working in the field of AI is to identify the exact nature of the neural pathways and their connections, and to replicate them in a computer program.
In the early days of AI research, there was a great deal of optimism about what computers would be able to achieve and how quickly they would be able to do it. One of the leading AI researchers, Marvin Minsky, was the technical consultant for the film 2001: A Space Odyssey. In the film, the computer HAL 9000 is a fully conscious and self-aware being engaging in conversation with humans, displaying an understated matrix of emotions and an ability to appreciate art. It is made clear in the film that HAL is treated as though he is a fully developed individual. Kubrick was keen to ensure that his rendering of HAL was scientifically plausible. The input of Minsky ensured that the depiction of HAL represented scientifically informed ideas of what computers would be able to become by the turn of the millennium.
What is clear from watching 2001 is that workers in AI in the 1960s regarded the development of intelligent, conscious computers as something that would be achieved imminently. This was because there was an assumption that the barrier to such self-conscious machines was the processing and computational speed of existing computers. It was rightly predicted that processing speed would increase exponentially over the coming decades, and it was therefore assumed that the ability to execute programs capable of replicating the complexity of human thought would follow.
But there was a flaw in this reasoning. A minority of thinkers refused to be drawn into the optimism of the early AI research community. One of the most prominent was Hubert Dreyfus, whose ideas concerning the limitations of computing power were mostly ignored at the time, but have subsequently been proven to be entirely astute. What Dreyfus realized was that the barrier to creating consciousness in a computer was not the deficit in processing power, but the very context of ”being” in which a computer existed.
In 1965, Dreyfus was asked to write a paper on the future of AI research for the RAND Corporation. He produced a document called Alchemy and Artificial Intellience which likened the research of contemporary AI researchers to the medieval alchemists’ attempts to turn base metals into gold. Dreyfus was convinced that the problems facing AI were not problems of processing power, size of memory capacity, or any other practical difficulties. He claimed that there were fundamental problems in principle with the claim that human intelligence could be reproduced in a digital computer. His paper attempted to show that there are certain capacities that human beings excel in which computers would never be able to master. In making his argument, he described three types of information processing which he claimed were “uniquely human.” They were fringe consciousness, essence/accident discrimination, and ambiguity tolerance.
Fringe consciousness is the unconscious awareness that human beings have for wider contextualizing states that exist beyond the area of present attention. For example, in a game of chess, “cues from all over the board, while remaining on the fringes of consciousness, draw attention to certain sectors by making them appear promising, dangerous, or simply worth looking into.” We do not make these cues explicit to ourselves, and often may be totally unaware of them, but they form a relevant background to those decisions we explicitly make.
Essence/accident discrimination describes the type of insight that human beings employ to instinctively distinguish which information is relevant and which is to be ignored in any given situation. Again, this process is carried out mostly unconsciously; it is never necessary for us to consider every piece of sensory input in our surroundings in order to decide what is inessential. The inessential is filtered out automatically.
Ambiguity tolerance is the ability that humans have to operate without explicit definitions or rules. This can be illustrated with a consideration of language use. It is perfectly normal for words to have secondary or tertiary meanings. In practice, we determine what sense of a word is being deployed with reference to its context. For example, if we were to say that one runner in a race was ”miles ahead” of the others, we would automatically know that the word “mile” is not being used in its primary sense of “a unit of linear measure equal to 1,760 yards,” but in its tertiary sense of “a very long way or a very great amount.” Even then we have no problem in understanding that “a very long way” might, in this context, refer to as little as two or three yards; amounts which in different contexts would be seen as very small indeed. In other words, we are capable of sufficiently reducing ambiguity to make meanings clear, but without the needing to explicitly define how we are doing this.
In a later work, What Computers Can’t Do, Dreyfus expanded even further on his skepticism regarding AI research. In this work, he identified four assumptions which he believed were uncritically and often unconsciously being utilized by AI workers to underpin research into AI. Dreyfus believed that the goal of achieving artificial intelligence in a computer could only be achieved if these four assumptions were correct but that, in fact, they were all false. These assumptions were the biological assumption, the psychological assumption, the epistemological assumption, and the ontological assumption.
The biological assumption is based on the fact that neural firings in the brain are “all or nothing” bursts of energy. This observation from neuroscience has been extrapolated to imply that such firings therefore correspond to bits of information in a digital computer, which operate in a binary “all or nothing” manner. In a computer, each bit of information is a discrete unit that has a particular symbolic function. But in the brain, Dreyfus argues, the neural firings that superficially resemble such bits of information are modified and “interpreted” according to many other localized conditions, such as rate of pulsing, frequency of pulsing along particular pathways, and interaction with other neurons. In short, the biology of the brain appears to be more analog than digital in character.
The psychological assumption prompts a somewhat philosophical treatment from Dreyfus. Researchers in AI usually assume that human psychology is a process that operates rather like a computer program, that is, that it is essentially an exercise in information processing. The problem for AI researchers is how to translate the physical properties of the brain into the higher-level intellectual concepts of the mind. As long as the brain is described in terms of its physical behavior, there is no problem; seeing a chair can be described as the presence of light waves on the retina causing a sequence of chemical reactions in the brain, all of which can be described quite precisely. But to speak of really “seeing” a chair, it is necessary to use a different sort of language, language which is more appropriate to the mind than the brain. AI researchers, according to Dreyfus, attempt to bridge this gap by suggesting that there is a level of information processing that occurs in the brain that can organize neuro-chemical bits of information into higher-level concepts. Unfortunately, there is no evidence that this is the case; in fact, in the absence of evidence, AI researchers postulate as yet unknown information processing functions of the brain, merely based on the supposed analogy with computers.
The epistemological assumption is concerned with the way in which humans know how to perform particular actions. It describes the belief that all non-arbitrary behavior can be formalized, and therefore can be reproduced. Dreyfus argues that any such formalization of human behavior, which would enable it to be programmed into a computer, would merely result in an imitation, rather than a reproduction, of that behavior. The computer would need to follow discrete stages of processing in order to perform any particular function, and Dreyfus is far from convinced that this is in fact how humans behave in practice. AI researchers assume that behavior must follow certain heuristic steps, and that where someone is unaware of following such steps that they must be being carried out unconsciously. Against this view, Dreyfus argues that human behavior is always rooted in a particular situation and orientated towards certain goals. Because of this, people effortlessly grasp the particular local aspect of any subject under consideration due to their experience in the situation. A computer has to work through all possible interpretations, discard those that are irrelevant, and focus on those that are relevant. Human beings do not follow such procedures due to their being located in a particular existential situation.
The ontological assumption concerns a fundamental problem for AI research. As Dreyfus notes, “the data with which the computer must operate if it is to perceive, speak, and in general behave intelligently, must be discrete, explicit, and determinate; otherwise it will not be the sort of information which can be given to the computer so as to be processed by rule.” Because computers must operate in terms of such discrete data, it has become habitual for AI researchers to make the assumption that this data is actually present as an aspect of the world; that we, in fact, perceive the world through such data. Contra such researchers, Dreyfus posits that, even where we are able to make explicit our perceptions of certain objects, any such fact is itself contextualized by its particular human situation: “Even a chair is not understandable in terms of any set of facts or ‘elements of knowledge.’ To recognize an object as a chair, for example, means to understand its relation to other objects and to human beings. This involves a whole context of human activity of which the shape of our body, the institution of furniture, the inevitability of fatigue, constitute only a small part.” Moreover, this situation cannot itself be reduced to isolated, context-free facts; it is colored by influences from the preceding situation, so that we build up associations and interpretations over time.
For a computer, this learning-through-time model presents a problem. If data can only be interpreted according to a situation, and if that situation relies for its meaning on the previous situation, then it seems to lead to an infinite regress. At some point, a programmer has to decide what information to give to a computer to begin with, and this will be based on the programmer’s own, human, situation; it will not arise naturally from the computer’s ”consciousness.” In humans, this paradox is avoided by the fact that we are, in Dreyfus’ words, ”wired genetically as babies” to recognize certain stimuli as positive and nurturing, and others as harmful. This appeal to genetics provides a powerful argument for the unique nature of human consciousness.
The Traditionalist View
According to Traditionalist teachings, the unique, inimitable aspect of the human being corresponds with the Divine. This aspect is mostly obscured by more transient aspects of the self, and in the modern climate it is precisely those ephemeral, shallow characteristics that find greatest resonance in their outer surroundings. Ananda Coomaraswamy writes about the supremacy of the Divine aspect of the self and entreats that it is necessary to experience the death of the personal self, that is, of the transient, passing attributes of one’s biography, and to realize the eternal aspect of the self which is identified with God. Thus:
“I” neither think nor see, but there is Another who alone sees, hears, thinks in me and acts through me; an Essence, Fire, Spirit, or Life that is no more or less “mine” than “yours,” but that never itself becomes anyone; a principle that informs and enlivens one body after another, and that which there is no other, that transmigrates from one body to another, one that is never born and never dies, though president at every birth and death.
This “spirit” that animates all beings, yet somehow transcends each, is the part of the self which is seen as being truly authentic. We recognize it throughout religion and folklore. When a child is born his spirit is delivered into him from the heavens, and we say that he has been delivered by a stork. When he dies, he ascends to sit with angels whose wings denote that he has returned to the same airy realm from whence he came. In heathen times, he would have been escorted by similarly-winged Valkyries.
Norse theology provides us with another insight into the way that this spirit partakes of many whilst retaining unity. When Odin sits on his throne, Hliðskjálf, he is said to be able to see everything in the world. He becomes omniscient by virtue of the fact that he is the All-Father, the fundamental generating principle in the world, and as such he is the spirit, or animating force, which expresses its particularities through each of us. He can see through our eyes because, to the extent that we are authentically alive, he is us.
Similarly, Odin’s ravens, Hugin and Munin, are said to leave him in the morning and return in the evening to tell him everything that they have witnessed around the world. The name Hugin means thought, whilst Munin means memory. Again, this attests to the omniscience of Odin. That Hugin and Munin are winged messengers again denotes that it is the realm of the spirit which is the key to understanding the nature of this transpersonal consciousness.
The omniscient aspect of Odin is retained even today through the attenuated figure of Santa Claus, who is able to see whether children are behaving or not. The importance of omniscience in these tales is not to signify that Odin is a different order of being than us, but to show that the Divine spirit is something which is transmitted through us. Odin is omniscient because he is all of us; he is the summation of being. In this sense, then, we can see why Coomaraswamy draws attention to the duality of being:
Our whole tradition everywhere affirms that “there are two in us”; the Platonic mortal and immortal “souls,“ Hebrew and Islamic nefesh (nafs) and ruah (ruh), Philo’s “soul” and “Soul of the soul,“ Egyptian Pharaoh and his Ka, Chinese Outer and Inner sage, Christian Outer and Inner Man, Psyche and Pneuma, and Vedantic “self” (ātman) and “self’s Immortal Self” (asya amrta ātman, antaḥ puruṣa)—one the soul, self, or life that Christ requires us to “hate” and “deny,” if we would follow him, and that other soul or self that can be saved.
The Traditionalist view of human individualism, as representing the ephemeral, transient aspect of a deeper and lasting higher self, is quite at odds with the sick ennui expressed by Larkin when contemplating the ultimate loss of the self. For Larkin, everything that can be experienced, all potentialities, are locked up inside the individual consciousness of a human being, and with his death all possibility is destroyed forever. For the Traditionalist worldview, this human consciousness is barely even the tip of the iceberg; in many respects, it is a positive obstruction to the proper apprehension of the authentic self. Hence Coomaraswamy’s approval of Meister Eckhart’s words, “the soul must put itself to death.” From the Traditionalist perspective, Larkin is in error in perceiving the entirety of his self to be contained within his individual personality. His terror of “the total emptiness forever” is the fear of one who has never suspected the presence of an animating life force anterior to the everyday personality. For such an individual, death is indeed a total cessation, as the self is only identified with its transient expression. The aim of Traditionalist teachings is to unveil the profound spirit which transcends the individual consciousness.
A Surprising Ally
In positing this dichotomy between the teachings of Traditionalist metaphysics on the one hand and the contemporary materialist paradigm on the other, it is surprising to note that some support for the Traditionalist view can be found deeply embedded in the enemy camp.
The most strident exponent of the modern atheistic, materialist paradigm is the neo-Darwinian evolutionary biologist, Richard Dawkins. In his classic neo-Darwinist text, The Selfish Gene, Dawkins argues that human beings, as well as other evolved organisms, are not the discrete, autonomous entities that they might consider themselves to be. Instead, such bodies are “gene machines,” organic vehicles for DNA molecules, whose purpose is to spread and protect the particular genetic sequences which inhabit them. For Dawkins, it is the gene, the unit of chromosomal material involved in evolution, that is really in the driving seat, acting according to the unfathomable dictates of millennia of blind selection. This process of selection, whilst without purpose, must necessarily favor genes which are good at surviving. So much is axiomatic. Those genes which survive well will do so by inhabiting organic material that is capable of providing good protection. Such organic material might range in sophistication from a thin cell wall to a human being.
The importance of Dawkins’ insight is in seeing the gene as the basic unit of life, and in understanding that it is the gene, not the organism in which it resides, which is the driving force behind evolutionary development. It might not be stretching it too far to say that, when viewed from this neo-Darwinian perspective, the purpose of life is the continuation of the gene, although Dawkins would not use the word “purpose.” It is also worth remembering that the survival of the gene is not a matter of individual genes establishing a line of heredity as organisms do; it is a matter of them perpetuating exact copies of themselves. DNA is a sequence of information which is used to shape and build organic material. Whilst the body it inhabits is born anew in each manifestation of a new generation, the genetic information present within the DNA remains exactly the same. In Dawkins’ words:
[The gene] is no more likely to die when it is a million years old than when it is only a hundred. It leaps from body to body down the generations, manipulating body after body in its own way and for its own ends, abandoning a succession of mortal bodies before they sink in senility and death. The genes are the immortals . . . We, the individual survival machines in the world, can expect to live a few more decades. But the genes in the world have an expectation of life that must be measured not in decades but in thousands and millions of years.
Whilst Dawkins’ detractors often focus on his secular, atheistic concerns, what is remarkable about the selfish gene idea is the way that it seems to support a Traditionalist view of the self. As Dreyfus pointed out in his objections to AI, the human being is born with certain genetic predispositions which orientate him into seeking certain cues from his environment, such as a mother’s nipple. From these simple cues, it follows that the human organism will potentially be able to achieve its primary aim of survival, and then develop into a more sophisticated individuated consciousness. But this individuated consciousness is not replicable in an artificial context, because it requires for its manifestation a prior genetic history of countless millennia of patient evolutionary development. “We are millions of yesterdays,“ wrote Austin Osman Spare, “and what appears autogenic is the work of unknown mediators who permit, or not, our acts by the mysterious chemistry of our believing.”
For Coomaraswamy, we are vehicles of the Divine spirit; for Dawkins, we are vehicles of pure information. Consider the similarity between Coomaraswamy’s words, “a principle that informs and enlivens one body after another, and that which there is no other, that transmigrates from one body to another, one that is never born and never dies,” and Dawkins’, “It leaps from body to body down the generations, manipulating body after body in its own way and for its own ends, abandoning a succession of mortal bodies before they sink in senility and death.” It should not be a source of great concern that one writer is considering the incomprehensible vastness of God whilst the other is considering infinitesimally tiny units of information. Both demonstrate the way in which our lives are suffused with an essence which is in us, but is more than us; and both show that we are expressions of something far greater than our egoic selves can truly comprehend. And if Dawkins is the primary atheist of our age, it should be of no consequence; after all, what is God if not the hidden intentions of eternity?
It might be objected that this brief consideration of the metaphysics of death gives no comfort. We will all die, “and soon” as Larkin mordantly reminds us. And for the outer self, the egoic personality, there is no hope: it will perish with the body. But this personality is already an illusion, a conceit of the organism. It is no more the real self than your name is the real you. It must be put to death by the real, immortal self. For anyone who can shift the locus of their being into this immortal self, reunion with the Divine awaits. Only the dead can know what it means to be dead.
1. Ananda Coomaraswamy, “The Meaning of Death,” in Coomaraswamy: Selected Papers, Volume 2: Metaphysics (Princeton: Bollingen, 1977), 429.
2. Hubert Dreyfus, Alchemy and Artificial Intelligence (Santa Monica, Calif.: RAND Corporation, 1965).
3. Hubert Dreyfus, What Computers Still Can’t Do: A Critique of Artificial Reason (Cambridge, Mass.: MIT Press, 1992), 206.
4. Ibid., 210.
5. Coomaraswamy, “The Meaning of Death,” 428.
6. Ibid., 428.
7. Richard Dawkins, The Selfish Gene (Oxford: Oxford University Press, 1989), 34.
8. Austin Osman Spare, Axiomata (London: Fulgur, 1992).