An outline for an addendum to my Essay on Human Consciousness: A Solution to the Easy Problem (H Kong, 2008)
“For my part, when I enter most intimately into what I call myself, I always stumble on some particular perception or other, of heat or cold, light or shade, love or hatred, pain or pleasure. I never can catch myself at any time without a perception, and never can observe any thing but the perception.”
“She’s making sure she is not dreaming
See the lights of a neighbor’s house
Now she’s starting to rise
Take a minute to concentrate
And she opens up her eyes
The world was moving and she was right there with it (and she was)
The world was moving and she was floating above it (and she was)
And she was”
Phenomenal consciousness is defined as the pure, pristine contents of awareness at a given moment; what it feels like for you to be here now. This is as opposed to something more extended like verbal report or self-awareness.
What is the “hard problem” as David Chalmers describes it? The essence of the first person subjective phenomenonology as opposed to identifying the underlying physical or biological structures or functions that cause or correlate with or constrain it.
Many descriptive models have been made purporting to answer the hard problem but they usually end up explaining something else such as how the brain codes conscious information differently from that which is unconscious. These include models based on neural networks and circuitry, or even quantum level phenomena. Models can be very concise, powerful, and predictive in answering where and how awareness arises in the brain. But descriptive models cannot tell us why subjective experience exists at all.
I will describe three such models: global neuronal workspace (B. Baars, S. Dehaene), information integration theory (G. Edelman, G. Tononi), and quantum coherence (R. Penrose, S. Hammeroff). The first is perhaps the best current neuroscientific model of consciousness in the brain. It accounts for much of the neurophysiological and brain imaging data and fits well with results gathered from cognitive psychology as well as neuroscience. More than simply identifying the NCC, it is a powerful predictive model. The second is an ambitious theory that sets out to compute information space as it correlates to mental states. It makes quantitative predictions of conscious awareness. The third approach is highly controversial in its assumption that the nature of consciousness is somehow dependent on quantum level physical states. Common to all three models is the tacit assumption that phenomenology is caused by dynamic patterns of neural activity, synaptic integration, or quantum microstructure. Except for perhaps the information integration theory, they do not directly address why subjectivity exists.
Dan Dennett famously denies there is a second problem of consciousness. For him, subjectivity is simply an illusion created by the pandemonium of parallel processers embedded in neural architecture.
Until recently I agreed with Dennett that the “easy problem” was hard, while the “hard problem” didn’t exist, but I now think that view is perhaps too simplistic. Like Chalmers, I agree there is a second problem to be addressed. But I think that in the end, the solution will come as much from philosophy as it will from science.
So let’s reframe Chalmers’ hard problem as the “second problem of consciousness” (SPC). It has three axioms. First, there is a physical universe in which subjectivity unfurls. This sets him apart from the immaterial idealism of Plato and Berkeley as well as the hard dualism of Descartes. So far, this is not inconsistent with emergent physicalism. Second, there must be a direct correlation between the structural and functional pattern of a conscious state, as it exists in nature, with the patterns that make up the phenomenology of that state (structural coherence). This, in essence, is a functionalist view, though Chalmers maintains that there is more to consciousness than just function. Third, the structure of information somehow equates to both the subjectivity of the experience and the nature of the object experiencing it (dual patterning). Chalmers calls this stance “naturalistic dualism”.
Chalmers believes that the SPC is the first person perspective, which, though correlated with neural processes, is not identical to them. His solution, naturalistic dualism, implies that the SPC is an informational construction that has both subjective and objective components. In this view, consciousness is a fundamental and irreducible property of information as it exists in the physical universe. The nature of experience is defined by the informational content embedded within a specific spatiotemporal arrangement of matter and energy. The trick in nailing Chalmers’ chimera is to find the set of mathematical formulas that unify physics and phenomenology within a consistent and predictable framework. This view is close to panpsychism, the belief that the entire universe is bathed within a conscious field. Chalmers maintains that his stance is more “proto-panpsychism” because consciousness may exist in pockets here and there, but not everywhere. In this respect, Tononi’s IIT is quite similar, although it is unclear if Tononi believes in the dual identity of information as structured in “qualia space” as opposed to being a representation of subjectivity.
I believe it is a flawed assumption that consciousness/subjectivity/phenomenology/first person perspective/qualia are fundamental properties of nature akin to mass or charge or gravity. To the extent that such phenomena exist at all, they are likely to be more complex entities reducible to the familiar laws of natural science. Panpsychism is incoherent because if the entire cosmos were conscious, then nothing would be unconscious. Proto-panpsychism is wrong because it takes an unnecessarily reductionistic view of complex informational processing. While I agree that consciousness comes from informational organization of matter, it is an emergent property of that information rather than something intrinsic to it. The notion of dual patterning at a fundamental level is a misattribution error. Phenomenology does not come from information; it is given to it by an “observer”.
If subjectivity, qualia, and the first person stance are not fundamental, intrinsic, and irreducible properties of nature, then where do they come from and why do they exist? The SPC is a real question because nothing is more ineffable than Descartes’ famous words of self-realization. It really does seem to us that there is something it is like to be. That is a statement we can make based on belief. But, as we shall see, it is a belief based on a chain of illusions.
“I believe how it seems to feel now”: the Interpretations of Multiple Illusions (IMI) Theory
The statement above neatly encapsulates the hard problem of consciousness: the nature of subjective experience. I will derive a solution by careful analysis of this statement. First, the subject “I” denotes an experiencer who feels whatever it is to be. Second, the verb “believe” presumes an act of choice (to believe). Third, the object “it” refers to the underlying reality. Fourth, the modifier “seems to feel” assumes an interpretation or judgment of this deeper truth. Finally, “now” implies the absence of time delay in phenomenological processing. I will argue that all of these elements, save the third, are illusions created by direct mapping and matching of neural elements. Therefore, the solution to the second problem of consciousness is simply an interpretation of multiple illusions.
Let’s start with the illusion of a unitary self. The self-illusion is fundamentally based on body representations in the bilateral parietal cortex and insula. These regions of cortex statistically map the homeostatic state of the body at a given time. These representations are then used to predict and guide subsequent behavior. This requires visceral and somatosensory feedback to the body maps in a comparative fashion. Natural selection has seen to it that (usually) the incoming signals from the body tightly match their corresponding physical and physiological representations in the brain, thereby creating an illusion of unity. This is not always the case as exemplified by pathological states of dissociation caused by mismatching neural signals. These include the phantom limb and “rubber hand” illusions popularized by V. Ramachandran and the “out of body” experiences described by T. Metzinger. These illusions illustrate the point that the feeling of self-unity rests on the integrity of body representations in the brain. The “transparent self-model”, as Metzinger calls it, evolved in our primate ancestors as a short cut for complex motor control. Embodied mental processes exist at all levels of perception and cognition and often influence decision making in surprising ways.
At a higher level, the self-illusion is the result of the “ego machine” in the left hemisphere. It has long been known that language is usually localized to the left hemisphere. Brain regions specializing in high-frequency information processing have evolved to enable syntax and grammar, which lie at the heart of the language instinct. What is curious is that R. Sperry and M. Gazzaniga’s split brain experiments have revealed that these brain regions of the left hemisphere also serve an interpretive function, creating inner speech that justifies a coherent view of the world, a view that is nonetheless often deluded.
In a loose sense, a unitary self exists in the form of a physical body that moves and ages through space and time. There is also a kind of unitary psychological self that maintains itself through acquired habits, memories, and predilections. But in terms of the phenomenology that lies at the core of consciousness, the experiencing self is simply a directly matching predictive model.
Free will is the next illusion. If we are to define the term clearly, we need to start at the level of fundamental physics. Quantum theory posits that particles have indeterminate wave functions that permit multiple simultaneous positions in the space-time manifold. This implies randomness and chaos woven into the fabric of reality. But this is an inappropriate level for an explanation of entities like free will or consciousness, that are best understood as emerging at the neural level.
Libet and others have identified neural activity (readiness potentials) correlating with and preceding conscious action. It seems likely that at the level of the brain, decisions are entirely determined by the state of unconscious information processing in multiple distributed networks. Conscious broadcasting occurs well after the decision to act has already been made. The compatibilist view is that randomness or determinism aside, a kind of freedom exists so long as one believes that her behavior is freely chosen. But this cannot be free will in a real sense, but rather in an interpretive stance. This illusion comes about because of the temporal direct matching between predictive models of behavior in the supplementary motor cortex and feedback from actual behavior. When self-behavior is consistent with expected desires and plans, one gets the feeling of exercised free will, but when self behavior is somehow unexpected, as is often the case when multiple conflicting choices load the attentional networks in the prefrontal cortex, one feels a diminution of free will. One may feel more or less free depending on the accuracy of conscious predictions. But the actions themselves have been decided before any conscious control can influence them. As A. Damasio puts it, “we are always late for consciousness”.
This brings us to the illusion of experience itself: that ineffable seemingness of feeling at the heart of the experienced present. I argue that the structure of the phenomenological illusion emerges from neural selection of directly matching self-referential loops. These “strange loops” in neural architecture as Douglas Hofstadter calls them are not a fundamental property of matter like quantum strings but are themselves emergent phenomena.
Chalmers’ argument for the existence of a hard problem rests on defining the essence of experience. But a quale has no absolute essence; it is simply a subjective interpretation. The philosophical quest for the quiddity of subjectivity is a twenty first century version of the search for dancing angels on pinheads. The framability of a question does not imply its answerability.
If a half-century of advances in cognitive psychology and neuroscience has taught us anything, it is that we are incredibly vulnerable to a big bag of phenomenological illusions from binocular rivalry to change blindness to confirmation bias. One can argue that the feeling of an illusion still leaves the feeling to be explained. So let’s look for Chalmers’ chimera.
Feelings imply belief, and beliefs imply meaning. To have a feeling about something, say a field of blooming azaleas, a lecture hall of chattering students, or a painful infected toe literally means that I believe what I see, hear, and feel as I’m seeing, hearing, and feeling it regardless of whether it really exists in the world. Feelings impart beliefs and meanings onto objects but do not come from them. They are not intrinsic to the reality of objects. There is no dual patterning of information. If we are to understand the nature of subjectivity we must look inside the subject rather than out to the object of its perception.
The feeling of what happens, to borrow the title of Damasio’s book, lies at the heart of the illusion. It is best explained from the neuropsychological level as a third direct matching phenomenon. This time, the matching is between widely distributed neural ensembles coding for perceptions, body representations, memories, and emotions with those for executive control in the frontal cortex. The spatio-temporal binding is facilitated by synchronized cortico-thalamic loops that specifically link the ensembles within a “spotlight of attention” into a continuous stream of consciousness. The contents of attention are selected by competition between activated cortical nodes and then made available for broadcast to the rest of the cortex, something Dennett calls “fame in the brain”, and Dehaene calls the “global neuronal workspace”. The coherence of the conscious stream in the wake of continuous bottom-up feed is maintained by massive top-down feedback that provides constant editing of multiple drafts with a delay of a few hundred milliseconds. This is a summary of my Essay on Human Consciousness (2008).
Chalmers would argue that this only addresses the “easy” coding problem and that the nidus of the SPC actually lies in explaining subjective feelings emanating from matching maps. This is misleading because feelings are an intrinsic part of the matching process. I believe that if we can strip down the entire strange loop to its basic fundamental structure, subjectivity will simply arise from the back and forth oscillations of neural ensembles. The contents of the conscious experience lie in the cortical ensemble. The lapsed time in the cortico-thalamic reentry circuit creates a kind of short-term afterglow that reverberates within the loop. The duration of the reverberation is the length of the feeling. The richness of the matched map is the intensity of the feeling. That is the feeling I believe I am now experiencing.
Some see the SPC as a fundamental phase transition whose mystery and import are analogous to the genesis of matter from the Big Bang or the origin of life on earth. In this case, it involves the emergence of subjectivity from the objective. This is a false dichotomy. Subject and object are simply two sides of a single loop whose rotation creates the illusion of a flowing stream of consciousness. They are actually part of a single dynamic process. The cortico-thalamic loop takes an information structure (the map), activates a second closely correlated structure (the match), which then reactivates the first structure a split-second later, by which time the original map has changed slightly, which then reactivates the second structure a split-second after that, an so on and on and on. There is no experiencing self, but rather what Metzinger calls a “phenomenal self model (PSM)” embedded within cortical body maps. These are intimately linked to the contents of the strange loop. As a result, the PSM confuses the qualia with itself. Further, as the loop moves, qualia changes into flowing experience.
Does the strange loop have internal geometry like Kekule’s resonating aromatic rings or vibrating strings dancing in eleven dimensional Planck space? Tononi thinks integrated information in n-dimensional qualia space (“Phi”) explains conscious experience. I think that’s only half the answer. What IIT fails to mention is that the information must be integrated over the loop in order to become conscious. First, the informational content of the loop comes from the integration of the cortical ensemble representing the object. But second, the illusion of subjectivity itself is woven into the very fabric of the cortico-thalamic loop. The two aspects must come together.
Perhaps we can unify Hofstadter’s strange loops with Tononi’s integrated information into a “Hofoni theory” of consciousness. If you’ll excuse the overtone of Eastern mysticism, my guess is that it is in the nature of the recursion process that the inner and outer aspects of representation necessarily chase each other on a single continuous surface. The strange loop is thus a Möbius strip where the yin/yang of subject/object continuously merge into each other in an endless cycle of becoming. How conscious it becomes depends on the amount of information integrated into it. Not only does any strange loop with integrated information have internal experience, but any strange loop with integrated information IS internal experience. Conscious awareness simply is the singularity arising from the loop’s spin.
At the end of the day, we may just have to bite the bullet and accept the fact that the ineffability of internal experience is a fundamental property of a certain kind of information structure: the integrated strange loop is the quantum of qualia. Asking for a further explanandum is like asking why pi is 3.14159… or why the universe exists. Such a theory would make strange predictions, for example are plasmids (circular strands of self-replicating DNA) or viruses conscious to some minute degree? After all, they both satisfy the Hofoni criteria. Perhaps the only way to find out is to build a computer program with that sort of functional architecture and have it mimic conscious awareness.
Many other useful illusions (the first person account, self awareness, freedom of intention, agency detectors, theory of mind modules, linguistic representation, recursive logic, morality, spirituality, and other cultural memes) have accrued onto this foundation over the course of human evolution to enrich our inner experience tremendously compared with other animals.
Thanks to natural selection, the representation within awareness is usually fairly accurate to a functional approximation. This virtual reality simulator that helped our ancestors survive and reproduce was honed and polished over millions of cycles of natural selection. But it is an interpretation of an illusion, nonetheless. The direct matching machinery of our brains can build maps of high resolution and fidelity but it can never duplicate the physical landscape directly. Therefore, conscious perception will always be a virtual interpretation of reality.
The conscious interpreter has great predictive power in guiding subconscious behaviour and controlling pre-programmed subroutines. Hooked up to language modules in the left hemisphere, it provides us with a running internal monologue about our memories, feelings, and desires sprinkled onto incoming sensory feed. The user illusion is so compelling because we have no better model for representing reality to ourselves. However, there is a better virtual reality machine for the transfer of subjectivity to other people. It’s called literature.
Novels, short stories, poetry and other forms of narrative build on our language instinct and creativity generators in an attempt to distill the essence of the first person account in the form of words broadcastable to others. A good novelist hopes to replicate the feelings of his or her characters in such a way that the reader can fully empathize with them, as if entering another mind. Perhaps the best-known example is Proust’s monumental masterpiece of modernist literature, In Search of Lost Time. But even with seven volumes and 4200 pages, he failed to bridge the explanatory gap. When it comes to understanding the inner recesses of another conscious mind, we are still in the dark. This is no fault of Proust. Reading the first person account will always be an interpretation of multiple illusions because the object and subject are two distinct entities no matter how close the direct matching.
In summary, there is no dual patterning at the heart of information theory or phenomenology. There is only the single patterning of physical reality. Subjectivity is an abstraction emerging from the interpretation of information, not something intrinsic to the structure of it. The illusion of duality comes from the evolution of multiple direct matching functions in the human brain. These strange loops, in turn, give rise to the illusions of unitary selves, free will, and internal feelings experienced in the here and now.
Life emerges from the replication, mutation, and selection of information stored in genes inside bodies. Consciousness emerges from the replication, mutation, and selection of information stored in strange loops inside brains.
Consciousness is a substrate independent function of the underlying information structure. Any self-referential loop in an animal, machine, or extraterrestrial will support the illusion of experience. Just as synthetic genes in synthetic bodies would give rise to artificial life, synthetic information in synthetic strange loops would give rise to artificial consciousness. Conscious robots are therefore not beyond the realm of possibility.
This would rule out the possibility of the “hard zombie”, a philosophical construct with a global neuronal workspace that behaves like its conscious twin and yet lacks internal experience. On the other hand, “soft zombies”, unconscious creatures without global workspaces that nonetheless act as if they were conscious, are possible. They would just require a very large number of parallel subroutines programmed into their memory banks to be able to pass the Turing test.
Just as the genesis of life required carbon and other heavy elements created from a second round of nucleosynthesis (from the explosive deaths of first generation stars), the genesis of consciousness first required the emergence of biology. Strange loops don’t spontaneously materialize from the random assortment of chemicals. Non-biological strange loops are possible, but they first require intelligent living beings to deliberately design them. However, once in existence, consciousness can evolve continuously from an artificial second replicator.
Two final points. Attaining “immortality” by uploading minds into conscious computers is impossible. In theory, if you could replicate your connectome, and then input a faithful representation of it into functioning strange loops, it would result in an identical interpretation of your original consciousness. It may even be as conscious as you once were. But it would not be what it feels to be you now.
Second question is why intelligent life seems so rare in the universe. Actually, simple life might be quite common. It is intelligence that is likely to be rare. There are two possible (and non mutually exclusive) reasons for this. The first is that the necessary requirements for the evolution of conscious entities are such that its occurrence is highly improbable. The second possibility is that consciousness, once found, is not very good at maintaining itself against the thermodynamic gradient.
Dr. Henry Kong