| The following perspective is based on Chapter 11 & 12 of Higher Orders by Sylvain Labattu. The notion of “concept” is covered in Chapter 2. Abstract: This article first clarifies the natural of sentience and qualia as the post-interpretation representation of stimuli and the physicochemical phenomenon that makes the experience of such representation possible. It then explore the potential evolutionary purpose of sentience before running through a range of hypotheses on the nature of qualia, including the plausibility of it being due to a new force, a hidden dimension or an unknown property of matter, including the derivative hypothesis of panpsychism. Finally, it investigates the idea of it being an emergent property of neural networks before landing on the concept of distributed intra-neuron phenomenology, which goes around or doesn’t suffer from the issues of combination and information storage and integration. |
a) What are Sentience and Qualia
Sentience is what underpins the nature of our experience of the world. Without it, life would be starkly different for humans and any other animal species. There would be no such thing as what it’s like to be me, no sound, no colour, no sense of touch.
The Wikipedia entry reads as follows: “sentience is the simplest or most primitive form of cognition, consisting of a conscious awareness of stimuli without association or interpretation”; however this is inaccurate in that our experience always involves a representation based on stimuli and representation requires interpretation, it is one of the outputs of the perception process. Indeed, we do not perceive the world as it is, we interpret it and our conscious experience is unlike the raw data coming from our sensory apparatus, it is always mediated by the respective perception processes.
Qualia, on the other hand, is a phenomenon that is part of sentience, it is the essence of our sensations. When you experience hearing, what is the nature of that sound? It isn’t sound or pressure waves. When you see red, what is the nature of that colour? It isn’t photons of light with a particular wavelength. Phenomenology, that is the generation of qualia, is what allows us to experience the world in a way that doesn’t match the true nature of reality.
There is a conceptual stumbling block in explaining the existence of qualia, namely “who” or “what” is doing the experiencing? The answer that pops in our mind is “you” or “me”. That’s not very specific though, what is “you” or “me”? It is easy enough to rule out the existence of an entity that would be watching colours projected on a surface within our visual cortex because we would then need to ask the question of how that entity can itself experience colours. This line of approach leads us straight into the land of infinite regress and so, using reductio ad absurdum, we can confidently state that there is no soul or other entity that watches colours and hears sounds.
As a result, the question really comes down to how a neuron or network of them (or other types of cells in the brain) can behave or be stimulated in such a way that we experience sound, colours, pain, balance, odours, time, and taste.
b) Likely purpose
Before we explore the “how?”, it may be interesting to ponder the “why?” as this may provide some helpful insights to answer the first interrogation. In this case, the why translates into a couple of central, related questions: what is the function that sentience fulfils that is not available to a non-sentient being and what is the evolutionary fitness attached to it?
The first rationale that comes to mind is that sensations are the bedrock of reward functions based on pain and pleasure; without sentience there would be no such thing as a painful injury and no pleasure coming from eating certain types of food. It therefore bears the imprint of natural evolution that has enabled organisms, over thousands upon thousands of generations, to adopt behaviours that favour their survival and reproduction, and therefore the replication of their underlying genes. Except that natural selection is always indirect. What happens in fact is that some organisms will, through mutations, adopt a behaviour that enables them to experience pleasurable sensations several times and avoid those tagged as painful. If those stimuli and behaviours enhance the relative odd of reproduction or survival of those organisms, then statistically this mutation will tend to spread and eventually become dominant in the gene pool of that species. The passing on of information through a reward function is thus a probable answer to the question “why did it evolve?” understood in the sense of “why was it a successful mutation?”.
Even so, that can’t be the whole story because sensations are not a requirement for highly skilled behaviours and, more generally, reward functions can achieve their desired objective without being accompanied of sensations. Computers provide a straightforward example since a machine learning algorithm can be trained based on reward functions such as scoring a maximum number of points or winning versus losing a game. The degree of expertise that can be achieved with such a reinforcement learning technique suffers no contest as AlphaGo demonstrated in 2016 when it beat one of the best professional Go players in the world. Likewise, some robots can accurately move around a warehouse, so they do not need to experience vision or other form of sentient representation to navigate spatially.
Following those remarks, it the point of sentience is likely to make the information processed much more salient, in a way that cannot be achieved otherwise, and therefore to increase the strength of the signal to either convey urgency or to make it more memorable, which favours learning in the human brain. A reasonable supposition that is hard to prove.
The second factor behind the development of sentience is probably that it is cognitively useful. By that I propose that sensations such as vision or the interpretation of air waves as sounds provide either additional information or computational efficiency compared to a purely algorithmic (non-sentient) treatment of the raw data. This could be advantageous for a broad range of skills, including but not limited to pattern recognition, the manipulation of higher-level concepts, and by extension for the modelling of the world around us. In turn, this would lead to a superior situational understanding and an ability to make better decisions. As usual, natural selection would ensure that any such comparative advantage becomes prevalent within that species.
Ultimately, we can’t totally discount the possibility that sentience serves no purpose, that it is a by-product of the array of mental operations constantly taking place in our head, though this seems very unlikely when we consider how central that phenomenon is to our life.
c) Implausible and unlikely hypotheses
There exists many high-level hypotheses regarding the origin of our subjective experiences. Among them are very few or none that are crisp and detailed enough to the point that they can be tested during experiments with an observable outcome that could, in principle, falsify that hypothesis. In other words, we lack a solid scientific hypothesis and seem to be mostly wading through the metaphysical.
In that context, I will group tentative explanations under broad categories, covering the spectrum of the possible, not just the probable, and then assess the merits and flaws at the category level. When doing so, we need to be cognizant that we will be limited in our ability to prove the inexistence of that which does not exist. Therefore, we will need to rely on inductive inference through an evaluation of what is plausible and what is improbable.
In the process of turning every conceptual stone in our investigation of the origin and nature of sentience, I have gathered candidate hypotheses under five overarching ideas that we will discuss in turn: (i) new force, (ii) new property of matter and panpsychism, (iii) hidden dimension of our universe, (iv) an emergent property of neural networks, and (v) distributed, synchronous phenomena occurring within individual neurons. The first three are explored in this section and the last two have their own dedicated sections.
- New Force
In physics, there are four fundamental forces of nature. In no particular order, these are gravity, electromagnetism, the strong nuclear force and the weak interaction. Gravity is an interaction that causes objects with masses or energy to be attracted towards one another. Although its effects are readily apparent to the naked eye, it is by far the weakest of all four fundamental forces. Electromagnetism causes electrically charged particles to attract or repel each other depending on whether they have opposite or similar charges, respectively. It is the dominant force at the inter-atomic and inter-molecular level.
Contemplating the introduction of a new force is no small matter, for that reason the bar needs to be set high in terms of explanatory power. Such force would need to shed light on a phenomenon that other forces cannot for this hypothesis to be taken seriously. Unfortunately the prospects do not look very good as I see three key reasons a fifth type of force is a very unlikely candidate behind the phenomenon that makes qualia possible. To be sure, these arguments are purely in the context of qualia and do not say anything, one way or another, about the likelihood of there being other fundamental forces yet undiscovered.
The first reason is the most obvious one: scientists have tried very hard to detect a new force, without success. It should be noted that in past centuries, the ability to detect a new force might not have been such a strong argument given the inherent limitations of the equipment and techniques of the time. On the contrary, nowadays the lack of detection is an important consideration since we can peer down to the atomic level and use incredibly large particle accelerators to smash them at ridiculously high energy levels in the hope of detecting new forces or elementary particles. To be precise, this isn’t an argument against the reasonableness about the hypothesis of a new force but one about its probability.
The second reason has to do with scale. If the effects of such a mysterious force were only meaningful at either the nanoscale or over cosmic spaces, it is very hard to imagine how our human body would have evolved to master it. Indeed, the fact that we can close our eyes and visualize in our mind’s eye a particular person or object indicates that our organism has developed the ability to a) encode the information required to create that qualia, b) store it (memories being a subset of that), and c) read that information at a later time through neuronal activation.
It is unthinkable that our brain circuitry would have acquired such minute control over forces that only exhibit meaningful effects at the subatomic level, let alone at the interstellar one. The corollary is that the most plausible range of that new force would be somewhere between the dimension of atoms and macromolecules. This happens to be the domain of electromagnetic force, and it is impossible to ignore the effects of such a force; it has a lot of explanatory power. Therefore it is difficult not to be sceptical; one would expect such a new force to be more readily observable or to answer many other everyday questions besides what lies behind subjective experience.
The third and last reason in my list is that assuming a new force would not really explain how a localized phenomenon, at the neuron or even molecular level, could give rise to an integrated experience such as an entire visual field. Another way to phrase this is that we would run into a combination problem.
This naturally leads us to observe that in actual fact there is not one but at least two phenomena we are yet to explain: qualia at the local level (atom, molecule or cell) and the information integration that can give rise to an experience seemingly spanning much more than the local level. Using the analogy of a screen: we need to understand how pixels are created and how we can experience an entire picture in our mind in the absence of a lone spectator.
- Panpsychism or new property of matter
Panpsychism is a theory postulating that every particle or atom has a mind-like quality, that consciousness is present everywhere like a force or property of matter and that, when assembled into larger, more complex organisms, it gains in depth and can explain our subjective experience. Contrary to animism, panpsychism does not attribute human-like qualities to every particle or every object.
The main issue of the theory is that it doesn’t really elucidate anything and that its proponents have not put forward any way to test their claims. A conspicuous practical hurdle this particular view of consciousness and the nature of the mind is faced with is the gap to be bridged between microphenomenal and macrophenomenal experiences, the second type being the one we are subject to as a multi-cellular organism. This challenge is known as the “combination problem”: how would particles of matter with a modicum of consciousness link together to derive a richer, deeper, ability to experience qualia?
The combination problem also raises another related yet different question: why do we experience one combined “I” or self rather than the simultaneous consciousness of all our atoms? As a matter of fact, most events within our body do not trigger sensations (or at least we are not aware of them) such as the ongoing repair of our DNA or the creation of energy by the mitochondria.
There is no disputing that matter is likely to have properties we are yet to discover and shed light on. Yet, consciousness is unlikely to be one of them. What is possible on the other hand, is that some properties of matter, undiscovered or incompletely understood, could be involved in the phenomenology of quale. This is a distinct idea from panpsychism because it doesn’t purport to be at the origin of our sense of self, awareness or self-awareness. It only advances the reason we can hear sounds is because the atoms within our nervous system basically have what it takes to create the sensation of sound, that matter is in some way imbued with qualia or proto-qualia. We will dub this the proto-qualia hypothesis.
Unlike panpsychism, this hypothesis is not faced with the combination problem. The idea of trillions of conscious (or proto-conscious) atoms somehow giving rise to one single much more complex consciousness but not being perceptible in their own right, as single entities, is more than perplexing. In comparison, the idea of trillions of bits of qualia information generating one single experience doesn’t raise any conceptual obstacle because there is no claim being made that the aggregation of individual bits creates a separate entity. On the contrary, it could partially explain how we can have many different bits of information experienced at the same time, such as an entire visual field with qualia taking on so many different values like green, red, intense or dim.
The two main consequential questions that the proto-qualia hypothesis raises are that of information integration and transmission. For context, our brain has billions of neurons, those neurons are each made of billions of proteins, and those proteins are each comprised of 1,000 to 10,000 molecules. The information transmission problem is to figure out how can qualia at the basic level of matter be transmitted through the various levels of biology, from atom or molecule to intermediary structures such as proteins and then to neuronal cells. The simple answer is that it is a non-sensical question because it loses sight of the fact that even though a macromolecule exhibits additional properties to that of a single molecule, it does not exist as a separate entity of matter. As an analogy, a house made of bricks has emergent properties like providing shelter and isolation, but it is not a bigger brick. The way around this would then be to argue that higher echelons of biology have evolved the capacity to capture this qualia information down at the level of atoms or molecules, integrate it and then pass it on to the next hierarchical level, all the way to the cell. However, this renders the proto-qualia hypothesis unnecessary: if the information contained by qualia can be controlled and transmitted without loss by non-fundamental level of matters then why can’t it be manufactured by those higher orders of biology altogether? In that sense, the hypothesis would not have incremental explanatory power and doesn’t deserve much attention. Just to close the loop, could we conceive that there is no transmission involved, that we are experiencing what all our atoms experience at the same time? Obviously, that goes against what we know of perception, the matching to concepts, and the fact that we are able to generate our own experience when we sing a song in our head. All of this implies some degree of control over qualia.
In addition, the proto-qualia hypothesis doesn’t self-evidently explain the multiplicity of senses that exist in nature, those experienced by humans as well as those of other animal species. There can’t be an exact one-to-one mapping to undiscovered properties of matter for each sense so one new property would need to do the trick for all. This doesn’t look very credible and it seems infinitely more likely that evolution has led our organisms to develop different senses for convenience by somehow mastering the ability to encode and store the information required to create and experience a particular qualia.
Finally, plants (or fungi) do not seem to experience qualia. Of course, we cannot be certain until we truly understand the underlying nature of qualia but the dynamics of natural evolution are very powerful and we could have expected an arms’ race towards upward or lateral mobility to capture more sunlight, even the capacity to move away from a predator or physically attack a competitor when nutrients are scarce. It could well be that the plant kingdom has not developed the ability to make use of an inherent property of matter, however it is more plausible that qualia is an artefact of the branches of the animal kingdom that have developed a nervous system.
- Hidden dimension
There are strong arguments against the hypothesis of a hidden dimension as the source of qualia. To begin with, it would imply that we are unable to observe such additional dimension, yet as sentient beings we are directly or indirectly experiencing one of its properties. Hidden in plain sight, literally. But here we must be unbiased and acknowledge that intuition is giving us a hard tug. After all, we are limited by what our senses tell us and the ability to observe or detect exactly what leads to qualia is not a skill that would give one a strong edge for survival, hence we would have derived little in terms of evolutionary fitness by evolving this capability
Furthermore, the hypothesis would imply that all we are aware of, which is qualia, relates to a dimension we do not perceive and we are unconscious about all other reactions or phenomena within our body even though they do not call on that hidden dimension. How ironical. Having a dimension that our body can manipulate, that is only involved in sentience and that we cannot personally perceive or detect through technological means is a somewhat awkward premise.
The second argument against the hidden dimension hypothesis is its lack of explanatory power. Perhaps it could address the information transmission issue but that would not make us any more knowledgeable about the nature of qualia, how it operates, and how our nervous cells are able to store the information required to generate it.
The third argument against (I am sure we could list more) is that, except maybe for the purpose of coming up with a satisfactory theory of quantum physics, we do not seem to be missing any dimension to explain our everyday world. It is thus challenging to make the case that quale, which as far as we know are only experienced by animals with nervous systems, is a fundamental property of the world. Consequently, it is difficult to give much weight to a theoretically world-altering hypothesis without any obvious manifestation on the laws of physics.
d) Emergent property of neural networks
The term “emergent property” has become quite fashionable and is often used liberally as an explanation behind phenomena that we are hard press to properly understand. Consciousness falls into that category and is sometimes described as an emergent property of neural networks.
Property emergence is about the sum having greater functionalities than the parts taken individually; to quote the Wikipedia entry, it occurs “when a complex entity has properties or behaviours that its parts do not have on their own, and emerge only when they interact in a wider whole”. Examples in nature include ant colonies and, certainly the most amazing of all, life itself.
If sentience or quale are indeed emergent properties of neural networks, then we should focus our attention on understanding what the individual parts consist of and how they interact with each other.
These are questions that ideas such as the integrated information theory of consciousness (IIT) somehow leaves unanswered. One of the theory’s axioms is the unified nature of consciousness, and its central proposition is that it is the brain’s way to integrate decentralized information and it arises as a result of feedback loops. IIT is phrased in ways that suggests this as a valid explanation but, erudite language notwithstanding, it falls well short of providing any meaningful reverse-engineering of the integration that takes place, nor does it explain the generation of quale.
This should not discourage us from exploring whether quale or sentience should be considered emergent properties of neural network. Starting with quale; could phenomenology be the outcome of an activity involving several neurons concurrently? The question of the information storage, which is required to create concepts or memories and generate quale when actively calling on such concepts during the perception process, doesn’t seem too problematic as it could be stored across one or several neurons within the relevant network. Likewise, what we know of neurons firing patterns (electrical impulses and neurotransmitters) provides a ready solution to the question of information transmission. We are thus left with the issue of integration and, once again, we must be asking the question of who or what is doing the experiencing. The ant colony or the television screen are perceived as emergent properties by an outside observer but neither the worker ant nor the individual pixel are having such experience. And if some form of single internal observer is required then we are once more stuck with the issue of infinite regress.
e) Distributed intra-neuron phenomenology
Let us now assume that quale is not generated and experienced at the neural network level but one level down, at the neuron level to be precise.
Just as in the previous section, we need to look into the matters of information storage, transmission, and integration. The question of storage doesn’t seem to be a conceptual issue; even if we do not know today how this storage would actually take place, there is enough space and sufficient organic materials in the neuron’s soma (which contains the nucleus) for the job, including in the axon hillock where membrane potentials are propagated and summed. In fact, it would seem like a logical place for qualia to be experienced, synchronously with the transmission of the action potential to the axon provided the firing threshold has been crossed. If that was the case, then this would bypass the transmission issue for the simple reason that the experience of quale would take place at the same location where the information is stored.
This leaves us with the issue of integration, as was the case in the previous section on neural networks. Except that there is an implied assumption underlying the integration issue for qualia, which is that it must be integrated! Fair enough, what is the alternative then? Simply that quale is experienced at the neuron level and that sentience, our feelings and the subjective impression of what it’s like to be X, although perceived as a unified phenomenon is in reality a synchronous experiencing of distributed quale.
When we recall that we do not perceive the world as it really is and that we experience such thing as the self, then it is not a stretch at all to suggest that what we see as a unified visual field is the simultaneous experience of visual “pixels”, each with a specific qualia and coordinates (the coordinates information can be passed on as data through neuron firing). This hypothesis would also nicely account for the rhythmic patterns of neural networks firing known as brainwaves (or neural oscillations): the network aspect would be required to combine quale and the repetitive firing for the experience to persist in time, to maintain a specific image in our mind for example.
The fact that computer vision works should give us great comfort that there is no need for a unified image to be seen and that simply having all the relevant qualia-type of information (in this case colour) for each spatial coordinate available at the same time is sufficient for the machine to then process this raw information, recognize patterns and eventually understand what it sees.
The distributed hypothesis has another significant explanatory power in that it allows for a dissociation of the various parts of perception. The low-level primary concepts can be tied to a specific qualia and experienced synchronously as part of the representation of higher-level concepts. Hence, when we see a red line, we ultimately experience several dots and the colour red for each of those points in our visual field. It is also an efficient way to store information since the alternative would be having to store every possible combination of concepts, which would increase exponentially. Through a process of elimination, this hypothesis is the one I strongly favour. It doesn’t address the key issue of how qualia is generated and how a neuron can experience it (the answer to one will likely also provide an answer for the other), for that more experiments and discoveries need to happen in the field of biochemistry but it gives us a better clue as to where we should search.