San Francisco CA 94115

cwt@ ski.org

Abstract

Having defined the concept of consciousness, its philosophical status as some transcendent activity of the human brain is evaluated relative to the emergent properties of complex physical processes. Consideration of the properties of complex systems via Godel’s theorem suggests that Darwinian evolution will always tend to engender such transcendent processes. A relativistic perspective of such processes provides insight into the Hard Problem. An experimental study of the attentional structure of human consciousness suggests that it is a single-channel process able to target a relevant processing mechanism in the visual field, and unable to take a broad overview of an array of such mechanisms.

An elaborated aspect of consciousness is the joint awareness of consciousness between two individuals, which may be evaluated by means of the artistic portrait. The salience of the eyes in portraits is a variable that seems to relate closely to interpersonal consciousness. Despite the variety of artistic compositions, artists almost always place one eye very close to the center line of the portrait frame and about 2/3rds of the height of the picture, providing hints as to the unitary structure of interpersonal consciousness.

Since "consciousness" means many things to many people, let me first define my use of the term (which is necessarily solipsistic because my consciousness is the only one that I know about. You will have to see whether you agree that these properties describe your own experience of consciousness). My viewpoint focuses on consciousness in the sense of the mental processes or their contents of which we are immediately aware at any moment in time. When I sense the environment, particular aspects of the environmental energy become the objects of consciousness in a never-ending flux of sensations. Memories of past sensations, or imagined sensations, may also be brought or bring themselves into consciousness. Beyond sensations, I am also aware of a limited set of internal manipulations of the relationships between such sensations, such as thinking that, if I dropped this egg, it would splatter all over the floor and my shoes. This kind of immediate awareness of sensations and events is experienced as a vivid "switched-on" activity that is peculiar to oneself; it is only by imputation that we assume that it is experienced by other humans and perhaps by other species to some phylogenetic level.

What I do not mean by consciousness is the sum total of all the things that have ever come to my awareness, or the cumulated store of my memories. I also do not mean the sense of myself in the world, which I would designate as "self-consciousness", except in the particular case when that sense comes to my immediate awareness. In other words, I define self-consciousness as that fleeting part of consciousness when I am aware of some aspect of my own thinking processes, not the underlying structure of self which forms and guides my responses. Finally, I do not mean the parts of my mental processes that seem to operate as subconscious workings of the thinking or emotional processes. My experience is that these subconscious workings often deliver thoughts, feelings or problem solutions to my consciousness without my being aware of the process by which the results were generated. I may have to engage in a conscious reconstructive process to understand the logic by which I reached a certain solution or emotional response. Such experiences convince me that much of the workings of the mind is, thus, unconscious. Most of my memories are also unconscious at any given time, lying dormant until called into active status in the arena of consciousness.

Another aspect of mental processes that is often confused with consciousness is understanding, or knowing. Understanding how something works or what someone means by their speech or their writing is a process that typically involves consciousness, often a heightened sense of consciousness. But this is not to say that understanding is consciousness. If I lie on my back in a field on a clear day, I experience a vivid sense of being surrounded by blue sky. I am highly conscious of the enveloping blueness but I do not feel that I understand it in any deep sense. I am simply aware of it. I may think about its origin in the scattering of light from particles of air, and its similarity to the Wedgwood blue of the china, or I may not. I do not have to think about the blueness of the sky to be conscious of it; I simply revel in its blueness as a pure sensation (or 'raw feel', in Bertrand Russell's term).

So consciousness may exist without understanding. Whether understanding may exist without consciousness is a thorny question, however. My view is that if a computer had a full range of responses to an object or structural entity, the ability to manipulate it on a variety of levels and the ability to predict the outcomes of such manipulations, it would be fair to say that the computer had an understanding of the situation. For example, when Big Blue (the computer) beat Gary Kasparov in their latest chess match, Kasparov was astounded at the deep level of ‘understanding’ of his strategy that the computer seemed to exhibit. I would say that it is fair to describe its strategems as showing an understanding of the game of chess, but the attribution of consciousness to this process of understanding is more problematic. The criteria for attribution of consciousness are not yet clear and the computer may well have had this level of understanding without being conscious in the sense that humans are.

Consciousness, in this immediate sense, as a transcendent state of the activity of the brain, because much brain activity is demonstrably unconscious (i.e., the activity proceeds during times when we characterize ourselves as unconscious). Some boundary state must be transcended for the activity to pass from being unconscious to becoming conscious. What is transcended for the activity to emerge into consciousness, we don't yet know, but there seems no reason to suppose that it transcends the activities of material entities such as neurons. Consciousness is often considered emergent in some deep sense, but the simple fact of being an activity or process imbues consciousness with evanescent qualities unlike the stable properties of material objects. It is important to draw this distinction between object properties, like liquidity or redness, and the dynamic quality of an activity such as the flying of a bird or the operation of a computer algorithm, which I regard as a better analogy to consciousness. Consciousness is too often characterized as a simple property or quality attached to thought processes whereas it is more probably a dynamic aspect inherent to the process.

The computer algorithm is a revealing example. The distinction is often drawn between the computer algorithm and the hardware that implements it, but there is a deeper distinction between the algorithm as a concept and the algorithm as a process. As a concept, the algorithm can be conceived and constructed in computer code, stored either as written symbols or as memory instructions in the computer. But when the computer is activated, the algorithm runs and produces its expected (or unexpected!) results. The conceptual algorithm has become a process, actually doing what it was designed to do as a dynamic operation rather than a static set of instructions.

In respect to its transcendent quality, consciousness seems more akin to cooperative processes such as fire, lightning, waves, tornadoes or nuclear reactions. In each case, a background level of activity intensifies into a self-organizing state that transcends the original activity state. For fire, the transcendent state is the colored region of the flame where the chemical recombination actually takes place. For lightning, it is the discharge of the accumulated electrical charge in a path connecting the cloud to the ground. For waves, it is the break-out of the wave into the foam of the surf. For nuclear reactions, it is the acceleration of the chain reaction into a nuclear explosion, when the increased energy transcends the bonds that hold the core material together.

The tornado is a particularly appealing example, where wind currents that can merely blow small particles across the earth's surface intensify into a transcendent structure, a self-organizing funnel that can suck large objects up into the clouds. This transformation characterizes the sense that I intend for "transcendence" in a physical process. It is the dynamics of the process that form the wind currents into the coherent entity of the tornado, which then develops the emergent power to levitate objects. The process is self-organizing, developing from the eddy currents of wind blowing past static objects in the environment. Winds blow most places on earth most of the time, but it is only under special conditions that the process transcends into an effective freight elevator, whose properties are effectively unpredictable from the preceding state. The fact that consciousness is so different from any other neural activity could have a genesis of the same general type.

Consciousness in humans, which requires a certain level of complexity, has obviously evolved from a state of non-consciousness in, say, single-celled organisms. Why did this happen? From a biological perspective, every contemporary organism has evolved to an equal degree to reach the current moment. Their evolutionary fitness is measured with respect to their adaptation to their ecological niche, and complexity is merely another strategy in the struggle for fitness within the random, non-directed environment that is the universe. Evolution has no special to develop complex, conscious organisms; it just arose as a chance adaptive strategy equivalent to any other. This cold evaluation flies in the face of the dominant human tendency to feel that there is a purpose in life and that human consciousness is the apotheosis of the evolutionary spiral. Can these two views be reconciled?

The key to a resolution is seeing that both individual organisms and their ecological nexus are complex self-organizing processes, which have the inherent property of state-optimization and stabilization. It follows that self-organizing processes will generally have, ipso facto, primacy over other processes that are not self-governing, or that have less effective self-regulation. They will tend to organize themselves into a stable state that is out of reach to a simple process. From this perspective, evolution toward greater complexity is an inevitable property of complex interactive systems. Although Darwinists generally dispute the existence of this hierarchy of complexity, it now seems evident from the properties of self-organizing systems.

Since a self-organizing system is a feedback system, this brings us into the territory of Godel's theorem that any self-referential system has states that are not linearly derivable from its axioms or reductionistic structure, which may be called transcendent states. Thus, any biological feedback system has transcendent states that could not have been predicted from its input and will, by the search processes of evolution, be evaluated for their fitness in a range of environments. It is therefore not hard to see how transcendent processes such as flight and consciousness will almost inevitably arise from the blind random processes of Darwinian evolution through progressively more complex systems. The key is that natural selection has an inherent vector towards the fittest. If the niches were simple, the result would be dull, but the world is complex to begin with and moreover the complexity grows because each species enriches it and forms part of the niches for other species, guaranteeing an increase in complexity. Once this ball is rolling, it must ascend the stream of entropy reduction like the homing of salmon smelt. The selection for flexible fitness will be likely to find any local path toward an advantageous state of greater complexity even though the flow of energy is ever to demolish it.

Even from the perspective of reductionistic Darwinism, therefore, we can derive the widely held beliefs that humans are the current pinnacle of creation, created by a God that is simply the incomprehensible synthesis of the laws of evolution of complex systems. The matter-energy ‘infundibulum’ of the universe is therefore neither casual, eternal or a formless collection. It is a self-organizing, developing ecology, with an inherent trajectory toward complexity. Perhaps this trajectory could be regarded as a fourth law of thermodynamics?

What do I mean by the "structure of consciousness"? Metaphors can be useful here. Is human consciousness more like lightning, where the entire cloud discharges through a single pathway, or more like fire, where many local elements can contribute to the flame simultaneously? In other words, is consciousness a unitary local process in the brain, scanning through a range of sensory inputs and memory contents by a rapid serial process, or is it a parallel distributed process, able to integrate the signals from many sources simultaneously?

This is not a new question: it has been addressed in the guise of studies of attention for decades. Now, attention is similar to but somehow distinct from consciousness. In my view, the difference between them is a relativistic one: in a traffic metaphor, attention describes the flow of mental traffic in relation to the highway architecture while consciousness is what it is like to be in the car driving through the traffic ahead. Thus, the reason that consciousness has an experiential quality unlike any other process that I observe is that I am the process of my brain, and being conscious is what it is like to be a process. The dramatic transformations that occur in the theory of relativity with a switch in reference frame from outside observer to the light-speed traveler give some insight into the power of a relativistic transformation to change the apparent properties of a process. Without an appreciation this power, no simple claim that consciousness is incomprehensible as a physical property can be evaluated. In other words, any philosophical evaluation of the identity between experiential consciousness and physiological brain states has to take account of the dramatic transformation inherent in the change of subjective viewpoint of the two processes. The fact, for example, that a brain state is localized in some piece of cortex while a thought may seem to have a universal, timeless validity does not imply a lack of identity.

A key question in the study of human consciousness is whether its structure is unitary or distributed, i.e., whether it has local serial or distributed parallel access to the contents of the working memory. In other words, are we conscious of one thing at a time, with attentional scanning over the contents of the stream of consciousness or are we able to spread our attention to monitor a broad range of neural signals simultaneously? This question is subject to direct experimental investigation by manipulating the size of the set of neural signals that need to be brought into consciousness to perform a detection task over the set. Basically, if consciousness is serial, performance is invariant with the size of the set, whereas if consciousness is parallel, performance improves with set size. (This logic is complementary to that for a search task with varying size of the set of distractors, where performance degrades for serial and is invariant for parallel processing.)

An established framework for such detection tasks is provided by the theory of signal detection in noise, which will be analyzed for that case of spatial vision where the detection task is specified for the neural set of visual signals across a uniform region of the retinal space. Signal detection theory provides contrasting predictions according to whether it is assumed that the subject acts a) as an ideal observer, with focal attention to an optimal receptive field that integrates the information linearly over the retinal area of each stimulus, or b) as a distributed attentional processor that can perform probability summation over the set of decision variables derived from the signals from the independent neural channels. The ideal observer prediction is that sensitivity grows with the square root of the number of signals. The theory of probability summation (which will be developed in exact form for Gaussian white noise here for the first time) has a very different prediction in which the increase in sensitivity for multiple neural channels asymptotes at a value no greater than four times that for a single neural channel.

These predictions were tested in human vision for the detection of grating targets varying in area over a range of nearly 1000:1. The analysis took into account a variety of confounding variables, such as nonlinear distortion and the effects of channel uncertainty. In the presence of a static full-field background grating, there was almost no increase in visual sensitivity over the value for detection of a single local spot (less than root 2 of the full range of size increase). This result implies that, under these conditions, attention must be focused at one point in the visual field, with no access to the parallel information across the retina that would allow the improvement predicted by theory of probability summation. It also implies that, in the presence of the static grating mask, the observer is unable to utilize any spatial summation mechanisms with or without selective attention among them. A control experiment without the static grating mask showed the expected conformity to ideal observer behavior up to an areal increase of about 20:1, implying the operation of a selective attentional process in this case, able to target the single most relevant detection mechanism for each particular stimulus block. .

One may interpret the structure of consciousness from these studies on the assumption that consciousness will employ any attentional mechanism that is available to optimize detection sensitivity. The implication of the lack of spatial summation in the presence of a static mask is therefore that there is no parallel processing by distributed attention across the visual field that would implement he probability summation process. Instead, consciousness is implied to be a unitary process that may be focused on one neural signal at a time. All previous reports of distributed probability summation would be interpreted as due to serial attention to mechanisms performing neural summation across their input signals.