20th WCP: Why Isn't Consciousness Empirically Observable? Emotional Purposes As Basis For Self-Organization

That there is something empirically unobservable about phenomenal consciousness follows from a modified knowledge argument. Traditional versions (Jackson 1986; Robinson 1982; Noren 1979) hold that if experiencing were equivalent with physical brain states, then complete empirical knowledge of brain states should constitute knowledge of everything about my experiencing; but complete empirical knowledge of brain states would not constitute knowledge of everything about experiencing (those alone wouldn't reveal 'what it's like' to have that experience); therefore, experiencing is not equivalent with physical brain states. This argument can be criticized for unwarrantedly assuming that everything 'physical' is empirically observable (from an experimenter's standpoint). E.g., Jackson assumes that the 'what it's like' aspect isn't "expressible in physical language" (291), but the reason for granting this assumption is that 'what it's like' is inexpressible in terms of possible empirical observations. Without the assumption that everything 'physical' is empirically observable, we can conclude, not that consciousness is non-physical (since there might be 'physical' processes that are observationally inaccessible), but simply that consciousness isn't identical with anything empirically observable. Still, given the intimate connection between each conscious event (C) and a corresponding empirically observable physiological event (P), what P-C relationship could render C empirically unobservable? If identical, they should be equally observable. I.e., if P ® EO and C ® not-EO, then P¹C. But if C were non-physical, why such systematic P-C correlations? Can C be physically instantiated, yet distinguishable from anything empirically observable?

One possibility is that C is a relation among Ps which is not identical to the Ps because it is 'multiply realizable' (Putnam 1993; Horgan 1992). If C hadn't been realized by P1, it could have been realized by P2, yet still have been the same relation. C might be a 'self-organizing' pattern of activity which appropriates and replaces physical substrata needed for its maintenance or evolution in self-motivated directions. Thus the P-C relation wouldn't be an identity, epiphenomenalism, or dualism, but a biological relation between an organismic process and its actively appropriated material components, continually replaced and reproduced through the organism's self-maintaining organization (Monod 1971; Kaufman 1993).

But how could this account explain the empirical unobservability of the relation C? As Bickle (1992) notes, something as simple as the temperature of a gas is multiply realizable by various movements of particles, yet temperature is not empirically unobservable!

Consciousness, unlike many self-organizing processes, is empirically unobservable because the emotions motivating the direction of conscious attention, partly constitutive of 'what it's like' to experience anything, are executed, not undergone by organisms (Newton 1996). Organisms feel motivations by generating them. But all phenomenal consciousness must be emotionally motivated; thus the 'what it's like' aspect of a phenomenal experience is inseparable from the emotions that permeate it. E.g., visual cortex activation is unconscious of red unless the emotional midbrain, limbic system, and anterior cingulate motivatedly "look for" a red object (Posner and Rothbart 1992; Damasio 1994; Aurell 1989). Directly experiencing someone's subjective consciousness entails executing her emotional motivations, thus being the person whose organism produces them. An experimenter observing my brain events does indeed know what those events 'are like,' but only for her organism, as motivated by her self-maintaining processes. Understanding or empathizing with another's emotions requires motivation by our own emotional processes, perhaps similar, but not numerically identical with the other's.

As discussed in section 2, a self-organizing process is a particular type of multiply realizable one in which the internal structure of the organism is the reason for its tendency to appropriate and replace its own substratum as needed to preserve functional continuity. My suggestion is not that the motivations stemming from biological self-organization are a sufficient condition for consciousness to occur, but only a necessary condition. Even this much can resolve a good bit of the mystery as to how consciousness can be in principle empirically unobservable without thus becoming a metaphysical substance split off from the physical realm.

1. How Can an Empirically Unobservable Process Have Physical Substrata? If the above modification of the knowledge argument entails that phenomenal consciousness is not identical with anything empirically observable, it also entails that consciousness must be a process or aspect of a process that appropriates the empirically observable physiological substrata needed for its maintenance, growth, and reproduction. This further conclusion follows from the process of elimination; every other theory of the mind-body relation would imply untenable conclusions in light of the in-principle empirical unobservability of consciousness. I.e., psychophysical identity and epiphenomenalism imply that consciousness should be empirically observable, whereas dualism and interactionism of the Popper and Eccles (1977) variety (the claim that conscious and physical events causally interact, but that neither necessarily requires the other as its underpinnings) both imply that consciousness is a non-physical entity or event, requiring no physical substratum. Each of these implications can be shown to be untenable.

We have already seen that, if consciousness were identical with empirically observable events, consciousness itself would be empirically observable. But it isn't, even in principle. No amount of empirical knowledge of physiological events, by itself, would entail knowledge of 'what it's like' to experience the consciousness in question. The experimenter may infer what it might be like by comparing the empirically observable events to what her own experience would be like when in an analogous neurophysiological state, but that knowledge in turn would not result from empirical observations of her own brain, but rather from subjective introspection. Thus it cannot be inferred from any set of purely empirical observations alone that phenomenal consciousness would be 'like' anything at all (Chalmers 1995). Thus consciousness in principle cannot be identical with anything that is either directly or indirectly empirically observable.

Even granted that most empirical observations are at least partly indirect, consciousness is not even indirectly empirically observable in this sense. Scientists 'empirically observe' temperature only indirectly, by first observing a thermometer reading, and then inferring the temperature from this reading, along with theories that presuppose inductive and deductive inferences from previous observations. But the point is that temperature is the kind of phenomenon that can be inferred at least indirectly from some set of empirical observations alone whereas what a state of consciousness 'is like' cannot, even indirectly.

If the empirical unobservability of consciousness conflicts with psychophysical identity, it also rules out a purely epiphenomenalist approach. If consciousness were simply caused by its empirically observable physiological correlates (as opposed to being identical with them), as Jackendoff (1996) and Searle (1984) propose, then consciousness itself would have to be either physical or non-physical. If physical, it ought to be capable of causing other physical phenomena, just as any other physical phenomenon is. Epiphenomenalism must reject this possibility because its central thesis is that consciousness is caused by physical processes, rather than causing them. On the other hand, if the conscious events caused by physical processes were non-physical, epiphenomenalism would have to posit the existence of non-physical entities, thus becoming a metaphysical dualism. It would be very mysterious what the non-physical entities might consist of, and how physical entities could make causal contact with nonphysical ones.

Metaphysical dualism and interactionism are untenable for similar familiar reasons. Dualism is unable to account for the extensively documented regular correlations between conscious and physical events; and interactionism, because it requires that some physical events be caused by mental events, would require that the chain of physical events should not be causally sufficient without input from the non-physical conscious events; yet we do not observe the frequent violations of chemical and physical patterns entailed here.

That consciousness is not empirically observable does not entail that it is a 'non-physical' entity. Of course, 'physical' here is not meant narrowly. Consciousness is not something that can be kicked, like a stone, but neither is temperature, a sound wave or a photon. None of these are 'non-physical,' since they have relations to observable physical events which ultimately can be described in terms of the behavior of things that can be kicked.

If we eliminate psychophysical identity, epiphenomenalism, dualism, and interactionism, then the relation between consciousness and its physiological substratum can be neither a causal relation between separate entities, nor a reducibility of one entity to another. What remains is a relation in which, as James (1968) said, consciousness is not an entity but a function; moreover, unlike most functions it is in principle empirically unobservable. This is possible only if consciousness is related to empirically observable physiological events as a process relates to the elements of the physical substratum for that process. E.g., a transverse wave takes physical particles and discrete movements of these particles as its substratum when the wave passes through that particular material medium (a sound wave through a wooden door, for example), but the wave is not identical with the door, nor is it caused by the door. The wave could have been the same wave even if some other material medium (e.g., a different door, or a volume of air) had been in a position to serve as its substratum.

Consciousness, like a sound wave, is multiply realizable in this sense; but unlike things like sound waves, it is not empirically observable. Sound waves are completely describable in terms of the observable movements of their substrata, whereas not everything about consciousness is. This feature of consciousness cannot be accounted for sheerly by its multiple realizability. Consciousness, if it is to be empirically unobservable without being 'non-physical,' must therefore be a special type of multiply-realizable process. I shall argue that, to account for the way consciousness differs from other multiply realizable processes, we must grant that it is a 'self-organizing' process, or an aspect of such a self-organizing process. But this will require a somewhat precise understanding of what it means for a process to be 'self-organizing.'

A process-substratum relation is the type of relation that 'supervenience' theorists have sought (Kim 1992, 1993; Van Gulick 1992; Searle 1992). Objections to supervenience (Bickle 1992; Newton 1996) insist that, if consciousness is to be a 'property' that 'supervenes' on physical things, then it ought to be a physical property as amenable to physical description and observation as any other. But consciousness does not seem describable in terms commensurate with physical parameters, and thus with the idea that it supervenes on physical things.

The process-substratum model provides a way of conceiving of consciousness that is commensurate with the physical realm, but without identifying it with anything that is objectively observable. Just as the velocity of an electron might be accessible only to someone not simultaneously observing its position, the process of consciousness, though it takes empirically observable events as substrata, may not be accessible to someone not living in the location of the organism that works as substratum for that process. The reason, I shall argue, is that all consciousness is permeated by emotional feelings; we 'have' these feelings only by actively generating them out of the total motivational structure of our own organisms.

2. Emotions, Organismic Purposes, and Self-organizing Processes. We must now address two important questions: How do 'self-organizing' processes differ from phenomena like temperatures and sound waves, which are indeed multiply realizable but not self-organizing? And how can the fact that consciousness is self-organizing account for its empirical unobservability?

Self-organizing processes differ from other multiply realizable phenomena in this sense: Temperature is multiply realizable in that the same temperature could have been realized by an infinite number of different combinations of movements of particles in the substance realizing that temperature. Similarly, a sound wave could have been the same sound wave even if it had travelled through a different medium. How do self-organizing processes differ from these? A self-organizing process, as Monod (1971) and Kaufman (1993) suggest, is one whose internal structure is what makes it especially prone to replace its own substrata if needed to maintain that structure. Typical examples of such an internal structure can be found at the cellular level. One example is a structure in which "The enzyme which catalyzes the first reaction of a sequence...is inhibited by the final product of the sequence. The intercellular concentration of this metabolite therefore governs its own rate of synthesis" (Monod 1971: 64) Another example is where "The enzyme is activated by a product of degradation of the terminal metabolite" (64). In each case, the internal structure of the process guarantees its strong tendency to be maintained, even across multiply realizable replacements of its own substratum elements.

Monod's concept of self-organization does not entail a causal interactionism: The process does not cause the behavior of its substratum elements. The behavior of each substratum element has causal antecedents at the substratum level that are both necessary and sufficient, under the given circumstances, to bring about that behavior. But the self-organization of the organism in which this behavior occurs is partly constitutive of the given circumstances under which those antecedents are necessary and sufficient for those consequences. This makes it possible that, if the needed antecedent for a behavior had not been available, the self-organizing organism is structured so that it could have changed some of its other functions in order to allow some other antecedent to be used as the necessary and sufficient antecedent of that same behavior. A typical example is the reorganization of brain function in mild stroke recovery. Even though the specific behavior of each substratum element has antecedents at the substratum level which are necessary and sufficient to produce it under the given circumstances, the structure of the self-organizing process as a whole is such that those given circumstances will tend to be changed when that is what is needed to maintain the general contour of the functioning of the overall process as such.

In other multiply realizable processes, such as temperatures or sound waves, it is true that the same temperature or sound wave could have been realized by different substrata if the given circumstances had been different; but only a self-organizing process is structured in such a way that it can change those given circumstances in order to ensure the continuity of the overall process. If the causal antecedents of the movement of a gas molecule are not such as to cause the temperature in the chamber to be 98 degrees, the chamber has no built-in structural tendency to change some other aspect of itself to ensure the maintenance of the 98 degree temperature; self-organizing organisms do this all the time.

Monod, like Kaufman, notes the similarity of living processes to crystal formations in this respect. Forming crystals are like growing biological organisms in that a property of their very structure as such is that additional materials coming into their vicinity have a tendency to enter into patterns that reflect that initial structure. But, unlike living organisms, crystals are not fluid or flexible enough to change aspects of themselves if needed to maintain continuity in some other aspect. If a particular element's causal antecedents had not been necessary and sufficient to produce the pattern under the given circumstances, the crystal as a whole would not have changed the given circumstances. This is just what is done by biological organisms.

A specific motivation, say the desire to raise my hand, results ultimately from the organism's self-organizing tendency. This self-organizing structure is therefore present, and embodies a tendency for me to want to raise the hand, even before the desire becomes pronounced enough to be a conscious awareness. Thus the 'expectancy wave' accompanying the decision to raise the hand is measurable before I am aware of a desire to raise it (Libet 1983; Young 1988: 164ff). Expectancy waves (also manifested in Libet's 'readiness potential') indicate that motivational feelings arise out of the organism's generally self-organizing nature.

Monod's concept of self-organization is essentially the same as Merleau-Ponty's (1942) concept of a purposeful organism -- i.e., in which a change in one part can be compensated for by changes in other parts where needed to maintain the continuity of the whole. Merleau-Ponty thus defines his notion that a part tends to be subordinated to the 'purposes' of the whole.

If self-organization is the same as purposefulness, and if emotional motivations are purpose-directed activities, it follows that only self-organizing processes can be characterized by emotional motivations. (Not all self-organizing processes are characterized by emotional motivations, of course, but just the converse.) And if there can be no consciousness without some emotionally motivated direction of attention, then it follows that only self-organizing processes can be conscious. (Here again, I am not claiming that all self-organizing processes are conscious.) Moreover, we have seen that the empirical unobservability of consciousness stems from the fact that to experience a state of consciousness entails generating the emotional motivations which are a crucial part of the constitution of conscious states, and this in turn requires being the organism that generates those emotional motivations. Thus the reason why the phenomenal character of a state of consciousness cannot be inferred purely from observations made from an external or empirical standpoint is that emotions are motivated actions which an organism performs, and to experience them is to perform them. A scientist observing a subject's brain, in order to be conscious of the observation, must motivatedly perform the emotions in herself which direct her conscious attention; but by doing so she does not experience the subject's emotions, but rather her own. Therefore her empirical observations do not yield knowledge of what the subject's consciousness 'is like,' although they do yield knowledge of what her own consciousness is like. The complex self-organizing process constitutive of the emotional motivations needed for the subject's phenomenal consciousness are experientially accessible only from the standpoint of the organism that executes them.