5 Arguments supporting an adaptive value of conscious processing
5.1 Evolutionary considerations
Given the continuity of evolution and the gradual increase in complexity of brains that reached a (perhaps preliminary) maximum in human beings, it appears likely that the ability to be aware of cognitive contents, of one’s own cognitive operations, and finally of oneself as an intentional agent, is not an all-or-none phenomenon but an ability that gradually emerged as we evolved. Paleoanthropological evidence supports this hypothesis by documenting correlations between increasing brain volume and increasingly refined artefacts that reflect gradual increases in cognitive abilities. Not much direct data are available on the evolution of the human brain, because our immediate ancestors are all extinct. Thus, we have to rely on evidence of comparative studies with brains of other species. In less evolved brains the paths from sensory to executive areas of the cerebral cortex are short. These relatively short sensory-motor loops are of course much more elaborate than simple reflex loops, because signals are processed extensively and transmission is made conditional on input from other systems, on past experience, and on context. As evolution proceeds and brains become more and more complex, one observes the addition of new cortical areas. A distinctive feature of this evolutionary process is the way in which these new areas are embedded in already-existing networks. These newly-added areas no longer communicate directly with the periphery, neither on the executive nor on the receptive side (see figure 3). Instead they receive their input exclusively from the phylogenetically older areas, and also distribute their computational results solely to other cortical areas and not to effector systems. This process is iterative, with more and more areas constantly added that communicate only with other higher areas. A neuron located in these more recent areas is connected exclusively with other partners in the cerebral cortex. Evidence indicates that all cortical areas, including older and more recent ones, operate according to the same basic principles, because they share the same intrinsic organization. These purely anatomical considerations suggest that the evolutionary-recent areas process the results of the older areas similarly to how they process signals from the outer world. This iteration of “cognitive” processes across several hierarchical levels could thus generate representations of representations, i.e., meta-representations. Information that has already been processed by the already existing areas becomes the object of yet another cortical computation, i.e., of a secondary cognitive operation. These iterative operations can even be circular, because all these areas are interconnected reciprocally. Thus, higher-order areas feed back to lower, order areas and can have their results reprocessed. In principle this recursive process should permit the generation of meta-representations of increasingly higher order. In other words, highly evolved brains can apply their cognitive functions not only to the outer world but also to processes that occur within the brain. Brain processes can therefore become the object of the brain’s own cognitive operations. This could be the basis of phenomenal awareness, the awareness of perceiving, to create a protocol for what one perceives, and in the case of human beings to create symbols for the perceived and for internal states and to communicate them to others. Animals probably share some of these abilities, because their brains are organized in a very similar way.
Curiously, the ability to be aware of the results of cognitive opertions provides no clue concerning the computational processes underlying these cognitive functions. We have no insight into the neuronal processes that bring about cognition. We are aware only of the results—just as we are aware of an action without being able to tell which neuronal processes in the motor centers of our brains caused this action. This fact is at the origin of most discrepancies between our intuitions and neurobiological evidence concerning the nature of agency, the experience of Free Will, and the ontological status of qualia and consciousness (see final paragraph).
These evolutionary considerations may provide some plausible explanation of the emergence of higher cognitive functions—including consciousness—but they do not suffice to counter the argument that the emergence of consciousness is an epiphenomenon without adaptive value. To address this problem, one must identify functions that can only be realized by conscious processing.
Figure 3: The evolution of complex brains is characterized by a massive increase of cortical areas. This renders responses to stimuli increasingly dependent on intracerebral processes and permits generation of meta-representations.
5.2 Functional considerations
Is the ability to be conscious of one’s own cognitive operations a fitness factor? And would it make any difference if brains lacked this ability? The philosopher of mind, David Chalmers, once stated: “[n]o, that wouldn’t make a difference. It’s just an epiphenomenon and if we wouldn’t have it we would do as well because the underlying brain processes would be the same and get us through life without us having to be aware of them.” I tend to disagree with this view for a number of reasons. The common denominator, however, is that there is something very special about the nature of conscious processes, and that this uniqueness does indeed constitute a fitness factor—in particular with respect to the ability to develop symbolic communication systems, a theory of mind, differentiated social systems, and, ultimately, culture.
As has been argued above, conscious processing allows for abstraction and symbolic representations due to its versatile binding of virtually all results of lower order cognitive operations in a unified representational space, capitalizing on the lingua franca (the homogeneous data format) of communication among cortical areas. This permits implementation of very effective—we call them rational—strategies for deliberation and decision-making that differ from and complement those of subconscious processes. The question then remains whether the results of these special processes affect brain functions and thereby affect behaviour.
It is difficult to see how the outcome of a conscious deliberation, that is, of an argument-based decision, could not affect future behaviour. A conscious decision leaves traces in declarative memory and so must the consequences of this decision. These traces are inscribed as modifications of the functional architecture of the respective networks. If a decision has averse or beneficial consequences, the experience of these consequences will also alter network properties, and the novel activation patterns generated by the modified networks will enter as a novel argument, or as a change of goal in future deliberations. Eventually, the newly-set goal, which initially had the status of a conscious rational argument, may change its status and become a habit that henceforth influences behaviour without having to appear as an explicit argument in consciousness. It can become one of the variables that act at a subconscious level. One then refuses another glass of wine not because one recapitulates all the rational arguments against alcohol consumption but simply because it does not feel right to drink too much.
Another, and probably the most important fitness factor of conscious processing is its capacity to support complex societies in cultural evolution. This suggests that there may have been a co-evolution of mechanisms supporting the emergence of conscious behaviour on the one hand and the formation of societies on the other, with the two developments mutually supporting each other.
Cognitive objects represented in consciousness are always bound together into a coherent experience. Whether this is also the case for subconsciously-processed contents is obviously difficult to ascertain, but the following argument suggests that subconscious processing may be less integrated, more modulary, and confined to subsystems. We subconsciously orient towards salient stimuli irrespective of their modality, visual, auditory, or tactile qualities, and these stimuli may be analyzed in the subconscious with respect to their behavioural relevance and thus give rise to action. However, if several stimuli compete for processing, usually the most salient will win. Compared to orienting responses, which are guided by conscious processing, there seems to be little evaluation of the embedding context here. Stimuli are processed more independently than they would have been had they entered consciousness, been bound together, and formed a unified coherent percept.
In order to achieve this integration and to relate the signals from various sensory systems to one another at the semantic level, the signals have to be encoded in a sufficiently abstract and homogenous format. As mentioned above, when describing brain evolution, the substrate for this integration of signals preprocessed by segregated sensory systems may be the evolutionary-recent cortical areas. By virtue of integrating and comparing signals from different modalities, it becomes possible to detect the similar in the seemingly different, and hence to extract invariant properties and to arrive at abstract descriptions. Thus the addition of the novel, so called “association areas” of the neocortex prepared the ground not only for a more unified, polymodal representation of cognitive contents but also for symbolic coding—which in turn is a prerequisite for the development of a symbolic language system and abstract reasoning. Thus one might consider consciousness, or the state of conscious processing, as a state where distributed computational results can be bound together into a coherent whole, establishing multiple, simultaneous relations between the various distributed items. This obviously allows for a more abstract, more symbolic, and more comprehensive description of conditions. By itself this is an advanced processing strategy, whose adaptive or fitness value is obvious. However, if this unified, condensed, and abstract information can be routed to a versatile communication system, as seems to be the case with contents that are processed consciously, the evolution of cooperating societies will be greatly facilitated. Not only will it be easier to communicate what one has perceived if the numerous signals from different sensory modalities have already been bound together into coherent wholes but, because of the reflexive nature of awareness, it will also be easier to convey information about one’s internal state. This, in turn, is an important prerequisite for society-building, because it nurtures trust in and predictability of the respective other. A condensed report of the actual contents of one’s conscious state and its storage in the episodic memory of the listener is an effective and parsimoneous way to couple brains with one another, to share experiences, and to foster cooperation. This interaction will modify the brains of the communicating partners and thereby act on their behaviour. In a sense this is an example of mental or top-down causation. The results of an information-processing strategy that can only be realized in the workspace of consciousness are stored in declarative memory—in this case not only in a single brain but in those of communicating partners—and henceforth influence future cognitive processes and behaviour. These considerations suggest that it might be useful and perhaps even necessary in consciousness research to consider the phenomena ascribed to consciousness not solely in the context of cognitive functions of individual brains but in the larger context of social interactions. In the following section this strategy is applied in an attempt to approach the “hard problem of consciousness”.