From Towards a Science of Consciousness 3         Section III: Neural Correlates       CogNet Proceedings

Toward a Cognitive Neuroscience of Consciousness

Antti Revonsuo

Consciousness as a Biological Phenomenon


An essential element of any scientific understanding of consciousness is a systematic, natural-science approach to consciousness, which links consciousness studies with cognitive neuroscience and more generally with the biological sciences. Let us call such a branch of science "Cognitive Neuroscience of Consciousness" (CNC).

Although a substantial part of current empirical work on consciousness, especially on the neural correlates of consciousness, is carried out within a natural science framework, there seems to be no unifying theoretical view of consciousness guiding such work. This is a serious weakness for the scientific approach to consciousness, for unless consciousness studies can be theoretically firmly linked with mainstream biological and cognitive science, there is little hope that it will become the widely respected, progressive multidisciplinary field of inquiry that it strives to be.

Developing a theoretical basis for CNC requires close interaction between theoretical and empirical issues. What we first and foremost need is a philosophy of consciousness that could be taken seriously even by empirical neuroscientists. However, if one looks at what several respectable philosophers currently say about the nature of consciousness, one finds a lot of rather peculiar views that could hardly be taken seriously by the craftsmen of a natural-science approach to consciousness. Some, like Dennett (1991), suggest that subjective phenomenal consciousness doesn't really exist at all, others, like Chalmers (1996), propose that perhaps consciousness is everywhere-even electrons, stones, thermostats and other very simple systems might possess some sort of consciousness. Dretske (1995) and Tye (1995) deny that phenomenal consciousness could be explained by studying the brain-in their view, phenomenology is not inside the brain. Instead of taking any of those highly exotic philosophical views of consciousness as a starting point, I propose that an empirically based CNC should start with the following simple, clear, and not highly implausible assumption concerning the nature and place of consciousness: "Consciousness (phenomenal experience) is a real, natural, biological phenomenon that literally resides in the brain." That sounds like a reasonable working hypothesis for CNC. Although there are some philosophers who would probably agree with this kind of a view (Searle 1992, Flanagan 1992), it seems that the majority of recent philosophy on consciousness is inconsistent with these assumptions. Indeed, one is inclined to agree with Bruce Mangan's (1998) observation that most philosophy today works against viewing consciousness as a biological system.

My intention here is not to analyze in detail all the various views that deny that consciousness is a biological phenomenon in the brain; suffice it to say that they all have deep philosophical problems of their own, in addition to the fact that they could never serve as the theoretical basis of an empirically based CNC. Instead of wasting ammunition for nothing-philosophers rarely give up their views anyway-I believe it is far more fruitful to try to see how we could actually start developing a theoretical basis for empirical CNC. What is needed for CNC to progress is an active interaction between the theoretical and the empirical points of view when trying to understand consciousness as a biological phenomenon, instead of endless arguments with the defenders of all kinds of far-fetched antibiological views.

Consciousness as a Level of Organization


How should we depict consciousness as a biological phenomenon? The general framework used in understanding complex biological systems conceives of them as composed of several different levels of organization. Phenomena at different levels of organization in nature (e.g., molecules, cells, organs) are usually quite dissimilar to each other, have distinct causal powers, and require different sorts of approaches to be studied empirically. For example, a single cell is quite dissimilar from the molecules that it is composed of and from the tissues and organs that it may be a part of. Levels of organization are seen as really existing out there in nature. Our biological theories attempt to capture these levels by postulating abstract models and conceptual systems that describe the levels and explain the relations between the different levels. Our theories thus have levels of description and explanation thought to correspond to the levels of organization actually existing in nature.

If this general framework is taken seriously in consciousness research, then CNC should reconceptualize consciousness as the phenomenal level of organization in the brain. This sort of characterization makes it perfectly clear that we are dealing with a real biological phenomenon; an integral feature of the brain as a biological system. This view leaves no room for arguments that try to separate consciousness from the brain by insisting that we can imagine a complete neurobiological description of the brain that does not tell us anything about consciousness. A complete description of the brain as a biological system necessarily includes a description of the phenomenal level of organization in the brain. If we fail to understand subjective phenomenal consciousness we will have failed to exhaustively understand the brain as a biological system.

Explaining consciousness involves finding answers to the following questions: What is the phenomenal level of organization like? Where can it be found? How could it be revealed to the methods of empirical science? How is it brought about by the underlying lower levels of organization? The first question is the one we have to start with, for it is the basic question concerning the systematic description of the phenomenon we are interested in. Any empirically based scientific discipline must start with systematic description, and it is the indispensable foundation of all explanatory research in biology as well (Mayr 1996).

Here we encounter one of the principal problems in current research on consciousness: it seems to operate at two levels of description only: first, the level of the cognitive (information-processing) mechanisms and second, the level of the neural mechanisms or neural correlates of those cognitive mechanisms. The most important level of description in consciousness research, namely that of phenomenal organization, surprisingly enough has no central role in current theorizing. Without a systematic description of the phenomenal level, however, it does not make much sense to chart the cognitive or neural mechanisms of consciousness, for it remains quite unclear what all those detailed mechanisms are supposed to be mechanisms of. The point is that a science of consciousness must first treat the phenomenal level of organization as a proper level of description. The lower levels of explanatory mechanisms can be invoked only after we have a clear conception of the phenomenon that these mechanisms are supposed to explain.

It is not too difficult to see why this lack of phenomenological description prevails in consciousness research: there is no well-established empirically based framework to turn to. In the history of psychology, introspectionism once failed, and many still feel that an empirically based scientific phenomenology is outright impossible. Phenomenology as practiced in philosophical circles seems to be too obscure and conceptually isolated from current cognitive neuroscience to be of any real value for the empirically minded scientist. Furthermore, some philosophers are not too optimistic about the prospects of an "objective phenomenology" (Nagel 1974). Thus, it is no wonder that in consciousness research conceptual frameworks are primarily taken from the empirically respectable and well-established branches of cognitive science and neuroscience. The problem, however, is that the science of consciousness is not simply a trivial further branch of those fields: standard mainstream cognitive science and neuroscience largely ignore consciousness; they will not provide us with adequate levels of description to handle phenomenal experience.

I suggest that the science of consciousness needs to develop a phenomenal level of description that systematically captures the phenomenal level of organization in the brain. This level of description cannot be imported from any other existing branch of science-it must be contributed by the science of consciousness itself.

Metaphors and Model Systems


How should we go about developing a better understanding of the phenomenal level? I propose that we compare this task with previous attempts to describe and understand other complex biological phenomena. At a stage when no well-developed theory of a given phenomenon is available, the researchers tend to describe it in terms of a suitable metaphor. For example at the time when biologists had little idea what the internal structure and chemical composition of genes was, they nevertheless were able to construct very useful models of genes by treating them as distinct units that are arranged linearly just like "beads on a string." This was a simple, powerful metaphor that guided the thinking in classical genetics for a long time. To take another example, cell cleavage patterns were represented in what were called soap-bubble models, and this analogy even allowed researchers to suggest theoretical explanations about cells based on knowledge about soap bubbles (Maienschein 1991). Thus, a good metaphor captures some of the essential features of the phenomenon in a form that is easily comprehensible and visualizable.

Metaphors have been used in current consciousness research, but I am afraid that most of them have not been constructed so as to capture the phenomenal level of organization specifically. The lack of phenomenology is reflected in the Theater Metaphor of consciousness (Baars 1997), which may be a fitting metaphor of the cognitive mechanisms of consciousness, but does not seem to catch the level of phenomenology at all that well. To be conscious does not typically feel like sitting in a dark theater and looking at characters or events appearing on a faraway stage. Rather, my moment-to-moment consciousness feels like being immersed into the center of a multimodal world that is present for me all at once, though I may pay focal attention only to a certain feature of it. The phenomenal level includes the externalized sensory events, my own body-image, and those events that I feel going on inside my body or my mind (emotions, inner speech). Therefore, we need a different kind of metaphor and a different conceptual framework in order to capture the level of phenomenal representation, or consciousness itself (but not necessarily its neurocognitive mechanisms).

A model system is a system in which the phenomenon of interest manifests itself in a particularly clear form. In the ideal case the phenomenon is clearly isolated from others with which it might otherwise be confused, and it is accessible to easy observation or manipulation by the researchers. The model system may otherwise not be prominent in any way; just consider the significance of the lowly fruit fly Drosophila melanogaster for the development of genetics.

I believe that the dreaming brain is an excellent source of both a model system and a metaphor of the phenomenal level of organization. We know from empirical dream research that the phenomenal level of organization is fully realized in the dreaming brain. The visual appearance of dreams is practically identical with that of the waking world (Rechschaffen and Buchignani 1992). When we dream, we typically have the experience of being in the center of a spatially extended world of objects and people, with all kinds of events going on around us. We have a body-image much like the one we experience during waking, and we apparently can control its actions and sense our own movement through dreamed space. We know that during REM-sleep, with which vivid dreaming typically is associated, the brain suppresses the processing of sensory information, but at the same time it activates itself internally. The motor output mechanisms are inhibited and voluntary muscles are virtually paralyzed, although motor commands are actively produced in the brain.

The dreaming brain shows us that sensory input and motor output are not necessary for producing a fully realized phenomenal level of organization. The dreaming brain creates the phenomenal level in an isolated form and in that sense provides us with insights into the processes that are sufficient for producing the phenomenal level. At the same time the dreaming brain is an excellent reminder of the subjectivity of conscious states: there is no way we can directly "see" another person's dream world. However, we have a large body of empirical data on the phenomenological contents of dreaming, thanks to the development of systematic collection of dream reports and quantitative methods in dream content analysis (Domhoff 1996). Quantitative dream content analysis may be the sole example of a field in which systematic empirically based descriptive phenomenology is practiced today.

So the dreaming brain is a good model system for consciousness research because it isolates the phenomenal level from other systems that it might be confused with, it underscores the subjectivity of the phenomenon, invites questions as to the possibilities of directly observing or imaging the phenomenal level, and it has resulted in empirical research with a systematic methodology and body of quantitative data based on phenomenological description.

The dreaming brain furthermore provides us with a proper metaphor of the phenomenal level itself. When the phenomenal level is fully realized, as it is in the dreaming brain, it is characterized by the sense of presence in or immersion into a multimodal experiential reality. These terms were originally launched to describe experiences created with the help of a virtual reality (VR) system. But of course these terms do not describe the computers or the programs in a VR system, but the subjective experience that such systems at their best can create. In developing such vocabulary I think the VR-community has done a valuable service to consciousness research, because what they have come up with are terms that capture the realization of subjective phenomenal organization from the first person's point of view.

The Virtual Reality Metaphor of Consciousness


I have proposed (Revonsuo 1995 1997) that we should take the concept of Virtual Reality as a metaphor for consciousness. To briefly summarize the ideas behind the VR-metaphor, when the brain realizes the phenomenal level it is actually creating the experience that I am directly present in a world outside my brain although the experience itself is brought about by neural systems buried inside the brain. The brain is essentially creating an "Out-of-the-Brain-Experience": the sense of presence in and the full immersion into a seemingly real world outside the brain. This is immediately obvious when we consider dreaming: there we are, in the middle of a strange dream world, but it never occurs to us to conceptualize it in any other terms than as a world or a place where we find ourselves in. Almost all dream reports begin by specifying the place in which the subject has found himself in the dream. We never come to think about the dream world as showing us how our own brain looks from the inside, although we know quite certainly that that's where all the fun is really taking place.

The phenomenal level of organization realizes what I call "Virtual Presence." Dreaming involves it in two different senses: first, there is the illusion that the experiential events do not take place inside my brain, but somewhere in an externalized perceptual world. Second, dreaming involves the further illusion that I am not present in the environment where my physical body actually is located and sleeping. In this sense dreaming creates a completely imaginary presence, just like the technological variety of virtual reality does. Waking perception, however, only involves the first kind of illusion and thus creates a sort of telepresence for the brain: the sense of direct presence in the world currently surrounding the body and modulating sensory input. The brain and the phenomenal level together with it, actually constituting this experience, reside deep inside the skull, never actually directly in touch with external objects.

The phenomenal level of organization can thus be seen as the brain's natural virtual reality system, a level of organization the purpose of which it is to construct a real-time simulation of the organism and its place in the world. This simulation is modulated by sensory information during waking perception, but during dreaming it is realized off-line, by recombining materials from experiences stored in long-term memory. In everyday thinking we rarely realize that what we directly experience is merely a clever simulation or model of the world, provided by the brain, not the world itself. In order to be an effective simulation, we are supposed to take it as the real thing, and that is exactly what we do even during dreaming.

Why does the brain bother to create a detailed model of the world for us (or rather for itself)? Obviously in order to guide the organism through paths that enhance its chances of survival and succesful reproduction in the real world. An excellent example of the importance of the phenomenal level in the guidance of behavior is a sleep disorder called REM Sleep Behavior Disorder (RBD). Patients with RBD do not become paralyzed during REM sleep as they ought to: the mechanisms of motor inhibition fail. Consequently, the patients act out the behavior that they dream about. If they are chased in the dream, they jump out of their beds and start to run for their lives. If they are attacked in the dream, they defend themselves and may kick around or throw punches at invisible enemies.

A subject with RBD is phenomenologically immersed into a dream world, but behaviorally interacting with the real physical environment, often with unfortunate consequences. Obviously, one cannot get very far in the real world if one has an entirely erroneous model of the world in consciousness. Thus, these patients often suffer all kinds of physical injuries during their attempted dream enactments and may even get seriously injured. For example, a 73-year-old man with RBD, when dreaming, attempted to catch a running man. His wife reported that he jumped off the end of the bed and awoke on the floor, badly injured (Dyken et al. 1995). If an epidemic of RBD suddenly were to spread, all of us would behave in bizarre ways every night, guided by whatever the nocturnal contents of the phenomenal level of organization would happen to be.

The brain constructs the phenomenal level of organization because that level is needed in mediating voluntary behavior. The content of the phenomenal level is ultimately the world as it is for the conscious organism: the world that the individual attempts to interact with and to which it attempts to adapt tothrough voluntary behavior. This crucial role in mediating voluntary, adaptive behavior, although functioning all the time in waking perception, becomes dramatically revealed when a full-scale hallucinatory world, such as we have during dreaming, guides our behavior.

An important implication of the Virtual Reality Metaphor is that this metaphor defines a framework for an empirically based phenomenology. The scope of a science of phenomenology could be depicted as the systematical description of the normal structure and the pathological breakdown of the phenomenal level of organization in the brain. We need to develop conceptual frameworks to systematically describe the different ways in which the structure of the phenomenal level can be distorted or break down in consequence of brain injury or in altered states of consciousness. The binding problem can be seen as the question of how the organization at the phenomenal level can become integrated or disintegrated as a result of normal or abnormal functioning of underlying integrative neurocognitive mechanisms. Similarly, the bizarreness of dream images could be conceptualized as specific distortions of the contents at the phenomenal level.

I believe that the critical question for the future of consciousness research is: Can we describe consciousness systematically even on its own terms? If not, the prospects for being able to understand its relations to other levels of organization look dim at best. Without a description of the phenomenal level of organization, cognitive neuroscience explanations of consciousness seem futile. The Virtual Reality Metaphor is my suggestion as to where we could start looking for such a description.

Empirical Approaches in the Science of Consciousness


Discovering the neural correlates of conscious states is certainly one of the most important tasks for an empirical science of consciousness. When engaged in such work we should, however, be quite well aware of what it is that we are actually measuring, and whether it has anything to do with the phenomenal level of organization at all.

Many highly interesting studies have been published lately, for example, on the neural correlates of binocular rivalry and on the neural representation of perceptual unity as high-frequency synchronization of neural activity. When it comes to the science of consciousness one annoying point is that most of this data is from animal subjects, mostly cats and monkeys. The unfortunate fact is that we are able to have any real access only to human phenomenology, and therefore the science of consciousness should mostly (but not solely) be based on results from human studies. In the rest of the present chapter, I review our studies on the neural correlates of human visual awareness.

We tested two hypotheses that were originally presented by Crick and Koch (1990, 1995). First, in which visual cortical areas are the neural correlates of visual awareness located? Crick and Koch (1995) argue that primates are not directly aware of neural activation in V1, but that they may be aware of activity in other visual cortical areas. We measured cortical magnetic responses (122-channel MEG) to visual awareness of objects in an object detection task (Vanni, Revonsuo, Saarinen, and Hari 1996).

In the task, the subjects (N = 8) were shown pictures of coherent and meaningful objects as well as disorganized and meaningless nonobjects (one picture at a time), using brief stimulus durations (30, 46 and 106 milliseconds) and masked stimuli. The subject's task was to detect the objects and to report (by lifting a finger) when he saw a coherent and meaningful object. If nothing at all or nothing coherent and meaningful was perceived, the subject lifted another finger. Less than 50 percent of the objects were detected at the shortest stimulus duration (30ms), but performance was very close to 100 percent of correct detections at the longest stimulus duration (106ms).

Source modeling of the evoked neuromagnetic signals was used in order to reveal the location of active brain areas. Although several different areas were more strongly activated by objects than nonobjects, the activation of only one area (the right lateral occipital cortex, possibly human V4) became stronger with increasing stimulus duration and directly correlated with the proportion of correct object detections (i.e., with visual awareness of the objects). Visual objects are represented and processed at multiple levels in the brain, but not all these representations are directly reflected in visual awareness. The right lateral occipital cortex seems to be specifically involved in the emergence of visual awareness of coherent objects. This is in accordance with the Crick and Koch hypothesis that the direct neural correlates of visual awareness are located in the extrastriatal visual areas, not in V1.

Second, the hypothesis that high-frequency neural oscillations around 40-Hz are associated with the binding of visual percepts into coherent wholes (Crick and Koch 1990) was tested by measuring EEG in a task in which the subjects perceived the same stimulus (a random dot stereogram) in one condition as an incoherent collection of random dots and in another as a coherent three-dimensional Gestalt (Revonsuo, Wilenius-Emet, Kuusela and Lehto 1997). Continuous viewing of the same stimulus in the incoherent vs. coherent condition was not associated with significant differences in 40-Hz synchronization. Thus, although there is a radical phenomenological difference between these two stable views of the stimulus, no corresponding difference at the neurophysiological level in 40-Hz power was detected. Next, we tested the hypothesis that the construction of the coherent Gestalt in visual awareness is perhaps accompanied by transient 40-Hz synchronization. The subjects free-fused the random-dot stereogram and pushed a button as soon as they saw the three-dimensional Gestalt clearly. In a control condition, they fused a stimulus from which no unified percept emerged. Event-related 40-Hz synchronization was observed 500-300 milliseconds before visual awareness of the coherent percept was reported, at right posterior and occipital electrode sites. 40-Hz activity thus seems to participate in the construction of the unified percept, perhaps by rapidly binding spatially distributed neural populations together. Consequently, the hypotheses presented by Crick and Koch on the neural basis of visual awareness were supported by our studies with human subjects.

Conclusions


The further development of an empirically based science of consciousness will, I believe, depend on whether we are able to take consciousness seriously as a biological phenomenon in the brain, and start creating the appropriate theoretical and empirical basis for such a science. If this attempt fails, or worse yet, if we fail to give it a serious try, we might see consciousness studies slowly fading from mainstream cognitive neuroscience, in spite of it being precisely the branch of empirical science supposed to explain how the brain generates the mind and consciousness. As I have attempted to show in the present chapter, the cognitive neuroscience of consciousness surely deserves to be given a serious try right now, instead of our resorting to increasingly bizarre philosophical speculations on the nature of consciousness.

Acknowledgment

The writing of this chapter was supported by the Academy of Finland (project 36106).

References

Baars, B. J. 1997. In the Theater of Consciousness: The Workspace of the Mind. New York: Oxford University Press.

Chalmers, D. J. 1996. The Conscious Mind. Oxford: Oxford University Press.

Crick, F. and C. Koch. 1990. Towards a neurobiological theory of consciousness. In Seminars in the Neurosciences 2:273-304.

Crick, F. and C. Koch. 1995. Are we aware of neural activity in primary visual cortex? In Nature 375:121-123. Dennett, D. C. 1991. Consciousness Explained. Boston: Little, Brown.

Domhoff, G. W. 1996. Finding Meaning in Dreams. New York: Plenum.

Dretske, F. 1995. Naturalizing the Mind. Cambridge, Mass.: MIT Press.

Dyken, M. E., D. C. Lin-Dyken, P. Seaba, and TYamada. 1995. Violent sleep-related behavior leading to subdural hemorrhage. In Archives of Neurology 52: 318-321.

Flanagan, O. 1992. Consciousness Reconsidered. Cambridge, Mass.: MIT Press.

Mangan, B. 1998. Consciousness, biological systems, and the fallacy of functional exclusion. InConsciousness Research Abstracts Toward a Science of Consciousness, Tucson III. p. 52.

Maienschein, J. 1991. From presentation to representation in E. B. Wilson's The Cell. In Biology and Philosophy 6: 227-254.

Mayr, E. 1996. This is Biology. Cambridge, Mass.: Belknap / Harvard University Press.

Rechtschaffen, A. and C. Buchignani. 1992. The visual appearance of dreams. In: The Neuropsychology of Sleep and Dreaming. Ed. J. S. Antrobus and M. Bertini, 143-155. Hillsdale, N.J.: Lawrence Erlbaum.

Revonsuo, A. 1995. Consciousness, dreams, and virtual realities. In Philosophical Psychology 8: 35-58.

Revonsuo, A. 1997. How to take consciousness seriously in cognitive neuroscience. In Communication and Cognition 30: 185-206.

Revonsuo, A., M. Wilenius-Emet, J. Kuusela, and M. Lehto. 1997. The neural generation of a unified illusion in human vision. In NeuroReport 8 18: 3867-3870.

Searle, J. R. 1992. The Rediscovery of the Mind. Cambridge, Mass.: MIT Press.

Tye, M. 1995. Ten Problems of Consciousness. A Representational Theory of the Phenomenal Mind. Cambridge, Mass.: MIT Press.

Vanni, S., A. Revonsuo, J. Saarinen, and R. Hari. 1996. Visual awareness of objects correlates with activity of right occipital cortex. In NeuroReport 8: 183-186.