"All parts of the brain may well be involved in normal conscious processes but the indispensable substratum of consciousness lies outside the cerebral cortex, probably in the diencephalon" Penfield 1937.
"The brain stem-thalamocortical axis supports the state, but not the detailed contents of consciousness, which are produced by cortex" Baars et al 1998.
One of the most exciting discoveries of neuroscience is that nearly all of the brain performs functions that are not part of conscious experience. In everyday life we are usually unaware of breathing or heartbeats yet there are parts of the brain dedicated to these functions. When we pick up a pencil we have no experience of the fine control of individual muscles yet large areas of cortex and cerebellum implement this. Things do not appear as greyscale and then have the colour poured into them although this strange colour addition is done in the visual cortex. Most of the brain is non-conscious but how is the "ghost in the machine", the mind, created by and linked into the non-conscious brain?
Although most of the processes in the brain are non-conscious there can be little doubt that the output of sensory processes contribute to experience. For example, although we do not experience the process of adding colour to visual data in cortical area V4 we do experience coloured forms and although we have little inkling of the hugely complex creation of words in the temporal/frontal lobes we do experience verbal thoughts. Our experience is an integrated output of most of the brain processes that deal with sensation as well as dreams, thoughts and emotions. But how and where does this experience occur?
It will be seen below that there is a large body of data supporting the idea of a non-conscious cerebral cortex and virtually no data that supports cortical consciousness. This data should guide neuroscientists away from the cortex in their search for the neurophysiology of experience; that it has not already done so is a testament to the intense depth of the belief that consciousness is a process rather than a phenomenon. The contents of consciousness are mostly an output of the cortex but the state is probably in the thalamus.
The Role of the Cerebral Cortex - The Big Grey Dream Machine
The cerebral cortex consists of a set of specialised areas that process different aspects of sensation and motor control. There are about ten times as many nerve fibres going from the cortex to the thalamus as there are from the thalamus to the cortex (Destexhe 2000). This means that the cerebral cortex should be conceptualised as a set of processors that intervene between the senses and a final, thalamic destination.
Histologically the cerebral cortex is a layer of greyish neurons overlying a huge mass of white nerve fibres, the cerebral medulla. The cortex consists of six main layers. The upper layers receive input from the relays in the thalamus such as the lateral geniculate, from the thalamus in general and from other areas of cortex plus a few specialised inputs from other locations. The lower layers give rise to output fibres that largely connect with the thalamus and other areas of cortex although particular specialised processors in the cortex may also have direct connections elsewhere such as to motor nuclei.
The cerebral cortex has many functions and is divided up into numerous separate processors. The most important function of the cortex from the point of view of consciousness studies is that it creates models. These models are most easily experienced when there is a lack of sensory input such as in dreaming, day dreaming, lucid dreaming or experiencing imaginary speech (thinking). In ordinary waking life the modelling processes create a model of the world around us and within us based on sense data and associated data. This model consists of overlapping sounds, images, smells etc. and is a combination of perceptual fields from all the senses.
There is considerable evidence that the parts of the brain that deal with imagining (modelling) things are also the parts that deal with perception (ie: modelling the world). The overlap between imagination and normal perception is not complete because, as Tong(2003), in a review of visual consciousness, put it: "Internally generated experiences share some, but not all, of the phenomenal properties of actual perception". There is also considerable overlap between the areas used for imaginary speech (thought) and actual speech, areas dealing with the control of sensation and of the tongue etc. being used in actual speech but not in imagined speech (Fu et al 2002). Kreiman et al (2000) investigated the activity of single neurons in humans and also found that the brain activity evoked by visual imagination overlapped that which occurs upon direct stimulation by the same image.
Our conscious experience consists of the output of the cortical modelling processes. The cerebral cortex itself appears to be non-conscious. The evidence for the non-conscious nature of the cerebral cortex is reinforced by lesion studies that show that up to 60% of the cerebral cortex can be removed without abolishing consciousness. Either hemisphere can be removed or much of the front or back of the cerebral cortex can be cut off yet consciousness persists. The cerebral cortex is often assumed to be the "seat" of consciousness because this collection of organs is relatively large in humans but the truth seems to be that the cortex is a collection of processors that provide an input to experience. There is also a substantial amount of neurophysiological evidence that the cortex is non-conscious.
Libet et al (1967) found that there could be cerebral cortical activity in response to weak stimulation of the skin without any conscious awareness of the stimulus. This work provides a neurophysiological basis for subliminal (non-conscious) perception and also shows that large areas of the cerebral cortex can be active without conscious experience. The insensitivity of experience to cortical activity has been further confirmed by Libet et al (1979). They electrically stimulated the cerebral cortex of conscious patients and discovered that the stimulus must be continued for about 0.5 seconds for subjects to report a conscious experience of the stimulation. What is the cortex doing in the 0.5 seconds between the start of stimulation and the report of awareness of the stimulation? It is probably synchronising its various processors and creating a waking dream, a structured set of events that accounts for the activity.
The 'Attentional Blink' (Raymond et al 1992) is also consistent with the concept of the cerebral cortex being a device that creates models. In the 'Attentional Blink' the identification of an object impairs the identification of a second object that is presented within 0.5 seconds of the first. Raymond et al used a stream of letters (11 letters per second) and the identification of a first letter impaired the identification of a subsequent 'probe' letter in the stream. If the probe letter followed the first letter within about 180 msecs it could easily be identified, suggesting that chunks of about 180 msecs of data stream are modelled together. Christmann & Leuthold (2004) have theorised that the 'Attentional Blink' involves perceptual and central components of visual processing. This is supported by the fMRI studies of Marois et al (2004) who presented subjects with faces mounted on scenes of places. The scenes of places often went undetected by subjects but they activated regions of the medial temporal cortex involved in high-level scene representations, the parahippocampal place area (PPA). When the scenes of places were detected by the subjects there was activity in the frontal cortex and the PPA activity was increased. These experiments are consistent with the idea of a cerebral cortex that is a multiprocessor system that creates consistent models of the environment for presentation to some other part of the brain.
Bregman's (1990) auditory continuity illusion is another example of how sensory events are modelled. If a pure tone is followed by broadband noise and the noise followed by the same pure tone it seems as if the tone occurs throughout the period of noise. If the noise is not followed by the pure tone there is no sound of the tone during the period of noise. This effect is similar to the results found by Libet because a delay of several hundred milliseconds between sensory stimulation and conscious experience is needed to account for the apparent rewriting of history after the second tone appears.
The 0.5 second delay required for the cortex to model an event has implications for the role of conscious experience in the control of our lives. If experience is always 0.5 seconds behind the true present instant then how can we be said to control anything? The brain must be acting automatically whilst performing most tasks. The 0.5 second delay also seems to contradict our everyday experience. We certainly feel like we are aware of things in less than 0.5 seconds, for example, the direct stimulation of sense organs seems to be experienced much more rapidly than the delayed experience of cortical stimulation. In fact subjects report that they are conscious of stimuli, such as being touched or seeing flashing lights, within 0.1 to 0.2 seconds of the event. So how can subjects report events within 0.2 seconds even though it seems to take 0.5 seconds for the cortex to generate activity that can be experienced? The simplest explanation is that the reaction occurs automatically within 0.2 seconds and then the conscious experience of this reaction occurs 0.3 seconds later. This gives a total 0.5 seconds delay before conscious experience whilst allowing fast reactions.
Libet et al extended their experiments by stimulating a "relay nucleus" in the thalamus that intercepts signals from the senses before they reach the somatosensory cortex. It was found that when this nucleus was stimulated for 0.5 seconds the subjects reported that the stimulus occurred 0.2 seconds after it had begun. When the nucleus was stimulated for less than 0.5 seconds the subjects did not report any sensation. This supports the concept of a 0.5 second delay whilst the cortex puts a stimulus in context before it is experienced.
These experiments show that our experience is an output of cortical processing rather than the processing itself. If our conscious experience is non-cortical then this raises the possibility that the non-conscious cerebral cortex can perform actions without conscious control. Of course, the cortex does this all the time when we are indulging in skilled or routine behaviour. The ability of the non-conscious cortex is quite remarkable; for instance car drivers sometimes discover that they have driven for several miles without conscious experience of driving, even at the level of having no recollection of the route.
Although it might be accepted that much of our everyday behaviour is automatic is there any behaviour that is definitely initiated by conscious experience? This is probably a pointless question because consciousness is about observation, not action; however, despite this there have been several experiments that have attempted to determine the relationship between consciousness and action.
In 1964 Kornhuber and Deecke performed a series of experiments that measured the electrical activity from the scalp (EEG) during voluntary actions. They averaged many EEG's from subjects who were about to move a finger and discovered that there is an increase in scalp potential before the movement takes place. The increase in potential can start as long as 2 seconds or so before the movement and is known as the "readiness potential" (Bereitschaftspotential). The readiness potential is strange because it seems to contradict our conscious experience; we do not decide to move a hand and then wait 2 seconds before the hand moves. It seems that the non-conscious brain may be taking things into its own hands.
Libet et al (1983) extended the readiness potential experiments by asking subjects to observe a Wundt clock whilst flexing a finger. The Wundt clock had a spot of light that moved around a circle every 2.56 seconds and allowed the subjects to obtain timings that were related to their mental experiences. When the subjects flexed a finger it was found that the readiness potential occurred about 0.5 seconds before the finger moved and the subjects reported they were going to move the finger about 0.2 seconds before the movement. This suggested that a subject's cerebral cortex was preparing for the movement about 0.3 seconds before the subject was conscious of this. Libet's experiments have been reproduced elsewhere (see Keller & Heckhausen 1990). (It is important to note that the subjects in Libet's experiment were asked to wait until they felt the urge to move the finger.) These results are consistent with the idea of the cortex as a modelling system that constructs a consistent model of events to pass on to whatever mediates conscious experience.
More about Models
Our dreams are clearly models that form a 'dreamworld' but the idea that perception might be like a dream that is updated by sensation is not so obvious. Experience seems to be an active model of the world (virtual reality) based on sense data rather than a simple mapping of retinal and other sensory data. This is demonstrated by visual illusions such as the Ames Room, Spoke Illusion and Muller Lyer illusions shown below:
Notice how the circle is distorted without any distortion in the 'spokes', it is as if the circle has been treated as a separate object by the processes in the brain that rearranged it. In all of these illusions the brain has rearranged large areas of the visual field and has managed the input as a collection of 'objects' that are manipulated separately.
The creation of a model is also demonstrated by the illusion of movement experienced when we watch the cinema or television. This is due to the cortical modelling that is known as 'short-range apparent motion' rather than flicker fusion or persistence of vision. It is intriguing that, although it has been known for decades that the joining together of static images in our minds is due to modelling activity in the brain the myth that it is due to persistence of vision or flicker fusion is universal. As Anderson and Anderson (1993) noted:
"Indeed, in the past decade, psychoanalytic-Marxist film scholars have retained the model implied by persistence of vision: theirs is a passive viewer, a spectator who is "positioned," unwittingly "sutured" into the text, and victimized by excess ideology."
Our experience of the cinema is like a dream updated by sensation rather than sensation updated by interpretation. In fact the most compelling evidence for the modelling power of the brain is the existence of dreams; our dreams are often models of worlds that do not exist and involve little or no sensory input yet can involve effects as powerful as any television drama.
The way that mental models may be the basis of ordinary reasoning was outlined by Johnson-Laird (1980), based on earlier work by Kenneth Craik.
Studies of 'change blindness' and 'inattentional blindness', where subjects fail to spot outrageous changes in their environment, also demonstrate that we experience a model and suggest that the brain must analyse an object to incorporate it fully into the model (See for instance Rensink (2000)).
Blindsight studies illuminate the relationship between the cerebral cortex and our experience. When the visual cortex is removed subjects become almost totally blind. If the visual cortex on one side is removed subjects become relatively blind in the contra-lateral hemifield. One of the most revealing studies of blindsight is Marcel's 1998 paper: " Blindsight and shape perception: deficit of visual consciousness or visual function?".
It is useful when considering blindsight to contemplate for a while the appearance of the world with both eyes closed and then with one eye closed. When both eyes are closed our experience is of a darkish space radiating out from our heads, with one eye closed we tend to ignore the darkish areas that cannot be seen even though they are still present. Marcel notes that patients who have a right blind field still have an underlying visual field on the right side and that this can even contain conscious visual experience. This sounds a bit like the darkish space that we all experience if deprived of visual input on one side. As Marcel says: "A question that naturally arises is whether the loss is a 'total' loss of visual consciousness in the blind field. It is often assumed to be so, especially by those who discuss blindsight without carefully reading the literature or working with the subjects. One can immediately respond negatively to the question.."
The consciousness of the completion of Kanizsa figures in blindsight patients is particularly indicative of the preservation of the field even though the content was largely missing. A Kanizsa figure is shown below:
If we put Marcel's observations together with cortical anatomy and function it seems that the space of our experience is located outside of the cerebral cortex. The cortex generates much of visual and other content but it does not generate the space. A reasonable hypothesis is that the field of brain activity that is the space of our conscious experience is located in the sub-cortical brain. This space is loaded with the output of the cortex.
Neuroscience of Consciousness continued...