Exploring the neural correlates of consciousness presents three kinds of challenges. The first has to do with the difficult notion of a neural—or of a cerebral—realization of a function. Although everyone agrees that every mental function supervenes on the activity of some part of the brain, some indeterminacy may arise regarding the kind of supervenience basis that is to be looked for. A common view is that every mental function, even highly modular ones, operates only on the background of other mental functions or cerebral systems. Therefore, when one looks for neural correlates of any particular function, one has to differentiate those neural or brain properties which are identifiable as directly correlated to that function from those which are only indirectly correlated to it. In other words, there are two ways of understanding the neural correlation of a mental state to a cerebral state. In the stronger sense, such a correlation should be both necessary and sufficient. The causal entanglement of cerebral functions makes such a demand implausible. What can be looked for is at most a weaker kind of correlation: only against the background of a normally functioning brain can some particular neural states be associated with particular mental properties. In the case of interest here, the effort to locate the supervenience basis of consciousness in one definite, well-circumscribed center is at odds with the notion that conscious states are information-processing states using widely distributed data analyses and retrieval.
The second challenge consists in specifying what the term "consciousness" is supposed to refer to. As Block (1995) shows convincingly, several different concepts are often confused under this word. Phenomenal consciousness refers to experiential states, like seeing green or feeling a pain. Access consciousness has to do with the rational control of speech and behavior. Whereas phenomenal states have a nonconcep-
tual content, access consciousness involves inference, and thus conceptual content.
This second challenge is made still more difficult when one realizes that conscious states of both categories (phenomenal and inferential) might be related to a specific function which itself depends both on implicit (unconscious) and on explicit learning (i.e., attention). Attention cannot itself be identified with some conscious state, because if it helps to determine which kinds of contents will be made phenomenally salient and inferentially promiscuous, it cannot without circularity be itself triggered by phenomenally and inferentially salient features of incoming or stored data (see Proust 1998).
The third challenge consists in a well-known, venerable set of worries linked to the fact that after all, there is nothing we can tell about the causal connection between some neural activation—more generally, some physical structure— and a particular phenomenal quality. Shepard (1995 ) gives an interesting twist to the problem. Given that nonhuman animals cannot report on their internal states, how are we to establish in any principled way what kind of organism can enjoy conscious states? Neither would the ability to report subjective feelings and experiences, as displayed by humans, suffice to establish that consciousness emerges at the agent level. Does, then, the ability to have conscious experiences depend on complexity in the underlying system? Is this a holistic capacity? Is the property of being conscious a quality emerging, say, at the level of the neuron or of the neuron assembly?
However laden which conceptual difficulties, these problems cannot be articulated and solved without relying on scientific inquiry. Folk psychology is a dubious adviser in such a task, which consists partly in understanding the possibility of folk psychology. One way to circumvent suggestions from folk psychology is to concentrate on the data of mental pathology. This domain, poorly described and understood by ordinary psychology, offers interesting correlations between functional alterations and perturbations of consciousness.
We will concentrate here on the subjective impression associated with the notion that one is the actor of one's own acts. Taking the subjective lead, we will try to look for a functional basis and for a neural correlate for this "feeling in charge.'' Although it is a platitude to say that one did something on the basis of one's own intention to act, it is not all that clear how an action is actually planned and executed; how a subject becomes conscious of intending, acting, and having completed the action; and how far prior or current conscious states are necessary for the whole process to develop. Studying a pathological case, where a patient may either misattribute some action as being his own or, on the contrary, disavow being the actor of a particular action—claiming that a foreign influence had him move—should shed some light on the relationships between functional data, subjective reports of agency or loss thereof, and neural correlates. Schizophrenia offers such cases.
The clinical symptoms of schizophrenia include disturbances in visual experience, blocking phenomena, and specific problems in speech production, speech understanding, and thought control. The patients often have delusions of control (they see themselves as endowed with specific, more or less extensive powers over other people) or delusions of reference (other people in the street are perceived as having something particular to communicate to them). A significant proportion of patients experience hallucinations, in general (verbal-) auditory, haptic, or olfactory. They also experience a number of difficulties related to action, also called "executive" difficulties. These perturbations are manifest both in the subjective reports issued by the patients and in their results on executive tests such as the Tower of London and the Wisconsin Card Sorting Test (WCST). Being requested to tell how it all started, a patient gives the follow ing account: "I get shaky in the knees and my chest is like a mountain in front of me, and my body actions are different. The arms and legs are apart and away from me, and they go on their own. That's when I feel I am the other person and copy their movements, or else stop and stand like a statue. I have to stop to find out whether my hand is in my pocket or not. . . . Sometimes the legs walk on by themselves or sometimes I let my arms roll to see where they will land'' (Chapman 1966, 232).
From the mosaic of reported perturbations in the domain of action there emerges a sense of imposed actions, of being subject to a foreign will or, conversely, of exerting influence on others; a disposition to copy others without pre-established intention of doing so; and a relative difficulty in resisting impulses triggered by contextual cues.
If we want to contrast the levels of analysis of the awareness of agency in these disorders, it is important to insist that this way of summarizing the clinical evidence is shaped by covert functional hypotheses, which will be made explicit below. A patient normally couches his own experience in qualitative, holistic terms involving not only a change in his own mind but also a severe change both in the external world and in several of his bodily functions. Delusions can be seen as normal inferences from puzzling new experiences, against the background of an existing system of beliefs. Any systematic change in the quality of perception can receive a subjective explanation in epistemic and motivational terms with a considerable impact on the subject's apprehension of his world and of his role in it.
The description just given of the schizophrenic syndrome is compatible with a functional hypothesis according to which the disease would be in part constituted by a perturbed monitoring of action. By "action monitoring'' is understood a set of complementary operations such as instigating an action in the correct time and place, exerting a feedback control on the movement toward the goal or target event, or stopping the movement when the goal is reached or when crucial preconditions fail to be met.
Before we examine various ways of articulating this hypothesis, we need to define an action. According to neuroscientists as well as philosophers sensitive to the issue of animal psychology (Jeannerod 1995; Frankfurt 1988), an action should be defined not on the basis of its source (in terms of relevant beliefs and desires) but as a process that develops from an internal model toward a goal with an appropriately monitored execution. In other words, feedback is the central notion for an action: A movement has to be guided in an internally controlled way up to goal attainment in order to be called an action. In this view of action, the source of the action (i.e., the actual cause that triggers it) may be external as well as internal. The fact that an action can be carried out consciously or not, in an automatic or in a deliberate way, is also an extrinsic property of action that does not need to enter its definition. As a consequence, one can also, at least as a matter of conceptual distinction, separate control of an action from conscious access to control mechanisms. On the other hand, an essential ingredient is a comparator, through which a system can modify the current steps toward the goal as a function of the difference between observed and predicted output.
According to an influential theory (Frith 1992), many of the clinical data reported above, besides the openly executive symptoms, result from impairments in processes related to the monitoring of action. Within this general framework, Christopher Frith has an interesting strategy for explaining schizophrenic symptoms. The first step consists in broadening the executive hypothesis in such a way that it encompasses action proper as well as the mental activity related to action. The second step consists in extending the hypothesis still further, by considering the previous perturbations as special cases of an overarching metarepresentational capacity. Let us examine these two steps in turn.
Was this article helpful?