Workshops

• Describe the principles of the four main brain imaging techniques: EEG, MEG, PET, fMRI
• Describe the theoretical and practical limitations of these techniques with particular reference to the environment in which experiments must be conducted and the measures that can be derived.
• Provide a framework for conducting brain imaging studies that distinguish neural correlates of consciousness from neural activity that is not related to consciousness and to provide examples of experimental designs that have been used for this purpose.

http://www.fil.ion.ucl.ac.uk/groups/Frith.main.html
http://www.klab.caltech.edu/~geraint/

Chris Frith - University College London
Geraint Rees - California Institute of Technology

Evidence for distinct neural substrates for two types of memory, declarative and nondeclarative, has been principally drawn from a large number of studies of patients with memory disorders due to neurological damage. Measuring the activity of the human brain can provide many clues about the neurophysiological mechanisms responsible for conscious and unconscious memory. In particular, the subjective experience of remembering appears to be closely associated with the ability to recall and recognize facts and events, often termed declarative or explicit memory.

This workshop will consider a rapidly growing body of investigations using ERPs and fMRI to explore brain mechanisms of multiple forms of memory and consciousness. Evidence that patterns of neural activity associated with conscious and nonconscious memory can be differentiated not only reinforces the neuropsychological dissociations but is also helpful for generating and testing new hypotheses about the operation of these types of memory. This work may thus lead to a better understanding of the enigmatic border between conscious and unconscious memory. The application of neuroimaging techniques to the study of memory systems should also serve to illustrate how neuroscientific study of the brain can provide insight into cognitive phenomena that are associated with conscious and nonconscious processing.

Paul J. Reber - Northwestern University
Ken A. Paller - Northwestern University

Hour One (Thomas Metzinger):

• Analysanda and explananda: Conceptual and empirical aspects.
• The four main levels of description: Conscious systems and their properties under phenomenological, representationalist, functionalist and neuroscientific descriptions. Conscious experience as a "level of organization".
• Six examples for fundamental constraints on the phenomenological level of description. Methodological issues concerning the epistemic asymmetry, ineffability and perspectivalness.
• Developing constraints on the representationalist level of description (current philosophical ideas about the "hegemony of representation"). The notion of "phenomenal content", and the strategy of developing a representationalist analysis of specific types of phenomenal content.

• Developing constraints on the functional and physiological levels of description: Potential explananda at the neurobiological level and goals of a theory of the NCC.
• Components of the NCC. Discussion of a tentative list of relevant neural processes. These include arousal, sensory segmentation, attentional, working memory, all of which seem required for sensory awareness as a basic form of consciousness, as well as processes like motivation, action planning, declarative memory and, in humans, symbol processing. It will be suggested that, from the viewpoint of contemporary empirical research, all items on this list must be envisaged as genuine explananda for future research, since for all these physiological processes current theories are, at best, incomplete.
• From correlation to explanation. Examples will be discussed for neurobiological evidence relating to basic mechanisms of conscious awareness. It will be suggested that for many of the processes listed above, binding mechanisms are required. Any consciousness theory must account for how activity in multiple neural modules can be integrated and how large-scale coherence can emerge within distributed neural systems. Furthermore, such a theory must specify mechanisms for the dynamic selection of subsets of neuronal responses, since only a fraction of all available information gains access to consciousness. We will propose that both requirements, cross-systems coherence and dynamic response selection, can be met by one and the same binding mechanism based on the synchronization of neuronal discharges. Recent evidence from both animal and human studies supports this hypothesis, demonstrating that neuronal synchrony covaries with perceptual integration, buildup of coherent representations, attentional selection, and awareness. These data suggest that synchronization, particularly if accompanied by fast oscillations in the gamma-frequency band, may be one of the necessary conditions for the emergence of conscious mental states.

• Discussion with participants: potential relevance of physiological mechanisms for a (1) 'access conscious' states, and (2) for constraining a representational theory of phenomenal consciousness.
• Discussion with participants: Directions for future research (e.g. role of animal models in NCC research, optimizing interdisciplinary cooperation, etc.). Necessary steps towards a systematic research program for consciousness.

Thomas Metzinger - Universität GH Essen
Andreas K. Engel - Max-Planck-Institüt für Hirnforschung

• Phenomenology: A phenomenological approach allows us to consider issues pertaining to body-image and to investigate the question of perceptual and non-perceptual awareness.
• Clinical Studies: We explore the contrasting characteristics of unilateral neglect and deafferentation. We detail one particular rare case (IW) involving loss of proprioception and touch from the neck down.
• PET and magnetic stimulation studies: The results of PET studies of IW involving self-movement and visual monitoring help to identify brain areas responsible for movements under visual control and those made without apparent feedback.
• Experimental studies: Results of experiments we conducted on IW's ability to gesture indicate to what extent gesture depends on proprioceptive feedback, and suggest precise conclusions concerning the relationship between motor and communicative aspects of gestural behavior.

Outline of workshop:

1. Phenomenological description of normal bodily experience. Awareness of the body during intentional action. Perceptual and non-perceptual awareness of the body. The status of proprioception. The distinction between body image and body schema. Questions and Discussion.


2. Empirical evidence for the distinction between body image and body schema. The case of unilateral neglect. Loss of body image. The case of deafferentation: IW. The effects of a loss of proprioception and touch from the neck down. Loss of body schema. Difficulties with motor control; the visual and cognitive monitoring of movement. Questions and Discussion.


3. Results of magnetic stimulation studies of IW. Visual and imaginative control as replacement for proprioceptive body schema. Task: small movement of thumb; superimposed magnetic stimulation. Evidence against perception of movement based on central corollary discharge. Questions and Discussion.


4. Results of PET scans of IW. The experiment involved a simple sequential finger/thumb apposition task. Four conditions of movement were investigated:
   
   (a) Self-movement and visual monitoring
   (b) Self-movement without visual monitoring
   (c) No self-movement and visual perception of other's movement
   (d) No self-movement and no visual perception of movement

   By comparison of the patterns of activation within the brain for these conditions in IW and controls conclusions may be made about the brain areas involved in movement without peripheral feedback. The areas activated during visual control of movement and during movement without visual or peripheral feedback may be analysed and from this the roles of parietal cortex and cerebellum in corollary discharge considered. Questions and Discussion.

5. Experiments on gesture. To what extent does gestural movement depend on conscious monitoring? What does the motor system contribute to language? Is it purely central or based on sensory feedback? Experiments with deaf unilateral neglect patients (Bellugi and Klima 1997) and with IW show to what extent and in what way embodiment is a necessary condition for linguistic behavior, and to what extent language transcends embodiment and shapes our thought.


6. Conclusions: The phenomenology of neglect and deafferentation as well as the experimental results allow us to draw conclusions about the need for body awareness during various motor tasks. Work from PET studies suggest the brain areas involved in movements under visual control and those made without apparent feedback. We discuss the implications of the empirical data for the question of the unity of consciousness, and outline a distinction between the phenomenal unity and the prenoetic unity of consciousness.

Jonathan Cole - Southampton University
Natalie Depraz - Université de Paris
Shaun Gallagher - Canisius College

There will be three sections, each of about an hour. The first will provide some general background in philosophy of mind. Building on that background, the second section will focus on specific questions about consciousness that have figured prominently in the recent philosophical literature. The third section will survey the major theoretical approaches to consciousness found in that literature and perhaps explore some connections these approaches have with empirical work about consciousness.

I. General Background in Philosophy of Mind

Recent work in the philosophy of mind clusters around four main questions. Two have a distinctively philosophical cast: the meta- physical problem of the relation of mind and body and the epistemo- logical problem of how we know about the minds of others. The other two problems are independent of such traditional philosophical concerns: the question of how to characterize the various types of mental phenomena (what I'll call the "characterization problem") and of how mental phenomena figure in psychological explanation (the "explanation problem").

All four problems are crucial for understanding current philosophical discussions of consciousness. (1) The mind-body problem has led to questions about whether conscious qualitative properties outstrip physical reality (Frank Jackson's "Knowledge Argument") and David Chalmers's "Hard Problem" about how conscious qualitative states can occur in physical systems. (2) The other-minds problem raises issues about the special, privileged access we seem to have to our own conscious states. (3) The characterization problem points to the question of whether the various kinds of mental state can occur without being conscious and, if so, in virtue of what conscious states differ from states that are not conscious. It also raises issues, highlighted by Ned Block and others, about whether there are distinct, independent properties we refer to under the heading of consciousness. (4) The explanation problem, finally, raises Joseph Levine's question about whether an explanatory gap separates physical processes from qualitative consciousness, as well as the question whether consciousness has any distinctive function.

II. Philosophical Questions about Consciousness

1. Are mental states all conscious? Or perhaps all qualitative states? Or are all mental states potentially conscious, as John Searle claims? When a mental state is conscious, is the property it has of being conscious intrinsic to the state? Or is it something about the way the state is related to other things?


2. Is there a property of being conscious (such as Block's "phenomenal" consciousness) that is unique to qualitative states, such as pains and perceptual sensations? And can we characterize the qualitative properties of sensory states independently of whether the states are conscious? Some discussion of the alleged problems about inverted and absent qualia may figure here.


3. Is what consciousness tells us about our mental states always correct? Is it complete? If not, how can we distinguish between our actual mental states and those consciousness represents us as having? Can introspective access be mistaken? Does introspection report our mental states or interpret our mental lives?


4. And on the topic of ASSC4: How are the contents of consciousness unified? Can mental states be conscious without being thus unified with other conscious states? What is it for a creature to be conscious, as against a mental state's being conscious? Does a creature's being conscious mean that its mental states are conscious? Does it mean that its conscious states are unified? Is the unity of consciousness more than mere appearance?

Recent discussions of consciousness in the philosophical literature fall into two main groups: those which represent consciousness as an intrinsic property of mental states and those which represent it as a relation conscious states bear to something else.

Intrinsicalist views describe consciousness by way of a relatively small circle of interdefined notions, such as subjectivity, perspective, point of view, and the like. So these views do not lend themselves to giving an informative, scientific account of consciousness with empirically testable consequences. Thomas Nagel's view, on which the subjective, perspectival character of conscious mental phenomena precludes any objective account of their nature, is usefully seen as an example of this kind of approach. (This is different from Colin McGinn's view that limitations on our mental nature itself preclude our developing or even understanding any informative account of consciousness.)

Relational models, by contrast, readily lend themselves to empirically testable predictions and explanations. The standard example of such a model is the "Inner Sense" view, on which a mental state's being conscious consists in one's sensing it. A second relational model discussed in the recent philosophical literature is Daniel Dennett's Multiple Drafts Model, on which consciousness is the continually revised result of many interacting subpersonal processes. A third version is the higher-order-thought hypothesis, on which a state's being conscious consists in one's having a non-conscious, noninferential thought that one is in that state.

These three models will be discussed, possibly in connection with such topics of current empirical research as metacognition, confabulatory introspective reports, blindsight and other dissociative disorders, confabulatory introspective reports, and the causal efficacy of conscious, voluntary decision. (Because our concern will be with how to characterize consciousness, we will not address questions about the neural correlates of consciousness.)

David Rosenthal - City University of New York

Antti Revonsuo - University of Turku

• What is the subjective experience of temporal unity?
• How does classical cognitivism address (or not) these experiences?
• Does connectionism, in the form of back-propagation networks, offer a more successful program?
• Does recent connectionist work in recurrent networks (Elman 1990, 1996) add significantly to the explanatory resources available to understand temporal awareness?
• Finally, in this progression of approaches are there new ways to interpret functional brain imaging as a component of consciousness studies?

• Models of working memory in recurrent neural networks, based on EEG "readiness" tasks in humans, and delayed matching tasks in humans and monkeys. The models are readily probed by methods analogous to population recording (EEG), "region of interest" recording (functional brain imaging), and single unit recording. But all of these methods are overly selective and thus distort the underlying neurodynamics. Multivariate analysis (cluster analysis, multidimensional scaling, and principle components analysis) affords a more comprehensive view, revealing the spontaneous emergence of rhythms and "cell assemblies," the internal representation of prospective and retrospective duration, and temporal categories.
• Models of dissociation following trauma. Trauma can be understood in neural network terms as single-trial learning with an abnormally high learning rate. This simple idea, implemented in a recurrent net, exhibits analogues of all the DSM symptoms of post-traumatic stress disorder, and even parallels exactly an early case history of Freud.
• Meta-Analyses of PET and fMRI data. The same techniques for multivariate analysis used with the artificial neural networks can be applied to the brain. Treating the brain as a distributed processor reveals underlying organization consistent with that found in the simulations, and analogous in many ways to phenomenological observations.

Workshop participants will receive complete specifications of the models discussed, and access to a public domain modeling software environment that readily permits construction of recurrent network models, and includes a built-in utility for cluster analysis and principle components analysis. Since I will have the models with me in a laptop, there will be time during and after the workshop to conduct new experiments as well.

Dan Lloyd - Trinity College

We will review the conclusions that can be drawn from change blindness and from related phenomena. Specifically, we consider what these phenomena can tell us about the structure and accessibility of our representations. In so doing, we will also examine recent evidence that change blindness may simply indicate our lack of awareness of the representations we do have. In other words, non-conscious, implicit representations of the details of scenes may at least partially underlie our experience of a stable, continuous visual world. We will also discuss how the techniques used to study change detection may help to determine the role of attention and effort in the formation of both conscious and non-conscious representations. At the end of the workshop, we will consider how the mechanisms underlying change blindness and change detection work together to provide a continuous, stable impression of our visual world.

The workshop will be segmented roughly as follows:

Hour 1: Intentional detection of change
Hour 2: Incidental detection of change
Hour 3: Representations, inattention, and non-awareness

Ron Rensink - Cambridge Basic Research
Dan Simons - Harvard University

1. Learning during general anaesthesia
Many studies of learning during general anaesthesia have shown implicit memory, on recovery, for stimuli presented during surgery. However, many studies have also failed to show learning. As many studies were opportunistic, there has been little control over factors such as the type of anaesthetic, type of surgery, or type of memory stimuli. Can patients really learn information while unconscious? What would be the clinical implications if they could? One possibility is that patients only formed implicit memories of intra-operative stimuli if they were inadequately anaesthetised.

2. Measuring depth of anaesthesia
Giving someone more anaesthetic makes them more unconscious, but the exact depth of anaesthesia attained depends also on the opposing effects of surgical stimulation. Depth of anaesthesia is therefore difficult to predict and, because many patients are given muscle-relaxants during surgery, it is also difficult to measure. I will describe some recent research into EEG measures of depth of anaesthesia that may provide useful tools for consciousness research and clinical practice.

3. The effects of impaired consciousness on cognition
Perhaps the most promising aspect of anaesthetics as a research tool is their ability to alter cognitive function in a gradual, dose-dependent fashion. Studies of volunteers receiving small, sub-clinical doses of anaesthetic may help reveal the "cognitive correlates" of consciousness. Contrary to other manipulations of consciousness, such as divided attention or subliminal presentation of stimuli, anaesthetics seem to affect implicit memory as much as they do explicit memory. What is it about being conscious which is so important for learning?

Jackie Andrade - University of Sheffield
Gareth Jones - Cambridge University

Hour 1. Current operational definitions and waking consciousness

Introduction. An historic PET scan obtained by Paulesu and colleagues (Paulesu, Frith & Frackowiak, 1993, shown in Baddeley, 1993) shows regions of high metabolic activity in the left hemisphere speech areas of a subject engaged in silent mental rehearsal. But it is not obvious from the PET scan which "hot spots" involve conscious events. This article helps to exemplify the kinds of questions and answers one can raise about the brain basis of conscious experience.

Current operational definitions. Accurate reportability of a wide range of stimuli is widely taken to be the most natural observable index of conscious sensory experiences. A remarkable study by Cowey & Stoerig (1995) made use of a commentary response to test whether macaques with cortical blindness lose the phenomenal visual qualia like color and motion, which humans report losing with similar brain damage. This approach may provide an equivalent to the reportability criterion in humans.

Properties of the conscious waking state. Waking consciousness has a distinctive anatomical, neuroelectrical, and neurochemical signature, separate from unconscious or low-conscious states. Consciousness during REM sleep resembles waking consciousness in many respects. Slow-wave sleep shares similarities with some comas, high-dose general anesthesia, and epileptic states of absence.

Much evidence supports the idea that conscious contents such as perceptual experiences or inner speech, require waking consciousness. This commonsense idea has been challenged by the finding that people woken up from slow-wave sleep may report mental activity, though different from typical REM dream reports (Foulkes, 1996; Cavallero et al, 1990).

Hour Two: Experiments contrasting conscious and unconscious brain events.

Studies showing differences between conscious and unconscious contents in the awake brain. These studies treat consciousness of a given content as an experimental variable. For example, binocular rivalry allows similar content to be presented to two eyes, though only one eye will be conscious. The brain responses to the conscious and unconscious input can therefore be compared. As Crick (1996) has written, "Binocular rivalry is a phenomenon of importance in its own right; but its real importance is that it may shed light on the baffling problem of visual awareness, a visual form of consciousness."

Local causal evocation of specific conscious contents. Local stimulation of isolated retina does not lead to conscious experiences. But local stimulation of visual cortex does lead to conscious phosphenes, which have recently been used as a prosthesis for the blind. Very local stimulation of visual neurons in area MT have also been shown to change the perception of visual events in the macaque (Newsome, 1993). Conversely, interference with normal cortical processes also blocks ordinary conscious visual contents.

Studies showing explicit neuronal correspondence between brain events and reported experiences. Crick and Koch (1995) have argued that area V1 does not explicitly represent visually conscious information, because its neurons do not respond to visual features that are lost from consciousness when V1 is lesioned. This suggests a more general criterion. Studies by Logothetis and coworkers illustrate this kind of evidence.

Hour Three: A set of criteria for brain proposals.

The five kinds of studies cited above can provide criteria to be met by any adequate theory of the brain basis of consciousness. Two more can be added.

Studies showing plausible relations to other known features of conscious processes. Any candidate brain basis for consciousness should be consistent with other well-established features, like limited moment-to-moment conscious capacity and a wide range of potential conscious contents. (Baars, 1988, 1997, 1999; Searle, 1994).

Discriminative validity.The brain basis for consciousness must be distinct from related constructs like selective attention, working memory, sensory knowledge and executive control.

Bernard Baars - The Wright Institute
Giulio Tononi - The Neurosciences Institute, San Diego
John Bickle - East Carolina University

A description of the cortical process of self-awareness is far from complete, though networks for cognitive processes that are related (e.g., Theory of Mind) are beginning to be elucidated. The integration of rTMS with traditional neuroimaging methods addressing the cortical correlates of self-awareness provides a background in which researchers interested in consciousness may develop an understanding of rTMS. This workshop will therefore provide an introduction into the methodologies of rTMS within the context of examining higher-order cognitive processes with a specific emphasis on self-directed awareness.

The topics as follows:

1. Introduction to rTMS
   • Basic Principles of Magnetic Stimulation
   • Early Historical Attempts: 19th and 20th Century
   • Successful Single-Pulse TMS and the advent of rTMS
   
   
2. Four Relevant Methodologies
   • Virtual Lesions
   • Cooling and Heating
   • Measures of Cortical Excitability
   • Paired Pulse Measures of Inhibition and Excitation
   
   
3. Integration of Functional Techniques
   • What Can Go Wrong
   • Direct Measures of Activity within the fMRI and PET environment
   • Stereotaxic Methods
   
   
4. Applications of rTMS in Higher-Order Cognition Not Addressed Previously
   • Language & Vision
   • Imagery & Memory
   • Self & Consciousness

The participants of the workshop should expect to leave the workshop with an firm and thorough introduction as to the methodologies of rTMS. Further, the participant will be in a position to implement his or her own questions of consciousness within the rTMS environment. Therefore the goals of the workshop are to a) Provide an introduction to those unfamiliar with rTMS, b) Demonstrate that rTMS can be a powerful tool in tests of hypotheses related to higher-order consciousness, c) Provide information and practical information regarding the integration of rTMS with fMRI and PET, d) Evaluate the shortcomings of the technique, and e) Provide enough information to the participant so that he or she can independently evaluate the validity of rTMS for use with his or her own research. This workshop is intended for persons with an interest in neuroimaging, and no previous experience with functional techniques is assumed.

Julian Paul Keenan - Harvard Medical School

1. What is consciousness?
2. What are the causal relationships between consciousness and the brain?
3. What is the function of consciousness?
4. What are the neural substrates of consciousness?

Similarly, it seems obvious that consciousness has a function. Indeed, according to evolutionary theory consciousness must have a function otherwise it would not have evolved to be so central in our lives. There have been many proposals in the scientific literature about what that function might be. Common suggestions are that consciousness is necessary to deal with novelty or complexity, to provide feedback, to enable memory and learning, to enable language and problem solving, to enable imaginal short and long-term planning in advance of carrying out acts in the real world, and so on. However these proposals face a central dilemma: Once one can specify how such functions work in information processing terms, one no longer seems to need consciousness to explain the working of the system which embodies that processing. One can envisage the same processes operating in mechanical or electrical systems unaccompanied by any subjective conscious experiences. So - what, if anything, does subjective experience add to effective functioning? Questions 1 to 4 also interconnect. If one is not clear about what consciousness is, how can one find its neural substrates in the brain? Nor can questions about causal efficacy be dissociated from questions about function. If consciousness has no causal influence on neuronal activity, it is not easy to see what its function in the brain's activity could be.

This workshop summarises the strengths and weaknesses of current attempts to deal with these problems within dualism, physicalism, and functionalism. Given the overall focus of the conference on information integration, we will give particular attention to the proposal that consciousness carries out the functions of a "global workspace" concerned with information integration and dissemination (a form of psychofunctionalism). The workshop will also introduce reflexive monism, an alternative approach to such problems, developed in depth in Understanding Consciousness(Routledge, 2000).

Max Velmans - University of London