- Please see our Call For Papers for details on submissions the deadline for which is 31st Oct 2004.
Background and Aim of the Symposium
There has been a recent resurgence of interest in machine consciousness. The goals of this field are both a) to create artifacts that have mental characteristics typically associated with consciousness (such as awareness, self-awareness, emotion and affect, experience, phenomenal states, imagination etc.) and b) to model these aspects of natural systems in embodied models. Recent machine consciousness workshops in Memphis (2002), Birmingham (2003), Turin (2003) and Antwerp (2004) have made it clear that this is a swiftly emerging field of international presence.
Independently, there have been several new developments in cognitive science and consciousness studies concerning the nature of experience and how it may best be investigated. Synthesizing results from embodied AI, phenomenology and hermeneutics in Philosophy, Neuroscience and enactive Psychology (among others), new paradigms for research into natural consciousness which transcend the limited behavioral/cognitive or neural/functional oppositions are being proposed and tested, with encouraging results.
Now is an excellent time, then, to entwine these two strands, to see how they might be of mutual benefit to each other. Not only can advances in consciousness research guide efforts into building conscious systems, but vice versa: the insights gained from attempting to build embodied, experiencing agents can provide important feedback to the various disciplines of consciousness studies - at the very least, by letting us know where our current theories are incomplete or inadequate for actual implementation.
What follows are some of the key recent developments in consciousness studies that we think should be investigated in relation to the goals of machine consciousness:
• Imagination-based Approaches: A key finding of the Birmingham and Turin meetings was the existence of a common theme in much of the important work in machine consciousness: the imagination or simulation approach to consciousness (Holland, Shanahan, Ziemke, Cotterill, Hesslow, Chrisley, Aleksander, Stein).
• Development: Development has in recent years become central to the understanding of how artificial cognitive systems can ever hope to achieve human level performance. The series of conference grouped around Epigenetic Robotics (2001, 2002, 2003) have considered the practical implications of how physical and social interaction can build more complex cognitive structures out of less complex ones. Although development has for long time been known to be a crucial component in the understanding of consciousness (Vygotsky 1993, org 1934) the connections between these areas remain largely unaddressed although work on the development of intersubjectivity (Trevarthen 1994) may point the way forward.
• Enactive Perception: Enactivists have argued that perceptual experience consists in the exercise of the mastery of sensory-motor contingencies, and that awareness consists in the application of this mastery to a reasoning process (e.g., O'Regan and Noë 2001). If this is correct, then central goals for machine consciousness research are a) establishing clear criteria for when a robotic system possesses such mastery and b) building robots which meet these criteria in a way which allows said mastery to play a crucial role in their deliberations.
• Heterophenomenology: It seems undeniable that phenomenological reports are a valuable source of data concerning consciousness, and yet a scientific theory of consciousness must be sensitive to the possibility that subjects may be mistaken in their sincere avowals concerning experience. (Dennett 1991) suggests, and (recent JCS issue) discusses a way to avoid the pitfalls of naïve or folk conceptions of consciousness without discounting phenomenological reports altogether: Heterophenomenology. Adopting this methodology with respect to machine consciousness seems promising, but poses difficult questions, such as: What kinds of communicative abilities need a robot possess in order to allow the direct application of this approach? An interesting possibility in this regard has recently been floated involving self-directed speech (Steels 2003).
• Synthetic Phenomenology: A science of consciousness, be it of natural or artificial agents, requires some ability to specify and refer to subjective, fine-grained experiential states, which, by their very nature, elude linguistic expression. One idea is that the states of artifacts-in-an-environment might themselves serves as ways of specifying the conscious states that they embody (e.g. Chrisley and Holland 1994; Chrisley 1995). The sub-field of synthetic phenomenology aims to investigate this idea by, e.g. constructing means of visualizing or otherwise communicating the (actual, or modeled) experiential states of robots.
• Second Person Approaches and Neurophenomenology : Crucially, social interaction, especially the notion that human consciousness develops from and is grounded in intersubjective precursors has been fundamental to the growth of first and second person studies in consciousness (Thompson 2001; Varela 2001). Since theorists are themselves social subjects one cannot ignore the intersubjective relationships between theorist and subject (or robot). The sophisticated, interactive protocols being developed in Neurophenomenology may prove to be a source of data and design intuitions for the construction systems that warrant the attribution of phenomenological states (Varela 1996; Thompson 2003).
• Ethics: Some would argue that MC (unlike 'mere' machine intelligence) has an inherent ethical dimension, since a genuinely conscious machine would perforce be capable of enjoyment, suffering, and thus apparently be a genuine ethical subject (Torrance 2000). Moreover, if the relationship between theorist and subject is crucial for the enaction of consciousness, then the ethical dimension can no longer be treated as external to the ambitions of machine consciousness. Rather, as we build increasingly complex artifacts in order to understand consciousness, normative concerns become essential both to our understanding of the constitution of subjectivity, and our appreciation of, and actions toward, the artifacts we create.
We hope that progress with respect to a significant number of these topics will be reported on and discussed at the Symposium.
Previous Workshops on Machine Consciousness |
| Can a Machine Be Conscious? 13-16 May, 2001. at The Banbury Center of the Cold Spring Harbor Laboratory |
| Panel on machine consciousness at Toward a Science of Consciousness, Skovde 2001 |
| Memphis (2002) |
| Models of Consciousness Workshop: In Search of a Unified Theory. Birmingham 2003. |
| Exystence conference on Machine Consciousness. Sep 29th - Oct 1st, Turin 2003 |
| Workshop on Machine Models of Consciousness. 29th June Antwerp 2004. |
For more information on the field see: Machine Consciousness
Citations
Berthouze, L. and T. Ziemke (2003). "Special Issue: Epigenetic Robotics: Modelling Cognitive Development in Robotic Systems." Connection Science 15(4).
Chrisley, R. and A. Holland (1994). Connectionist synthetic epistemology: Requirements for the development of objectivity, COGS CSRP 353: 1-21.
Chrisley, R. (1995). Non-conceptual Content and Robotics: Taking Embodiment Seriously. Android Epistemology. K. Ford, C. Glymour and P. Hayes. Cambridge, AAAI/MIT Press: 141-166.
Clowes, R. W. (2003). Action Oriented Adaptive Language Games. Proceedings of the Third International Workshop on Epigenetic Robotics, Boston, Lund University Cognitive Studies.
Dennett, D. C. (1991). Consciousness Explained, Penguin Books.
Holland, O., Ed. (2003). Machine Consciousness, Imprint Academic.
Iizuka, H. and T. Ikegami (2003). Adaptive Coupling and Intersubjectivity in Simulated Turn-Taking Behaviours. proceedings of ECAL 03, Dortmund, Springer.
Ikegami, T. and M. Taiji (1999). Imitation and Cooperation in Coupled Dynamical Recognizers. Advances in Artificial Life. D. Floreano, Springer-Verlag: 545--554.
O'Regan, J. K. and A. Noë (2001). "A sensorimotor account of vision and visual consciousness." Behavioral and Brain Sciences 24.
Sanz, R., A. Sloman, et al. (2003). Models of Consciousness Workshop: ESF/PESC Exploratory Workshop Scientific Report, Birmingham.
Sloman, A. and R. Chrisley (2003). Virtual Machines and Consciousness. Machine Consciousness. O. Holland. Exeter, Imprint.
Sloman, A. and R. Chrisley (2003, in preparation). The architectural basis of affective states and processes. Who needs emotions?: The brain meets the robot. Oxford University Press. J. Fellous and M. Arbib.
Steels, L. (2003). Language Re-Entrance and the 'Inner Voice". Machine Consciousness. O. Holland. Exeter, Imprint.
Thompson, E. (2001). "Between Ourselves: Second-Person Issues in the Study of Consciousness."
Thompson, E. (2003). Neurophenomenology: Integrating Subjective Experience and Brain Dynamics in the Science of Consciousness. Toward a Science of Consciousness, Prague.
Torrance, S. (2000). Towards an Ethics for Epersons. Symposium on Artificial Intelligence, Ethics and (Quasi-) Human Rights: AISB2000, Birmingham.
Trevarthen, C. (1994). The self born in intersubjectivity: The psychology of an infant communicating. The Perceived Self. U. Neisser, Cambridge Univ. Press.: 121-173.
Varela, F. J. (1996). "Neurophenomenology: a methodological remedy to the hard problem." Journal of Consciousness Studies 3: 330-350.
Varela, F. J. (2001). "Intimate distances. Fragments for a phenomenology of organ transplantation." Journal of Consciousness Studies 8(5): 259 -- 271.
Vygotsky, L. S. (1993, org 1934). Thought and Language, MIT Press.