My research is about embodied, "socially intelligent agents". I am interested in how autonomous agents (hardware agents, robots, as well as simulated or virtual agents) explore their environment and interact with the world, including interaction of agents of the same "matter" as well as "multi-species" interactions. As soon as a human is involved in these interactions (in the role of a user or just as a co inhabitant of the same real or virtual environment), all interactions are "cross-species" (e.g. between humans and robots or between humans and intelligent agents). The question is whether for all these different agents, contexts, situations and modes of interaction, different mechanisms for interaction and communication have to be realized, or whether a common "social interface" exists, which is applicable to interaction and communication situations in general. So, is there anything "universal" which makes up "successful" communication situations, "successful" in terms of being efficient, enjoyable and "natural"? (With respect to artifacts "natural" means "acceptable", i.e. whether humans like to become engaged in interactions or not. Humans cannot avoid contact with other humans, but artifacts and new technology can, to some extent, be used or not). I assume that there is such a common basis underlying interaction, communication and "understanding". If we talk about interaction and communication in an entertainment or business context, agents have to understand each other. And all nature (and studies in natural sciences confirm this) tells us is that understanding cannot be separated from a "living body", an embodied mind. Understanding means re-experiencing and reconstructing past experiences on the basis of the current context and bodily state of an agent.
Accordingly, the issue of bodily functions in cyberspace should be seen in the border context of embodiment in natural and artificial living and complex systems. The question of embodiment has always been linked to the question "What is life and intelligence?" My point is that parallel to the discussion on what embodiment (and intelligence) means in cyberspace, a close look is needed at what it means in real life. What do we know about how animals and humans use their body in real life? How do they live with their bodies within a given environment? In public, but also in the scientific community, "cyberspace" is often used as a metaphor for a world where the human body only plays a minor role, where we are in danger of becoming a creature without a body, experiencing things by direct stimulation of the nervous system, where we will lose the sense for reality and get lost as a virtual creature in a virtual world. A world where we become isolated, ending up as extremely individualized selfish beings. Some people in the artificial intelligence and virtual reality community are very enthuastic about this bodiless future; they are looking forward to "getting rid" of the ("imperfect", mortal) body, i.e. they like the idea of transmitting their personality and intelligence to a computer memory and leading an immortal and perfect life. I do not believe in such a future, for a number of reasons which I can touch upon only briefly:
Life and intelligence only develop inside a body. This body is coevolved and adapted to an ecosystem which the creature is living in. We can only talk about the organism in a given environment. Intelligence can only be studied within a c omplete system, embedded and coupled to its environment. Nature gives us the mos t impressive examples of adaptations and dynamic coupling of bodies to an enviro nment. Imagine that you are an engineer and that you want to build robots. The c lassic bionics approach takes nature as a model and tries to imitate what animal s do. In contrast, Artificial Intelligence researchers are mostly interested in what humans do, in human problem solving. A bat for example is a very complex system which uses sonar radiation to navigate in its environment. A particular bat is able to catch fish. It somehow bridges the gap between the domain of the water and the domain of the air. Imagine that you are a robot engineer and want to build such a robot - and we are talking about robots as complex systems. Another example - it is not taken from a science fiction movie - is a fish living deep in the sea (midwater zone from about 100 meters to a few kilometers below the surface) where it is almost dark. Here we find animals which use bioluminescence (a chemical process for producing light) for various purposes: burglar alarm, camouflage or - like the dragonfish - attracting prey with a bioluminescent trap suspended from its skin. The dragonfish has a very bizarre-looking body. From an engineering point of view you would have said: "This is impossible, it cannot work." But these systems can be found in nature and you cannot understand the bodily functions of this animal and what it is doing without knowing where it lives and how it interacts with its social and non-social environment, including other species. Let me give you another example. Perhaps you are more familiar with various ways how the bodies of insects are adapted to the environment in order to be protected against predators, for instance, through a kind of mimicry. A tropical milliped has several hundreds of legs. They are very beautiful animals. Again, you cannot understand why it looks like it does and why it behaves in a particular manner if you don't understand the environment, the interactions with the environment and the history. All animals, and humans too, have a historical aspect. We can, to some degree, modify our behaviour and appearance. But we are always limited to our history and the evolution of our body, to the concrete environment where our body evolved and developed.
Natural bodies are individual and unique. Even twins do not have a completely identical body. They might have a identical set of genes but their bodies are not identical. Bodies are physical entities. They occupy a distinct volume in space and time. No other individual can be at the same point in space and time. Our perception of reality uses the metaphor of a geometrical three- dimensional space within which our body is located. This is also a metaphor usually used in cyberspace that you locate interactions and people, or representations of people, in three-dimensional space. This is not just by accident or by a technological bias. This is because it due to a straightforward way of transferring metaphors of our natural thinking to metaphors of our thinking in artificial worlds. The body's shape changes during active movements or contacts with other objects. It is not something fixed. The body shows characteristic symmetrical and asymmetrical features. Individuals take a certain perspective according to the orientation of the body. Even if you are very close to another individual when, for example, you are watching a movie or dancing, it will never appear that you both experience things in exactly the same way, but simililarly. You can discuss the similarity of the experience and various forms of interpretation. You have to be aware, however, that you are always discussing ways of finding similarities. Language is our tool with which we communicate and find these common, similar aspects in our life and perception.
The big advantage of having a physical body is that it gives you a first-person experience by actively exploring the world. This is why we often find it boring to watch TV or read a book. We can have very lively mental experiences by doing this. What is missing is a certain interaction, a direct feedback from t he world. This is why virtual reality applications are sometimes much more appealing to children and also to adults, because they can provide this kind of feedback and interaction to some extent. In my view virtual environments and reality should not be regarded as two different "worlds". In the same way we could make the difference between "real worlds" and "fictional worlds" (which we know from novels, movies etc.), or "ideological worlds" (worlds of religion, politics, i.e. cultural constructions etc.). What matters are the ideas which are constructed in our embodied mind and which have an important meaning to us. A real person can have the same importance as a cartoon figure (e.g. to a child), or a football team (e.g. to adults). Technical devices which let humans enter virtual worlds manipulate the human body. They do not replace the body, they try to make virtual experiences (as we know them say, from watching TV) more realistic to us by allowing active exploration and interaction. This interactive aspect is the crucial difference to other media like TV or reading books. A movie or TV film tells a single story, we can re-experience according to a fixed timing, we are spatially and temporarily "enslaved", watching passively, consuming. Reading a novel is much more "powerful", but energy-consuming. A "script" is given, but multiple potential realizations can be created inside or mind, we can re-write the story and choose the timing ourselves. But it is not "realistic", sensory-motor feedback is still missing. Since actively exploring the world is a crucial aspect in (animal) intelligent behavior and learning, virtual environments could potentially bridge this gap. Various metaphors in different languages show the strong interrelationshiop between understanding and physical activity, e.g. the German word "begreifen", touching with the mind requires touching with the hands. This "interactive" agent-environment coupling is the mechanism by which the world around us makes sense to us.
Intelligence is linked to a social context. In other words: all intelligent creatures are social beings. Social interactions in real and virtual life are not only necessary to "entertain" humans or for solving a specific task (e.g., business meetings). Social interactions are important for the feeling of being there (the presence effect), to have "social" reality, to be "real". Social interactions are important for the development and maintenance of a personality and for behaving intelligently. Cyberspace can potentially enlarge the social network of people; it dramatically increases the chances to find "related souls", to get feedback and reinforcement from other people. Technology can and should support this social networking aspect. The connection between the body and social interaction is that humans are experts to take the body as a social tool. It is not just of our way of being in the world but also about our way of interacting with others living in the world. In human species we find the most elaborated forms of self-manipulation, using the body as a means to express personality and social roles, also as a means of manipulating others.
This kind of bodily experiential interaction is also important for understanding, for instance empathy. It requires a kind of openness toward others which is not possible without a body. This is one reason why I think that a rational model of a person being transferred to a computer system is not plausible. I am not saying that it is not possible. It might be possible at some point in the future that we will be able to transfer our brain activity to a computer or somewhere else, but I don't think that what we will get is in any way equivalent to a person or even to an intelligent creature. If we create anything analogous to the dynamics of nervous activities in our brain in a computer memory, then we should not hope that this activation pattern "is me". If intelligence and personality is so tightly connected to an active body, then why should this representation be human, and why should it behave intelligently at all? Moreover, all what we know about how meaning comes into the world, e.g. when a child grows up, is linked to a physical body, to physical experiences, to the full variety of pain and pleasure felt with a body. Even if we could (hypothetically) extract the essence of our personality in computational terms and transfer it to a computer memory, then who or what should "enjoy" immortality and "admire" the computational beauty of its "life"? We as humans are able to "feel" the beauty of algorithms or data structures; the data structure itself definitely cannot!
To conclude, in my view, we are not in danger that the physical human body will become extinct in cyberspace. It will be re-interpreted, somehow re-invented. Its meaning and expression will be adapted to changing conceptions of (social) "reality". But it is the very same process which has been going on since members of the homo genus discovered themselves as embodied minds, as "I" and "you". People have been discussing for a long time what human memory is. A popular metaphor sees human memory as a kind of database. There are items stored in a database and retrieved in a similar way they were before. Recent discussions point out that this is an illusion. We do not have a database in our head but a story generator. We are constantly writing and re-writing scripts, stories about ourselves and other people. Cyberspace is a attractive medium because it gives us ways to enact and to replay these kinds of stories. It seems to be a universal human feature that we like stories, that we need stories in our life. My interest in cyberspace came from the idea that this technology may help us to write and to enact good stories, better stories than before. The body is the point of reference for the reconstruction of old "stories" and the reinvention of new ones.
Aronson, E. 1994: The social animal, New York.
Byrne, R. 1995: The thinking ape: evolutionary origins of intelligence, Oxford.
Dautenhahn, K. 1995: Getting to know each other - artificial social intelligence for autonomous robots, in: Robots and Autonomous Systems 16, 333-356. Dautenhahn, K. 1997: I could be you - the phenomenological dimension of social understanding, in: Cybernetics and Systems 28(5). Special issue on epistemologocal issues of embodied artificial intelligence.
Johnson, M. 1984: The body in the mind, Chicago/London.