In Michael Tolson's installation "Las Meninas", two observation stations are supposed to tempt the participants into voyeuristic behavior. From a low podium the artificial creatures can be fed with a sensor. The result can be viewed through a pair of binoculars which are facing the reflection of the monitor.

"Las Meninas" is a piece about intimacy, observation and the observer's interactions with active, living space. It has two components, which represent two views of the same virtual space. This virtual space exists in a computer which is running a simulated (artifical life) ecosystem. This system is populated by creatures which have evolved within it. Their co-evolved emergent behaviors are an image of the dynamics of the system, which include the contributions of observers.

The dynamics of the system are driven by the flow of energy through it, in the form of abstract food which viewers can inject interactively. Our participation in the space is simultaneously active and voyeuristic. Our experience of the space is double, with each view literally an inversion of the other. This double space resembles the double space in stereotypical late-renaissance portraits, where the subject is in an intimate enclosed space with a window behind them which reveals an infinite projective landscape. As viewers we participate in both spaces. The simulation will be running on a Silicon Graphics workstation. A population of mobile creatures roam about a volume of space. Their behavior is governed by artificial neural nets which have been evolved using genetic algorithms. They eat and move. The food they eat is provided by the observer.

The video output of the system will be field-sequential stereoscopic (120 HZ). There will be two observation stations. The first will be in the form of a low (approx. 1m) pedestal. Within the pedestal will be a monitor facing upwards. Attached above this monitor will be a pair of stereoscopic (liquid crystal shutter) goggles. Thus a virtual space will be created between the surface of the monitor and the goggles. Also attached to the pedestal will be a Polhemus 3D sensor, which the observer can pick up and move around within this space. The observer will see simultaneously the creatures moving around in their virtual world and his own hand, holding the sensor, immersed in that world. This sensor will determine where and when food is injected into the creatures' world.

The stereo video output is also sent to a second monitor at the opposite end of the installation space which is mounted in a corner of the ceiling. This monitor is only visible reflected in a mirror which is mounted in the opposite corner. A pair of binoculars is mounted on a tripod, roughly in the center of the space, pointed at the mirror, and focused on the monitor. This pair of binoculars is specially modified to include liquid crystal shutters, and is thus stereoscopic. Because the monitor is reflected in a mirror, the parallax is inverted. Thus the projective space becomes exterior instead of interior, and the monitor becomes a window instead of the bottom of a pool. The gallery space is otherwise empty.

An important element of the piece is discomfort. Both viewing stations are designed to force the observer into a slightly uncomfortable position, typical of the voyeur. At one station, the observer must bend over or kneel; at the other he must stoop to look through the binoculars. Virtual reality traditionally seeks to liberate the cartesian eye from the body; this piece is designed to force that eye back into the body.

Thinking Skin -­ Neural Mediation of Reaction-Diffusion

In 1952 Alan Turing published a seminal and far-reaching paper: "The Chemical Basis of Morphogenesis". In this paper he postulated a mechanism for pattern generation in the embryonic skins of animals and in other generative processes. This mechanism was based on "reaction-diffusion" ­ a coupled ensemble of differential equations each having a nonlinear reaction term and a diffusion term. He postulated abstract chemical reactants called "morphogens" which were diffusing through a medium at different rates and interacting with one another, and demonstrated how such a model could reproduce patterns found in nature. This "Thinking Skin" work presented here replaces the explicit reaction terms in Turing's equations with neural networks having inputs and outputs mapped onto concentrations of abstract reactants diffusing over a substrate. These networks are "bred" using genetic algorithms. Networks are bred in large co-evolving populations of mobile creatures which are "feeding" off of abstract reactants, and metabolizing those reactants, accumulating internal energy and transforming this energy to work and ultimately back to "food" which is returned to the environment. As species co-evolve, patterns of macroscopic behavior emerge out of the microscopic but indirectly coupled behaviors of individual organisms. The intention of this research is to develop new techniques for generating "reactive surfaces" and for synthetic morphogenesis, or "A-Form".