Imagistic reasoning

Imagistic reasoning is a new paradigm for understanding sensory data and controlling environments based on the construction, interpretation, and manipulation of image-like, analogue representations of physical systems. The reasoning is primarily perceptual and only secondarily symbolic. In the past decade, we have built several imagistic reasoners that perform at an expert level on scientific problems that defy current analytical methods, including helping us solve open problems. We hypothesize that much of scientific reasoning is imagistic and that this reasoning is best automated by imagistic algorithms. The classical artificial intelligence architecture—a central deductive reasoner operating on symbolic predicates delivered by low-level perceptual preprocessors—is unsuitable for these tasks. Imagistic reasoners are faster and more efficient because they trade many inferences for sensing and action. Their behavior is easier to understand and debug because they deal directly with geometric structures and their interactions. Imagistic reasoning organizes computations around information-rich, analogue representations of physical systems. The analogue representations behave in a manner analogous to a system or phenomenon under study. The parameters of the representations are continuous so that varying the parameters will normally lead to continuous changes in behavior, except at special points where qualitative changes occur. The reasoner can learn about the system by adjusting the parameters of the representation and observing the subsequent changes, Typical examples of analogue representations are the light-intensity array of a scene, the temperature field of a room, the velocity and vorticity of a fluid flow, the configuration space of a mechanism, and the phase space of a dynamical system. As a new problem-solving paradigm, imagistic reasoning draws on techniques from computer vision, qualitative reasoning, scientific computing, and computational geometry. The central research pro-blem in imagistic reasoning is to identify and codify the many different analogue representations and imagistic algorithms that have proven useful in implementing the bidirectional mapping between sensing and action. As an example of imagistic reasoning research, we have developed a computational method called spatial aggregation that systematizes many imagistic reasoning algorithms. Like computer vision routines for lowand intermediate-level image analysis, spatial aggregation pro-