|Understanding and Creating the Technology
of Expertise: The Role of Perceptual Learning
Phil Kellman, UCLA
Understanding and Creating the Technology of Expertise: The Role of Perceptual Learning
A widely held belief is that advances in computer and multimedia technology offer opportunities to improve education in mathematics and science. To achieve real improvements, we need to grapple with the "bells and whistles" problem: how to distinguish technology that truly impacts the learning process from meaningless (and expensive) colorful animations, simulations and sounds. Two useful strategies in meeting this challenge are 1) to emphasize the collection of appropriate, objective data, and 2) to understand and apply principles of cognitive science in exploiting the capabilities of technology. In this talk, I will illustrate these strategies in applications of principles of perceptual learning to training and education.
Perceptual learning -- changes in the way information is picked up as a result of experience -- is a major component of expertise. Miraculously, with appropriate experience in any domain, human attentional systems seem to zero in on the relevant details and structure needed for important classifications in that domain, while becoming able to ignore irrelevant variation. Research in a number of domains has demonstrated dramatic differences between novice and expert information pick-up: Experts extract larger chunks of information, detect details better, discover higher-order relationships and become automatic in extracting patterns. Although the importance of expert pattern extraction has been recognized, there have been few attempts to directly train these skills. They are not well addressed by traditional methods.
We have been exploring the conditions that accelerate the development of pattern extraction skills. We incorporate the conditions for rapid training of pattern pickup into domain-specific perceptual learning modules (PLMs). PLMs depend heavily on computer technology for timing, trial sequencing and display variation. Advances in simulation, multimedia and virtual reality technology are making possible even more realistic and useful PLMs for many learning situations.
PLMs in a 5-year project in aviation training have been highly successful. I will present data from these efforts as an illustration of PLM concepts and outcomes. In more recent efforts, we have applied PLMs to mathematics learning, and we have begun to study systematically the conditions to optimize learning and transfer. We will consider data from a series of experiments in which students grapple with the algebra-geometry connection -- the relations between the structure of symbols in equations and graphical representations of functions. The results indicate strong training effects but raise a number of questions about optimal strategies for obtaining transfer using PLMs. We will also examine current attempts to apply PLMs to science learning, specifically the development of pattern recognition in chemistry.
There is a natural fit between modern computer technology and concepts of perceptual learning. Methods for developing expertise in pattern processing are drastically in lacking in traditional instruction. This expertise can be trained, but it requires certain conditions and formats. Combined with an understanding of cognitive principles of expertise, the ability of computers to store large display sets and to display, vary and animate them in real time, makes possible unprecedented advances in addressing neglected components of learning. To realize these advances, substantial research efforts will be required. Among the broad research questions are at least these three: 1) What are the conditions that optimize perceptual learning? 2) What are optimal ways of combining perceptual classification training with instruction about facts and concepts? 3) What are the optimal ways of structuring perceptual learning interventions to motivate and interest students as they participate in tasks that exercise the cognitive machinery leading to expert pattern extraction?
for Education Research