elearnspace

Chaos theory can provide a useful model for learning: a limited range of inputs can provide a significant variety of outputs. Because the output range is so diverse, it’s easy to assume that the process itself must be astonishingly complex. It’s not. Systems that are complex and even chaotic function according to a few simple rules (see agent-based modeling). Scientific American reports on how scientists are using the multiple inputs driven by basic rules in order for a robot to “learn” (it’s not exactly learning, it’s more about negotiating and reacting to chaotic stimuli…learning would require some pattern storage, which the video states is a future research task).