How the concept of a pattern, as understood in information science and applied in contemporary AI, can address deep questions in science and philosophy.

The explosive growth of AI and machine learning in recent decades is predicated on the recognition and exploitation of patterns in data. Of course, scientists have engaged in their own—less automated—processes of pattern recognition since the birth of science itself, and biological organisms evolved their own neural networks for pattern recognition long before people and their technology came along.

In his seminal work, “Real Patterns,” philosopher and cognitive scientist Daniel Dennett laid out a road map for connecting the idea of “patterns” as understood by information theory to the practices of scientists and to our own cognitive capacity to model and predict the world around us. In this book—the first dedicated to the topic of real patterns—Tyler Millhouse, Steve Petersen, and Don Ross follow this road map. They explore the relevance of patterns to important aspects of both science and nature, including the emergence of high-level structure in physics, the nature of biological species, the measurement of welfare in economics, the evaluation of causal models, and the possibility of understanding in large neural networks.