Sensory Prediction: Engineered and Evolved

Engineers try to predict future input from past input; this can take the form of prediction of natural video, natural audio, or text, which has famously led to such products as Generative Pre-trained Transformer 3 (GTP3) and proprietary algorithms for stock market prediction. Organisms and parts of organisms may have evolved to efficiently predict their input as well, and the hypothesis that they do are cornerstones of theoretical neuroscience and theoretical biology. How one can design systems to predict input is still a matter of debate, especially when one has continuous-time input—input that has a state at every point in time, not just at specially sampled points. We aim to bring together researchers from across the globe who approach the question of how to design systems to predict input through the lens of biology with machine learning, information theory, and dynamical systems. We will gather for a five-day workshop meeting in July 2023 at the Santa Fe Institute, a mecca for complex systems research. In gathering together this group of experts, we aim to develop a unifying framework for sensory prediction—establishing metrics for when a system has predicted well, and an understanding of how it has done so.  This knowledge will help establish a foundation of theoretical neuroscience and theoretical biology, to enable the scientific community to better calibrate and understand prediction products.

This workshop will be the first time these experts on sensory prediction have gathered together to collaborate on an interdisciplinary approach to understanding both the brain and how to improve prediction algorithms. Workshop participants bring a range of valuable perspectives for considering sensory prediction. Several of these experts are devoted to examining evolved systems, including the study of neurons in the retina, hippocampus, and the visual cortex. Several others are devoted to studying engineering systems to better predict input through reservoir computing and training recurrent neural networks, in which reservoir weights are trained as well. The common thread across all participants’ research is that there is a definition of what one needs to infer in order to predict. This definition has inspired biological work and many prediction algorithms, which work together in tandem: an understanding of the biology has inspired improvements to prediction algorithms, and improvements in prediction algorithms have led to new testable hypotheses about parts of biological organisms. 

AGENDA LINK