2019 Benjamin Franklin Medal in Physics presented to John J. Hopfield, Ph.D.

After conducting highly-regarded theoretical research in solid state physics at Cornell University, AT&T Bell Laboratories, Princeton University, and the University of California, Berkeley, John Hopfield turned his attention to biology. Of a number of problems on which he worked, two are perhaps most prominent. The first was how certain biological processes, such as expressing the genetic code, could achieve the accuracy measured in experiments when simple physics predicted a much lower accuracy. By theorizing an energy source that drove one reaction in the sequence to be irreversible, thus giving more time for another reaction in the sequence, Hopfield was able to explain how higher accuracy is possible. He then applied this theory of kinetic proofreading to three biological functions: DNA synthesis, linking an amino acid to its specific tRNA, and assembling proteins. This basic concept has turned out to be of fundamental importance for all systems that have an intrinsically low ratio of incorrect to correct products. The second of his prominent achievements addressed how neural networks might be capable of associative memory, devising a way that a specific network state could be "remembered" through the strengths of the connections between neurons. Such a network could "remember" multiple states of the network and recall a specific one by starting in a somewhat similar initial state and firing neurons randomly. Instead of the state being remembered locally by each neuron, Hopfield's network distributed the memory across the entire network. This model not only turned out to be quite important in the fields of neuroscience and cognitive science, but also became a prime launching point for the development of the field of machine learning.