John Hopfield (born 1933) is an American physicist and neuroscientist who introduced Hopfield networks in 1982, a form of recurrent Neural network that established a rigorous link between statistical physics and associative memory in neural systems, and who shared the 2024 Nobel Prize in Physics with Geoffrey Hinton for foundational discoveries enabling machine learning with artificial neural networks.
Background
Trained as a physicist, with a PhD from Cornell University, Hopfield spent much of his early career working on the physics of solids before moving into biological and computational problems, holding positions at Princeton, Bell Labs, Caltech, and later Princeton again. His interdisciplinary path, from condensed-matter physics into neuroscience and computation, shaped the distinctive approach he brought to neural-network research.
Hopfield networks
In a 1982 paper, Hopfield described a type of recurrent neural network, now called the Hopfield network, in which neurons are fully connected to one another and update their states based on the weighted inputs from all other neurons, converging over time toward stable patterns, or "attractors." Hopfield showed that this dynamic could be understood using the mathematics of spin glasses from statistical physics, where the network's stable states correspond to local minima of an energy function. This let a network function as content-addressable memory, able to recall a complete stored pattern from a partial or noisy cue, a capability distinct from the simple feedforward pattern recognition of the Perceptron. The energy-based framing Hopfield introduced influenced later generations of models, including David Rumelhart and Geoffrey Hinton's Boltzmann machines, and is considered a foundational contribution to the mathematical study of neural networks and, more broadly, Deep learning.
2024 Nobel Prize in Physics
In October 2024, Hopfield and Geoffrey Hinton were jointly awarded the Nobel Prize in Physics "for foundational discoveries and inventions that enable machine learning with artificial neural networks," part of the broader 2024 wave of Nobel recognition for AI research that also saw Demis Hassabis and collaborators win the Chemistry Prize for AlphaFold. The physics prize committee cited Hopfield's use of tools from statistical physics to model associative memory, and Hinton's extension of those ideas into the Boltzmann machine and subsequent backpropagation-trained networks. The award drew some debate within the physics community over whether neural-network research constituted physics in the traditional sense, though the committee framed the honor around the physical methods Hopfield employed rather than the field of application.
Legacy
Hopfield networks are considered one of the founding architectures of modern deep learning, and the energy-based perspective they introduced remains an active area of research, including modern variants sometimes called "modern Hopfield networks" that have been linked mathematically to the attention mechanisms used in the Transformer (architecture) architecture underlying today's large language models. Hopfield's career, bridging physics and computation, exemplifies the interdisciplinary origins of much of contemporary Artificial intelligence research.