Room 3 · The Winters · 1982

Hopfield Networks

Hopfield Networks (John Hopfield, 1982): A memory that rebuilds itself.

In 1982 the physicist John Hopfield described a network of interconnected neurons that works as an associative memory: if it is started with an incomplete or noisy pattern, its activity settles until it recovers the stored pattern. His work brought physicists and mathematicians closer to neural networks.

In April 1982 John Hopfield, a physicist then at Caltech, published a short paper in the Proceedings of the National Academy of Sciences: "Neural networks and physical systems with emergent collective computational abilities." It brought a new kind of researcher to neural networks at a time when few were working on them.

A Hopfield network is a set of simple neurons, each connected to all the others, that can be on or off. You store patterns in it, such as simple images, by setting the strengths of the connections. Then you start the network from a damaged or incomplete version of one of those patterns. Each neuron updates according to the signals it receives, and the whole network settles, step by step, into the closest stored pattern. That is associative memory: a fragment brings back the whole, the way a few notes can bring back a song.

Hopfield's key insight came from physics. He showed that the network has a quantity that behaves like energy and that only goes down as the network updates, so the stored patterns are like valleys in a landscape and the network rolls into the nearest one. That link to models of magnetic materials, studied in statistical physics, gave researchers a precise mathematical way to analyze what networks of neurons can do, including how many patterns they can store before memories start to blur.

The paper helped revive interest in neural networks in the 1980s, together with the work on Boltzmann machines and backpropagation that came soon after.

In 2024 Hopfield shared the Nobel Prize in Physics with Geoffrey Hinton for this work. You can see that prize in Room 6.

Ask your AI: explain the energy landscape of a Hopfield network using a ball rolling into valleys, then tell you where the analogy breaks.

Did you know? Hopfield received the Nobel Prize in Physics in 2024, together with Geoffrey Hinton.

Source: doi.org/10.1073/pnas.79.8.2554

See it in the museum →

← The Fifth Generation Project · Backpropagation →

All of Room 3: The Winters

Cite this: Nazareno, I. (2026). Hopfield Networks (1982). New Mind Museum. Retrieved November 17, 2026, from https://newmindmuseum.com/museum/hopfield-networks-1982/