Announcing the 2025 Sejnowski-Hinton Prize
The NeurIPS organizing committee is pleased to announce the winner of the 2025 Sejnowski-Hinton Prize: Timothy Lillicrap, Daniel Cownden, Douglas Tweed, and Colin Akerman for their groundbreaking 2016 paper “Random synaptic feedback weights support error backpropagation for deep learning” published in Nature Communications (arXiv version).
The 2025 Winner: Feedback Alignment
Theories of the brain have long sought to explain how neural circuits learn efficiently using only local information—how synapses adjust based on signals available at each connection rather than through explicit, global error propagation. Inspired by this challenge, researchers have explored ways to train artificial neural networks that perform gradient descent using local learning rules.
This paper is recognized for its contribution to that pursuit through the discovery of “feedback alignment”. The authors demonstrated that multi-layer networks can learn effectively using fixed, random feedback weights, rather than requiring exact backward weight symmetry as in backpropagation. Remarkably, the network’s forward weights naturally align with these random feedback signals over the course of learning, yielding a biased yet useful estimate of the loss gradient. This insight provided the first concrete, biologically grounded solution to the long-standing weight transport problem, the question of how real neurons might follow loss gradients without non-local information transfer.
The work had a significant impact, helping to establish a new sub-field of “biologically plausible” learning rules in the NeurIPS community and beyond.
About the Prize
The Sejnowski-Hinton Prize is awarded annually to a paper published within the past ten years that has made major contributions to computational theories of the brain drawing on insights from artificial intelligence, and has had a significant impact on the NeurIPS community. A seven-person committee—Blake Richards, Sebastian Seung, Eva Dyer, Razvan Pascanu, Maneesh Sahani, Terry Sejnowski, and Geoffrey Hinton—selected this year’s winner. The prize carries a $10,000 award to be shared among the authors.
Origin of the Prize
First awarded in 2025, the Sejnowski-Hinton Prize is rooted in collaboration and generosity. In 1983, Geoffrey Hinton and Terry Sejnowski made a pact: if one of them received a Nobel Prize for their work on Boltzmann machines and the other didn’t, they would share the prize money. In 2024, when John Hopfield and Geoffrey Hinton were awarded the Nobel Prize in Physics—with Boltzmann machines cited among the contributions—Terry declined his share. Geoffrey then donated these funds to NeurIPS to establish the Sejnowski-Hinton Prize, honoring their longstanding partnership and commitment to advancing computational theories of the brain and collaborative work in the community.