Imagine hearing a familiar sound and expecting something to happen. Before the event arrives, the brain can predict what it will be, when it will occur, and how likely it is. Yet, computational models often treat these questions separately or use learning methods that are difficult to reconcile with biological circuits. A new study proposes an alternative in which one population of spiking neurons learns all three together.
The research team was led by Associate Professor Zenas C. Chao along with Mr. Yohei Yamada, Academic Specialist, from the International Research Center for Neurointelligence (WPI-IRCN), UTIAS, The University of Tokyo, Tokyo, Japan. The researchers developed a recurrent spiking-network model to test whether a single neural population could learn event identity, timing, and probability using local learning rather than backpropagation or a globally broadcast error signal. The study was published online in the journal Communications Biology on August 26, 2026.
The team designed a Multi-Event Expectation Task in which a brief cue predicted one of two events. Each event had its own expected delay, while its probability could vary. The model was trained in 100-trial blocks and tested with the cue alone, allowing researchers to examine activity generated by the network’s internal prediction. The network contained 1,000 spiking neurons, while learning was confined to readout connections.
The model successfully combined the three dimensions of prediction. When an event became more likely, its predicted activity became stronger; when its expected timing changed, the network shifted its anticipatory response. Most improvement occurred during the first 50 trials. Identity and timing were represented by separable patterns within the same neural population rather than by separate modules.
“Prediction in everyday life is inherently multidimensional ,” said Prof. Chao. “ Our results show computationally that a single recurrent spiking population can learn what is expected, when it is expected, and how likely it is, while updating these predictions when environmental statistics change.”
The network also adapted when probabilities or event timings were abruptly switched. Prediction errors rose at each change before falling again. Compared with alternative approaches, including a global least-squares method and versions without online timing updates or shared readouts, the local learning approach maintained more stable representations and faster recovery.
Further analyses showed that probability and timing formed factorized patterns in the network’s readout weights while remaining within an overlapping population. The model also generalized across numbers of possible events, timing variability, and teaching-signal shapes, suggesting that the framework was not limited to one task.
The work has potential implications for brain-inspired artificial intelligence and neuromorphic computing. Local learning rules may suit hardware that must adapt efficiently in real time. The model also generates testable neuroscience hypotheses about shared populations and neuromodulatory signals that could help update predictions.
“We wanted to connect a simple everyday idea about prediction with a learning mechanism that could operate locally ,” said Prof. Chao. “ The model provides a computational foundation for investigating how flexible prediction might arise without globally coordinated learning.”
Overall, the study provides a computational demonstration that one recurrent spiking population can jointly encode event identity, timing, and probability and rapidly update these predictions as conditions change. It does not directly show how the brain implements prediction, but offers a grounded framework and experimentally testable ideas for future studies of cognition and adaptive artificial systems.
........
The article, “Joint encoding of ‘what’ and ‘when’ predictions through error-modulated plasticity in biologically plausible spiking networks,” was published in Communications Biology at DOI: 10.1038/s42003-026-10836-2
***
Reference
DOI: https://doi.org/10.1038/s42003-026-10836-2
International Research Center for Neurointelligence (WPI-IRCN), The University of Tokyo
The IRCN was established at the University of Tokyo in 2017, as a research center under the WPI program to tackle the ultimate question, “How does human intelligence arise?” The IRCN aims to (1) elucidate fundamental principles of neural circuit maturation, (2) understand the emergence of psychiatric disorders underlying impaired human intelligence, and (3) drive the development of next-generation artificial intelligence based on these principles and function of multimodal neuronal connections in the brain.
Find out more at: https://ircn.jp/en/
About the World Premier International Research Center Initiative (WPI)
The WPI program was launched in 2007 by Japan's Ministry of Education, Culture, Sports, Science and Technology (MEXT) to foster globally visible research centers boasting the highest standards and outstanding research environments. Operating at institutions throughout Japan, the 18 centers that have been adopted are given a high degree of autonomy, allowing them to engage in innovative modes of management and research. The program is administered by the Japan Society for the Promotion of Science (JSPS).
See the latest research news from the centers at the WPI News Portal: https://www.eurekalert.org/newsportal/WPI
Main WPI program site: www.jsps.go.jp/english/e-toplevel
About Associate Professor Zenas C. Chao
Prof. Zenas C. Chao is an Associate Professor at the International Research Center for Neurointelligence (WPI-IRCN), UTIAS, The University of Tokyo, Tokyo, Japan. He earned degrees in Life Science and Chemistry in Taiwan before pursuing Biomedical Engineering at Georgia Institute of Technology, USA. During his doctoral work, he grew neurons in petri dishes and interfaced them with robots, demonstrating that machines equipped with an organic brain can learn behaviors. He worked at RIKEN, the National Institute for Physiological Sciences, and Kyoto University. Since 2019, his research at IRCN explores predictive coding, brain-controlled systems, and mechanisms of creativity and intelligence.
Funding information
This work was supported by World Premier International Research Center Initiative (WPI), MEXT, Japan (to Z.C.C.).
Communications Biology
Computational simulation/modeling
Not applicable
Joint encoding of “what” and “when” predictions through error-modulated plasticity in biologically plausible spiking networks
26-Aug-2026
The authors declare no competing interests.