Paper Accepted into IOPscience Neuromorphic Computing and Engineering

We are pleased to share that the article "Phasic attractors for flexible and adaptive working memory in spiking recurrent neural networks" has been accepted into the IOPscience journal Neuromorphic Computing and Engineering. This open access, multidisciplinary journal publishes leading research on artificial neural systems with a focus on neuromorphic hardware and computational neuroscience.

The paper, authored by Willian S. Girão, Thomas F. Tiotto, Jelmer P. Borst, Niels A. Taatgen, and Elisabetta Chicca, introduces a novel spiking recurrent neural network model for working memory that leverages phasic attractors—short-lived, energy-efficient activation states. Inspired by biological mechanisms, the model utilizes synaptic delays to enable context-sensitive and probabilistic computation, leading to more adaptive and efficient information processing.

By shifting from persistent to phasic attractor dynamics, the approach significantly reduces energy consumption while maintaining flexibility and robustness, marking a step forward in the development of biologically plausible and computationally powerful neuromorphic systems.