Overview

The use of animals in neuroscience research is a fundamental tool to understand the inner workings of the brain during perception and cognition in health and disease. Neuroscientists train animals (often rodents) in behavioural tasks over several months, however training protocols are sometimes not well defined and this leads to delays in research, additional costs, or the need of more animals. Finding strategies to optimise animal training in safe and ethical ways is therefore of crucial importance in neuroscience. In this project, we will develop a computational framework based on artificial neural networks (ANNs), to predict the behavioural output of animals trained in neuroscience tasks. We will use existing behavioural data to train our ANNs in the same way that animals are trained, and study whether our models can predict the behaviour of the animals. Our results will lead to more efficient training protocols and improve neuroscience practices.

Outcomes

The ANNUBeS / NeuroGym project featured in dedicated and thematic workshops. The first dedicated ANNUBeS workshop took place in October 2024 at the Netherlands eScience Center, where invited experimental neuroscientists and computational modellers gave multiple presentations and interacted with the ANNUBeS team. This was followed by the Recurrent Neural Networks Making Decisions workshop that took place in August 2025 at the University of Amsterdam. The workshop covered the topic of decision making using recurrent neural networks and the neural mechanisms underpinning the decision-making process. A second dedicated workshop took place at the Netherlands eScience Center, marking the official end of the project. The workshop included presentations outlining the progress made on the NeuroGym platform and ongoing research on behavioural and cognitive neuroscience that could form the basis for new neuroscience tasks.

A preprint of a paper presenting the NeuroGym platform was submitted in June 2026. The paper introduces the design and implementation of NeuroGym and highlights its many useful features. It is backed with representative examples of well-studied classical neuroscience tasks whose findings have been faithfully reproduced with Neurogym.