Dr Rui Ponte Costa

Rui leads the Neural & Machine Learning group that leverages AI principles to develop a new generation of computational models of learning in the brain. In particular, the group focuses on understanding how (i) cortical circuits, (ii) neuromodulation and (iii) subcortical regions enable efficient credit assignment in the brain.

During his undergraduate studies in computer science at the University of Coimbra (Portugal), Rui became fascinated by our ability to solve and learn complex problems. This led him to pursue a PhD in Computational Neuroscience and Machine Learning at the University of Edinburgh (UK) where he established close collaborations between theoretical (Mark van Rossum) and experimental groups (P. Jesper Sjöström). During this time Rui was also a visiting PhD student at the University College London (UK) and McGill University (Canada). 

Next, Rui conducted postdoctoral research in computational neuroscience & machine learning at the University of Oxford (UK) with Tim Vogels where he established collaborations with the groups of Nando de Freitas (Oxford/Google Deepmind) and Nigel Emptage (Oxford). Next, Rui did a short postdoc with the group of Walter Senn (Bern, Switzerland) in collaboration with Yoshua Bengio (MILA, Canada). 

In 2018 Rui moved to the University of Bristol to start his own group (Neural & Machine Learning group), which in 2023 moved to the University of Oxford. In 2025 he was awarded the title of Associate Professor.

Computational Neuroscience, Learning, Neural Circuits, Systems Neuroscience, AI.

At LMH Rui tutors Neuroscience to Biomedical, Medicine and Psychology undergraduates.

Self-supervised predictive learning accounts for cortical layer-specificity
Nejad KK, Anastasiades P, Hertäg L* and Costa RP* [* equal supervision]
Nature Communications, 16, 6178, July 2025

Cerebellar-driven cortical dynamics can enable task acquisition, switching and consolidation
Pemberton J, Chadderton P and Costa RP
Nature Communications, 15, 10913, December 2024

Cerebro-cerebellar networks facilitate learning through feedback decoupling
Boven E*, Pemberton J*, Chadderton P, Apps R and Costa RP [* equal contributions]
Nature Communications, 14, 51, January 2023

Single-phase deep learning in cortico-cortical networks
Greedy W*, Zhu HW*, Pemberton J, Mellor J and Costa RP [* equal contributions]
Advances in Neural Information Processing Systems, New Orleans, USA, 2022

A deep learning framework for neuroscience
Richards BA*, Lillicrap TP*, (…) Costa RP, (…) Therien D* and Kording KP* [* equal contrib.]
Nature Neuroscience, 22, pages 1761–1770, 2019