About us

Responsive image Despite huge advances in artificial intelligence (AI), the mammalian brain is still unrivaled in terms of sustainability and speed of learning, and robustness in inference. One central goal of AI research is to build intelligent systems that exceed the capabilities of biological brains. However, to date we know very little about how computations in neuronal circuits give rise to biological intelligence.

Our group uses AI both as a testbed and a tool on large scale neuro-physiological and -anatomical data to better understand the constituent elements of neuronal intelligence. We are inspired by the idea that a deeper understanding of computational motifs in cortical circuits can help build the next generation of intelligent systems.

We are based at the University Göttingen and the University Tübingen as part of the Cybervalley initiative. We closely collaborate with experimental and computational neuroscientists to develop new tools and experimental paradigms to discover principles of biological intelligence.


News

February 2024

On Tuesday, Mohammad successfully defended his thesis. Congratulations Dr. Bashiri!

Mohammad defending

October 2023

A special episode about our workshop “How can machine learning be used to generate insights and theories in neuroscience?” is now online. Thanks for Paul Middlebrooks for moderating the panel discussion and making it into a podcast episode. The workshop was joint work with the Ecker lab. Thanks for Mohammad, Michaela, and Pavi for leading the effort!

September 2023

Our paper “Energy Guided Diffusion for Generating Neurally Exciting Images” was accepted at NeurIPS 2023. Joint effort with many contributions from the lab, led by Pawel.

September 2023

Suhas’ and Konstantin’s paper “Taking the neural sampling code very seriously: A data-driven approach for assessing generative models of the visual system” was accepted at NeurIPS 2023.

January 2023

Konstantin’s and Mohammad’s paper “Bayesian Oracle for bounding information gain in neural encoding models” was accepted at ICLR 2023.


Collaborators


Affiliations