Somehow, our minds make sense of sensory data in a way that eludes current machine vision systems. We study visual perception in humans using psychophysics and computational models. A secondary interest is in machine vision systems, including deep neural networks. We compare the inferences of humans and machines to gain stronger insights into both biological and artificial seeing.
Topics of current interest include
Ultimately, we aim to understand the computational principles that underlie our brains’ ability to combine sensory inputs to generate holistic perceptual experience.
Why things look as they do
July 1 2022: joined the lab as a doctoral student. She will work with Tom and be part of the teaching staff at the Institute of Psychology and Centre for Cognitive Science. Welcome Swantje! Swantje Mahncke
March 1, 2021: Tom joined the TU Darmstadt as Professor of Perception.
Zimmermann, R. S., Borowski, J., Geirhos, R., Bethge, M., Wallis, T. S. A., & Brendel, W.. (2021). How Well do Feature Visualizations Support Causal Understanding of CNN Activations? Neural Information Processing Systems (NeurIPS). Accepted as a Spotlight (< 3% of submissions). Link
Funke, C. M., Borowski, J., Stosio, K., Brendel, W., Wallis, T. S. A., & Bethge, M. (2021). Five points to check when comparing visual perception in humans and machines. Journal of Vision, 21(3), 16. . Link
Kümmerer, M., Wallis, T.S.A., Gatys, L.A., & Bethge, M. (2017). Understanding Low- and High-Level Contributions to Fixation Prediction. The IEEE International Conference on Computer Vision (ICCV), 2017. LinkModel web service
Wallis, T.S.A., Funke, C.M., Ecker, A.S., Gatys, L.A., Wichmann, F.A., & Bethge, M. (2017). A parametric texture model based on deep convolutional features closely matches texture appearance for humans. Journal of Vision, 17(12):5. Link
We collaborate honestly: we are comfortable telling each other when we don't understand something, we patiently teach through misunderstandings without judgment, and we challenge each other's ideas when we think they're wrong.
Failures and mistakes are an integral part of the scientific process. Be open and honest, transparently discuss, and learn from these for next time. You should never feel pressure to fudge or falsify data.
We strive to be inclusive and equitable. We do not make inappropriate jokes or comments about others based on gender, race, or other personal factors.
We encourage healthy work/life balance and inclusion of lab members with carer duties through clear separation of work and non-work time, and fostering strong time management skills. Lab members are discouraged from being responsive outside their normal work hours.