Jozwik Lab

We want to understand how the brain processes information about the visual world, makes meaning of what we see, and how visual and semantic perception of the world may differ across people, including those with mental health conditions. The lab’s research has focused on probing specific visual dimensions in the context of face, animacy, and object representations more generally.

We study visuo-semantic cognition combining cognitive science, neuroscience, and computational modelling. We collect and analyse human behavioural (including virtual/augmented/mixed reality), physiological, and brain imaging (fMRI and M/EEG) data. We also analyse macaque electrophysiology data obtained through collaborations and perform cross-species comparisons. We use machine learning techniques for neural data analysis and computational modelling with a special interest in biologically-inspired deep learning and AI models (NeuroAI). The computational models we work with include vision deep learning models (including topographical, recurrent, or developmentally inspired models), multimodal vision and language models, and Large Language Models.

We are also starting to use non-invasive neurostimulation – Focused Ultrasound Stimulation – FUS to understand the causal role of brain regions involved in visual and semantic perception. We are also exploring the use of generative AI and mixed reality for exposure therapy.

What do behaviour and computational models miss about how the brain processes what we see?

Journal of Neuroscience, 2023

To what extent can models guide experiments and predict which faces we consider more similar?

PNAS, 2022

Is “animacy” one dimension in the brain — or many, and where and when is it represented?

Communications Biology, 2022

Open questions

Which behaviourally-relevant visual and semantic dimensions shape visual perception?

How does the brain represent animacy when we interact with people and objects?

How does the visual perception change when we are immersed in the environment?

To what exent are visual and semantic representations different in mental health conditions?

Your question