Models and Methods of cognition an perception
We aim to formulate our theories as models that make concrete predictions for human behavior and develop statistical methods to fit and evaluate those models on data. To do so, we leverage the large methodological overlap between modern machine learning and statistics tools and models how humans might understand the world.
Models and methods
Models
We develop models of human perception and cognition, aiming to develop models that work for complex situations. For perceptual processes this implies models that can take arbitrary images or videos as input. This allows us to study how we are able to reduce the complexity of the world around us to a level that allows prediction, memory, planning and cognition in general.
To achieve this we use methods from deep neural networks and machine learning and artificial intelligence in general.
Methods
To evaluate models of complex behavior quite advanced statistical methods are necessary. In particular, the high dimensionality of the data and the flexibility of the models we use require special attention.
When we develop better methods, we always aim to publish them as useful toolboxes for others to use in their research as we have done with psignifit and the RSA toolbox before.