While psychonomic research would ideally be theoretically grounded in neural process accounts, practical methods to deliver such accounts remain limited, recent advances in neural AI notwithstanding. Dynamic Field Theory provides such methods that have a broad reach from psychophysics to grounded, embodied, and higher cognition. In a gentle introduction, the workshop will give introduce the core ideas of DFT and will illustrate how DFT models and architecture can be built and simulated to account for cognitive function.

A neurally grounded understanding of perception, cognition, and action must be based on theoretical concepts that capture the constraints of neural processing. Neural populations with strong recurrent connectivity provide the level of description with the strongest explanatoy power for understanding the laws of behavior and thought. These populations take the form of neural fields that represent low-dimensional feature spaces which are critical to cognition. Activation states are stabilized by the strong within-population connectivity, enabling neural fields to represent mental states without the need for continuous feedforward input.  The detection, selection, and memory instabilities form the basis for all cognitive processing. DFT architectures are built by coupling different neural fields. Together with the dynamic instabilities, different coupling pattenrs enable cognitive processes such as binding, unbinding, cued selection, as well as transforming representations into different reference frames. DFT architectures reach a wide range of cognitive and sensory-motor processes and may account for experimental signatures in a wide variety of sub-field of psychonomics. 

After a gentle introduction into these core ideas, we will take participants through the process of building and parametrically tuning a DFT architecture using a Jupiter Notebook that participants can follow along. The relatively simple architecture models the autonomous construction of the representation of a visual scene in working memory using spatially bound feature dimensions. 

12-12:30 Gregor Schöner: The core ideas of Dynamic Field Theory

12:30-1:20 Minseok Kang: A step-by-step guide to buiding and simulating a DFT architecture

1:20-1:30 Outlook and discussion