Both in an evolutionary and a developmental perspective, cognition emerges from sensorimotor behaviors that become increasingly invariant and abstract. These sensorimotor origins of cognition explain why mental processes are so intimately intertwined with perceptual and motor processes, and share key properties, including, most prominently, the continuity of processing in state and time, and the dynamic stability of functionally meaningful states.
The mathematical framework of Dynamic Field Theory (Schöner, Spencer, and the DFT research group, 2016) makes these hypotheses explicit by postulating that cognition is based on the activation dynamics of neural populations organized as strongly recurrent neural networks that stabilize neural representations. Instabilities of such neural representations generate the state transitions that build sequences of mental and motor acts.
DFT enables models of cognition in two flavors. (1) Psychophysical experiments can be accounted for in neural process models that instantiate specific cognitive and behavioral competences such as visual attention, working memory, change detection, executive control, and many more. Current research probes how far toward higher cognition such embodied neural process accounts may reach (Richter, Lins, Schöner, 2021). (2) DFT models can also be used to generate behavior and thought in autonomous agents that are situated in structured environment and ultimately produce motor behavior (Tekülve et al., 2019). Learning the concepts of DFT may enable cognitive scientists to broaden the reach of their theoretical work toward neurally grounded models. Integration across many different types of processes from the sensorimotor level to higher cognition is a key element of DFT. This may provide a useful complement to students typical training in computational concepts of system integration. Ultimately, understanding the need for stability in neural approaches to cognition will be a unique insight from this short workshop.