ZiF Research Group
Cognitive Behavior of Humans, Animals, and Machines:
Situation Model Perspectives
October 2019 – July 2020
Convenors: Werner Schneider (Bielefeld, GER), Helge Ritter (Bielefeld, GER)
Recently, a position paper entitled "Enabling cognitive behavior of humans, animals, and machines: A situation model framework" (PDF) by Schneider, Albert, and Ritter has been published.
Recent advances in cognitive neuroscience (CN, the combination of psychology and brain science) have given us new insights about likely core components of cognitive behavior that exhibits the striking flexibility and context-sensitivity that we see in humans and many animal species (e.g., rodents, monkeys). At the same time, progress in artificial intelligence (AI) and robotics, particularly through deep learning and its connection with other machine learning approaches, along with the availability of sophisticated robots, scenarios and datasets, have opened up new routes for synthesizing intelligent functions. These advances have created a strong basis for a converging and cross-disciplinary challenge: to understand how the emerging functional modules need to be connected in order to enable flexible context-sensitive behavior for both natural cognitive agents as well as for robots to live up to what we would expect from truly intelligent systems.
The ZiF Research Group – a "think tank" for AI and CN – brings together an interdisciplinary group of researchers from pertinent fields to approach this challenge of cognitive behavior from the conceptual framework of situation models: a situation model details the required processes and the computational space that together connect perception and memory in the service of the current behavioral demand (task, exploration). Thus, predictions and other forms of manipulating perceptual or memory-based information are considered as key processes of situation models, allowing flexible and context-sensitive forms of action decisions, planning and learning. In order to foster a productive dialogue between research fields and disciplines we will concentrate on basic non-language mediated forms of behavior (e.g., manual manipulation, navigation, search). By the establishment of the ZiF Research Group, we plan to pursue the following research goals: Elucidating the processing architecture (representations and operations) of situation models, clarifying of how task- and exploration-driven behavioral demands interact, advancing highly-controlled experimental paradigms for studying situation models, and spelling out the scientific and societal implications of the generated insights for medicine, philosophy and technology. Four focus perspectives on situation models will guide and organize the research of the group: (1) Working memory as a central gateway for cognitive behavior (2) Situation models and efficient context-sensitive learning (3) Two-systems approaches to the control of cognitive behavior (4) Real and imagined flexible context-sensitive behavior by cognitive maps. Each of these focus perspectives highlights new questions and potential approach directions for answers to our key issues (goals) of the situation model framework. Crucially, we expect strong benefits by tackling these questions from the CN as well as from the AI side, especially when it comes to take to the claims of mutual inspiration of ideas seriously. Organization-wise, besides having fellows and associate group members present at the ZiF (between a few days and several months), two conferences and four workshops are planned.