It is becoming increasingly clear in many areas of science that reduction must be complemented by synthesis. While reduction emphasizes decomposition and localization, synthesis is focused on understanding how the organization of the interactions between these components give rise to the original phenomenon of interest. Thus, synthesis is more systemic in nature. Perhaps the best-known example of synthesis in modern science is the burgeoning field of systems biology. After more than 50 years of dissecting living systems into their molecular components, it has become increasingly clear that the challenge for biology in the 21st century is putting these pieces back together into organisms. Accordingly, synthesis requires a rather different skill set from reduction, with modeling and mathematical analysis playing central roles in understanding the complexity of interactions between large numbers of nonlinear dynamical components.
Nowhere is this need for synthesis more apparent than in the behavioral and brain sciences. Converging lines of evidence from psychology, neuroscience, and robotics suggest that brain and mind are embodied, situated, and dynamical. The brain is a highly dynamic complex system, naturally interfaced with the body through sensors and effectors. In turn, our bodies have their own intrinsic dynamics which serve as the spatiotemporal locus of our sensorimotor and cognitive identity, and provide the foundation for even our most abstract metaphors. In addition, both the physical and social environment in which we live is highly structured by, and in turn structures, our cognition. For example, we arrange our desk to offload memory and problem-solving demands, and cooperating groups of people can solve problems that no individual could. There can be no understanding of the nature of human intelligence – how it works, how it develops, and how it can be made – if one “dissects away” the clutter in our environments or the other intelligent agents with whom we are continually engaged. Finally, evidence is accumulating for the importance of continuous temporal processes and interactions across nested levels of analysis and nested time scales.
These developments suggest a new perspective in which behavior and cognition are seen as arising from the dynamical interaction between a nervous system, body and environment (which includes other agents) and can only be fully understood within this broader context. From this perspective, behavior and cognition are conceptualized as arising from the closed-loop interaction of a nervous system with the body and environment in which it is embedded, rather than as the sole product of any one component of this coupled system, such as the brain. Certainly from an evolutionary perspective, it is the behavior of the complete system that is selected for or against, not the individual components. It might seem that taking such a coupled perspective would make an already very difficult problem simply intractable. Studying any one component of a brain-body- environment system is difficult enough, but studying all three components and their interactions simultaneously in any animal is simply beyond our current experimental capabilities, let alone our ability to understand theoretically. However, the trick is to work with simplifications that cut across the traditional brain-body and agent-environment boundaries instead of along them, as disciplinary approaches normally do.
Such an integrated approach requires bringing together many important theoretical threads from across the cognitive, behavioral and brain sciences. Cognitive science and neuroscience have largely been bounded by the human skin, failing to take advantage of the way that feedback through the environment of an agent’s own actions actively structures the sensory information that it receives. On the other hand, ecological psychologists who emphasize the mutuality of organisms and their environments have often failed to address the neural support for perceiving and acting. Many areas of cognitive science, psychology and neuroscience have been slow to embrace the mathematical tools offered by complex systems theory in general and dynamical systems theory in particular as a way to come to grips with the fundamental nonlinearity of the interactions occurring across temporal and spatial scales. Finally, if physical embodiment and situatedness in an actual environment are to play essential roles, then computer modeling must be augmented by robotic models of real systems interacting with real environments. Of course, such theoretical and modeling efforts must be developed hand-inhand with experimental studies of human subjects, with constant feedback between modeling and theoretical work on the one hand and experimental analysis on the other.
This emerging new perspective in the cognitive, behavioral and brain sciences requires a new kind of training experience for graduate students. The central vision for this proposal is to bring together cognitive science faculty at Indiana University doing cutting-edge research at the neural, behavioral, and social levels using both experimental and modeling approaches to develop a unique training program in the dynamics of brain-body-environment systems. Just as symbolic and connectionist cognitive science require proficiency in computation and linear algebra, respectively, in order for students to contribute to research in situated, embodied and dynamical cognitive science, they need to become familiar with the concepts and tools of dynamical systems theory, the mathematical language of change. In addition, they need to learn to think across multiple spatial and temporal scales, so they can understand how phenomena at one level arise from the interactions between components at another level. They also need to learn how to make use of robots as a new modeling medium for embodied cognitive science. Finally, students need to become comfortable with analyzing behavior and cognition across the brain-body and agent-environment boundaries.Objectives
The specific objectives of our training program are:
- To provide trainees with the theoretical and mathematical tools necessary to understand behavior across nested levels of time and space
- To provide trainees with the experimental and methodological tools to study cognition across the brain-behavior, agent-environment, and agent-agent boundaries
- To build a training infrastructure based on continuing interactions and collaboration across leading scientists who use different methods to study related phenomena at different levels of analysis;
- To develop a model training program whose structure of collaborative multi-leveled and multi-method research may serve as a transformative program capable of transcending boundaries of disciplinary culture
- To recruit, train and place in top research institutions a cohort of students who though originally diverse in their starting disciplines become able to think across the brain-bodyenvironment boundaries
- To increase the participation of underrepresented groups in cutting-edge science – by recruiting those individuals to this program and providing the support necessary for success in the program and beyond;
- To systematically evaluate our success in meeting these objectives and to continually adjust the components of the program in response to those evaluations