BrainSpace: Relating Neuroscience to Knowledge About Everyday Life Newton Howard • Henry Lieberman Received: 28 August 2011 / Accepted: 8 July 2012 Ó Springer Science+Business Media, LLC 2012 Abstract Knowledge about the brain and the human nervous system ultimately relates to human thought, per- ception, and behavior. People use natural languages like English to talk about ‘‘common sense’’ concepts, but the brain processes that produce them are described in a highly technical vocabulary. BrainSpace tries to relate technical concepts in anatomy and chemistry to a general body of common sense knowledge, Open Mind Common Sense. It uses the novel inference technique of Blending, to perform joint inference between technical neuroscience knowledge and common sense knowledge, and vice versa. The current paper reports preliminary results showing that Brain- Space’s inference organizes concepts such as the ‘‘visual brain,’’ ‘‘dynamic brain,’’ and deep brain stimulation in an intuitively plausible manner, indicating that it can serve as a foundation for interpreting more specific experimental and medical data. Keywords Brain disorders Á Neuroscience Á Information retrieval Á Data mining Introduction BrainSpace is a new computational method for organizing concepts in brain science and neuroscience. It utilizes novel AI techniques to collect knowledge from natural language sources, as well as inference techniques to semantically organize this knowledge and reason about it. This can permit reasoning from specific experimental results and concepts to descriptions of everyday life and promises to shed additional light on other relevant exper- imental concepts and data. We hope it will, among other things, yield new insights into the impact of specific stimuli, drugs, surgery, or other interventions, on behavior and perception. The proposed research agenda aims to provide researchers and doctors with a diagnostic and advising tool that presents complex information regarding the structure and function of the human brain in a more intuitive manner than is currently available. The overall approach is based on unifying theories, techniques, and tools to exploit the synergy in the information acquired from multiple sources: sensors, databases, expert systems, individual practitioners, and others. In addition to pro- viding the common platform for collection and presenta- tion of brain-related information, we propose to develop a full-scale simulation capability and incorporate data from novel methods of brain research. It is important to be able to causally connect the spe- cifics of brain anatomy and chemistry to the specifics of human perception and motor control. Better tools for integrating, correlating, and interrelating the wealth of knowledge that does exist about the brain, and relating it to human perception and action, hold the promise of con- tributing greatly to our knowledge of neuroscience. As a whole, seeing and making the much needed data available in real-time calls for creating multiple levels of data fusion and general algorithms that enable understanding of the data at hand. With this program accessible, we hope that one will be able to better study this amazing organ gifted to us humans. N. Howard (&) Mind Machine Project, Massachusetts Institute of Technology, Cambridge, MA, USA e-mail: nhmit@mit.edu H. Lieberman Massachusetts Institute of Technology, Cambridge, MA, USA 123 Cogn Comput DOI 10.1007/s12559-012-9171-2