Science of Learning and Augmented Intelligence (SL)

Science of Learning and Augmented Intelligence (SL) supports potentially transformative research that develops basic theoretical insights and fundamental knowledge about principles, processes and mechanisms of learning, and about augmented intelligence — how human cognitive function can be augmented through interactions with others or with technology, or through variations in context. The program supports research addressing learning in individuals and in groups, across a wide range of domains at one or more levels of analysis, including molecular and cellular mechanisms; brain systems; cognitive, affective and behavioral processes; and social and cultural influences. The program also supports research on augmented intelligence that clearly articulates principled ways in which human approaches to learning and related processes, such as in design, complex decision-making and problem-solving, can be improved through interactions with others or through the use of artificial intelligence in technology. These could include ways of using knowledge about human functioning to improve the design of collaborative technologies that have the capacity to learn to adapt to humans. For both aspects of the program, there is special interest in collaborative and collective models of learning and intelligence that are supported by the unprecedented speed and scale of technological connectivity. This includes emphasis on how people and technology working together in new ways and at scale can achieve more than either can attain alone. The program also seeks explanations for how the emergent intelligence of groups, organizations and networks intersects with processes of learning, behavior and cognition in individuals. Projects that are convergent or interdisciplinary may be especially valuable in advancing basic understanding of these areas, but research within a single discipline or methodology is also appropriate. Connections between proposed research and specific technological, educational and workforce applications will be considered as valuable broader impacts but are not necessarily central to the intellectual merit of proposed research. The program supports a variety of approaches, including experiments, field studies, surveys, computational modeling, and artificial intelligence or machine learning methods. Examples of general research questions within scope of Science of Learning and Augmented Intelligence (SL) include: What are the underlying mechanisms that support transfer of learning from one context to another or from one domain to another? How is learning generalized from a small set of specific experiences? What is the basis for robust learning that is resilient against potential interference from new experiences? How is learning consolidated and reconsolidated from transient experience to stable memory? How do human interactions with technologies, imbued with artificial intelligence, provide improved human task performance? What models best describe the interplay of the individual and collaborative processes that lead to co-creation of knowledge and collective intelligence? In what ways do the capacities and constraints of human cognition inform improved methods of human-artificial intelligence collaboration? How can we integrate research findings and insights across levels of analysis, relating understanding of cellular and molecular mechanisms of learning in the neurons, to circuit and systems-level computations of learning in the brain, to cognitive, affective, social and behavioral processes of learning? What is the relationship between assembly of new networks (development) and learning new knowledge in a maturing or mature brain? What concepts, tools (including Big Data, machine learning, and other computational models) or questions will provide the most productive linkages across levels of analysis? How can insights from biological learners contribute and derive new theoretical perspectives to artificial intelligence, neuromorphic engineering, materials science and nanotechnology? How can the ability of biological systems to learn from relatively few examples improve efficiency of artificial systems? How do learning systems (biological and artificial) address complex issues of causal reasoning? How can knowledge about the ways in which humans learn help in the design of human-machine interfaces?