There are at least three powerful insights for educational technology researchers and designers from recent neuroscience studies of the brain and from cognitive science research findings: First, our brains learn and process two very different types of knowledge; non-conscious, automated, procedural, or implicit knowledge, and conscious, controllable, declarative knowledge. Evidence also suggests that we believe we control our own learning by conscious choice, when in fact nearly all mental operations are highly automated, including learning and problem solving. Thus, first, educational technology designers must focus more on the teaching of procedural (application) knowledge. Second, human beings have a very limited capacity to think during learning and problem solving, and when that capacity is exceeded, thinking and learning stop without us being aware. Thus, designers must strive to avoid cognitive overload by focusing all presentations on essential information to be learned. Third, nearly all of our instructional design and learning theories and models fail to account for the influence of non-conscious cognitive processes and therefore are inadequate to deal with complex learning and performance. Evidence for these points is described and their implications for instruction and the learning of problem-solving and higher order thinking skills are discussed. Models of learning and instruction that appear to help overcome some of these biological and cognitive barriers are described. In addition, suggestions for new research questions on interactive technology-based learning environments that take account of the three insights are also described. (Contains 1 table.)