This paper focuses on the notion of human learners as being structure-determined. Alternate views of educational phenomena and educational research of G. Cziko (1989), D. Bohm and F. D. Peat (1987), H. Maturana and F. Varela (1980, 1987), and S. E. Pirie and T. Kieren (1989, 1990) are considered. A mathematical theory of chaos, random behavior/order, and how growth in knowing takes place are discussed. It is proposed that because human behavior is determined at the level of any unity, it is unpredictable. What this unpredictability means for education research is considered. According to structure-determinism, since knowing is recursive and not simply iterative and since the knower is a closed system, a person can know things on many levels and in many ways at once. Thus, a child's behavior (or that of any knower), although determined, is unpredictable both because it is not caused by a stimulus and because the level of response or knowing action cannot be foretold by an observer. It is suggested that education research should shift its focus from viewing educational phenomena as linear successive orders to viewing them as implicate orders. A person's understandings/knowledge can be observed as a recursively numerous set. A theoretical pathway exists whereby an observed understanding can be modeled or placed in a set of understandings. Researchers and teachers should look for patterns over time that might give them a sense of the structures of the children/students with whom they work. Although knowing is unpredictable, the hope and aim of cognitively guided instruction is not a lost cause. Educators cannot seek to cause things by instruction, but to provide experiences (and evaluations), which show their regard for children as individuals, recognize the complexities of their structure, and point them in potentially useful directions. (RLC)