A number of examples are presented to illustrate a common flaw in the published research on learning, memory, and instruction. Experimental subjects--often college students--have certain expectations about the problems they will be asked to solve and about the questions that will appear on reading comprehension or recall tests; these expectations influence what is learned, noticed, and remembered during the experiment. If the experimental questions, then, are very different from the subjects' expectations, or if the instructions given to the subjects are ambiguous or misleading, poor performance will result. If the learners' expectancies are not consistent with the final testing, then interpretations based upon the research may be flawed. This author has noticed that about half of the papers and published studies of learning have ignored this problem, in spite of previous warnings by other researchers. Perhaps the subtlety of the problem in some situations allows it to be overlooked. (Specific examples are described for situations in which different experimental groups receive different task presentations, and in which they receive identical content but with different instructions. Recommendations for further research emphasize the importance of the subjects' role in an experiment). (GDC)