Before the present century, the primary means of studying animals was observation of the form and effects of their behavior combined with presumption of their intent. In the present century, ethologists continued to emphasize observation of form and replaced presumption of intent with the study of proximate function and evolution. In contrast, most learning psychologists minimized both observation of form and presumption of intent by defining behavior in terms of simple environmental effects and establishing intent by deprivation operations, We discuss advantages of the use of observation in the study of learning, examine arguments that it is unnecessary, irrelevant, and unscientific, and consider some practical considerations in using observation. We conclude that observation of the form of behavior and concern with its ecological function should be an important part of the arsenal of techniques used to study learning.Observation has been the dominant method of studying the behavior of animals since the beginning of recorded history (Warden, 1927 (1918) and Heinroth (1910Heinroth ( /1985, and experimentalists (see Lubbock, 1882; C. L. Morgan, 1900;Small, 1899;Spalding, 1872). Further, these written accounts represent only a fraction of the daily observations made by hunters, gamekeepers, animal trainers, herders, farmers, and amateur naturalists (Barber, 1980;Mountjoy, 1980). In one sense, the historical prominence of observation is uninteresting because of a lack of alternatives. Drawings, imitations of movements and sounds, and stories of intentionalactions by animals have given way only recently to the precise recording of behavior by the use of microswitches, photocells, cameras, video and audio recorders, and neurophysiological techniques. Observation, though, was considerably more than a less exact precursor to precise, automated measurement of behavior. Observers influenced the nature of their data by frequently focusing on the form of behavior and by inferring the function or intent. Many were drawn to the perceived similarities between human behavior and the behavior of nonhuman an- (Warden, 1927), anthropomorphically viewing the latter as partially disguised people enmeshed in a web of human goals, social relations, and rules (Aesop, approximately 620 B.C.; Selous, 1908). Other observers adhered to a more animal-centered description and inference of function (Craig, 1918;Huxley, 1914).The purpose of the present paper is to briefly review the historical and current treatment of observation as it relates to the study of learning in animals, focusing particularly on the observation of response form and the inference of function. We will consider advantages of the observation of form and inference of function and important objections and impediments to their use. Our conclusion is that observation should be encouraged in the study of learned behavior, not at the price of rejecting the rigor of traditional laboratory techniques, but for the advantages that observation of form and inference of function hol...