A fundamental distinction in the analysis and interpretation of behavior.

Abstract
In 1935 the editor of this journal called attention to limitations inherent in the presentation of grouped or average data (Dunlap, 1935/1982). Because of contemporary misconceptions analogous to those Dunlap discussed, we clarify and emphasize the important distinction between the use of grouped data to demonstrate learning in a species and the reliable measurement of learning in identified individuals, as well as how best to assess the outcome of selective breeding with conditioning as the phenotype.