摘要：The concept of generalized, higher-order, or overarching operant classes has been invoked by a number of researchers when dealing with complex behavior. For example, identity matching, generalized imitation, and relational framing all rest firmly on this concept. However, if the use of the term generalized (or higher-order/overarching) is to carry any explanatory value it needs to be clearly defined. The current article examines two approaches to clarifying this definition. The first approach suggests that generalized operants may be defined in terms of the different orders of contingencies involved, relative to nongeneralized operants, but fails to specify the exact nature of these different orders of contingencies. The second approach suggests that the term generalized should be used in a nontechnical way to emphasize that certain operant classes can only be specified in purely functional terms. This nontechnical definition appears to circumvent some of the thorny problems that arise in attempting to define generalized operants in terms of different orders of contingencies. However, other issues are raised by the latter, nontechnical definition of the generalized operant when it is applied to complex human behavior. These issues are examined and are found to be largely unproblematic.
