A developmental analysis of rule-following.

The development of rule-governed behavior begins when a class of verbal forms called instructions, in conjunction with other variables, systematically occasion functional consequences for compliance (or noncompliance). These instructional forms then function as specific stimulus controls for compliance, with the other variables potentially taking on conjoint or conditional stimulus control of that compliance. Experience with numerous instructions occasioning functional consequences for compliance can establish generalized compliance to an indefinitely large stimulus class of instructions, a stimulus class that readily and naturally includes those forms that describe relationships between events and response to them that we also call rules. The important conjoint variables may well be the instruction giver (rule stater), any of their other systematic characteristics that have occasioned functional consequences for compliance in the past, and the topography of the response requested. Generalization of compliance across people, settings, and topography can occur; this generalization can extend to self-instruction, and can fail to extend to certain topographies ("Thou shalt not kill"), certain rule staters (nonpeers who say "Don't use drugs" when peers say it is fun), and certain settings ("Don't play with John in class"). Finally, children's experiences with describing behavior can lead to new self-instructions that fit the already established discriminative-stimulus class of rules and instructions for compliance, so that the novel self-instruction produces compliance. Compliance-oriented research has demonstrated how instruction givers can establish instructions as a discriminative-stimulus class for compliance, and how generalized compliance to new instructions and new instruction givers can be accomplished. Correspondence-training research has demonstrated that self-instructions can become discriminative for later performance of the described behaviors. Self-instruction research suggests that these forms of instructions also can control subsequent behavior, and that instructions and rules may be produced covertly as well as overtly. What has yet to be demonstrated is an ability to produce a novel self-instruction that controls subsequent behavior. The argument has been that the development of a behavior class, compliance to instructions, could provide a reasonable explanation for the acquisition of rule-governed behavior. Experiences within the natural environment, as well as research on compliance, correspondence, and self-instruction, suggest that a behavior class, compliance to instructions, could provide a reasonable model of the acquisition and ge