Two experiments were conducted in order to investigate the retardation of autoshaping following pretraining with intermittent US (reinforcer) presentations, and to examine the "competing response" hypothesis from Staddon's ideas about "adventitious reinforcement". Six homing pigeons were given 4 stages of training: (1) Unsignalled food delivery on a fixed time (FT) schedule of 12, 24 or 36 sec. (2) Autoshaping with a fixed intertrial interval (ITI). (3) Automaintenance on the fixed ITI with contextual cues changed. (4) Automaintenance on a variable ITI. Experiment 1 (stages 1 and 2) was done in a black wooden chamber, and Experiment 2 (stages 3 and 4) was in a green Plexiglas chamber. The average values of the interfood interval (IFI) for each subject were fixed throughout the 4 stages. Ten response classes that occurred during the IFI were recorded. It was found that: (1) Autoshaping and automaintenance of key-approach responses were retarded by prior FT training. (2) When retardation was found, "superstitious" responses, developed during the FT schedule, continued to occur even when food delivery was signalled. Most of the responses were magazine-directed behaviors. (3) Even in the presence of competing responses, the effects of the signal (CS) on behavior were clear (e.g. emergences of key-directed responses, concentrations of terminal responses during signal presentation). (4) A control group of 6 additional pigeons were given only stages 2, 3 and 4. They autoshaped faster and responded at a higher rate than the pretrained pigeons during stages 2 and 3 but not during stage 4. The key-approach responses of the pretraining group increased on the variable ITI, not context-changed, automaintenance. Implications for theories of retardation and "performance" of autoshaping were discussed.
We propose a notation system for displaying the meaning structure of a narrative story. As a first task the basis for representing narrative events and states in memory was investigated by analyzing readers' identifications of types of basic statements and their causal inferences between statements. Results suggested that the notation should contain six types of nodes (statements), three types of directed arcs (possible causal connections), and the embedded 'mental worlds', each of which incorporated the desire, intention, belief, memory, and other mental activities of each of the narrative characters.
Autoshaping, defined as the development of the stimulus-directed action through positive stimulus-reinforcer contingency, has several novel features which have necessitated a serious reconsideration of certain principles of both classical and operant conditioning. These features include: (1) The stimulus-reinforcer relation plays a major role in both acquisition and maintenance of the autoshaped response. (2) The autoshaped response is directed toward the signal of reinforcer delivery. (3) The autoshaped response is similar in form to the response elicited by the reinforcer. It was shown that these features are difficult to explain on the basis of operant reinforcement alone. Although all of these features can be considered within a framework of the classical conditioning of skeletal actions, the Pavlovian notion of stimulus substitution is not applicable to certain autoshaped responses. Two recent hypotheses were discussed which extend the stimulus substitution principle and further emphasize the phylogenetic origins of autoshaped responses. In conclusion, autoshaping research has not only challenged the operant reinforcement principle as a general account of all instrumental learning but also expanded the scope and applicability of classical conditioning to various signal-directed skeletal actions.