C.B. Ferster and Marian K. DeMyer, “A Method for the Experimental Analysis of the Behavior of Autistic Children,” American Journal of Orthopsychiatry 32, no. 1 (1962):89-98.
Complete original source available here.
This was among the very first research reports claiming to show that the behaviors of autistic children could be shaped and changed according to the principles of reinforcement and operant conditioning. Applied behavior analysis originated in these principles. The subjects of the experiment described by Ferster and DeMyer were three children—10-year-old Thomas, 11-year-old Margie, and 3-year-old Patrick—who all lacked meaningful speech. They were part of a special program for autistic children in a children’s ward of a psychiatric hospital.
The experimental room contains a large number of devices which when operated either by a coin or direct key provide some rewarding consequence for the child. These devices include: a pinball machine; a pigeon and trained monkey both trained to perform only when the animals’ compartments are lighted; a color wheel giving a kaleidoscopic effect; a television set; a phonograph; an electric train whose speed and direction the child can vary; an eight-column candy vending machine with a separate light and coin slot in each column so that the child can choose the particular candy; a second vending machine which can deliver small trinkets or small packages containing part of the child’s lunch (both the trinkets and food were varied from day to day and from subject to subject depending upon the subject’s preference); a telephone handset with music through the earpiece; an electric organ; and a 35 mm. slide viewer. Figure 1 is a schematized drawing of the experimental room. The room contains a one-way vision screen on the wall facing the experimental devices.
… Food and candy appear to be the major reinforcers available, and candy was therefore the reinforcer used during the child’s first introduction to the experimental procedures….
Once it proved possible to sustain a simple activity for substantial period of time and under conditions of infrequent reinforcement, we began to develop more complex forms of activity. The general plan here was to choose a variation in the child’s activities in the direction of the desired repertoire and shift the reinforcement contingency in that direction. We attempted to choose slight variations so that reinforcement would not occur too infrequently. Once the child’s performance changed to conform with the new contingencies of reinforcement, reinforcements were delivered only for further variations in the direction of the required performance….
In a first procedure, the child’s behavior was placed under the control of the lights behind the plastic panel where the key was mounted, by delivering keys only when the panel was lighted. This control was developed without difficulty in all of the subjects with whom it was attempted. A second kind of stimulus control, developed by making the coin slots and associated lights inoperative when the coin slot was not lighted, was developed with more difficulty. This control was carried out by lighting the coin clots after the delivery of every nth (2nd to 5th) coin…..
The further development of more complex repertoires was carried out by reinforcing “matching to sample”. Instead of a simple key, the child faced three windows (see Fig. 1), each producing an electrical connection when pushed. A device behind the windows programmed a strip of paper on which could be painted or pasted any kind of visual material. The subject was first trained to respond to the sample appearing in the center window. Touching the sample in the center window tended to force the child to attend to the stimulus and produced the second frame. The sample reappeared in the center window with a matching figure either to the left or to the right and a nonmatching figure in the remaining position. If the child touched the matching figure, a coin was delivered. If he touched a nonmatching window, the apparatus was disconnected electrically (time out) for a period variously ranging from 1 to 20 seconds…
The early results of this experiment, using the techniques of operant reinforcement to sustain and widen the repertoire of autistic children, show that it is possible to bring the behavior of these children under the close control of an artificial environment by means of a conditioned reinforcer possibly generalized. After sustaining simple performances it was possible to widen the behavioral patterns of the child by the normal processes by which behavior is sustained and altered in normal humans and in other species. While the behavioral repertoires developed in these children are still not nearly as complex as those involved in normal social repertoire, they indicate at least the existence of normal processes at a very basic level….
Perhaps these analyses can serve as guides for attempts to use the same processes of developing behavior in social situations where the performances sustained and altered would be activities in respect to other persons (social) and where the important consequences sustaining the activities would be the social effects of these performances.