I see, time after time, instructors using physical prompting to teach children with autism, who even given considerable training on the job, still overuse the physical contact, making learning slower and fostering inappropriate stimulus control; prompt dependency an accident waiting to happen. I see that even in videos on the internet that are supposed to show good examples of applied behavior analysis being implemented to teach children with autism. I had guessed a long time ago that it may help if instructors receive training in shaping a behavior "with their own hands", that is, observing the organism behave and making several decisions about exactly when to present the reinforcer contingent on a response that is not yet the final one, but is the one that a) is the closest approximation to that one seen until then; b) with such timing that the approximation itself will not be extinguished until a better approximation is seen and caught, reinforced.
One way to teach well using physical prompting is to use graduated guidance. But graduated guidance is difficult to teach instructors. I first came across this difficulty when developing a self-instructional manual to teach instructors to conduct discrete-trials teaching with children with autism. Even when the participants quickly learned the more rule-governed components such as graduated prompt delay, they had major difficulties combining the rules with the implementation of the physical prompt fading of the graduated guidance. If graduated guidance cannot be implemented well, one might as well use most-to-least or least-to-most so that at least there is consistent control over the prompting and the fading by the instructor.
This is from our manual, Discrete-Trial Teaching with Children with Autism, by Daniela Fazzio and Garry Martin:
GRADUATE GUIDANCE (A VARIATION OF MOST-TO-LEAST)
A prompting procedure called graduated guidance can be used to help a child to perform actions, as opposed to vocal responses. With graduated guidance, hand-over-hand guidance (i.e. physical guidance) is adjusted from moment-to-moment during a trial as needed to help the child perform the task, and subsequent trials typically begin with less guidance than preceding trials. Thus, the goals of graduated guidance are:
a) Within a teaching trial, to use the least intrusive level of prompting necessary to make the behavior occur, which can be adjusted moment-by-moment during the prompting.
b) Across successive teaching trials, to fade out the level of prompts as quickly as possible from most-to-least.
Guidelines for Graduated Guidance:
a) Preliminary assessment to determine the guidance that is required. You should first determine the amount of physical guidance needed to help the child perform each step of the task.
b) During a trial, adjust prompting as needed. Within a trial the level of prompting is adjusted as needed. With the example of Mary above, the guidance within a trial was adjusted from full guidance to partial guidance to a light touch.
c) Across successive trials, fade out the prompts. Across trials, prompts (that continue to be adjusted within each trial as needed) are faded out as quickly as possible from most-to-least.
This post has been the most read in the ABA Blog for a very long time. Years later it appeared on Special Education Advisor: http://www.specialeducationadvisor.com/prompting-using-graduated-guidance/