The three most common training methods of teaching sequences of behaviors are forward chaining, backward chaining and whole task training. In chaining methods, training involves gradually teaching the sequence by teaching increasingly longer subsequences until the entire sequence has been learned. In forward chaining, training involves gradually teaching the sequence from the beginning. Backward chaining involves gradually teaching the sequence from the end. Training is continued in this manner until the entire sequence is learned. Whole task training does not involve gradual learning. Instead, on each trial, the sequence is attempted in its entirety. Trials of this kind continue until the sequence is learned. It has been hypothesized that backward chaining is the superior method to teach sequences of behaviors because reinforcement is given at the end of each subsequence (Martin & Pear, 1988; Gilbert, 1962a & b). An example of this reinforcement would be the ball going through the hoop at the end of the sequence of behaviors involved in making a basketball shot. Random chaining consisted of teaching randomly chosen behaviors within the subsequences. Component behaviors were placed next to each other within the subsequence according to sequence order. Random-2 chaining was similar to random chaining except that the component behaviors were taught within the subsequences in the order in which they were selected. The final subsequence is an exception as it is the sequence taught in sequence order. Random-3 chaining was similar to random-2 chaining except component behaviors were selected with replacement. For each subsequence, selection starts fresh and the required number of behaviors are selected out of the eight possible behaviors. Whole task training produced more true errors than each chaining method except random-3 chaining. Random-3 chaining produced more true errors than forward, backward, random and random-2 chaining. No significant differences were found between random-3 chaining and whole task training. The results clearly did not support the hypothesis that backward chaining is superior to other training methods for the sequence of behaviors taught in the experiments. The discussion involves an analysis of the training methods into a number of factors which may influence their effectiveness and some suggestions for future research. (Abstract shortened by UMI.)
Identifer | oai:union.ndltd.org:uottawa.ca/oai:ruor.uottawa.ca:10393/5614 |
Date | January 1990 |
Creators | Scott, Janine Mary. |
Publisher | University of Ottawa (Canada) |
Source Sets | Université d’Ottawa |
Detected Language | English |
Type | Thesis |
Format | 551 p. |
Page generated in 0.0017 seconds