Using multiple-exemplar training to teach a generalized repertoire of sharing to children with autism.

Marzullo-Kerth et al. (2011) wanted to teach kids with autism to share toys. They used a package called multiple-exemplar training. The package had three parts: watch a short video of sharing, get a gentle prompt to copy it, and earn praise or stickers for doing it.

Three children joined the study. Staff first counted how often each child shared during play. Then they ran the training with many toys. They kept track of sharing every day to see if it rose.

Sharing jumped up for every child right after training began. The kids also shared new toys that looked like the trained ones. One child even shared toys that looked very different, showing wide generalization.

Parents and staff did not have to remind the children. The sharing stuck without extra rewards.

Bauman et al. (1996) used the same three-part package to teach facial and vocal emotions. Both studies got strong generalization, showing the combo works for different social skills.

Rutherford et al. (2003) ran a small group class for greetings and play. Gains stayed in the clinic but barely moved to home. Denise et al. got broader carry-over, hinting that one-to-one MET plus home probes beats a short group format.

Murphy et al. (2010) used extra exemplars when equivalence training alone failed to produce new mands. Denise et al. started with exemplars from day one, so no added rounds were needed.

You can copy this package next week. Pick a social skill your learner lacks, film a 10-second clip of the right move, show it, prompt once, and reinforce. Rotate at least five examples so the child sees the pattern. Track sharing with new toys and peers to be sure the skill travels. If generalization stalls, add more exemplars like Murphy et al. (2010) instead of jumping to a new program.

Multiple exemplar training (MET) teaches a skill using a range of varied examples of the same concept, rather than one fixed instance. Exposure to multiple exemplars helps the learner respond correctly to new, untaught examples.

The goal is generalization. By varying stimuli, settings, materials, and people during teaching, MET makes the learned response more likely to transfer beyond the exact items used in training.

In this study, children with autism were taught to share using a package of video modeling, prompting, and reinforcement across several categories of stimuli. Offers to share increased for all three children and maintained over time.

Generalization was strongest within trained stimulus categories, and sharing appeared in a novel setting with new materials, adults, and peers. Across-category generalization was weaker, a useful reminder to program exemplars from every category you want the skill to cover.