Back-chaining, teaching a skill by starting at the end and working back to the beginning, is one of the training tools that clicker trainers use to build highly reliable behaviors. It is a very efficient way to teach, a method that limits the potential for error and leads to fluency with less training time.
Back-chaining has many applications for animal training. For example, the retrieve is an important skill that can be taught "backward" easily. Start by teaching the dog to take and hold the retrieve item, then to give it, then to bring it and give it, then to pick it up, bring it and give it, then to go out, get it, bring it, and give it! Teach each step separately and then put the steps together in reverse order.
The above video by Tonji Stewart, grand prize winner of the 2011 Canis Film Festival, is an entertaining example of training a modified retrieve with back-chaining. This article from clickertraining.com contains a more detailed description of how to back-chain a retrieve. Finally, come to ClickerExp0 2013 and attend the Chaining Behaviors Session with Cecilie Køste to learn more about back-chaining.
The principles of back-chaining are equally applicable to teaching humans. Humans are more confident if they learn the last part of the skill first and if they are always moving toward the part of the behavior with which they are most familiar. For instance, the pianist who learns the last part of the piece first will be moving toward the part he has practiced most and with which he has the greatest confidence. In Click to the Music, a Letter from Karen Pryor, Karen wrote about her experiences with a choir director who understood back-chaining and applied it most effectively.
If you want to learn more about back-chaining in teaching humans, join Joan Orr for a webinar on back-chaining on October 10, 2012, from 2:45 p.m. to 4:00 p.m. (recording available after this date). Register at the TAGteach site.