People ask clicker trainers the same questions, over and over. Why can't I just use my voice? Why can't I tell the animal what to do? When can I get rid of the food? Why should I bother to shape the behavior inch by inch when I can just make the animal do it?
Clicker trainers, of course, have practical answers, the answers that fill the pages of Melissa Alexander's wonderful new book, Click for Joy, coming out next month. But behavioral scientists ask the same questions: Is the clicker better? Why can't you just use your voice? Is there really any advantage to shaping a behavior rather than modeling it? ( i.e. pushing the animal through the moves physically). And scientists want more than pragmatic answers from experienced trainers. They want data. They want measurements. They want research designs and results that can't be poked full of holes by other scientists.
The first scientific clicker paper: trailer loading difficult horses
Designing a study that will stand up to scrutiny of one's scientific peers is surprisingly difficult, even with the simplest of clicker questions; I know, because with various other researchers I've been struggling to do so for the last five years.
I'm happy, therefore, to be able to tell you about the publication of what I believe is the first data-based clicker-training research paper to appear in a peer-reviewed scientific journal. The paper describes the thesis research done by Dawnery Ferguson for a master's degree in behavior analysis from the University of North Texas. The authors are Ferguson and her supervising professor, Jesus Rosales-Ruiz. The title is "Loading the problem loader: the effects of target training and shaping on trailer loading behavior of horses," and it was published in the Journal of Applied Behavior Analysis.
Anybody who's watched people trying to load a frightened and unwilling horse into a trailer can testify that it can be quite a scene. To quote the authors, "The combination of a horse that fights loading and an owner who uses physical force can produce a very dangerous situation. Injuries to the trainer can include rope burns, lost fingers, broken bones, or bruises and bleeding. Injuries to the animal can include lacerations to the head from banging into the trailer, scrapes and cuts on the legs, broken legs from falling, or even a broken back if the animal falls backwards while rearing."
How they did it and what the horses learned
Ferguson selected for her study a group of five pedigreed quarter horse broodmares, from 5 to 18 years old, who were all "bad loaders," i.e. taking up to three hours to get into a trailer. Two people were involved, one to handle the horse and clicker and the other to take data, deliver reinforcers, and jot down observations. They also recorded every session on video, and compared the video to the paper records to establish reliability. The ultimate goal was to be able to walk each mare up to the trailer, throw the lead rope over her back, tell her to get in, and stand back while she loaded herself.
They started by establishing, in multiple five-minute sessions, the baseline data, demonstrating that before training none of the horses would go in the trailer when cued to do so. (The authors tersely comment that fighting the lead rope and rearing were ignored.) Working with one horse at a time they clicker conditioned them and trained them to go to a target (a red potholder on a string.) Then they used the target, and the cue "Touch," to teach each horse, in progressive increments, to get into the trailer (a narrow, dark, step-up trailer, the worst kind).
Working every available afternoon and weekend for many weeks, recording every click and every cue and response, they reached their goal with every mare. Most of them got into the trailer in about twenty sessions, but training to confirm and extend the behavior went on for more than 70 sessions in at least one case; a very thorough job. The authors wrote: "Target training and shaping were effective in training the five horses to load into a trailer without the use of punishment or negative reinforcement. The horses' loading generalized to other trailers and to other trainers, including the owner. Although no procedures were implemented to decrease the undesired behaviors that occurred during baseline, they disappeared soon after trailer training began."
Furthermore, the authors point out, undesirable behaviors also decreased outside of the training situation. The mares who once fled to the far side of the field when they saw a person with a halter became more than easy to catch: they came up to the gate and put their heads in the halters voluntarily. They started cooperating around the barn and in other situations. They became, one might say, friendly.
To me, one of the most exciting things about this paper is that the authors end the paper by calling for more research on "these important side effects of training with positive reinforcement."
Curious about clickerwise horses?
- Delightful sampler horse video only $14.95
Want to click your own horse?
- Learn more
Jesus Rosales-Ruiz requires all his students in learning theory and behavior analysis to clicker train an animal as a regular part of their course work. Visiting the UNT campus last year I learned about clicker trained cats, clicker trained dogs of course, and one school teacher who, with her kids, satisfied the course requirement by clicker training a goldfish.
- Behaviour Program
The next masters' thesis from a Rosales-Ruiz student to be excerpted in published papers will be E.J. Fernandez' work with goats in a zoo, identifying in exquisite detail the effect of halter training with food, kindness, and repetition (not much success) and then after adding the clicker (100% success.) The graphs are sensational. When the papers have been published we'll hope to put some of that data on the website.