This is a follow up post to Do I have to treat every time I click?
The clicker is commonly thought of as one or more of the following:
–a marker signal
–a bridge signal
–a secondary reinforcer
–a cue
The clicker is often thought of as a marker signal, meaning it marks good behavior. However, just marking a behavior doesn’t strengthen it. This is why the click is paired with a treat. The treat is the positive reinforcer that then strengthens the behavior. (Note: the reinforcer doesn’t have to be food, it just has to be something the animal is willing to work for. For my dog, I can throw a tennis ball instead of giving her a treat, works the same.)
The clicker as a bridge signal. Meaning, the click bridges the gap between when the behavior occurs and when the reinforcer will be delivered. The theory behind this is that the clicker improves training because we can’t be shoving food in the animal’s mouth at exactly the moment the behavior occurs. However, the clicker is a weak bridge, at best. The food still needs to follow fairly close after the click. If the food is delivered 5 seconds after the click, as opposed to immediately, you’ll get all sorts of extra and unwanted behavior developing in those 5 seconds. The clicker doesn’t do a great job as a bridge.
Many people also refer to the clicker as a ‘conditioned reinforcer’ or ‘secondary reinforcer,’ meaning, usually, that the click itself has become reinforcing. If the clicker is actually functioning as a reinforcer, than you should be able to build and strengthen behavior using just the click.
So, try it with something simple. Get an object that you know your animal will take interest in. A feed pan or bucket would work well for a horse. Start by clicking and treating 20 times or so every time the horse touches or approaches the object. Work on building duration if you can. Give the horse a break and come back later or the next day and try it with just the clicker, no treats (or treating sporadically). Can you use the clicker to reinforcer and strengthen behavior for your horse? If so, possibility two is great. If not, I’d stick to possibility one.
A Cue is good description for the clicker. This is based off of fairly recent research being done on animal training at the University of North Texas.
The clicker in itself has absolutely no meaning to the horse. (except, of course, for the occasional horse who it terrified of the sound). We establish that the clicker means something good by pairing it with food. Pretty quickly, the horse learns that the click means–that was great! now here’s a treat. The click becomes a predictor, or cue, for the delivery of food.
So, what happens if we don’t treat every time we click?
We are telling the horse that sometimes the click means “here comes food” and sometimes the click means “keep doing what you were doing, lets keep on training.” A cue will have the strongest meaning if it is always paired with a behavior. (This is why, in clicker training, we usually wait to add the cue until after we are consistently getting the behavior.)
Here’s an example.
For my dog, the leash is a cue for getting to go on a walk. (And she loves walks!) Get out the leash, and she gets excited because she knows exactly what is about to happen. The leash is a reliable predictor of getting to go on a walk.
When I put on my tennis shoes, sometimes it means I’m going to take the dog for a walk. However, other times it means I’m going to go to the gym, or the grocery store, or go to school. The dog pays attention to me, but not in the same way. There is not a clear association between putting on my shoes and taking the dog for a walk.
How does this relate to the clicker? If the clicker is defined as a cue, then for it to have the strongest effect, it always needs to be paired with food. Otherwise, we introduce ambiguity into our training and disrupt the link between the click and the delivery of food. Everytime we click without delivering a treat, the click loses a bit of its meaning. Eventually it becomes like me putting on my tennis shoes, slightly interesting background noise. Since the click in itself has no reinforcing power, we want to make sure we maintain a strong link between the click and delivery of food.
A weakening of this link might not be visible at first, either because the reduction in effect is so slight, or as often is the case, prior training and routines are strengthening and maintaining behavior instead. It can be especially difficult to see if the clicker is teaching or maintaining behavior if we are combining it with prior training, such as methods based on punishment or negative reinforcement (such as pressure/release).
So, the real question boils down to, what function do you want the clicker to serve in your training and what function is it actually serving?
Behavioural Psychology is interesting – I wonder who ‘first’ discovered that method of training horses and other animals (pairing some random thing, eg. a clicker, with both desired behaviour and with treats).
Hi Sarah,
Great question! You’ve reminded me that I really need to do a post on the history of clicker training.
The pioneer of operant conditioning was B.F. Skinner, who worked mainly with pigeons and lab rats.
Some of the first people to work with a variety of domestic species were two of his students, Keller Breland and Marian Breland Bailey.
Mary
great posts Mary- this is a sticky wicket and you’ve done a good job of clarifying. I especially like your tennis shoe analogy and the resulting conclusion!
thanks Jane!
Thanks for the help I was always a little confused when some trainers had indicated the clicker by itself was a reward.
Thanks for stopping by and leaving a comment, Nan. I’m glad to hear that you enjoyed the article.
great posts Mary- this is a adhesive wicket and you've done a acceptable
job of clarifying. I abnormally like your tennis shoe affinity and the
consistent conclusion!