In the language of modern dog training a clicker is a “secondary reinforcer.” Sorry, but the term “secondary reinforcer” is simply eight syllables of blather. Why? Because reinforcers must, by definition, strengthen a behavior connected to them. That isn’t what happens when you click a clicker (or say the word, “NO”) and then not provide the tangible event that is supposed to follow. I can prove it. No, I have proven it – at MIT no less. You can prove it too, in the privacy of your own home. No lab coats, no rat or pigeon cages needed. Just you, a clicker and a hungry dog.

Starters: Get some treats that your dog likes. Click the clicker and feed the dog a treat. Repeat about 20 times. Now get some rest.

Day 2: Start teaching any behavior. Preferably not “sit” – you’ll find out why later. You can choose something as simple as having the dog touch your finger or roll-over or anything you want. Try to build up some duration – about five seconds. So, if it’s down, you will click at the five second mark.

Day 3: Repeat what you did on day two exactly as you did the first time. Now try something new. After a string of successful clicks and treats, wait until the next repetition and just click – no treat. Now wait. Watch the next repetitions closely. You aren’t going to give any more treats for awhile. Watch what happens. The behavior changes. A click without a treat doesn’t “reinforce” anything. That is because the function of the click is to identify/ limit the behavior, not to strengthen it.

The problem with the nomenclature of behavioral science is that it doesn’t match the phenomenon known as behavior. Because of this, they fail to understand the processes connected with learning and maintaining behavior. As they do not actually train animals, they are free to invent words and arbitrarily toss them around regardless of whether they work. Consider that most “modern” trainers who use “scientific” methods use the term secondary reinforcer as if it means “surrogate” reinforcer. That is a red herring – a false assumption that leads away from the real point of a clicker.

The Real Function:
The best metaphor I can find to explain the clicker is the word “filter” combined with the word “hitch”. The clicker takes a snapshot of a momentary event and all the things uniquely associated with that event. Over a series of repetitions, that snapshot is compared and the number of unique items is reduced – that is the filter aspect. The end goal is to connect a specific behavior to a specific event – to “hitch” the unique factors to a specific consequence. EG: If I “sit” I get a treat, regardless of whether a bird flew over or a siren sounded in the distance. If anything in the environment consistently precedes the opportunity to sit for a treat, it will also be connected in the sequence. Scientists call this a “discriminative stimulus” because by being able to discriminate some aspect the environment apart from others, the dog increases its chance of getting a good outcome. Most folks think of it as a command, cue or signal. I like that. (Ogden R. Lindsley, in my opinion the most brilliant behavior analyst in history, cautioned his colleagues in his last communication to refrain from using unfamiliar words. Og said to stick to plain English if you want to communicate with people. Sage advice.)

So, what does the clicker do? It creates associations fast. It is not the force that connects it or the force that gives the behavior retention in the dog’s repertoire. It defines the behavior. Period. When you try to use it as a surrogate for tangible rewards, it causes an effect with another fancy science term – operant variability. I just call it “wiggle.” Meaning if the dog has learned that the click is associated with tangible rewards, then a click that is not followed by the pay-off triggers an adaptive wiggle. The animal immediately starts to vary its behavior to adapt to the perceived lack of function of the learned association. Meaning you put a quarter in a Coke machine and punched the select button and nothing came out. Will you automatically put another quarter in the machine? Three? Four? Nope. That is because the disconnect between the learned signals and the actual reward causes you to do something different. You might put money in another machine. You might kick the machine or hit it. You will do something different. But wait! The secondary reinforcers of the sight of the machine and the lighted button are supposed to strengthen your behavior, right? Dead wrong.

The reason that behavioral scientists never figured this out is because they never study animals doing more than one behavior. Even studies of operant variability are done on rats pressing levers in mathematical combinations. They can’t see the wiggle because they don’t record the variations of behavior other than a single press of the lever. To read about an experiment by B.F. Skinner and Og Lindsley that proves this statement, take a look at this blog post. http://clickandtreat.com/wordpress/?p=191 and to understand operant variability, take a look at this – http://clickandtreat.com/wordpress/wp-admin/post.php?post=33&action=edit. To understand the clicker without a treat better, here’s another one. http://clickandtreat.com/wordpress/?p=197

By golly, follow that set of links and you are going to be so smart your friends will hate you. If you are wondering about the MIT reference, I was a consultant there on an artificial intelligence project. I also participated in a symposium where I demonstrated how a click without a treat trashed a well crafted behavior. I do it all the time. That is because I know how this stuff works apart from arbitrary terminology invented for rats and pigeons in tiny little boxes. i.e. It’s my job.