The illusion of choice: Frequency, Quantity, Quality, Chronology…

Frequency, quantity, quality, chronology. State of being, state of mind, status of needs… The variables influencing our dog’s behavior are innumerable. When we can be aware and thoughtful of the variables, however, our training will improve dramatically.

In a given situation, where & when multiple behaviors are possible, The Matching Law is the equation which predicts the relative frequency of each behavior. It is the underlying mechanism which drives all operant conditioning. In laborotory situations, where variables can be measured and controlled, the relative frequency of behavior, even with complicated variables, can be measured quite precisely in non-human animals (humans too, but not as precisely; we do things for reasons that are harder to quantify, like religion, morals, politics, guilt, spite, pride, etc). Even in real life, with all the different possible variables, the matching law determines how often, relative to other related possibilities, different behaviors occur.

Choice is an illusion. How animals perceive their environmental options, and which among those options they do, or do not, engage with, is a result of their biology (which they did not, and cannot, “choose”), and their environment, which is not within their control.

When we inaccurately (and commonly; I do it, too) describe an animal’s action as “choice” or “decision,” we are referring to the sum of a behavioral equation, the answer to which is theoretically (and often, actually), knowable. This equation includes multiple factors:

What is subject’s hereditary proclivity? If we throw a tennis ball in front of two dogs, and the retriever retrieves, while the Malamute ignores it, did the dogs makeninvidual choices? Or was their behavior a result of breed traits caused by unnatural selection?

How many times have the possible behaviors been reinforced?

What was the relative quality of the reinforcers?

What was the relative quantity of the reinforcers?

How recently were these behaviors reinforced, relative to the other possible behaviors?

How it applies:

From the ABA Online Glossary: “CONCURRENT SCHEDULES OF REINFORCEMENT. Two or more schedules of reinforcement (e.g., FR, VR, Fl, VI) are simultaneously available. Each alternative is associated with a separate schedule of reinforcement and the organism is free to distribute behavior to the schedules.”

So, a dog training example would be:

available schedules of reinforcement:

–eating high value food from a puzzle feeder.

–performing a recall to the dog’s main handler.

–countless other possibilities including sniffing, FRAP/zoomies, reclining, napping, spinning, rolling, licking, cavorting through nearby tunnels, moving as far away as possible from handler, etc, etc.

A few variables (there are countless) to consider regarding the puzzle behavior:

How hungry is the dog? Are they sated? Ravenous? How often, and how recently, have they eaten from puzzles? How about puzzles like this one? How about this one in particular? How valuable was the food they worked out of the puzzle(s) in the past? How frustrating were the experiences? Did it hurt? How valuable is the food in the puzzle at this moment? Does the dog know it? Is the puzzle near anything scary? Is it near anything that is attractive?

A few variables (there are countless) to consider regarding the recall behavior: How many times has the dog been recalled successfully? How many times was their recall unsuccessful? Were the successes reinforced, and if so, how many times, how strongly, and how recently? Was it this handler who reinforced them? How many times did the dog not come? How were they reinforced for whatever they did instead? How much energy is expended by running to the handler?Are there other current possible behaviors they have been reinforced for in the past when called? How often, how recently, and how severely has the dog been punished in general, and by this handler in particular? How many times has the dog been punished after coming when called? How many times by this handler? What was the severity? How recently?

Regarding other possible behaviors, all of the above questions from either category might apply. The more free the animal is to “distribute behavior,” the more accurate the equation will be.

So:

*dog has eaten lots of great stuff from puzzles including this one, not generally been hurt, and frustration has been low. The dog eats something good, from some puzzle or another, every day. The puzzle is currently filled with dehydrated buffalo liver, enough to keep the dog busy for 30+ minutes.

*dog is recalled successfully 15-30 times each week, and is always reinforced with high value food. Dog is not generally subjected to intentional P+ by this handler or anyone else. Overall approximate success to failure recall rate in last 5 years is 100:5. Subject has not missed a recall in 10 months. Handler plans to reinforce recalls in this exercise with portions of dehydrated buffalo liver large enough to keep dog chewing for 3-5 seconds per recall.

Discounting the possibility of some other unpredicted schedule having enough history to compete with the two predicted most likely behaviors, which will the dog perform, and under what circumstances?

so…

If the dog described above is given the puzzle, and called away 20′ a minute later, will the dog recall? If they are rewarded with the same liver that is in the puzzle, will they recall again? How many times, if any, will the dog recall from the puzzle? What is the predicted ratio of success?

The possible relative reinforcers and punishers in the above scenario are innumerable. In the accompanying video, it appears the dog is enjoying the trot in each direction, so may be experiencing reinforcement for the travel, but at another time, he may have experienced the trot as a punisher.

While the matching law does not inform what precise behavior will occur in any one sequence, it does predict the relative frequency within a set of possible behaviors.
If we learn to understand the science, if we can account for our subject’s history, and control elements of our environment, training is more predictable, more efficient, and less stressful for the dog and handler alike.

Rudy is not “choosing” to leave the puzzle, and come to me. His recall behavior is the result of training over the course of many hours, spread across months of practice. If he does not recall, that was not a choice, either. His training (for which I am responsible), is not yet sufficient for the circumstances into which his handler (me) placed him.