The Stages of Learning

A series of articles for professional dog trainers, those who want to become professional dog trainers, and those who want to become certified dog trainers.

A concept that is particularly useful for trainers is what we call the stages of learning: acquisition, fluency, generalization and maintenance. Familiarity with these stages of learning will help you be a more efficient trainer because you will have a deeper understanding of the learning process.

Acquisition is a term for the change that takes place when learning happens. This is a little tricky, because you can certainly argue that change (and consequently learning) is taking place during all of the stages of learning. However, when trainers refer to acquisition, they are talking about the beginning stages of an animal learning a new behavior – the time during which the animal begins to understand that certain antecedents predict certain consequences for a particular behavior.

Chance (Learning & Behavior, 5^th ed., pg 36-37) prefers the word “change” to acquisition because he feels that learning doesn’t always require the acquisition of something. Chance uses the example of the pecking order in a flock of chickens. He asserts that by pecking at the other chickens in the flock you establish your position within the flock; once you’ve established your position, you give up pecking at the higher ranking chickens. Perhaps one could argue that the chicken has acquired the knowledge that pecking at higher ranking chickens is futile and possibly dangerous.

Regardless, in training, acquisition is a very commonly used term and one we should understand.

The next stage of learning is fluency. “Fluency is a combination of errors and rate; it is the number correct per minute.” (Chance, Learning & Behavior 5^th Ed., pg 46)

Fluency tells you how well the animal can perform the behavior under given circumstances. For trainers, this is where record keeping comes in very handy, as you can always determine the fluency of a behavior by taking a rate-per-minute measurement. The work of Breland and Bailey informs us that the optimum time to raise criteria is when you have an 80% fluency rate.

So let’s do a “case study.” Your dog, Fido, has been acquiring the “rollover” behavior under the following circumstances:

• trained by you
• hot dogs for reinforcement
• trained 7 days in a row
• trained in your living room
• trained every morning between 8:30 and 8:45
• trained with low-to-no distraction (i.e., no people, no other pets, etc. in the room)
• current criteria = trainer in upright position, stimulus is a circling motion with the index finger pointing out, compliance within 2 seconds of stimulus

One evening you go to dinner at your in-law’s house and decide to take Fido so you can show off the new trick. Just before leaving, you do a few practice rollovers with Fido, while your wife waits patiently. Below is a chart showing Fido’s fluency under the various circumstances.

What this tells you is that your dog is fairly fluent in “rollover” at your house in the morning, somewhat fluent in the evening, and not fluent at your father-in-law’s house.

“Generalization is the tendency for a learned behavior to occur in the presence of stimuli that were not present during training.” (Chance, Learning & Behavior 5^th Ed., pg 451)

Fluency is closely related to generalization. Fluency will tell you how well generalized the behavior is. From the above example with Fido, we know that Fido is fluent in the behavior at 8:30am in your living room, somewhat fluent at 5:00pm in your living room, and not fluent at 6:00pm in your father-in-law’s living room.

What’s different about each of these situations? Well, let’s analyze it:

• Out of six variables, three are constant – you as the trainer, the stimulus for the behavior and the reinforcement for the behavior.
• One variable – location – is the same in Scenario #1 and Scenario #2.
• Two variables are different in every scenario – time and outside stimuli. We could probably come up with a lot more outside stimuli, but these are the obvious ones, so we’ll stop with them to keep it simple.

Fido performed best in Scenario #1, which is the scenario under which the acquisition stage and most of his training took place, and under which he is most familiar with the behavior. He does reasonably well in Scenario #2, where the only differences are the presence of the wife and the time of day. In Scenario #3, he pretty much falls apart – different location, different time of day and lots of distraction.

By measuring Fido’s fluency in each scenario, we can determine how well generalized he is in the rollover behavior. Ideally, you will “re-train” Fido under each new scenario; as he learns the behavior under new circumstances, it will begin to generalize, and he’ll learn to perform the behavior on cue regardless of circumstances. Once he’s generalized the behavior, he should be reliable under a variety of circumstances – familiar and new.

This does not mean that he will be reliable under all circumstances – there will always be situations that are too distracting. However, we can train for many circumstances, and certainly for the circumstances to which our dogs are most often exposed.

Last, but not least, the behavior must be maintained. “Maintenance is the continuation of the conditions that generated a performance. The analysis of maintained performance, as a subject matter, is different from but not incompatible with that of acquisition (e.g., many experiments concerned with effects of schedule parameters on performance do not really begin until acquisition has been completed).” (University of South Florida Behavior Analysis Glossary, http://www.coedu.usf.edu/abaglossary/glossarymain.asp?AID=5&ID=2238)

All behavior must be maintained. Remember – behavior has function. We behave for a reason, and if there is no consequence to our behavior, it’s going to extinguish. If, after the debacle at your in-law’s house, you don’t ask Fido to rollover for six months, chances are, when you do ask him to rollover (even under Scenario #1), he won’t perform. His behavior has not been maintained. It shouldn’t be as hard to elicit the behavior as it was originally, because he has already learned it, but it will still take some of practice to get it to fluency.

Raising Canine has a school for dog trainers which focuses on operant training for dogs, dog behavior, working with clients and addressing client compliance, and the science behind behavior modification.