Wednesday, March 9, 2016

How do dogs learn?

It's a reasonable - even essential - question to ask before embarking on a journey to help them learn our language, live in our world, and coexist peacefully in our quirky social structure. The fields of learning theory, canine cognition, and canine ethology are all constantly-evolving sciences, but there are some fundamentals of learning theory that every pet owner should familiarise himself with in order to properly apply the techniques taught in basic classes.

Classical conditioning: by pairing a neutral stimulus to an existing stimulus that already elicits an emotional response, we can change the subject's emotional response to the formerly neutral stimulus. Does Pavlov ring a bell?

Operant conditioning: altering the frequency of a behaviour by controlling its consequences. It all boils down to five categories. Four of these are commonly referred to as the four quadrants of operant conditioning because they are easy to represent in a chart. The fifth is extinction: a behaviour that is not reinforced will decrease in frequency.
Artwork by Lili Chin

We can't mention the work of B.F. Skinner (widely considered to be the father of operant conditioning) without taking into account nature's role in shaping behaviour. Like many a good PhD student, Keller Breland's claim to fame arose from questioning the shortcomings of Skinner's work. He wrote The Misbehavior of Organisms as a direct argument against Skinner's The Behaviour of Organisms, both excellent reads. The gist of it? Nature plays a role as well as nurture, so always be aware that dogs will be dogs, cats will be cats, and humans will be humans.

A primary reinforcer is something that an organism innately desires and is motivated to obtain. The primary example for almost any species - certainly for dogs and humans - is food. The earliest distinguishable behaviours in newborn puppies and human infants all revolve around the acquisition of milk. Other notable primary reinforcers include social interaction and play in its various forms.

secondary reinforcer (bridge/marker) is a neutral stimulus, such as a distinct sound, that has been paired with a primary reinforcer via classical conditioning so that the secondary reinforcer begins to elicit the same response as the primary reinforcer. In dogs, the most commonly used secondary reinforcers are a verbal marker and a clicker, but other stimuli can also be used, such as a gentle touch for blind dogs or a light flash for deaf dogs.

General process of training behaviours

  1. Choose a marker that you can use to precisely mark a behaviour the instant it happens. You can use a clicker or a verbal marker, such as "yes", "good", or "click".
  2. Pair the marker with a primary reinforcer.
  3. Obtain or wait for a behaviour or a loose approximation of the behaviour (more on that later!).
  4. Mark the instant the behaviour occurs, then reward promptly, causing the behaviour to become more readily offered.
  5. Repeat until the dog is offering the behaviour without prompting after each reward, at least 8/10 times. 
  6. Add a verbal (or other) cue after each reward so that it precedes the behaviour that the dog is now offering.
  7. Practise in new environments, only changing environments or adding distractions once your dog has mastered the new challenge with 8/10 reliability. Don't expect your new dog to respond to his name while romping about the dog park even though he does it perfectly in class! Set him up for success.
Let's put this together into a practical scenario:
  1. Gather your dog, clicker, and treats (such as his regular meal) in a non-distracting environment at breakfast time.
  2. Wait for your dog to make eye contact. The instant he does, click, then treat. Ensure that the click is the earliest predictor of the treat - do not make any motions toward the treat bag or bowl before clicking, or these stimuli will "block" the effectiveness of the click.
  3. Once the dog reliably looks back up at you after eating each treat, at least 8/10 times, begin to say his name before he looks at you.
  4. Practise the name game with very mild distractions until you again reach 8/10 reliability. The first step you will take is to say his name while he is not looking at you but also not particularly interested in anything else going on. By reinforcing baby steps and adding in some advanced concepts later, you will eventually be able to command your dog's full attention if he takes off after a deer or is engaged in a wrestling match with a dog at the other end of the field!

No comments:

Post a Comment

About the Harenbergs

Audrey, Aipa, and Steve Harenberg on the Blue Ridge Mountains

We build strong working relationships between dogs and handlers through the use of modern learning theory and the development of technological aids.

Audrey has been involved with numerous non-profit organisations, as foster mom, therapy dog handler, and obedience instructor. In addition to teaching private lessons, group classes, and in-home training, she's busy creating free blog articles and books to help make progressive dog training accessible to all.

Steve is finishing his PhD in computer science and will be taking over technological development once he is freed from academia.