Here’s another remark often addressed to reinforcement based trainers, sometimes in a mocking tone, sometimes seriously:
The writer often further implies that to do that with children would be the worst sort of bribery, indulgence, and permissive parenting (and-by-the-way-it’s-responsible-for-all-the-current-evils-of-society). And we’re being just as weak willed when training our dogs!
But the “cookie” objection is so easy to address. People who write this, with all due respect, don’t have much of a clue about how positive reinforcement works. And the misunderstanding has existed for decades.
Let’s let B.F. Skinner handle it this time. Yes, this (mis)perception has been around that long! In the following excerpt from his Review Lecture: The Technology of Teaching, 1965, Skinner is referring to a case study about a boy with “childhood schizophrenia” (the terminology of that time) who had had cataract surgery. He was resistant to wearing his glasses and was taught to wear them through positive reinforcement with an external treat.
Another objection is to the use of contrived contingencies of reinforcement. In daily life one does not wear glasses in order to get food or point to circles in order to receive chocolate. Such reinforcers are not naturally contingent on the behavior and there may seen to be something synthetic, spurious, or even fraudulent about them….
It was necessary to use a “spurious” reinforcer to get the boy to wear glasses, but once the behavior had been shaped and maintained for a period of time, the natural reinforcers which follow from improved vision could take over. The real issue is whether the teacher prepares the student for the natural reinforcers which are to replace the contrived reinforcers used in teaching. The behavior which is expedited in the teaching process would be useless if it were not to be effective in the world at large in the absence of instructional contingencies.
…In American education it is commonly argued that a child must be taught nothing until he can reap natural benefits from knowing it. He is not to learn to write until he can take satisfaction in writing his name in his books, or notes to his friends. …Unfortunately the teacher who confines himself to natural reinforcers is often ineffective, particularly because only certain subjects can be taught through their use, and he eventually falls back upon some form of punishment. But aversive control is the most shameful of irrelevancies: it is only in school that one parses a Latin sentence to avoid the cane.
There is a whole world of relevance to dog training in those few paragraphs. Skinner is making a distinction between “contrived” reinforcers (treats) and what we would call life rewards in dog training, and describing making the transition between them. And he also points out that those teachers and trainers who do not successfully use “contrived” reinforcers when natural reinforcers are not available fall back on one thing: punishment.
This applies perfectly to animal training as well.
So, let’s discuss a little more the criticism of the very idea of giving children a treat for tying their shoes (and applying that disapproval to dog training). On top of everything else we have the societal nervousness about commonalities in learning between humans and other animals to contend with.
First, the kids. Newsflash: it is not really such a bad idea at all to use external reinforcement for tasks that are not intrinsically meaningful or rewarding to children.
Tying shoes is a great example. During the first stages of learning this mechanical skill, the process may not be intrinsically rewarding. The child may be too young to understand or experience the advantages of knowing how to tie her own shoes. The physical task can be quite a challenge. Reinforcing the individual steps with something more potent than praise, like bits of a favorite food or counters that can be swapped for an item or a privilege, can aid the process. And once a child can successfully and consistently tie her own shoes, the act will be reinforced by physical comfort and independence, both potent reinforcers. She will not need or be expecting raisins, Skittles, or chocolate chips anymore.
There is, of course, a whole discipline of positive reinforcement training for humans using markers: TAGteach. Teaching with Acoustical Guidance is clicker training for humans. It is used in a variety of contexts, frequently with children with autism but also all sorts of older kids and adolescents learning a sport or other physical skill. The youngest children are indeed given treats for performing behaviors; slightly older kids may get counters or chips they can cash in, and teenagers are generally working for the satisfaction of succeeding at the task. (It certainly appears on the videos that getting to select a counter–a bead or a bean or a pebble–and put it in a bowl is pretty fun for kids all on its own.)
Here is a boy being taught to tie his shoes using TAGteach, and he counts a bean into a bowl for every click. But my favorite TAGteach video shows the exact moment of a child transitioning from Skinner’s “contrived” reinforcers to intrinsic reinforcement. Be sure and watch to the very end of the video of the little boy learning to get in the swimming pool. He is somewhat fearful. At the beginning he is getting Skittles, but then something very interesting happens.
However. Many people find the idea of using food treats, stickers, or tokens when teaching children shocking, even repellent. Like dogs, they are supposed to eagerly apply themselves to tasks that have no intrinsic value to them just because we want them to. In the world of cultural fog described by Susan Friedman, people tend to react to punishment as a perhaps unfortunate but inevitable part of bringing up children, but run screaming at the idea someone deliberately might use reinforcement to affect a child’s behavior. Oh, the horror!
Second, the dogs. There is one particular difference between teaching tasks to children and behaviors to dogs. The important tasks we teach to children will eventually become socially or intrinsically reinforced. These reinforcers are often not even recognized as such by the critics of teaching with reinforcement. A smile or nod from a parent. A “thanks” from a stranger. Physical comfort. Control over one’s environment. Encouragement from a teacher. Passing a test. Performing well in a job interview. We are social animals and sensitive to social success and acceptance. Folks who quote the “cookie” remark above generally do not recognize the reinforcement that is naturally available and going on all the time for us social humans. (That includes negative reinforcement, of course. A child may choose to make his bed to get his parents to stop nagging him about it. But the bigger point is that many things that are chores for children are naturally positively reinforcing when they get older.)
Dogs do not “grow up” to get the social reinforcement or the joy of fitting into our culture and society. However, Skinner’s plan to switch to non-contrived reinforcers works for them, too. Teaching a dog to walk on a loose lead using positive reinforcement may take a lot of treats (or even play) at first, but a skilled teacher can show the dog that learning to walk nicely on a leash expands their world. The dogs can transition to the life rewards of going places, exploring, and sniffing new things.
Here is Skinner again in the same article, about children:
The application of operant conditioning to education is simple and direct. Teaching is the arrangement of contingencies of reinforcement under which students learn. They learn without teaching in their natural environments, but teachers arrange special contingencies which expedite learning, hastening the appearance of behavior which would otherwise be acquired slowly or making sure of the appearance of behavior which might otherwise never occur.
Adapted by me for dog training:
The application of operant conditioning to dog training is simple and direct. Training is the arrangement of contingencies of reinforcement under which the dogs learn. Dogs learn without human trainers in their natural environments, but trainers arrange special contingencies which expedite learning, hastening the appearance of behavior which would otherwise be acquired slowly or making sure of the appearance of behavior which might otherwise never occur.
Doesn’t that sound familiar? The un-doggy behaviors we want from our dogs would not initially have natural reinforcement without our intervention. Most would be acquired “not at all.” So we arrange reinforcement for them. But in many cases, just like for kids, some natural reinforcement will fall into place as well as our dogs enjoy their lives with us.
Thanks for reading. Have a cookie.
This post is part of a series:
- But Purely Positive is a LIE!
- But It’s Unhealthy to Protect Your Dog From All Stress!
- But Every Dog is Different!
- But What if Your Dog Runs Out Into Traffic?
- There’s No Such Thing As Positive Training Because….
- Why We Need to Keep Talking about “The Quadrants”
- When Management Succeeds
- Level 1 Breakfast (quick behavior drills)