“Don’t Shoot The Dog” is Karen Pryor’s 1984 book laying out the fundamentals of operant behavioral conditioning for non-scientists, with an emphasis on practical applications in the everyday world.
Like most of the books I’m reviewing in the “52 Self-Help Books in 2019” project, I’ve had this paperback on my shelf forever. I think I came across it in a thrift store (by the way, guys, thrift stores are absolute gold mines of secondhand self-help paperbacks for, like, a quarter apiece). I remember being enticed and intrigued by the cover blurb, a quote of praise from none other than one of the most badass American psychologists ever, B.F. Skinner.
B.F. Skinner was one of the most fascinating figures in American psychology. At a time when clinical psychology was dominated by the psychodynamic theories of Freud, which attributed most human behavior to unconscious sexual and aggressive drives, Skinner sought to get more practical and hands-on in explaining why we do what we do. Skinner strongly believed that, insofar as we can’t directly observe, say, the id, ego, or superego, they weren’t the most useful things to study and speculate about.
Behavior, however, is almost always observable and often precisely quantifiable. Skinner thus centered his experimental efforts on what happens to an animal during and immediately after behavior occurs, and the impact those happenings have on the potential for that behavior to occur again. His work evolved into a uniquely American approach called behaviorism, which, as its name implies, concerns itself with the prediction and control of behavior.
In “Don’t Shoot the Dog,” Pryor, an animal (and people) trainer, sets out to explain to her audience how they might practically apply the principles of behavioral psychology in their everyday lives. She points out that we’re frequently attempting to influence the behavior of the animals around us— including other people— and that there’s nothing inherently manipulative or exploitative about this; it’s just the way society works.
We want our spouses, our kids, our subordinates (or supervisors), our pets, and almost every other living thing around us to do certain things and refrain from doing other things. Thus, why animals do what they do, and what we can do to influence them, is a very practical matter to understand.
Pryor points out that most people, in trying to influence others, clumsily rely upon variations of punishment. We yell; we sulk; we withhold; we lash out; all in response to behavior that has already happened, in hopes that we can keep that behavior from happening again.
Alternatively, we tend to rely on variations of bribery— promising payoffs to behavior that hasn’t yet happened yet, but we want to happen— hoping that we can induce the people and animals around us to do something we’d prefer.
Both punishment and bribery, Pryor maintains, have absolutely lousy track records, both in the laboratory and real life.
The problem with punishment is largely that it happens after behavior has occurred (in the case of the criminal justice system, often times months of years after the fact). Similarly, the problem with bribery is that it happens before behavior has occurred. In both cases, we’re counting on the organism in question to connect, in their minds, the punishment or bribery with behavior that is not currently happening.
The thing is, every scrap of research we have on how behavior works tells us, that’s not how effective conditioning happens.
Effective conditioning happens when the organism that is behaving can alter currently occurring behavior in response to the good thing or bad thing that it is evoking.
Thus: if you want punishment to stick— it needs to happen WHILE the behavior is happening.
Similarly, if you want reinforcement to stick— it needs to happen WHILE the behavior is happening.
There needs to be no question, in the organism’s nervous system, what behavior the good thing or bad thing they’re experiencing is in response to.
Reinforcement or punishment, in other words, need to carry INFORMATION to the animal about whether they should keep doing what they’re CURRENTLY doing— not whether they should do something in the future or whether they should have done something in the past.
Everybody with me?
Pryor goes on the note that, between reinforcement and punishment, it is reinforcement that is by far the more effective conditioner. The research establishes that it is far, far more effective (not to mention, usually more practical, ethical, and pleasant) to condition behavior through reinforcement than through punishment.
Pryor then spends the rest of her concisely written book laying out the principles of effective conditioning via reinforcement. As it turns out, there are rules for how and why reinforcement works— which very few parents, bosses, or pet owners have troubled themselves to learn, as it turns out.
Pryor introduces us to the concept of “stimulus control.” There are certain reinforcers that are “unconditioned stimuli,” things that animals tend to naturally like. Treats and attention are examples of unconditioned stimuli— i.e., stimuli that need no conditioning for an animal to crave or enjoy.
However, it’s not always practical to hand out treats and belly rubs every time you want to reinforce a behavior. The solution to this is to pair, via classical conditioning, an unconditioned stimulus with a “conditioned stimulus”— i.e., making a nose, giving a hand signal, flashing a light, or something else the organism can easily register every time you offer up the unconditioned stimulus.
Soon you’ll find the organism responding to the conditioned stimulus as readily as it responded to the unconditioned stimulus— and, viola, you’ve established stimulus control.
Now, you can reinforce a behavior you want to encourage while it’s actually happening via your conditioned stimulus. That is, when your cat is doing what you want it to do, you don’t have to give it a treat right then; you can use your cool little clicker, the sound of which you’d already linked to treats.
The research tells us that, for whatever reason, once animals make the link in their heads between the unconditioned stimulus (treats) and the conditioned stimulus (clicker), they’ll respond to the conditioned stimulus (clicker) just as readily as the unconditioned stimulus (treats).
This leads us to the concept of behavioral “shaping.” As it turns out, we can’t always get an animal to do exactly what we want it to do, so we can reinforce it for doing what we want. The solution to this is to reward the animal for getting closer and closer to the behavior we want, and reinforcing each step along the way with our conditioned stimulus.
As you reinforce the baby steps, the animal gets closer and closer to doing what you want it to do— all without expecting too much of it at any given time. Viola: behavior shaping.
The research shows that behavior shaping with reinforcement via conditioned stimuli speeds up learning exponentially…and is way, way, WAY more effective than bribery or punishment will ever be.
Pryor tells us about “variable schedules of reinforcement.” All this means is that, in teaching an animal a new behavior via reinforcement, you don’t need to reinforce EVERY correct response after awhile. It’s good to mix it up— keep reinforcing correct responses, but instead of reinforcing every single one, vary the intervals between reinforcement.
This will keep the animal interested and energetic in its responses, insofar as it never knows when, exactly, the treats are coming.
As it turns out, animals that are reinforced every time tend to become lazy and complacent— their motivation dips as they realize they’ve figured this puzzle out.
Pryor offers practical tips for behavior shaping, and most of them boil down to more or less common sense. Keep your expectations within range of the animal’s capability. Try to teach only one part of a behavior at a time, so the animal doesn’t get confused. Know where you’re going, so when the animal learns what you want it to learn you’re not scrambling to figure out what to teach them next. Be willing to be flexible and responsive to what the animal is actually doing, rather than what you wish it was doing. Be consistent and patient, try to end training sessions on a high/encouraging note, and don’t be afraid to go back to basics with the animal when you need to.
In one of my favorite bits, Pryor explains what she calls “The Training Game,” which you can do with your friends. Send one friend out of the room, designate a “trainer,” and decide what you want to train the subject to do. Have the subject return to the room and start doing stuff; reinforce the subject (with something like a bell or a whistle) when the subject gets closer to what you want them to do, thus “reinforcing” them. See how long it takes you to “shape” the behavior of the subject, using only reinforcement.
(If you’ve ever played the “warmer…warmer…colder…colder…warmer…warmer…HOT!” game with someone— that’s a variant of The Training Game. Who knew you were ALREADY using behavior shaping via reinforcement, just like a real behavioral psychologist?!?)
Pryor wraps up her book withs some tips on how to condition an animal OUT of behavior they’re doing that you don’t want them to do, and it’s an interesting collection of techniques.
Turns out, you can train an animal to perform a different behavior that is incompatible with the undesirable behavior; or you can reinforce the animal for doing anything BUT the undesirable behavior; or, perhaps most cleverly, you can actually condition the animal to perform the undesirable behavior on cue from…then just never give the cue.
So…what’s the bottom line here?
I very much believe in positive reinforcement. I completely buy Pryor’s contention that it works way, way better than bribery, punishment, or even negative reinforcement (introducing a negative stimulus that the animal can make go away by performing the behavior you want).
If we lived in a world that was completely and only governed by the power of positive reinforcement, I’d be all on board.
The problem is: we don’t always have access to things other organisms find reinforcing.
Without access to treats, you can’t condition an animal to a conditioned stimulus.
Without access to a conditioned stimulus, you can’t reinforce as a behavior is occurring.
Seriously, think of more than two things you can use to reinforce an animal. You basically have food and attention. Can you think of more than that? I can’t off the top of my head.
Now, think of more than two things you can use to condition a human. You have food, maybe attention, MAYBE sex. And you certainly can’t use all three of those with any given human you want to condition.
I have no quibble with the behavioral science that Pryor is describing. I think the principles of operant conditioning and reinforcement theory are important for every human to understand. Breaking free of our reliance on punishment and bribery is a highly desirable outcome for MOST people who want to influence behavior or any animal, human or pet.
That said: you’re going to have to get creative and observant with those reinforcers.
“Don’t Shoot the Dog” is a quick read, and Pryor is a very straightforward and conversational writer. It’s a good book.
Its only constraints are the constraints of reinforcement theory itself.
Welcome to the real world.