Book summary “Don't shoot the dog!”: summary

Introduction

You often struggle to manage stubborn behavior in pets, children, and colleagues using ineffective scolding and punishment. These coercive methods create fear, confusion, and resistance without producing lasting changes in habits.

You can transform human and animal behavior reliably and humanely through the scientific principles of positive reinforcement and shaping rather than punishment and coercion.

Karen Pryor demonstrates how operant conditioning provides precise tools to modify actions without pain or anger. The book explains the mechanics of reinforcement, variable schedules, and complex behavior chains. You will learn the art of shaping new skills through successive approximations, establishing strict cue control, and applying non-coercive methods to eliminate unwanted actions.

How can you time a reinforcer for maximum impact? What are the ten laws of shaping? How do you eliminate bad habits without using punishment?

1. The Power of Reinforcement

A reinforcer is any event that increases the probability of a behavior recurring, provided its timing and size are carefully managed.

Karen Pryor demonstrates that reinforcement operates like a feedback loop in real time. It is distinct from vague rewards or punishments because it directly strengthens the preceding behavior when timed correctly.

Defining Reinforcers

Memorize the definition of a reinforcer as the secret of good training.

A Wall Street lawyer in his late forties started saying attaboy for every good squash shot and even patted himself on the back during solo practice, rising four rungs higher on his club ladder. This simple shift from self-criticism to positive reinforcement took two weeks to yield noticeable improvements in his game.

Negative reinforcement involves removing an aversive stimulus when the desired behavior occurs, which can tame skittish animals.

Llama trainers use a clicker to mark standing still, using the retreat of the scary human as the negative reinforcer until the animal accepts touch within five to ten minutes.

• Ask the subject to stand still as you approach to a certain distance.
• Click the clicker when the subject stands still.
• Turn and go farther away to remove the aversive stimulus.
• Gradually decrease the distance as the subject tolerates the approach.

Timing and Size

The timing of a reinforcer provides critical information to the learner about exactly which action earned the payoff. Delayed reinforcement reinforces whatever action the subject is performing at the moment the reinforcer arrives, often leading to confusion or reinforcing the wrong behavior.

Zoo keepers failed to move a gorilla out of a doorway because they tossed food when it sat in the path, essentially bribing it rather than reinforcing spontaneous exiting.

• Check whether you are reinforcing too late whenever you encounter difficulties in a training situation.
• Have an observer watch for late reinforcers if you are caught up in the action.

Reinforcers should be kept as small as possible to minimize eating time and allow maximum repetitions before satiation. Large reinforcers slow down training sessions because the subject spends too much time consuming them.

A keeper at the National Zoological Park hired as a consultant in 1979 gave a panda whole carrots, resulting in only three reinforcers in fifteen minutes, whereas single carrot slices would have sped up the process.

• Use reinforcers as small as you can get away with, such as a grain of corn for a chicken or a quarter-inch meat cube for a cat.
• Divide normal daily rations into about eighty reinforcers if conducting multiple training sessions a day.

A jackpot is a surprisingly large reinforcer that can mark major breakthroughs or reset resistant subjects. An unexpected, massive reward breaks through behavioral inertia or uncooperativeness, abruptly restarting active responses or cementing a difficult new behavior.

When a dolphin researcher's subject spent twenty minutes offering no responses, tossing two fish for nothing startled the animal into becoming active again.

• Deliver a reward roughly ten times larger than normal as an unannounced surprise to mark a breakthrough or revive a stalled training session.

Conditioned Reinforcers

Conditioned reinforcers are initially meaningless signals paired with real reinforcers to communicate exactly which behavior is being rewarded when immediate delivery of a primary reinforcer is impossible. Because a trainer cannot deliver food instantly during midair behaviors or underwater actions, a neutral sound or motion acts as a bridge.

Marine mammal trainers use a police whistle because it is heard underwater and leaves hands free, while dog and dolphin trainers frequently use a dime-store party toy called a cricket.

• Start practical animal training using positive reinforcement by establishing a conditioned reinforcer.
• Pair the meaningless signal with food, petting, or real reinforcers while the subject is doing nothing in particular.
• Observe when the subject visibly startles upon perceiving the signal and begins seeking the real reinforcer.

Dr. Ogden Lindsley called the click an event marker because it identifies the exact behavior being reinforced. A conditioned reinforcer acts as an event marker that puts control into the hands of the learner and signals the termination of a behavior.

Police officer Steve White used a click to signal a German shepherd searching for a six-foot-tall bush to raise its head, helping it catch the scent at head height and locate the target.

While a terminating click ends the behavior, complex tasks require an interim reinforcement stimulus that tells the trainee they are right but must keep going. A keep-going signal is an interim reinforcing stimulus that indicates a correct action without terminating the behavior chain.

In a dog agility competition, an owner yelled Yes! as a dog looked toward the correct jump while running an obstacle course, causing the dog to veer instantly and take the right obstacle.

• Insert an interim reinforcing stimulus somewhere before the terminating click.
• Use the signal to communicate specific information within a behavioral chain without stopping the chain.

A conditioned aversive signal warns the subject that something bad will happen unless they stop their current behavior, making corrections more effective than sudden threats.

A friend accidentally cured her cat of clawing the couch by shouting No! right as she dropped a large brass tray next to the cat, startling it severely. Two repetitions ended the behavior permanently.

• Say no as the dog does the wrong thing when using a choke chain.
• Pause before yanking on the chain to give the dog a chance to avoid the aversive by changing its behavior.

2. Schedules and Complex Behavior Chains

Constant reinforcement is only needed during learning stages. To maintain a learned behavior reliably, reinforcement must be delivered randomly. Variable schedules prevent extinction and drive persistence, which is the psychological foundation of gambling and slot machines.

Once a behavior is learned, switching from continuous reinforcement to an unpredictable variable schedule maintains the behavior far more effectively.

An old clunker car that rarely starts on the first try creates long, variable schedule behavior where a driver tries for half an hour, unlike a reliable new car whose behavior extinguishes immediately if it fails once.

• Provide a high frequency of reinforcers to beginners during the learning stage.
• Stop reinforcing every successful performance once the behavior has been acquired.
• Switch to using reinforcement only occasionally on a random and unpredictable basis.

When subjects must work for a predetermined time or number of actions for reinforcement, the early responses in the series are never reinforced, causing effort to dwindle. On long tasks, this creates the delayed-start phenomenon where subjects procrastinate before beginning.

James Thurber avoided starting his writing by lying on a couch all morning reading a book with one hand while tapping typewriter keys at random with the other.

• Introduce a reinforcer just for getting started on long-duration tasks.
• Break down long journeys or tasks into initial sequential steps and reinforce the completion of each early step.

Both humans and animals develop superstitious habits when a reward happens to coincide with an unrelated action, such as chewing a pencil during an exam or performing an elaborate sports windup. Traditional trainers and professionals often rely heavily on superstitious practices because they cannot separate working methods from accidental traditions.

A baseball pitcher performs a strict nine-step sequence including touching his cap, touching the ball to the glove, wiping his ear, and scuffing his foot before every pitch.

• Introduce variations in all circumstances that do not matter once a behavior is partially trained to prevent accidental conditioning.
• Identify and systematically examine personal habits, gear attachments, or dependencies to eliminate superstitious behaviors.

Behavior Chains

Complex sequences of behaviors are maintained because each step's cue acts as a conditioned reinforcer for the previous action. Behavior chains break down if any step lacks stimulus control. Training them from the last step backward ensures the learner always moves from unfamiliar tasks into well-learned, highly reinforced territory.

A New York friend taught his golden retriever to play Frisbee by training the retrieve first, then catching, and finally the chase.

• start with the last behavior in the chain and make sure the signal is recognized
• train the next-to-last behavior
• continue working backward until the entire sequence is linked

3. The Art of Shaping

Shaping is the process of developing new or complex behaviors by taking a small existing tendency and shifting it step-by-step toward an ultimate goal. Because living things exhibit variable behavior, you can establish intermediate goals. Start with any presently occurring behavior and reinforce small increments in the right direction. Raise criteria gradually through successive approximation.

Shaping develops complex new behaviors step by step through successive approximations governed by ten strict operational laws.

The author trained a chicken to dance by first reinforcing any movement to the left. She then selectively reinforced quarter turns, and finally half circles until the chicken performed full turns rapidly. Effective training relies on understanding underlying principles rather than just following established methods. Arthur Godfrey, an expert horse trainer, became angry when the author suggested that a horse could be trained to bow without whips and ropes using positive reinforcement and targeting. This illustrates how tenaciously people stick to traditional methods.

The Ten Laws of Shaping

You must follow specific rules to keep training efficient and prevent frustration. The ten laws of shaping govern pacing, setbacks, and criteria:

• Raise criteria in increments small enough that the subject always has a realistic chance for reinforcement.
• Train one aspect of any particular behavior at a time; don't try to shape for two criteria simultaneously.
• During shaping, put the current level of response onto a variable schedule of reinforcement before adding or raising the criteria.
• When introducing a new criterion, or aspect of the behavioral skill, temporarily relax the old ones.
• Stay ahead of your subject: Plan your shaping program completely so that if the subject makes sudden progress, you are aware of what to reinforce next.
• Don't change trainers in midstream; you can have several trainers per trainee, but stick to one shaper per behavior.
• If one shaping procedure is not eliciting progress, find another; there are as many ways to get behavior as there are trainers to think them up.
• Don't interrupt a training session gratuitously; that constitutes a punishment.
• If behavior deteriorates, go back to kindergarten; quickly review the whole shaping process with a series of easily earned reinforcers.
• End each session on a high note, if possible, but in any case quit while you're ahead.

A father confiscated his teenage son's motorcycle due to poor grades. When the boy improved to Ds and Cs, the father withheld the bike because it wasn't enough improvement, causing the boy to stop working entirely.

Shaping Shortcuts

Professional trainers use shortcuts to accelerate learning. Targeting involves getting an animal to touch its nose to an object and moving that object. Mimicry leverages the natural tendency of certain species to copy others. Modeling involves physically pushing a subject through a motion, though it requires combining with shaping to be effective.

A chimpanzee watched a poodle being taught to sit on a child's rocking chair and rock it for a piece of ham. After that, the cat unprompted performed the exact same behavior to earn a reward.

4. Stimulus Control and Cues

Stimulus control means that a learned behavior is reliably executed immediately and precisely upon presentation of a specific signal.

True stimulus control requires meeting four strict conditions: the behavior always occurs immediately upon the signal, never occurs in the absence of the signal, never occurs in response to a different signal, and no other behavior occurs to that signal.

Establishing Cues

In operant conditioning, a cue should be introduced only after the behavior is already securely established and offered reliably. Instead of forcing an animal into a position before it understands, trainers shape the behavior first and then introduce the cue as a green light signaling an opportunity to earn reinforcers.

At a seminar, an owner took a clueless four-month-old black Labrador puppy and successfully taught it sit, down, roll over, come, a high five, and a retrieve within twenty-four hours using clicker training.

You can establish the cue effectively through specific actions:

• produce the cue just as the behavior is starting and reinforce completion, gradually backing up the cue in time;
• alternate between cue and no cue, reinforcing on-cue sits and extinguishing off-cue sits;
• shape response to the cue as if it were behavior in itself.

A temporary drop in performance to zero often precedes mastery as the subject shifts from unaware conditioned response to conscious awareness of the cue. When learners realize a signal determines reinforcement, they may experience a frustrating transition period marked by nonresponses, leading to anger or a prelearning temper tantrum. Researcher Michael Walker found that tuna in sensory-discrimination experiments got so upset when left unreinforced for more than forty-five seconds that they jumped out of the tank.

Signal Fading and Limited Holds

Once a learned signal is established, it can be made smaller and smaller until it is barely perceptible while still eliciting the full response. Unlike unconditioned stimuli where response intensity matches stimulus size, a learned cue only needs to be recognized, enabling trainers to fade signals down to subtle movements. A parrot at the San Diego Wild Animal Park cackled in hysterical laughter in response to a tiny, imperceptible movement of the trainer's hand.

Teaching a learner to touch a physical target with its nose or paw provides an easy discriminative stimulus to guide movement and location. Targeting is widely used by marine mammal trainers, zoos, and special educators to move animals or humans voluntarily without physical guidance. A cat owner in Holland clicker-trained her house cat to target on a coffee spoon, enabling her to lead it all around the breakfast table.

You can implement targeting with these steps:

• hold out a target stick or object for the learner to touch;
• reinforce the touch, then raise criteria in small steps;
• move the target around to lead the learner through obstacles or locations.

A limited hold sets a specific time window in which a cued behavior must occur to be reinforced, drastically reducing response time. By timing the normal response interval and progressively shortening it while withholding reinforcement for sluggish responses, trainers can achieve maximum promptness. At Sea Life Park, trainers used a stopwatch and limited holds to shrink the performance interval for spinner dolphin aerial acrobatics from fifteen to twenty seconds down to two and a half seconds.

You apply limited holds through these actions:

• estimate the normal interval in which the behavior usually occurs;
• reinforce only behavior that occurs within that interval;
• gradually tighten the time window as responses adjust.

5. Untraining Without Punishment

Any attempt to stop an unwanted behavior is a variation of eight core methods, ranging from extreme measures like capital punishment to positive ones like training incompatible behaviors or changing motivation.

You can eliminate unwanted behaviors effectively using eight specific methods ranging from extinction and incompatible behaviors to motivational changes.

The Failure of Punishment

Method 1 eliminates behavior by physical restraint or removing the subject, but it teaches nothing and allows no relearning. Karen Pryor took her cat to the pound to be put to sleep because it urinated on stove burners despite having outdoor access.

Use temporary confinement like shipping boxes or crates as a standard aid to housebreaking puppies when you cannot watch them.

Punishment is a clumsy training method that often fails and leads to an endless escalation of severe penalties and negative side effects. The author's parents punished her exactly twice with scoldings, once at age six for pilfering and once at age fifteen for skipping school.

People often resort to punishment not to modify behavior, but out of a desire for revenge or to establish and maintain dominance.

Ask yourself if you want to alter a given behavior, if you want revenge, or if it is a dominance dispute.

Negative reinforcement involves halting an unpleasant stimulus through behavioral change, but every instance inherently contains a punisher. A cow touches an electrified fence, feels a shock, and pulls back, learning to avoid the fence to prevent future shocks.

Disconnect the TV or withhold dinner until the laundry is picked up, ceasing the negative reinforcer when compliance is obtained.

Non-Coercive Elimination Methods

Extinction allows a behavior to die down by itself for lack of any reinforcement, working best on verbal behaviors like whining, teasing, or quarreling. Conductor Thomas Schippers ignored woodwind and violin foolery during a New York Philharmonic rehearsal, and the naughtiness quickly extinguished.

Inform a whining child that whining does not work, and make haste to reinforce with praise or a hug when they stop.

Method 5 uses positive reinforcement to train a behavior that is physically incompatible with the unwanted action. At Sea Life Park, the author stopped a six-hundred-pound bottlenose dolphin named Apo from harassing a swimmer by training Apo to repeatedly press an underwater lever for fish rewards.

Train a dog to lie in the dining-room doorway during meals instead of begging at the table, rewarding the behavior with food in the kitchen afterward.

Bringing an unwanted behavior under stimulus control means it can be triggered by a cue and tends to extinguish in the absence of that cue. The author trained a dolphin named Makua to sink to the bottom on cue, after which sinking stopped being a problem during blindfold training.

Make as much noise as possible in the car on command, repeating two or three times until the noisy behavior palls and ceases.

Method 7, or DRO, shapes the absence of an unwanted behavior by ignoring the misbehavior while reinforcing any other positive or neutral activity. The author used Method 7 and Method 4 on weekly phone calls with her invalid mother, letting complaints extinguish with noncommittal sounds while enthusiastically rewarding any non-complaint topics after twenty years of conflict and within two months.

Ignore complaints and distress with neutral responses like ah and hmm, while enthusiastically responding to queries, weather, books, or friends.

Method 8 eliminates unwanted behavior by identifying and altering the underlying motivation, which is often the kindliest and most effective approach. A mother stops supermarket tantrums by feeding her hungry child before or during shopping rather than jerking on the child's arm.

Feed kids before or while going to market to prevent distress caused by hunger and the proximity of food.

10 Key Ideas

1. A reinforcer is any event that increases the probability of a behavior recurring, provided its timing and size are carefully managed.
2. The timing of a reinforcer provides critical information to the learner about exactly which action earned the payoff.
3. Conditioned reinforcers are initially meaningless signals paired with real reinforcers to communicate exactly which behavior is being rewarded when immediate delivery of a primary reinforcer is impossible.
4. Once a behavior is learned, switching from continuous reinforcement to an unpredictable variable schedule maintains the behavior far more effectively.
5. Shaping develops complex new behaviors step by step through successive approximations governed by ten strict operational laws.
6. Complex sequences of behaviors are maintained because each step's cue acts as a conditioned reinforcer for the previous action.
7. Stimulus control means that a learned behavior is reliably executed immediately and precisely upon presentation of a specific signal.
8. Punishment is a clumsy training method that often fails and leads to an endless escalation of severe penalties and negative side effects.
9. You can eliminate unwanted behaviors effectively using eight specific methods ranging from extinction and incompatible behaviors to motivational changes.
10. You can transform human and animal behavior reliably and humanely through the scientific principles of positive reinforcement and shaping rather than punishment and coercion.