FUNDAMENTALS OF ANIMAL LEARNING LECTURE NOTES

Transcription

1 FUNDAMENTALS OF ANIMAL LEARNING LECTURE NOTES phone fax Suzanne Hetts, Ph.D., CAAB, CPDT Dan Estep, Ph.D., CAAB 1

2 LECTURE NOTES FOR FUNDAMENTALS OF ANIMAL LEARNING TELECOURSE PHONE FAX Suzanne Hetts, Ph.D., CAAB, CPDT Dan Estep, Ph.D., CAAB COURSE OUTLINE Week I What is learning and how do we know it s happened? Defining operant conditioning Contrast briefly to other types of learning such as habituation, sensitization and classical conditioning The Four Quadrants Positive reinforcement Negative punishment Positive punishment Negative reinforcement Practice categorizing events Week II Using Positive Reinforcement What to use for reinforcemen t Primary reinforcers Secondary reinforcers Premack Principle When to deliver reinforcement Schedules of reinforcement Interval schedules Ratio schedules Duration schedules Techniques using positive reinforcement Prompting and fading Lure reward Target training 2

3 Catching Shaping Week III Using negative punishment Types of negative punishment Time out Extinction DRO differential reinforcement of other behavior DRI differential reinforcement of incompatible behavior Counter commanding Operant counter conditioning Week IV Using positive punishment Criteria for effective use What to use for punishment Conditioned punishers Social punishers vs. remote punishers The ethics of using punishment A decision tree Summary and Wrap up 3

4 LECTURE NOTES FOR FUNDAMENTALS OF ANIMAL LEARNING TELECOURSE PHONE FAX Suzanne Hetts, Ph.D., CAAB, CPDT Dan Estep, Ph.D., CAAB WEEK I INTRODUCTION In order to change an animal s behavior, it is necessary to know how an animal learns. Understanding how an animal learns requires more than just common sense. It often also requires a shift in perspective from How can I get my pet to stop-----, to How can I get him to do what I want. The former usually gets people thinking about what they can use for positive punishment (which is usually NOT the tool of choice), while the latter forces people to think how they can use positive reinforcement to encourage the right behavior. If your goal is to modify behavior for any reason, you must have an understanding of both operant and classical conditioning. This course will cover only the former; classical conditioning and other behavior modification techniques will be covered in future telecourses. GLOSSARY Learning - A process by which a relatively permanent change in behavior is produced as a result of specific experiences. Learning can t be observed directly, only its effects on behavior. You must see a change in behavior (performance) to infer that learning has occurred. Physical injury, maturation and fatigue can all produce changes in behavior as a result of experience but they are not learning. Latent learning from being exposed to a particular situation without any apparent benefit (no reward or punishment), an animal is then able to use this experience later in a beneficial way. The animal learned something from the exposure, but the learning wasn t expressed until the benefit was made available. Habituation - A decline in a response as a result of the repeated presentation of a stimulus. Example - A dog in a cage at an animal shelter initially jumps or cowers when he hears unfamiliar sounds. After several hours or days in the cage, he no longer jumps up or cowers when he hears the noises. This process allows animals to actively ignore things in the environment that aren t important. 4

5 Sensitization - The opposite of habituation, an increase in a response as a result of shaking only to very bad thunderstorms, now shows fear and shaking to even very mild thunderstorms after several experiences with them. Sensitization and habituation are the simplest forms of learning. No new behavior is learned nor does the animal learn new associations. A behavior increases or decreases as a result of repeated exposure. Animals generally habituate to things that aren t important and sensitize to things that may be harmful or important to survival. Classical or Respondent Conditioning - Stimulus substitution learning, where a new stimulus comes to elicit an automatic response. Example - A strong clap of thunder (unconditioned stimulus) produces a fearful startle response in a dog (unconditioned response). After experiencing a number of storms where the smell of rain preceded the thunderclaps, now the smell of rain by itself (conditioned stimulus) with no thunder, produces the startle response (conditioned response). The dog has learned to associate rain with thunderclaps and now rain by itself produces fearful behavior. The dog s response and the consequences of that response have no effect on the learning. Classical conditioning is most often seen in very emotional behaviors such as fears and aggression. Operant or Instrumental Learning Response-consequence learning. The animal learns a relationship between its behavior and the consequences of that behavior. The consequences can be pleasant or unpleasant. Example - A dog that learns to bark or speak on command to get a food treat. The word speak is the antecedent or cue for the behavior, barking is the response, and the treat is the consequence. After repeated trials, Rover learns a treat follows when he barks after he hears the word speak, but no treat follows when he barks at other times. Target behavior the behavior that is being influenced by a learning event. In real-life situations, several of these kinds of learning may be going on at the same time. For example a dog being taught to sit for a food treat, may habituate to the other things going on around him when he is being trained. The cue Sit will become classically conditioned to the presentation of the food treat. The response of sitting will be operantly conditioned to its consequence, the presentation of food. 5

6 OPERANT CONDITIONING Operant conditioning, also known as instrumental or Skinnerian conditioning, is based on the principle that the consequences of animal s behavior will influence its frequency. This is known as Thorndike s Law of Effect, which says behavior that results in pleasant outcomes will be repeated, and that which results in unpleasant consequences will decrease in frequency. The outcomes are positive and negative reinforcement, and positive and negative punishment. These terms are illustrated in the following table. Consequence (What Follows the Behavior) An enjoyable, good thing An unpleasant, bad thing Add It, Present It, Give It (Positive) Positive Reinforcement (R+) Behavior Increases Example: Dog sits, he receives a food treat Positive Punishment (P+) Behavior Decreases When he barks, a dog receives a spray of citronella from an anti-bark collar Remove It, Withhold It, Take It Away (Negative) Negative Punishment (P-) Behavior Decreases Person stops petting a dog when the dog jumps on her Negative Reinforcement (R-) Behavior Increases Example: A person backs away from a fearful dog who is growling. 6

7 PRACTICE EXERCISES FOR BEHAVIORAL CONSEQUENCES Event Target Behavior Consequence 1. Dog enters off limits area of house; owner shakes a can filled with pennies 2. Dog stops on command, and gets to return to herding sheep 3. Dog barks at a cyclist, but then stops barking when he is frightened by a group of cyclists following the first one 4. Dog barks at window to get owner s attention; owner leaves the room 5. Person stops at red light and avoids getting a ticket 6. Dog comes when called while playing on the beach and gets to play tug with seaweed. 7. Dog runs into dog house to avoid getting pounced on by exuberant puppy 8. Child wakes up parent by playing stereo too loud; parents take away stereo for a week 9. Dog barks from his crate, owner squirts him with squirt gun 10. Puppy doesn t jump on person and gets petted 11. Puppy tries to snatch toy from owner s hand; owner puts toy away 12. Dog runs and hides to avoid a bath when shampoo and towels appear 7

8 PRACTICE EXAMPLES TO HELP YOU UNDERSTAND OPERANT CONDITIONING Classify whether each event is positive reinforcement, negative reinforcement, positive punishment or negative punishment. More than one outcome may be applicable if you identify more than one target behavior in each example. We ll work through a number of these in each lecture. 1. Dog jumps on you, you step back and dog lands on floor 2. The dog brings back the dumbbell and gets a treat 3. A dog owner works hard on training her dog, and trainer praises her for her efforts 4. A dog owner works hard on training her dog, and trainer points out the things she did wrong 5. A dog owner works hard on training her dog, and trainer says nothing 6. A dog, chewing on a bone growls at his owner as she walks by. Owner takes the bone away. 7. Owner takes the ball away from her dog who has been pestering her to play. 8. When her daughter comes home late, mom takes away her phone privileges 9. Owner rattles the shake can when her dog begs at the table 10. Driver honks at a dog who is chasing his car. Dog is frightened and runs away. 11a. When woman doesn t balance her checkbook, the checks bounce 11b. Woman balances her checkbook to avoid bouncing the checks. 12A Student studies and gets an A 12B. Student studies to avoid failing. 12C. Student doesn t study and fails 13. A woman s child whines for a chocolate bar in the grocery store, and the woman gives in and buys one 14. Your parrot screams and you scream back 15. A trainer teaches her dog to down-stay by spraying water in his face when he gets up 16. Your parrot says a word you like and you give him a treat 17. You look outside and see it might rain so you take your umbrella 8

9 18. A novice rider wants the horse to turn right so he pulls on the right rein until the horse turns. 19. Your dog picks up a food wrapper and when he sees you coming, runs away, gulping it down. 20. Your horse gets very excited and paws at the ground as you prepare to feed it. When it does this, you close the grain bin and leave the barn for a few minutes. 21. Your dog is in the crate, he barks too much, you throw a blanket over his cage until he stops barking. 22. You teach your dog to kiss by sticking your face in the dog s face until eventually he licks your face and then you move your face away. 23. A dog growls at someone who is approaching him, and the person backs up 24. A dog barks at the mailman, who then leaves the yard 25. The horse balks at a pole on the ground and the handler backs off momentarily and then tries again. When the horse steps over the pole, the handler feeds the horse a carrot 26. Your dog presses his nose against the patio glass door and you let him out to chase a squirrel 27. In order to avoid an argument, I don t harass my husband about leaving the toilet seat up 28. When he brings back the wrong scent article, I don t give my dog a treat 29. Husband cooks dinner, wife gives him a foot massage 30. If dog lies down quietly by owner s side, she periodically gives him a piece of popcorn 31. When your dog is loose in the park, you call him and he comes, then you throw a ball for him 32. You call the dog and he doesn t come, you shout and yell at him until he does 33. You call the dog and he doesn t come, you go to him and put him in the car for 10 minutes 34. Dog barks at the doorbell and you put a muzzle on him. 9

10 35. Dog licks your face and you hide your head in a pillow for 10 seconds 36. You walk up and say hi to two people talking and they ignore you 37. Driving carefully allows you to arrive alive. 38. When dog tries to enter an off limits room in the house owner shakes a shaker can 39. Wear a seat belt to avoid being hurt in an accident 40. Dog starts to enter a crate and owner (inadvertently!) drops a bell on it - dog is afraid and won t enter the crate. 41. Dog attempts to go to front yard to play Frisbee, owner takes Frisbee away. Owner play Frisbee with dog when he goes to back yard. 42. Dog successfully stops on command, dog gets to return to herding sheep 43. When he misbehaves, dog is shown prong collar and stops misbehaving. 44. Husband forgets anniversary and has to sleep on the couch 45.Husband remembers birthday and spends the night in his own bed 46. Student trainer spends evening watching Animal Planet instead of doing assignment on operant conditioning and gets yelled at by instructor. 47. Dog brings owners his slippers and owner shares his sandwich with him. 48. Dog starts barking at a cyclist and is then frightened by a pack of cyclists and runs and hides. 49. Dog barks at window to get owner s attention. Owner leaves the room 50. Dog growls at mail carrier who then leaves 51. Dog runs after rabbit and gets caught in the blackberry bushes. 52. Two boys are arguing over whose turn it is to play Nintendo. Father comes in and tells them they have to turn off the game because they are arguing. 53. Student acts inappropriately during class. Teacher makes student write on board I will not misbehave in class 100 times. 54. Pet trainer works hard to increase sales and customer service. Trainer receives coveted superlative customer service pin. 10

11 55. Family of 5 goes river rafting down the Colorado river. Mother insists that everyone wear life jackets so they won t drown. 56. Robber is convicted and sentenced to life imprisonment. 57. Dog jumps on person when she walks out the back door. Person goes back inside. 58. Kids play videogames instead of cleaning the house so mom grounds the kids. 59. Little dog tries to steal big dog s bone. Big dog growls at little dog who runs away 60. Child gets yelled at for making a scene at the airport. 61. Dog comes when called while playing on the beach and owners lets him play tug with seaweed. 62. Dogs comes when called and avoids being pounced on by exuberant puppy who is running loose 63 Dog doesn t come when called and gets pounced on by grouchy other dog 64. Dog doesn t come quickly when called and when he finally wanders over to owner, owner withholds his cookie 65. Puppy stops playing to come when called, and is given a cookie and allowed to return to play. 66. Husband makes breakfast and wife goes out and warms up his car for him. 67. Son wakes parents up by playing stereo too loud, loses stereo privileges for a week 68. Dog is lying quietly by owner while owner works on a computer. Owner praises dog for lying quietly. 69. Dog barks and runs fence when children are passing, children shriek and run away. 70. Dog barks when in crate, owner squirts him with water from a squirt gun. 71. Child gets A on report card and gets to buy a new toy 72. Child doesn t get to eat dessert because she didn t eat her peas. 73. I brush my teeth to avoid getting cavities. 74. Child eats her veggies and gets ice cream for dessert. 11

12 75. Dog pesters husband while he is lying on couch. Husband throws the ball for the dog. 76. Dog pesters husband and husband yells at dog 77. Husband yells at dog until dog goes into another room 78. Husband calls dog and then takes his ball away. 79. Dog barks at door, husband gets up off the couch to let him outside, dog runs to couch and lays down where husband was. 80. Puppy doesn t jump up on person and is praised and gets to go for a walk. 81. Dog tries to snatch toy out of owner s hand, owner puts toy away 82. Dog fence runs with another dog and is put into kennel run so he can t see other dog. 83. Dog runs and hides when shampoo and towels are brought out. 84. Dog rings the bell hanging next to the door and owner opens door and lets him outside. 85. Child runs (rather than walking) on the playground and teacher gives him a time out. 86. Student doesn t do homework and doesn t get to go to recess. 87 Student does homework and gets a piece of candy. 88. Student finishes homework early and gets to go to recess early. 89. Student attends class, mom doesn t get a call from the school principal and doesn t yell at student. 90. Student doesn t do homework and gets his knuckles rapped by the teacher. 91. Student comes to class late, teacher says if continues will make student take an extra exam. Student becomes punctual. 92. Horse is shocked by electric fence and stops trying to break out of paddock. 93. A manager of a store loses bonus for not cooperating with new sales program. 94. When two dogs fight over a toy toy is taken away. 95. Woman feels compelled to respond to chain letter because she is afraid of all the bad things the letter says will happen if she breaks the chain. 12

13 LECTURE NOTES FOR FUNDAMENTALS OF ANIMAL LEARNING TELECOURSE phone fax Suzanne Hetts, Ph.D., CAAB, CPDT Dan Estep, Ph.D., CAAB WEEK II INTRODUCTION USING POSITIVE REINFORCEMENT What to use for (types of) reinforcement When to reinforce Training techniques using reinforcement GLOSSARY Continuous reinforcement each correct response is reinforced Cue a discriminative stimulus signaling the availability of reinforcement or punishment for a behavior Differential reinforcement of excellent behavior (DRE) selectively rewarding only those responses that are of increasing quality. The dog must be able to understand what feature (e.g. straighter sits, faster recall) of the response causes reinforcement to happen. Duration schedule of reinforcement a behavior is reinforced after it has endured for a particular length of time. Used to reinforce behaviors such as heeling, staying, maintaining eye contact, or holding an object. Can be used on either a fixed (not recommended) or random duration schedule. Fixed interval schedule of reinforcement a reinforcer is given for the first correct response after a fixed period of time, like one minute. Fixed ratio schedule of reinforcement - a reinforcer is given after a fixed number of responses. In contrast to a VR of 1 in 10, a FR will reward the 10 th and the 20 th responses. Intermittent reinforcement only some correct responses are reinforced, on a schedule the animal cannot predict. There are several different patterns, or schedules, for the delivery of intermittent reinforcement. 13

14 Interval schedules of reinforcement the first correct response after a specified period of time has passed is rewarded. Motivation - an external or internal force that influences the likelihood an animal will engage in a particular behavior. Premack Principle any high probability behavior can be used to reinforce a low probability behavior. Everyday activities a dog enjoys can be used to reinforce a behavior you are trying to teach. An example would be rewarding a dog for coming when called by releasing him to go play in the park. Primary (unconditioned) reinforcer something the animal naturally enjoys (without any conditioning) and will work to obtain. Examples include food, water, sex, and in some animals, the opportunities to explore new areas or things and opportunities to play, and social contact with others. Random ratio schedule of reinforcement a random portion of a dog s correct responses are reinforced (practically, there is usually an upper limit to this). Ratio schedule of reinforcement frequency of reinforcemen t is based on a specified portion of a dog s responses. Secondary (conditioned) reinforcers - an initially neutral event which, through classical conditioning, predicts the presentation of a primary reinforcer. A clicker is an example of a secondary reinforcer, as is verbal praise. For people, examples of secondary reinforcers include money, jewelry or fast cars. Conditioned reinforcers may better mark the target behavior, and serve as a bridge until the primary reinforcer can be delivered. Variable interval schedule of reinforcement the dog is reinforced for the first response after a variable period of time but where the average time is specified, such as 30 seconds. So one time the interval may be 20 seconds, the next it may be 45 seconds, but the average interval will be 30 seconds. Variable ratio schedule of reinforcement a reinforcer is given after a variable number of responses but the average will be some number, such say 1 in 10. For example, of 20 responses, 2 will be reinforced, perhaps the 5 th and the 18 th. 14

15 What To Use For Positive Primary Reinforcement Positive reinforcement is anything an animal will work to obtain. For most dogs, primary reinforcers things dogs inherently like and will try to acquire for themselves are food, toys (or at least the opportunity to play) and petting. Also consider the use of so-called life rewards, a term coined by Dr. Ian Dunbar. Examples include the opportunity to chase a squirrel, go for a walk in the park, get on the bed and cuddle. The technical term for this is the Premack Principle. It means that a behavior that is highly likely to occur can be used to reinforce one that is much less likely. You can help clients figure out what things in life are important to their pets and then use these things to reward good behavior. In order for this to work effectively, owners must be able to control their pet s access to these resources. For example, if there are 4 tennis balls lying around on the floor all the time for the dog to play with, playing with a tennis ball is not going to be something the dog will work for because he can do it anytime he wants. This doesn t mean dogs should be deprived of things they want, but many of the things they want should be under the owner s control. Food As A Reinforcer. Tidbits should be small, very palatable, soft and easy to swallow without much chewing, and be a mixture of several types of food the pet doesn t receive outside of training. Toys As Reinforcers. As already described, if a toy is used to reinforce good behavior, the pet cannot have free access to this toy. Contrary to popular belief, tug toys make great reinforcers. Research has shown that playing tug of war is not correlated with the development of aggression problems. Two main rules are that the dog s teeth must never touch skin during the game, and that the dog will release the toy when requested. For some dogs, it may be better to delay tug until a release word is learned. Petting As A Reinforcer. Most pets and many cats enjoy petting, however its value as a reinforcer is often diminished because most everyone tends to pet and touch their pets all the time. Since this is a resource that is usually impractical to control access to, its best use is probably as an additional reinforcement, rather than the sole reinforcer. Vary The Reinforcer. A particular reinforcer will remain more motivating if it isn t given every time. If a small piece of cheese is the only thing an owner uses for reinforcement, the pet is going to tire of cheese pretty quickly. Owners should vary their reinforcers by using a mixture of food, or a selection of toys. The Eager To Please Belief This is an anthropomorphic description of dogs for whom social reinforcements and punishments are very important. Eager to please could either mean that petting, praise and attention are very strong inducements, or that verbal discipline and threatening body postures are very unpleasant consequences. 15

16 Positive Reinforcement Techniques To Elicit Behavior Prompting (and Fading) There are any number of ways that a behavior could be prompted to occur. One way is the lure-reward method. This method typically makes use of a food treat as a prompt to lure the dog into a desired position such as sit or down. Once the dog is responding consistently to the lure, the prompt is gradually faded and the verbal cue sit attached to the behavior. The tidbit is then used only for positive reinforcement and not as a prompt. In the beginning stages of training when the behavior is being acquired the tidbit is given each time the dog performs the correct behavior. As the response becomes reliable, the tidbit is reduced to an intermittent reinforcement schedule. Steps must be taken to ensure that the dog cannot predict when the food reinforcement will be given and when it will not. If a) the food lure is properly faded, b) the food reinforcer is put on an intermittent schedule, and c) the dog cannot predict food availability, the problem of the doing only obeying when he sees the food will be avoided. Another type of prompting and fading is target training. With this procedure, the dog is taught to touch a target with his nose or paw. The target can then be moved around to manipulate the dog into various positions, such as jumping up to touch a light switch. If the goal is to teach the dog to turn off the switch, both targeting and shaping are being used. The target must eventually be faded so it is no longer the cue for the behavior. Catching Catching is the process of waiting for the dog to spontaneously perform a desirable behavior and then reinforcing it. If the behavior spontaneously occurs fairly regularly, catching can be very effective. If the behavior is relatively uncommon, catching may not be the best method to use. Catching requires the trainer/owner to consistently and closely observe the dog so as to not to miss an occurrence of the desired behavior which can be reinforced. Although probably not the most efficient method for teaching basic control behaviors like sit and down, catching can be a great way to teach unusual, trick-type behaviors. Shaping By Successive Approximations This is a technique that reinforces responses which approximate, or resemble, the final behavior that is desired. As training progresses, each succeeding response must be more and more like the desired behavior in order for it to be reinforced. If shaping were to be used to teach a dog to sit for example, perhaps more and more bending of the back legs might be reinforced for example, until the dog is finally sitting. Shaping requires good timing and observations skills, plus practice to be done efficiently and effectively. The procedure can be somewhat frustrating for the dog, especially if the behavior to be shaped is complex. Dogs who have been previously trained using aversive methods may not respond well to shaping, at least initially, because they will be reluctant to offer new behaviors for fear of making a mistake and being punished or corrected. 16

17 Cueing the Behavior What is referred to as the cue for the behavior in learning terminology, is what is commonly termed the command. Many training methods make the mistake of attempting to use a cue for the behavior before the behavior is actually learned. This only serves to confuse the dog, because the cue has no meaning. In addition, using the cue before the dog is reliably displaying the behavior causes the dog to tune out the cue because it is not relevant. When shaping, catching or prompting behaviors, the cue shouldn t be used until there is a high probability the dog is going to display the behavior at the appropriate time. The cue can usually be added relatively quickly with prompting (lure-reward) procedures, but with both catching and shaping, the cue generally can t be added until training has progressed significantly. The cue tells the dog that when he performs the behavior associated with the cue, reinforcement will follow. No reinforcement follows if the behavior is performed without the cue. This is what putting the behavior on cue means. This technique doesn t work if the owner can t control the reinforcement for the behavior. 17

18 LECTURE NOTES FOR FUNDAMENTALS OF ANIMAL LEARNING TELECOURSE PHONE FAX Suzanne Hetts, Ph.D., CAAB, CPDT Dan Estep, Ph.D., CAAB WEEK III USING NEGATIVE PUNISHMENT INTRODUCTION Negative punishment is under appreciated and utilized by most pet owners. It results in a decrease in behavior. Taking away the opportunity to do something rewarding can be a powerful way to change behavior. Negative punishment is the other side of the coin of positive reinforcement. GLOSSARY Classical counter conditioning a procedure which changes an emotional response by creating a new conditioned emotional state that is counter to, or opposite of the first one. The stimulus that elicits the unwanted emotional response is paired with (predicts) another event that is pleasant. Counter commanding signaling, or cueing an animal to perform a behavior that is incompatible with another behavior. Differential reinforcement of incompatible behavior (DRI) - withholding reinforcement for the unwanted behavior and reinforcing an opposing behavior (the two behaviors are mutually exclusive) Differential reinforcement of other behavior (DRO) reward occurrences of all other behaviors but the unwanted one. Extinction burst in increase in either the frequency or intensity of a response, prompted by the frustration of removal of reinforcement Operant counter conditioning a procedure by which an animal learns to perform a behavior that is incompatible with an unwanted behavior. The cue for the counter conditioned behavior is the same cue that originally elicited the unwanted behavior. Partial reinforcement extinction effect (PREE) behaviors on an intermittent schedule of reinforcement take longer to extinguish, because the dog initially can t discriminate 18

19 between extinction (complete removal of the reinforcement) and a lean intermittent schedule (the behavior must be performed many times before it is reinforced) Spontaneous recovery a resumption of a previously extinguished behavior. An extinguished response can be fully recovered after just a few unintentionally reinforced trials. 19

20 The following table compares negative punishment with positive reinforcement and punishment. Behavior Pos. Rft. R + Neg. Pun. P - Pos. Pun. P + Jumping up Pulling on leash Lying down Pestering behavior Mouthing Reward when sits instead (doesn t jump) Reward when in position (continue walking, treat, pet) Treat, pet when does Pay attention to when quiet Pet when removes mouth from hand Walk away, stop petting, ignore Stop moving forward, move in other direction Withhold rft. Ignore, time out, leave the room Walk away, stop petting, time out Kneeing in chest, stepping on toes, holding feet Jerk on collar, shock Jerk collar, step on leash Yell at, collar correction Hit on nose, put fingers down throat There are several different types of negative punishment, which are defined in the table below. Negative Punishment Techniques for Decreasing Undesirable Behavior and For Stimulus Control of Behavior Extinction - Withholding or removal of positive reinforcement for a previously reinforced behavior A dog is no longer let inside when she barks at the door. Differential reinforcement of other behavior dog is reinforced for displaying any other behavior than the target behavior Operant counter conditioning Counter commanding is initial step, becomes OCC after dog responds automatically without cue. A type of DRI (see glossary) Time out Conditioned negative punishment Can be mis-used and become a conditioned positive punisher. Definitions and Details Behavior must occur frequently, must be able to control the reinforcement, establish an alternative behavior, be aware of extinction bursts and spontaneous recovery, intermittently reinforced behaviors must resistant to extinction Frequent DRO in beginning. Danger of reinforcing another undesirable behavior. Relatively slow process. Establish a behavior that is incompatible with undesirable one, using any P+ technique. Reinforce sitting rather than jumping up; laying on his bed rather than begging, watching you rather than barking at a dog. Access to reinforcement is prevented either by removing the dog from the reinforcing environment or removing the reinforcing environment from the dog. Timing may be a problem, as is finding the proper environment. Duration should be relatively brief. A signal that predicts the removal or delay of positive reinforcement. Go to your room precedes and predicts a time out. Nope means a treat is not forthcoming if dog sits when told to lay down. That s enough means I m finished petting you for awhile. 20

21 LECTURE NOTES FOR FUNDAMENTALS OF ANIMAL LEARNING TELECOURSE PHONE FAX Suzanne Hetts, Ph.D., CAAB, CPDT Dan Estep, Ph.D., CAAB WEEK IV USING POSITIVE PUNISHMENT Even though punishment is often the least effective way to modify behavior, it is often the first thing that is tried, if a person is thinking How can I get my pet to stop misbehaving. When the focus is on decreasing behavior, punishment is often the first thought. Instead, the focus should be How can I get my pet to do what I want so I can reinforce him for desirable behavior. Basic training procedures should not set the dog up to fail so he can be punished, but rather set him up to succeed so he can be reinforced. GLOSSARY Conditioned punisher a stimulus that signals the onset of a positive punisher. It should be able to mark the unwanted behavior just as a conditioned positive reinforcer does. A warning tone on an electronic fence is another example of a conditioned punisher. However, with a conditioned signal, the dog may still show the unwanted behavior, but stop when the conditioned punisher is presented. An example is squirting a dog with water for barking. Picking up the squirt bottle becomes a conditioned punisher that may cause the dog to stop barking. Types of punishers it is common to refer to remote punishers and interactive punishers. Remote punishers are triggered by the environment or the dog s behavior (or at least appear to be). Examples include a Snappy Trainer, a Gentle Spray citronella bark collar or a Scat Mat. Examples of remote punishers that require the owner to activate them are a squirt gun or an airhorn. Depending on how they are used, they could be perceived as interactive. Interactive punishers involve the owner directly in the punishment process. Examples include verbal reprimands, punishment delivered using a leash and collar, and intimidating body postures 21

22 Criteria for Effective Positive Punishment 1. Must be immediate the aversive stimulus must occur within a few seconds of the behavior. Punishment after the fact is ineffective and unfair to the animal. Punishment delivered at the beginning of the unwanted behavior is more effective. 2. Must be consistent be delivered after every occurrence of the target behavior. Contrary to intermittent reinforcement, intermittent punishment is not a tool for effective learning. 3. Must be of appropriate intensity should be sufficiently unpleasant to suppress the behavior within a few trials. If it does not, then by definition, the aversive event is not punishing. The minimum intensity required to suppress the behavior should be used, which is difficult to determine ahead of time. Weak punishers may not be perceived as punishment. Punishers that are too strong create fear which interferes with learning 4. Must be of brief duration the aversive event must terminate quickly and must not continue after the animal is engaging in another behavior 5. Remote punishment is usually advantageous over interactive the aversive event should not come from the owner. Scat Mat, SSSCat, Snappy Trainers, bark collars are all examples of non-interactive punishers. 6. Best if provide alternative response which can be reinforced suppressing a highly motivated behavior creates a behavioral vacuum that animal will fill with his own choice if another is not provided. Reinforce the dog for the behavior you want him to do instead. For example if you choose to positively punish jumping up, the dog should be reinforced for sitting to greet people. 7. Best if animal does not have opportunity to enjoy the fruits of his labor e.g. punishing a dog after he s stolen food from the counter will be less effective because he s enjoyed eating/tasting what he s stolen Because these criteria are difficult to meet, punishment alone has limited value in resolving behavior problems. In addition, to suppress some highly motivated behaviors, very aversive stimuli may be required. Highly aversive events have a strong possibility of causing fearful and/or aggressive responses, which are often more problematic than the behavior needing to be suppressed. It is sometimes hard for owners to understand why an animal continues to engage in a behavior despite their attempts to punish it. By definition, punishment decreases the frequency of a behavior, so if this hasn t happened, technically the behavior hasn t been punished. Ineffective attempts at punishment usually mean the criteria for its use have not been followed. In some cases, although owners may not recognize it, the goal is not to suppress the behavior, but redirect it onto another target. For example, housetraining should not involve suppressing elimination, but merely targeting where it should occur. 22

23 Disadvantages of Using Punishment 1) Criteria are difficult to met 2) It can elicit aggression. 3) It can elicit fear. 4) If the punishment is associated with the owner s presence, the dog only learns not to do the bad behavior under that one condition. 5) It never teaches the dog what he should be doing. The Ethics of Using Aversives A question you should consider is when, or if aversives should ever be used. A decision tree reproduced from Professional Standards for Dog Trainers: Effective, Humane Principles, published by Delta Society is included in these notes. Some of these criteria are borrowed and modified from the mental health field that also has guidelines for when aversive techniques should be used with people. Consider the following: Aversive techniques should be used only: after positive alternatives have failed if a problem behavior presents a danger to the dog or the safety of others at the minimum intensity that is reasonably expected to suppress the behavior by skilled professionals (if the aversive is something like shock) Aversives should not be used in anger, from force of habit, or because someone is unfamiliar with how to use positive alternatives. Positive punishment should meet the criteria for effective use. 23

24 Reproduced from: Professional Standards for Dog Trainers: Effective, Humane Principles. 2001, Delta Society, Renton, WA. 24

25 FUNDAMENTALS OF ANIMAL LEARNING TELECOURSE PHONE FX Suzanne Hetts, Ph.D., CAAB, CPDT Dan Estep, Ph.D., CAAB REFERENCES AND BIBLIOGRAPHY Burch, M.R. & Bailey, J.S. How Dogs Learn. New York: Howell Book House. Donaldson, J., Dogs Are From Neptune. Lasar Multimedia Productions, Montreal. Hetts, S. Pet Behavior Protocols. What to Say, What to Do, When to Refer Lakewood, CO: American Animal Hospital Association Press, pp Pryor, K., Don t Shoot the Dog. Bantam Books, NY. Schwartz, B., and S. J. Robbins Psychology of Learning and Behavior, W.W. Norton and Company, NY Reid, P.J Excel-erated Learning. Explaining in Plain English How Dogs Learn and How Best To Teach Them. Oakland, CA: James and Kenneth Publ. 25