how dogs learn - hshv€¦ · health food safety! if these needs aren’t met, most of dog’s...
TRANSCRIPT
How Dogs Learn HSHV Volunteer Continuing EducationJan. 28, 2018
Angela Schmorrow, CPDT-KA
Agenda
How animals learn: science of learning theory
Making training decisions
Using positive reinforcement effectively
Empowerment-based training
Case study
The Science:How Animals Learn
What is Behavior “For”?
Changing one’s environment
Getting needs met
Communication
Not inherently “good” or “bad”
Unlabel Me
Think beyond “reactive,” “aggressive,” “fearful,” “hyperactive,” “stubborn,” etc.
What is the animal actually doing?
Behavior is modifiable using scientifically-established principles
How do dogs learn?
By association (Classical Conditioning) What is safe? What is scary?
By consequences (Operant Conditioning) What happens when I do this?
Classical Conditioning
Creating an association between two stimuli Primary/unconditioned – animal doesn’t need to learn to like or
dislike it (food, pain)
Secondary/conditioned – animal learns to react to it based on its association with the primary stimulus
This is occurring ALL the time.
Pavlov’s dogs:
Food triggers salivation
Bell began predicting food
Soon, bell alone could trigger salivary response normally only caused by food.
Food Salivation
Bell Food Salivation
Bell Salivation
Operant Conditioning
Forming an association between a behavior and a consequence.
Triggered by an antecedent in the environment.
Behavior is changed by changing the antecedent or consequence.
Antecedent Behavior Consequence
Four Possible Consequences
Reinforcement(behavior increases)
Punishment(behaviordecreases)
Positive(something is added)
Positive Reinforcement:Something added increases behavior
PositivePunishment:Something added decreases behavior
Negative (something is taken away)
NegativeReinforcement: Something removed increases behavior
Negative Punishment:Something removed decreases behavior
You don’t need to remember all the technical terms . . .
Just ask: Do I want to stop this behavior, or make it continue?
Am I going to give the dog something he likes, or do something to him he doesn’t like?
Dog training examples
Reinforcement(behavior increases)
Punishment(behaviordecreases)
Positive(something is added)
Dog sits and is given a treat (+R)
Dog jumps on guest and is kneedin the chest (+P)
Negative (something is taken away)
Handler shoves dog’s rear down until dog sits on ground(-R)
Dog receives no attention and is ignored when jumps on guests (-P)
How Do We Choose?
All four quadrants work – all can change behavior.
But there are important issues to consider when choosing which method to use . . .
Reinforcement vs. Punishment
Reinforcement BUILDS behavior. Behaviors that are reinforced become inherently rewarding on
their own.
Example: Dog that is reinforced enough for going to a mat will begin to seek out that mat to relax on, even on his own.
Punishment TEMPORARILY stops behavior. Behavior may even stop for a long time, but in the absence of
punishment, it will reappear, and require additional punishment.
The Problems With Punishment
Requires impeccable observation skills and timing that most of us don’t have. (Too easy to correct your dog for the wrong thing)
Desensitization: learners become used to a level of punishment, so require harsher punishments over time to be effective. How many of us really want to do this?
Using punishment is like dropping a bomb: imprecise and unpredictable
Remember the lessons of classical conditioning – a punished dog may associate anything in its environment at the time with the punishment, including you. Common fallout from punishment: increased fear, aggression, and
anxiety
Something doesn’t have to be painful to be aversive.
Avoidance: dogs try to prevent punishment from happening by stopping all behaviors, which ultimately slows learning.
We’re not solving the underlying problem. It’s like pushing down a beach ball in the water. Sooner or later, it’s going to pop up somewhere!
Advantages of Positive Reinforcement
Creates eager learners who enjoy the behaviors being taught (classical conditioning at work again)
Builds a relationship between teacher and learner
Mistakes do no harm (just a couple extra calories for the “oops” cookies!)
Creates strong behaviors
Easier to focus on what we DO want than all the things we DON’T. Rather than punishing a behavior, we can replace it with
something better (sitting instead of jumping)!
Creates positive associations
“Pavlov is always sitting on your shoulder.”
Shelter Example: Dog Jumping on Walker in Kennel
Basic Needs: Has dog been able to eliminate recently?
Is kennel dry and clean?
Does dog feel safe in kennel?
Antecedent Arrangements: change the environment to promote a different behavior Toss treats on ground before entering, cheese whiz on floor or wall
Shelter Example: Dog Jumping on Walker in Kennel
Positive reinforcement: reinforce dog when feet remain on ground
Reinforcement of alternate behavior: teach dog to sit when walker touches kennel door and enters
Negative Punishment: Ignoring dog, leaving kennel when jumps (NOTE: This can still be very frustrating for the dog, and therefore aversive!)
What DON’T we want to do?
Negative Reinforcement: Shout at dog continuously until stops jumping
Positive Punishment: Shoving dog down, spray bottle, shaker can, kneeing in chest, etc.
What could happen if we did these things? How does dog feel about walks? About people?
Do “difficult dogs” really need a heavier hand?
Not if our goal is to solve underlying problems or to prevent problems from getting worse
Let’s go back to learning theory Reinforcement builds behavior
Punishment temporarily suppresses behavior and can have fallout that will make things worse/more dangerous
Animals are constantly making associations. Not enough to stop a behavior – need to also change the emotional response.
Same for all animals, and all behaviors regardless of severity
"From both a practical and a personal point of view, perhaps the most significant thing to remember about the first side effect of coercion is that people who use punishment become conditioned punishers themselves . . . All the side effects that shocks generate, we, too, will generate. Anyone who uses shock becomes a shock." - Murray Sidman, "Coercion and Its Fallout"
The Art:Teaching Behaviors
Get the Behavior!
Capturing: Catch the animal naturally doing the behavior Advantages: Useful for behaviors that are offered frequently (sit), or that are
natural behaviors that may be hard to elicit otherwise (stretching).
Considerations: Need to be prepared and observant. May capture unintended behaviors.
Luring: Using food to lead animal into desired position. Advantages: Fast way to get certain behaviors.
Considerations: Animal may be following food, not as aware of behavior. Need to get food out of hand quickly or food may become the cue for the behavior.
Shaping: Rewarding successive approximations (baby steps) on the path to the desired behavior. Advantages: Builds strong behaviors. Empowering. Can get complicated
behaviors that you couldn’t elicit otherwise.
Considerations: Requires higher level of trainer skill (observation skills and mechanics)
How does dog know what behavior earned the reward?
Reward Markers Communication tool that has already been associated with a
primary reward (food, toys)
Examples: clickers, saying “yes”, thumbs up signal for deaf dogs, touch on specific part of body for deaf/blind dogs.
Reward marker indicates the behavior that we are looking for –reward still will always follow.
Used for teaching a new behavior. Not necessary once dog understands.
Get it on Cue!
Cue: Signal to the dog to perform the new behavior.
Five Dollar Rule: When you will bet $5 that the dog will perform the behavior, then add a cue.
Physical Cues Often first learned
May come naturally from what we were doing to elicit behavior (ex. Hand lure towards the ground for down, body position).
Verbal Cues Must be given right before you would expect dog to perform the
behavior.
Generalization and Proofing
Can animal perform behavior in different contexts, settings, etc.?
When one element gets harder, may need to back up and reduce criteria (reward more frequently, less duration, etc.)
“It is effortful to train for situations you don’t normally have to cope with.” – Susan Friedman
Dealing with Incorrect Responses
Plan A: Be proactive and set dog up for success! Manage environment, split behaviors, reinforce
Calmly interrupt if necessary (the behavior is dangerous, destructive, self-rewarding, etc.) Removing reinforcement is enough – no need for “NO!” or other
corrections
Change the environment Lower criteria after more than 2 incorrect responses Behavioral momentum: Ask for a highly likely behavior (a
favorite trick, etc.) before continuing Increase rate of reinforcement and/or value of reward
What if it isn’t working?
Basic physiological needs must be met Health Food
SAFETY! If these needs aren’t met, most of dog’s behavior will be about trying
to get them met. They aren’t in a place to learn new things – they are trying to survive.
The dog decides whether something is reinforcing or not. If a behavior is not increasing, the reward is not really a reward. Need to know what the dog likes.
Need to identify ways that dogs still get reinforcement for undesired behaviors. Example: petting the dog that is jumping up on you.
Dog unable to determine what is relevant in environment.
Empowerment-Based Training:
The Power of Choice in Training
Importance of Choice
“Being able to choose what to do, freely and without compulsion, is an incredibly important aspect of life.” - Irith Bloom
“The power to control one’s own outcomes is essential to behavioral health, and the degree to which a behavior reduction procedure preserves learner control is essential to developing a standard of humane, effective practice.” - Susan Friedman
Animals need to have opportunities to make choices, and see that their behaviors can control the outcomes.
If you remove choice and predictability, effects can include anxiety, stress, depression, and decreased immune function as well as other problems.
What choices are already taken away from dogs?
Choice-Based Training
Create situations where dog can make a decision
Make it easy/desirable to perform the decision you like
Control access to the rewards so that undesired behaviors won’t get rewarded
Reward the decisions you like
Wait! As long as something isn’t destructive, dangerous, or rewarding to the dog, have patience to let them figure out what will work instead.
Shelter Case Study:Fish
Why Would We Train Dogs in the Shelter?
Teach desired behaviors
Modify behaviors that may be undesirable
Build confidence
Create predictability
Form relationships
Mental enrichment
“Animals deserve the best care we can possibly provide. Training should not be considered a luxury that is only provided if there is time; it is an essential part of good animal care. Just as one would never consider developing an animal care program without a veterinary component, a nutritional component, a social component, and an environmental component, nobody should consider caring for an animal without a behavioral management component integrated into the program” -Ken Ramirez
Fish
Adult male lab mix
Transferred to Michigan shelter from high-kill animal control in southern state
No history
Heartworm positive and intact at time of intake
Scarring on ears from excessive fly bites, thick scarring around neck (history of embedded collar?)
Fearful and avoidant of staff, growling at some male staff
Notes from first attempt to assess:
“Unable to test today - too fearful. Ran from me in exterior kennel, tail tucked, body low, wide eyed. M. leashed up and brought into real life room. Paced and stood at door - very stressed. Would take treats I dropped but not from hand. Would avoid if I tried to touch.”
First two weeks:
Worked on pure desensitization/counter conditioning – tossing treats from outside kennel.
Eventually would approach and allow me to leash and take out.
Other than leashing, would avoid any contact. Startled, cowered and fled with any slight movement.
Started Clicker Training
Worked on shaping two foundation behaviors: settle on mat and hand target
Goal was to build relationship with him where he was empowered to make decisions and earn reinforcement.
His behaviors mattered and he could control the interaction.
Predictable ways of engaging with a person.
Go to Mat, First Session:
Clicking for any movement toward or on to the mat, then remaining there.
Treat delivered on mat.
Re-set cookies after several seconds.
Go to Mat: Second Session
Waiting for body fully standing on mat.
Go to Mat: Third Session
Shaping a default sit, building to a down.
Go to Mat: Fourth Session
Settling in down position on mat.
Now use this as his “consent” behavior to work on DS/CC to my movement. Begin standing up and moving gradually.
If he stands up, I will sit back down.
Empowers him to communicate if he is ready.
Hand Target
Goal: shape Fish to touch nose to my hand as a way to voluntarily approach, take in information, and interact with me.
The problem: Any movement of my hand would cause a fear/avoidance response.
Fish began HW treatment and subsequently came down with kennel cough. Could not go to play yards due to treatment, or to real life room due to his infection. Worked with kennel door between us, which made my hand less intimidating to him.
Fish Now
Was able to be assessed, did well, and is up for adoption.
Approaches all staff and others eagerly for attention.
Continuing to work on settling on mat during movement and environmental change.
Working on chin rest behavior as consent for touching for husbandry.
Shaping games.
What Did Fish Learn?
A couple basic skills
That people equal good things
That his behaviors and choices matter
That he can control his environment
Recommended Reading
Karen Pryor, “Reaching the Animal Mind.”
Karen Pryor, “Don’t Shoot the Dog.”
Patricia McConnell, “The Other End of the Leash”
Grisha Stewart, “The Ahimsa Dog Training Manual” and “Behavior Adjustment Training 2.0”
Pamela Reid, “Excel-erated Learning: Explaining How Dogs Learn and How Best to Teach Them.”
Contact Us!
Website: www.harmonyk9.com
Facebook: www.facebook.com/harmonyk9s