Post by RealPitBull on Mar 19, 2010 15:30:20 GMT -5
Good article about the importance of dealing with observable behavior and working to modify that behavior, instead of falling into the trap of assigning assumed emotions and motivations to an animal's behavior.
From a course I am taking in canid behavior and wanted to share.
+++++++++++++++++++++++++++++++++++++++++++++
But What Does it Look Like?
Using Observable Phenomena to Interpret Behavior and Solve Training Challenges
Emily Insalaco, Behavior Programs Manager
Natural Encounters, Inc.
Abstract
Animal training utilizes ideas from various branches of science, including animal behavior, behavior analysis and
cognitive psychology. By definition, science deals with describing physical phenomena and the environment in
measurable terms. As trainers, we are encouraged to read and interpret our animals’ actions, and even go as far as
to try to figure out what the animal is thinking. However, this can lead to trainers using terms that may seem like
clear descriptors, but are actually open to interpretation by others, and difficult to operationally define, which in
turn can often make it more difficult to determine the next step in the training plan. This paper will explore how,
by concentrating on only the behaviors we actually observe in our training, and by using measurable, observable
behavioral terminology, we can greatly improve communication among staff members, and empower the trainers
to approach their training challenges with more effective tools.
Introduction and Historical Perspective
I have heard training described as an art. I have heard it called a science. I have also heard that it is both at the
same time. Whatever camp you may sit in, I doubt anyone can argue that training strongly relies on scientific
principles. The field of animal training as we know it in zoos and aquariums was built from behavior science. We
look back on the work of figures such as Thorndike, Pavlov and Skinner for guidance in defining what it is we are
doing. As trainers, we routinely use terms like operant conditioning, extinction, and schedules of reinforcement.
We read textbooks and attend workshops to learn how to apply these terms.
If we read carefully enough, we also find a historical struggle in behavioral science to measure things like thought
and emotion. One of the earliest works in comparative psychology is Animal Intelligence by Georges Romanes
(1882), which attempted to define just that. In his work, he makes reference to mental abilities and processes
but uses various feelings, such as jealousy, to illustrate some of these processes. The response from the rest of
the psychological field was one of backlash, and spurred a movement that led to the improvement of scientific
controls and interpretation of data. (McFarland, 1999) Later studies showed much more discipline in defining
relevant information. Thorndike’s puzzle box experiments were praised because he “showed strict avoidance of
anthropomorphic interpretation of the behavior he observed.” (Domjam, 2003) David McFarland describes the
behaviorist’s view in this way: “Psychology is the study of behavior itself, rather than the mental events. In its
[strongest] form, behaviorism rejects all reference to inner processes in the explanation of behavior.” (McFarland,
1999) Why, then, if we are truly using science to train our animals, do we consistently try, whether consciously
or not, to get inside the animals’ heads? More importantly, is it helping us, or merely making our jobs more
challenging than they need to be, by leading us to misinterpret the problem, and making our end goal more
difficult to identify?
Some Definitions, Quotes, and How They Apply to Animal Training
Susan Friedman PhD stated in her paper Straight Talk About Parrot Behavior that “Science employs methods of
systematic observation of measurable phenomena” and “Science seeks explanations that are testable.”
Using these guidelines, we can ask ourselves some questions that can help get us on the right track. How does the
way we look at behavior fit in with “systematic observations of measurable phenomena?” What, then, is behavior?
Also, how can we test if the animal’s behavior has changed, or if we have met a particular training goal?
First, let us look at the scientific definition of behavior.
Behavior: Anything a person or animal does that can be measured. (Reber, 1995)
Paul Chance, PhD adds this: “In practice, the term usually refers to publicly observable overt behavior. However,
behavior that is available only to the person performing it (such as thinking) may be included if it can be reliably
measured.” (Chance, 2003) In other words, a behavior must be something that is operationally defined and so
is recognizable, observable, and must possess some quality that makes it easy to quantify. To truly know if a
behavior is being elicited by stimulus, we need to define what that particular behavior is. If this is the case, the
behavior should not be open to interpretation by any of the observers. In theory, the observers would all see the
same thing. A behavior could, then, easily be a head shake, a specific vocalization, or a raising of the eyebrows.
Can it be jealousy, stubbornness, or hatred? In other words, can thoughts or emotions be behaviors? According
to the guidelines we have looked at, they can, but only if we can operationally define these terms in such a way
that we can quantify them, if all the observers would be able to recognize them by these operational definitions,
and if we can test these observations.
In most cases we can argue that mental processes are behavior. There are even laboratory methods for measuring
things like thoughts and emotion (e.g. neuron activity, change in heart rate, etc.) but the problem comes when we
go ahead and try to label what the thought or emotion is. Actually, there are at least 3 problems.
First, we don’t tend to have these laboratory-based methods of measuring thought available to us in our institutions.
Even if we did, “physiological manifestations of emotion occur in many animals, but this does not mean that
animals experience emotion in the way that humans do.” (McFarland, 1999) In the same way the animals might
experience emotion differently, observers might observe it differently as well. What one trainer may see as
aggression, another might see as fear, while yet another might see it as hunger. If an animal consistently tries to
bite us, is it because he “hates” us, or could there be another option? For example, could the behavior possibly
have been reinforced, and is now being offered repeatedly by the animal to earn that reinforcement?
Second, even if all of the observers saw the same emotion or thought, when we go ahead and apply a thought or
emotional label we are really just making an assumption as to what the animal is thinking based on some other
behavior we are seeing. Sometimes we do it because it is easy and comfortable. This pitfall also leads us to
the dreaded “A-word;” anthropomorphism, or what Ken Ramirez calls “A trainer’s biggest stumbling block.”
(Ramirez, 1999) Anthropomorphism encourages, and justifies us in thinking for the animal, when in reality, we
should be training behaviors, not thoughts. It also allows us to make excuses for our animals and their responses.
We may often find ourselves reinforcing behaviors that are below criteria, because of an excuse we made.
The other problem with labels is that we limit ourselves in the ability we have to change the behavior we are
seeing. For example, if an animal is “stubborn” or “lazy,” it may be difficult to know how to go about changing
that. How will we know when he is no longer stubborn? How do we measure that? However, if the animal is
simply slow to respond, or is refusing to perform a specific behavior on cue, we have a clear goal to achieve in
order to solve this particular training problem. We can look in our tool bags for the tools that can help us speed up
his response time, or we can put together a plan to reshape the behavior that he is no longer performing. We will
be able to easily measure whether the animal is now responding quickly or performing a cued behavior.
Conclusion
So why do we do it? Why do we apply labels to behavior instead of just acknowledging the specific behavior we
are observing?
Some may say that it is because we are human. The labels we assign and the conclusions we jump to are ones that
reflect the way humans might think or feel in that particular situation. We identify with the labels, and understand
them, which makes them convenient. As a result, we become comfortable with these labels.
Some may say it is just using common sense. The problem with common sense is it often focuses more on the
common part, and less on the sense. As Dr. Friedman said, “Common sense maintains the status quo so we can continue to do what we know best, rather than seeking out the best we can do.” It is a use of anecdotal evidence;
a useful way to develop leads, but only that. Common sense and anecdotes do not, on their own, lead us to a
scientifically based, clear solution.
So how do we avoid it? There are some easy ways to examine whether we are looking at the behavior.
• The simplest way, would simply be to use the verb “to do” when asking yourself questions about the
animals’ behavior. What is the animal doing? When I gave him this cue, what did he do?
• If that still does not give a clear enough picture of what actually happened, I like to ask “What does that
look like?” I often ask this when a fellow trainer relates a story to me and gives a behavioral description
that I cannot interpret clearly enough to feel confident that I know exactly what was going on.
• Another important component is whether we are looking at the environment, and how it contributes to the
behavior we are seeing. If the entire process is simply taking place in the animal’s head, there is little we
can modify in order to see a more desirable behavior from the animal.
Here is a hypothetical conversation between two trainers.
A: “That animal hates me.” (Emotional assumption)
B: “What do you mean by that? What was he doing?”
A: “He tries to kill me when I walk by the cage.” (Thinking for the animal)
B: “Kill you? What does that look like?”
A: “Every time I walk by the cage, he runs to the front, growls, and hits the door with his paw. (A behavior we
can visualize. It still may not be the whole story, but it states observable behavior, takes into account the animal’s
environment and is something we can now work with.)
In this example, The “behaviors” of hating the trainer, or trying to kill the trainer would be hard to modify, and test
in the first place, but the behavior of running at the front of the cage or pawing at the trainer could be modified,
for example by training incompatible behaviors such as stationing in the back of the cage or lying down when the
trainer is present in the environment. These incompatible behaviors could be easily observed, which would allow
us to measure whether or not the animal was still performing the undesired behavior of running and pawing at
the cage front. In other words, the behavior that led to the trainer’s conclusion in the first place can be modified,
which is the real measure of success.
Does this mean the animal actually hated the trainer in the first place, or learned to no longer hate him? We have no
idea. Most importantly, it does not really matter. When we simply define the behavior we are seeing, (What does
it look like?) we can avoid hazards that lead to misinterpretation of the real problem, and as a result, a decreased
inability to solve the real issue. We can also avoid limiting ourselves in being able to solve the training problem,
by being able to now ask “What do we want it to look like?” If we can answer both of these questions, we then
empower ourselves to more easily and successfully interpret and solve our behavioral and training issues.
References
Chance, Paul. Learning and Behavior, 5th Edition. Belmont, CA: Wadsworth Publishing, 2003.
Domjam, Michael. The Principles of Learning and Behavior, 5th Edition. Belmont, CA: Thompson Wadsworth
Publishing, 2003.
Friedman, Susan. Straight Talk About Parrot Behavior. Utah: Utah State University Department of
Psychology. Originally Presented at the Stop PDD Conference, November 2004.
Chance, Paul. Learning and Behavior, 5th Edition. Belmont, CA: Wadsworth Publishing, 2003.
Domjam, Michael. The Principles of Learning and Behavior, 5th Edition. Belmont, CA: Thompson Wadsworth
Publishing, 2003.
Friedman, Susan. Straight Talk About Parrot Behavior. Utah: Utah State University Department of
Psychology. Originally Presented at the Stop PDD Conference, November 2004.
From a course I am taking in canid behavior and wanted to share.
+++++++++++++++++++++++++++++++++++++++++++++
But What Does it Look Like?
Using Observable Phenomena to Interpret Behavior and Solve Training Challenges
Emily Insalaco, Behavior Programs Manager
Natural Encounters, Inc.
Abstract
Animal training utilizes ideas from various branches of science, including animal behavior, behavior analysis and
cognitive psychology. By definition, science deals with describing physical phenomena and the environment in
measurable terms. As trainers, we are encouraged to read and interpret our animals’ actions, and even go as far as
to try to figure out what the animal is thinking. However, this can lead to trainers using terms that may seem like
clear descriptors, but are actually open to interpretation by others, and difficult to operationally define, which in
turn can often make it more difficult to determine the next step in the training plan. This paper will explore how,
by concentrating on only the behaviors we actually observe in our training, and by using measurable, observable
behavioral terminology, we can greatly improve communication among staff members, and empower the trainers
to approach their training challenges with more effective tools.
Introduction and Historical Perspective
I have heard training described as an art. I have heard it called a science. I have also heard that it is both at the
same time. Whatever camp you may sit in, I doubt anyone can argue that training strongly relies on scientific
principles. The field of animal training as we know it in zoos and aquariums was built from behavior science. We
look back on the work of figures such as Thorndike, Pavlov and Skinner for guidance in defining what it is we are
doing. As trainers, we routinely use terms like operant conditioning, extinction, and schedules of reinforcement.
We read textbooks and attend workshops to learn how to apply these terms.
If we read carefully enough, we also find a historical struggle in behavioral science to measure things like thought
and emotion. One of the earliest works in comparative psychology is Animal Intelligence by Georges Romanes
(1882), which attempted to define just that. In his work, he makes reference to mental abilities and processes
but uses various feelings, such as jealousy, to illustrate some of these processes. The response from the rest of
the psychological field was one of backlash, and spurred a movement that led to the improvement of scientific
controls and interpretation of data. (McFarland, 1999) Later studies showed much more discipline in defining
relevant information. Thorndike’s puzzle box experiments were praised because he “showed strict avoidance of
anthropomorphic interpretation of the behavior he observed.” (Domjam, 2003) David McFarland describes the
behaviorist’s view in this way: “Psychology is the study of behavior itself, rather than the mental events. In its
[strongest] form, behaviorism rejects all reference to inner processes in the explanation of behavior.” (McFarland,
1999) Why, then, if we are truly using science to train our animals, do we consistently try, whether consciously
or not, to get inside the animals’ heads? More importantly, is it helping us, or merely making our jobs more
challenging than they need to be, by leading us to misinterpret the problem, and making our end goal more
difficult to identify?
Some Definitions, Quotes, and How They Apply to Animal Training
Susan Friedman PhD stated in her paper Straight Talk About Parrot Behavior that “Science employs methods of
systematic observation of measurable phenomena” and “Science seeks explanations that are testable.”
Using these guidelines, we can ask ourselves some questions that can help get us on the right track. How does the
way we look at behavior fit in with “systematic observations of measurable phenomena?” What, then, is behavior?
Also, how can we test if the animal’s behavior has changed, or if we have met a particular training goal?
First, let us look at the scientific definition of behavior.
Behavior: Anything a person or animal does that can be measured. (Reber, 1995)
Paul Chance, PhD adds this: “In practice, the term usually refers to publicly observable overt behavior. However,
behavior that is available only to the person performing it (such as thinking) may be included if it can be reliably
measured.” (Chance, 2003) In other words, a behavior must be something that is operationally defined and so
is recognizable, observable, and must possess some quality that makes it easy to quantify. To truly know if a
behavior is being elicited by stimulus, we need to define what that particular behavior is. If this is the case, the
behavior should not be open to interpretation by any of the observers. In theory, the observers would all see the
same thing. A behavior could, then, easily be a head shake, a specific vocalization, or a raising of the eyebrows.
Can it be jealousy, stubbornness, or hatred? In other words, can thoughts or emotions be behaviors? According
to the guidelines we have looked at, they can, but only if we can operationally define these terms in such a way
that we can quantify them, if all the observers would be able to recognize them by these operational definitions,
and if we can test these observations.
In most cases we can argue that mental processes are behavior. There are even laboratory methods for measuring
things like thoughts and emotion (e.g. neuron activity, change in heart rate, etc.) but the problem comes when we
go ahead and try to label what the thought or emotion is. Actually, there are at least 3 problems.
First, we don’t tend to have these laboratory-based methods of measuring thought available to us in our institutions.
Even if we did, “physiological manifestations of emotion occur in many animals, but this does not mean that
animals experience emotion in the way that humans do.” (McFarland, 1999) In the same way the animals might
experience emotion differently, observers might observe it differently as well. What one trainer may see as
aggression, another might see as fear, while yet another might see it as hunger. If an animal consistently tries to
bite us, is it because he “hates” us, or could there be another option? For example, could the behavior possibly
have been reinforced, and is now being offered repeatedly by the animal to earn that reinforcement?
Second, even if all of the observers saw the same emotion or thought, when we go ahead and apply a thought or
emotional label we are really just making an assumption as to what the animal is thinking based on some other
behavior we are seeing. Sometimes we do it because it is easy and comfortable. This pitfall also leads us to
the dreaded “A-word;” anthropomorphism, or what Ken Ramirez calls “A trainer’s biggest stumbling block.”
(Ramirez, 1999) Anthropomorphism encourages, and justifies us in thinking for the animal, when in reality, we
should be training behaviors, not thoughts. It also allows us to make excuses for our animals and their responses.
We may often find ourselves reinforcing behaviors that are below criteria, because of an excuse we made.
The other problem with labels is that we limit ourselves in the ability we have to change the behavior we are
seeing. For example, if an animal is “stubborn” or “lazy,” it may be difficult to know how to go about changing
that. How will we know when he is no longer stubborn? How do we measure that? However, if the animal is
simply slow to respond, or is refusing to perform a specific behavior on cue, we have a clear goal to achieve in
order to solve this particular training problem. We can look in our tool bags for the tools that can help us speed up
his response time, or we can put together a plan to reshape the behavior that he is no longer performing. We will
be able to easily measure whether the animal is now responding quickly or performing a cued behavior.
Conclusion
So why do we do it? Why do we apply labels to behavior instead of just acknowledging the specific behavior we
are observing?
Some may say that it is because we are human. The labels we assign and the conclusions we jump to are ones that
reflect the way humans might think or feel in that particular situation. We identify with the labels, and understand
them, which makes them convenient. As a result, we become comfortable with these labels.
Some may say it is just using common sense. The problem with common sense is it often focuses more on the
common part, and less on the sense. As Dr. Friedman said, “Common sense maintains the status quo so we can continue to do what we know best, rather than seeking out the best we can do.” It is a use of anecdotal evidence;
a useful way to develop leads, but only that. Common sense and anecdotes do not, on their own, lead us to a
scientifically based, clear solution.
So how do we avoid it? There are some easy ways to examine whether we are looking at the behavior.
• The simplest way, would simply be to use the verb “to do” when asking yourself questions about the
animals’ behavior. What is the animal doing? When I gave him this cue, what did he do?
• If that still does not give a clear enough picture of what actually happened, I like to ask “What does that
look like?” I often ask this when a fellow trainer relates a story to me and gives a behavioral description
that I cannot interpret clearly enough to feel confident that I know exactly what was going on.
• Another important component is whether we are looking at the environment, and how it contributes to the
behavior we are seeing. If the entire process is simply taking place in the animal’s head, there is little we
can modify in order to see a more desirable behavior from the animal.
Here is a hypothetical conversation between two trainers.
A: “That animal hates me.” (Emotional assumption)
B: “What do you mean by that? What was he doing?”
A: “He tries to kill me when I walk by the cage.” (Thinking for the animal)
B: “Kill you? What does that look like?”
A: “Every time I walk by the cage, he runs to the front, growls, and hits the door with his paw. (A behavior we
can visualize. It still may not be the whole story, but it states observable behavior, takes into account the animal’s
environment and is something we can now work with.)
In this example, The “behaviors” of hating the trainer, or trying to kill the trainer would be hard to modify, and test
in the first place, but the behavior of running at the front of the cage or pawing at the trainer could be modified,
for example by training incompatible behaviors such as stationing in the back of the cage or lying down when the
trainer is present in the environment. These incompatible behaviors could be easily observed, which would allow
us to measure whether or not the animal was still performing the undesired behavior of running and pawing at
the cage front. In other words, the behavior that led to the trainer’s conclusion in the first place can be modified,
which is the real measure of success.
Does this mean the animal actually hated the trainer in the first place, or learned to no longer hate him? We have no
idea. Most importantly, it does not really matter. When we simply define the behavior we are seeing, (What does
it look like?) we can avoid hazards that lead to misinterpretation of the real problem, and as a result, a decreased
inability to solve the real issue. We can also avoid limiting ourselves in being able to solve the training problem,
by being able to now ask “What do we want it to look like?” If we can answer both of these questions, we then
empower ourselves to more easily and successfully interpret and solve our behavioral and training issues.
References
Chance, Paul. Learning and Behavior, 5th Edition. Belmont, CA: Wadsworth Publishing, 2003.
Domjam, Michael. The Principles of Learning and Behavior, 5th Edition. Belmont, CA: Thompson Wadsworth
Publishing, 2003.
Friedman, Susan. Straight Talk About Parrot Behavior. Utah: Utah State University Department of
Psychology. Originally Presented at the Stop PDD Conference, November 2004.
Chance, Paul. Learning and Behavior, 5th Edition. Belmont, CA: Wadsworth Publishing, 2003.
Domjam, Michael. The Principles of Learning and Behavior, 5th Edition. Belmont, CA: Thompson Wadsworth
Publishing, 2003.
Friedman, Susan. Straight Talk About Parrot Behavior. Utah: Utah State University Department of
Psychology. Originally Presented at the Stop PDD Conference, November 2004.
I look forward to learning much here