“Champagne Training on a Beer Budget”
Traci Belting, Point Defiance Zoo and Aquarium
During this conference, many of our colleagues will be
discussing the application of operant conditioning techniques and
particularly the role that positive reinforcement plays in modern
polar bear management in zoos and aquariums.
This paper will take a more reflective approach in order to highlight the evolution of the formal polar bear training program at the Point Defiance Zoo and Aquarium. It will focus on some of the behavioral challenges that we have faced and mention some of the strategies that we have found helpful.
While many of you have decades of experience with polar bears, I had not had the pleasure prior to working at Point Defiance. I had trained marine mammals and birds, but this part-marine/part-terrestrial carnivore was beyond the realm of my experience. I still remember my first day as we toured the polar bear holding area. As I walked past the fence, a bear sat calmly watching the procession. Just as I turned my head to watch the keeper in front of me, the bear reared up on his hind legs, thunderously hit the fence with his front paws, and remained erect, standing a full three feet taller than me. All the while he was emitting a loud “SSSSHHHHH” sound that reverberated straight through me. I took my cue, respectfully exited the area, and waited for my heart rate to return to normal. After more than seven years now working with polar bears, I can proudly say that their behavior has become more predictable and I no longer have to bring a change of underwear to work!
To be quite honest, there was a time in the not-so-distant past when the staff at Point Defiance believed that they didn’t have enough personnel or time to train polar bears. The general philosophy seemed to be that it was appropriate to put time and energy into training some animals, while others – because of their disposition – should not or could not be trained. Polar bears fell into the latter category. In hindsight, this thinking seems ridiculous because in general all animals behave for one of two reasons:
Given that the bears were aware of every aspect of their environment, we were training them whether we wanted to admit it or not. As animal caretakers, we influence how and what our animals learn. There are times we make a conscious effort to train a particular behavior for husbandry, research, education, or entertainment purposes. But, more often, we influence an animal’s behavior inadvertently through our actions or our husbandry routines. In effect, we are always training and this fact needs to be kept at the forefront of strategic planning when it comes to behavioral issues.
Our formal bear-training program was initiated just over six year ago when the zoo’s keeper teams were restructured and the marine mammal training staff was integrated in with the existing polar bear keepers. At the time, our collection consisted of 3.2 bears. Two of the bears were older (a 30-year-old male and a 20-year-old female) while the other 3 bears were each less than two years of age. None had any previous training history. We decided to use the zoo elephants’ well-established protected-contact training program as a model. We initially thought that training the bears would be a long, arduous process because we had very little time to devote to it. Our staff was relatively small and, because we had other animals to care for, we could only carry out one or two training sessions per day, each rarely exceeding 10 minutes. But the bears responded quickly. Within a matter of months, we succeeded in training all 5 bears in basic management behaviors including:
Once upon a time we thought we didn’t have enough staff or time to train. In reality, we didn’t have enough staff or time NOT to train.
While training obviously benefits the staff with regard to time management and husbandry goals, the greatest benefit is to the bears. Fundamentally, operant conditioning relies on an animal’s ability to choose. The behavioral choices that an animal makes will impact the consequences that follow. Providing an animal with the opportunity to choose replicates to some extent the pathway for learning that occurs in the wild. One obvious difference is that a bear’s incorrect choice in the wild could result in death, whereas in a controlled environment like a zoo or aquarium, a more likely consequence is simply the removal of positive reinforcement.
The cornerstone of protected-contact training is the understanding that you can’t make an animal do anything it doesn’t want to do. Our job as trainers is to make the learning process so positive and so interesting that the bears are motivated to participate. As a training program grows, sometimes the goals can become so focused on specific behaviors that we lose sight of the bigger picture. We need to create a positive attitude toward learning. Remember that it is the behavior that is being reinforced or punished, not the animal! Because we as trainers are the ones delivering the reinforcement, we have the power to shift the balance of learning towards success. Although it’s normal for an animal to fail when learning a new behavior, it’s up to us to make sure the victories far outweigh the failures.
In order to set a bear up to succeed in a training program, it is important to:
Two male bears were already living at Point Defiance when Boris and Kenny arrived. Suddenly we were faced with the challenge of managing a group of four adult male bears. The new arrivals had been previously trained using operant conditioning, only they were used to making choices to avoid punishment rather than acquire rewards.
During their initial quarantine, Boris and Kenny would flinch in avoidance and offer submissive postures in response to any quick movement from the staff. It took time to gain their trust. Eventually they began to accept food from us through the fence and building relationships began. We needed extra time to habituate the bears to the presence of a target pole, then to condition it as a positive stimulus, because the bears’ initial response to the sight of the pole was avoidance.
I think that it is a testament to the power of positive reinforcement that within a matter of weeks we were able to overcome the bears’ negative responses and lay the foundation for successful acclimation into our existing bear management program.
Once we had the bears tentatively following us, we attempted to get voluntary weights. While most of the surfaces in our holding areas are relatively level, our scale is slightly raised. The bears had identical reactions to stepping on the platform for the first time: each one immediately sat down and remained perfectly still. Despite our best efforts to encourage them along, they remained on the platform. We tried tempting them with food but they would only eat what they could reach without moving their feet from the platform. Apparently their previous experience with remaining still on a circus platform was stronger than their brief reinforcement history with us. We had to regress back to spending more time on basic follows in other parts of the holding area in order to shift the balance in our favor and overcome their predilection to stall on the platform.
February 6, 2003 was a landmark day for Boris and Kenny. They finally passed quarantine and were introduced onto the main exhibit. For the first time in decades, they were able to submerge themselves and swim in natural seawater. Initially we had planned to exhibit Boris and Kenny together, but their aggressive interactions intensified as their health improved. Although Boris was initially dominant over Kenny, this dominance hierarchy switched within a few days of their introduction onto the main exhibit. Kenny became so dominant over Boris that Boris appeared reluctant to move and remained relegated to the far corner of the exhibit. Despite trying various strategies – including varying the shift order, adjusting the time they spent together on exhibit, and reinforcing Kenny for allowing Boris to move – Boris remained submissively motionless as far away from Kenny as possible. We decided to separate the bears and rotate their time on exhibit. Although it took several days for Boris to trust that Kenny wasn’t out there hiding somewhere, he is now an incredibly active and playful bear. This change in his demeanor has reinforced the importance of our goal of providing all the bears in our care with the most positive environment possible.
As trainers, we all agree that we should focus on the positive and avoid circumstances that elicit undesirable results. It is inevitable, however, that animals will develop or offer behavior that we don’t want. Animals will make mistakes on occasion. When this happens, it is essential that we not overreact and draw attention to the problem.
When a bear offers undesirable behavior, certain things should be considered:
Despite our best efforts and intentions, animals still will learn undesirable behaviors. Therefore it is important to know some effective techniques for reducing the frequency of these behaviors. Operant extinction is a process by which the reinforcement of a previously learned behavior is discontinued. If a behavior no longer produces any reinforcing consequences, it will fade over time. For example, if a bear bangs on a door and the door opens, the behavior is reinforced. But if the bear bangs on the door and the door no longer opens (i.e. door-banging no longer produces the desired reward), the frequency of the behavior will decrease. A behavior will often get worse before it gets better, but you can bet that if the behavior is never reinforced, it will disappear eventually.
Studies have shown that it is much easier to extinguish a behavior if an animal has an alternative way to get reinforced. A technique often referred to as Differential Reinforcement of Incompatible behavior, or DRI, allows for this. With DRI, a response that is incompatible or competitive with the undesirable response is reinforced. Let’s say, for example, that a bear is shifting through a gate before being cued. Before opening the gate, ask the bear to lie down. A bear cannot lie down quietly and rush out of a gate at the same time. These behaviors are incompatible.
Solving behavioral problems is an ongoing process. As trainers, we must constantly consider a variety of training strategies that might prove helpful. This is especially true at a facility that would be crazy enough to manage a collection of four adult male polar bears. Given that conservation is one of the primary missions of the Point Defiance Zoo and Aquarium, we felt privileged to participate in the rescue effort of the Suarez Brothers bears and to provide sanctuary for Boris and Kenny. As animal care staff, we are incredibly proud of the progress they have made in a relatively short amount of time. Regular visitors to the zoo can no longer distinguish between Boris and Kenny and our resident bears Blizzard and Glacier. This is testament to how much Boris and Kenny’s behavioral and physical health have improved.
We currently exhibit only two bears at a time, but it remains possible that through careful behavioral evaluation and the application of positive reinforcement we can increase the number of combinations of bears that can be exhibited together. We have begun to introduce all of the bears informally through a fence in our holding area. With careful planning, we hope to expand those small windows of social interaction that occur naturally in the wild and provide our bears with a diverse and enriching experience. With the information gathered at this conference and the opportunity to learn from each other, we will continue to provide our bears with champagne training.
Acknowledgements: I would like to thank our animal care staff: Lisa Triggs, Margaret Gaspari, Mike Messersmith, Rebecca Stocker and Jean Joseph. Jean deserves additional accolades for her photography assistance. I would like to thank our Administration for their ongoing support of our bear training efforts and of course…a huge thank you to Blizzard, Glacier, Boris and Kenny.
References: Klein, S.B. (1991). Learning: principles and applications (second ed.) New York: Mc Graw-Hill, Inc.
Kazdin, Alen E. (1994). Behavior Modification in Applied Settings. California. Brooks/Cole
Goldblatt, Allen. (2002). Extinction: Some Facts for Trainers. Soundings. Magazine of the International Marine Animal Trainers Association. Volume 27, Number 3 12-15.
Sullivan, Tim. (2002). The Yin and Yang of Positive Reinforcement Training. Soundings. Magazine of the International Marine Animal Trainers Association. Volumer 27, Number 2 28-30.
Back to Behavior Training
This paper will take a more reflective approach in order to highlight the evolution of the formal polar bear training program at the Point Defiance Zoo and Aquarium. It will focus on some of the behavioral challenges that we have faced and mention some of the strategies that we have found helpful.
While many of you have decades of experience with polar bears, I had not had the pleasure prior to working at Point Defiance. I had trained marine mammals and birds, but this part-marine/part-terrestrial carnivore was beyond the realm of my experience. I still remember my first day as we toured the polar bear holding area. As I walked past the fence, a bear sat calmly watching the procession. Just as I turned my head to watch the keeper in front of me, the bear reared up on his hind legs, thunderously hit the fence with his front paws, and remained erect, standing a full three feet taller than me. All the while he was emitting a loud “SSSSHHHHH” sound that reverberated straight through me. I took my cue, respectfully exited the area, and waited for my heart rate to return to normal. After more than seven years now working with polar bears, I can proudly say that their behavior has become more predictable and I no longer have to bring a change of underwear to work!
To be quite honest, there was a time in the not-so-distant past when the staff at Point Defiance believed that they didn’t have enough personnel or time to train polar bears. The general philosophy seemed to be that it was appropriate to put time and energy into training some animals, while others – because of their disposition – should not or could not be trained. Polar bears fell into the latter category. In hindsight, this thinking seems ridiculous because in general all animals behave for one of two reasons:
- To acquire desirable outcomes
- To avoid undesirable outcomes
Given that the bears were aware of every aspect of their environment, we were training them whether we wanted to admit it or not. As animal caretakers, we influence how and what our animals learn. There are times we make a conscious effort to train a particular behavior for husbandry, research, education, or entertainment purposes. But, more often, we influence an animal’s behavior inadvertently through our actions or our husbandry routines. In effect, we are always training and this fact needs to be kept at the forefront of strategic planning when it comes to behavioral issues.
Our formal bear-training program was initiated just over six year ago when the zoo’s keeper teams were restructured and the marine mammal training staff was integrated in with the existing polar bear keepers. At the time, our collection consisted of 3.2 bears. Two of the bears were older (a 30-year-old male and a 20-year-old female) while the other 3 bears were each less than two years of age. None had any previous training history. We decided to use the zoo elephants’ well-established protected-contact training program as a model. We initially thought that training the bears would be a long, arduous process because we had very little time to devote to it. Our staff was relatively small and, because we had other animals to care for, we could only carry out one or two training sessions per day, each rarely exceeding 10 minutes. But the bears responded quickly. Within a matter of months, we succeeded in training all 5 bears in basic management behaviors including:
- Responding to a verbal “good” or visual point as a bridging stimulus
- Targeting
- Follows
- Sit. This behavior was especially important for positioning a bear on our scale.
- Down
- Front paw presentation
- Mouth inspections
- Gate Desensitization. The movement of a gate no longer cued an animal to shift; instead, they waited for a signal from their trainer. This provided us with much greater flexibility in moving bears to and from the main exhibit and throughout our holding facility.
Once upon a time we thought we didn’t have enough staff or time to train. In reality, we didn’t have enough staff or time NOT to train.
While training obviously benefits the staff with regard to time management and husbandry goals, the greatest benefit is to the bears. Fundamentally, operant conditioning relies on an animal’s ability to choose. The behavioral choices that an animal makes will impact the consequences that follow. Providing an animal with the opportunity to choose replicates to some extent the pathway for learning that occurs in the wild. One obvious difference is that a bear’s incorrect choice in the wild could result in death, whereas in a controlled environment like a zoo or aquarium, a more likely consequence is simply the removal of positive reinforcement.
The cornerstone of protected-contact training is the understanding that you can’t make an animal do anything it doesn’t want to do. Our job as trainers is to make the learning process so positive and so interesting that the bears are motivated to participate. As a training program grows, sometimes the goals can become so focused on specific behaviors that we lose sight of the bigger picture. We need to create a positive attitude toward learning. Remember that it is the behavior that is being reinforced or punished, not the animal! Because we as trainers are the ones delivering the reinforcement, we have the power to shift the balance of learning towards success. Although it’s normal for an animal to fail when learning a new behavior, it’s up to us to make sure the victories far outweigh the failures.
In order to set a bear up to succeed in a training program, it is important to:
- Design a training plan for every behavior with as many steps as necessary to clearly communicate the goal.
- In any one training session, move one or two steps toward a desired response and stop. Sometimes, as trainers, we get too exuberant when we get a positive response and we want to move quickly to the next approximation. So often, however, this increases the likelihood of confusion and/or frustration and could potentially result in the inadvertent reinforcement of less-than-desirable criteria.
- Be consistent. Ideally, a single trainer would work on a particular behavior throughout the behavior acquisition phase and multiple trainers could then maintain the behavior. It’s important to recognize that some animals can tolerate more trainers than others; when multiple trainers are being used, good communication among those trainers is critical.
- Take into consideration the animal’s hunger level. We found it helpful to feed the bears prior to training sessions in order to take the edge off. We all know how much a bear’s appetite can fluctuate seasonally. If a bear is too hungry, it can become so intent on obtaining food rewards that it loses focus on the training task and can become easily frustrated when food isn’t delivered.
- Avoid the use of aversive stimuli. I am sure that you will all agree with me that except in extreme emergencies, we need to avoid the use of aversive stimuli. While negative reinforcement undeniably can be used to shape behavior, it does nothing to build trust between bear and trainer and certainly doesn’t support a positive learning environment.
- Keep things interesting and varied. Once a
training program is established, some factors that can be varied
include:
- Location for performing behaviors
- Order of behaviors (When doing paw presents, for example, don’t always ask for the right paw first, etc.)
- Number of training sessions per day
- Length of sessions
- Time of day sessions occur
Being unpredictable can include:
- Shifting a bear onto exhibit and then right back off
- Opening a shift door and then cueing the bear to follow you in the opposite direction
- Starting a training session but, rather than asking for a behavior, feeding the bear instead for calm responses.
The possibilities are endless.
- Consider each bear’s individual history so as to select the most effective and appropriate techniques to proceed with training.
Two male bears were already living at Point Defiance when Boris and Kenny arrived. Suddenly we were faced with the challenge of managing a group of four adult male bears. The new arrivals had been previously trained using operant conditioning, only they were used to making choices to avoid punishment rather than acquire rewards.
During their initial quarantine, Boris and Kenny would flinch in avoidance and offer submissive postures in response to any quick movement from the staff. It took time to gain their trust. Eventually they began to accept food from us through the fence and building relationships began. We needed extra time to habituate the bears to the presence of a target pole, then to condition it as a positive stimulus, because the bears’ initial response to the sight of the pole was avoidance.
I think that it is a testament to the power of positive reinforcement that within a matter of weeks we were able to overcome the bears’ negative responses and lay the foundation for successful acclimation into our existing bear management program.
Once we had the bears tentatively following us, we attempted to get voluntary weights. While most of the surfaces in our holding areas are relatively level, our scale is slightly raised. The bears had identical reactions to stepping on the platform for the first time: each one immediately sat down and remained perfectly still. Despite our best efforts to encourage them along, they remained on the platform. We tried tempting them with food but they would only eat what they could reach without moving their feet from the platform. Apparently their previous experience with remaining still on a circus platform was stronger than their brief reinforcement history with us. We had to regress back to spending more time on basic follows in other parts of the holding area in order to shift the balance in our favor and overcome their predilection to stall on the platform.
February 6, 2003 was a landmark day for Boris and Kenny. They finally passed quarantine and were introduced onto the main exhibit. For the first time in decades, they were able to submerge themselves and swim in natural seawater. Initially we had planned to exhibit Boris and Kenny together, but their aggressive interactions intensified as their health improved. Although Boris was initially dominant over Kenny, this dominance hierarchy switched within a few days of their introduction onto the main exhibit. Kenny became so dominant over Boris that Boris appeared reluctant to move and remained relegated to the far corner of the exhibit. Despite trying various strategies – including varying the shift order, adjusting the time they spent together on exhibit, and reinforcing Kenny for allowing Boris to move – Boris remained submissively motionless as far away from Kenny as possible. We decided to separate the bears and rotate their time on exhibit. Although it took several days for Boris to trust that Kenny wasn’t out there hiding somewhere, he is now an incredibly active and playful bear. This change in his demeanor has reinforced the importance of our goal of providing all the bears in our care with the most positive environment possible.
As trainers, we all agree that we should focus on the positive and avoid circumstances that elicit undesirable results. It is inevitable, however, that animals will develop or offer behavior that we don’t want. Animals will make mistakes on occasion. When this happens, it is essential that we not overreact and draw attention to the problem.
When a bear offers undesirable behavior, certain things should be considered:
- Is the behavior being bridged properly? The bridge is designed to deliver positive feedback with pinpoint accuracy in order to let an animal know instantaneously that it is performing a behavior correctly. Behavioral breakdown often stems from confusion associated with inconsistent bridging.
- Are all staff consistently using the correct SD (signal)? As trainers, we need to watch each other and discuss any variations in SD’s and make sure that we are being as clear as possible.
- Are the behavioral criteria clear? If one trainer is accepting one set of criteria and another is expecting something else, an animal easily can become confused or frustrated.
- Has stimulus control broken down due to predictability? This can be especially challenging in the case of animals that are shifted on and off exhibit at set times.
- Do overriding social situations exist? Boris and Kenny’s dominance shift exemplifies how significantly social pressure can impact behavior.
- Is there a physical problem causing the animal to fail?
- Are there environmental influences or changes? Because these bears have such keen senses, any major change in their environment can have an impact on behavior, both positive and negative.
Despite our best efforts and intentions, animals still will learn undesirable behaviors. Therefore it is important to know some effective techniques for reducing the frequency of these behaviors. Operant extinction is a process by which the reinforcement of a previously learned behavior is discontinued. If a behavior no longer produces any reinforcing consequences, it will fade over time. For example, if a bear bangs on a door and the door opens, the behavior is reinforced. But if the bear bangs on the door and the door no longer opens (i.e. door-banging no longer produces the desired reward), the frequency of the behavior will decrease. A behavior will often get worse before it gets better, but you can bet that if the behavior is never reinforced, it will disappear eventually.
Studies have shown that it is much easier to extinguish a behavior if an animal has an alternative way to get reinforced. A technique often referred to as Differential Reinforcement of Incompatible behavior, or DRI, allows for this. With DRI, a response that is incompatible or competitive with the undesirable response is reinforced. Let’s say, for example, that a bear is shifting through a gate before being cued. Before opening the gate, ask the bear to lie down. A bear cannot lie down quietly and rush out of a gate at the same time. These behaviors are incompatible.
Solving behavioral problems is an ongoing process. As trainers, we must constantly consider a variety of training strategies that might prove helpful. This is especially true at a facility that would be crazy enough to manage a collection of four adult male polar bears. Given that conservation is one of the primary missions of the Point Defiance Zoo and Aquarium, we felt privileged to participate in the rescue effort of the Suarez Brothers bears and to provide sanctuary for Boris and Kenny. As animal care staff, we are incredibly proud of the progress they have made in a relatively short amount of time. Regular visitors to the zoo can no longer distinguish between Boris and Kenny and our resident bears Blizzard and Glacier. This is testament to how much Boris and Kenny’s behavioral and physical health have improved.
We currently exhibit only two bears at a time, but it remains possible that through careful behavioral evaluation and the application of positive reinforcement we can increase the number of combinations of bears that can be exhibited together. We have begun to introduce all of the bears informally through a fence in our holding area. With careful planning, we hope to expand those small windows of social interaction that occur naturally in the wild and provide our bears with a diverse and enriching experience. With the information gathered at this conference and the opportunity to learn from each other, we will continue to provide our bears with champagne training.
Acknowledgements: I would like to thank our animal care staff: Lisa Triggs, Margaret Gaspari, Mike Messersmith, Rebecca Stocker and Jean Joseph. Jean deserves additional accolades for her photography assistance. I would like to thank our Administration for their ongoing support of our bear training efforts and of course…a huge thank you to Blizzard, Glacier, Boris and Kenny.
References: Klein, S.B. (1991). Learning: principles and applications (second ed.) New York: Mc Graw-Hill, Inc.
Kazdin, Alen E. (1994). Behavior Modification in Applied Settings. California. Brooks/Cole
Goldblatt, Allen. (2002). Extinction: Some Facts for Trainers. Soundings. Magazine of the International Marine Animal Trainers Association. Volume 27, Number 3 12-15.
Sullivan, Tim. (2002). The Yin and Yang of Positive Reinforcement Training. Soundings. Magazine of the International Marine Animal Trainers Association. Volumer 27, Number 2 28-30.
Back to Behavior Training