Tuesday, August 6, 2013


 I went to see Blackfish this past weekend.  What a fascinating film.  This documentary is about the Orca Tillikum, and what has happened to him.  There are a few periods of his life that I think are particularly relevant to dog trainers, especially when it comes to handling more than one dog at a time.  There is a right way to do that and a wrong way to do that, and understanding what happened to Tillikum can help us to learn more about the right way to work with multiple dogs at the same time. 

The short version of Tillikum’s story is that he was abducted (yes, I am going to use that word, and yes, I realize that I am biased!) from his pod or family when he was between two and four years old; I think the film claims that he was four, but he could have been a little younger.  He was brought to Sealand in Vancouver, British Columbia, Canada where he was housed in a small net structure with two mature female whales.  At night they were herded into a small indoor pool so that animal rights activists couldn’t cut the net and let him get away.  They used a very unusual training procedure on him in the hopes of training him to do tricks, and then he killed a trainer.  Sealand was closed down and he was sent to SeaWorld in Florida and several years later he was somehow or another involved with the death of a drifter who may have climbed the fence and gotten into his pool with him overnight.  In 2010, he dragged another trainer down during a show and killed her.  In part at least, the documentary claims that Tillikum was psychotic from the weird training process that was used on him in Vancouver.

I am not a marine mammal trainer, and I don’t even play one on TV, but I do understand training very, very well.  The first vignette I want to examine involves what happened to Tillikum at Sealand in Vancouver.  The treatment that the film described Tillikum experiencing in Vancouver was NOT training.  What they did was to set this whale loose in their tank with two unknown females, and if he did what the trainers asked the other whales to do on cue, they all got fed a treat.  If he didn’t, none of them got anything.  In order to get them to go into the cramped smaller enclosure at night, the whales were all food deprived before they were put in, and fed only when they were all in for the evening.  On the surface, this sounds like something that might work.  When you dig deeper there are some serious problems, and they are common problems I see when people try and get multiple dogs to do things in groups.

The experienced whales were able to do the known behaviours, and when the trainer cued them to do those, they would hear a bridging signal (a whistle) and they would get a fish.  So whale A is cued to jump up and touch a target, and if she does that she hears the whistle and she gets a fish.  So far so good.  Along comes Tillikum who doesn’t know the behaviour and when the trainer cues target touching and the female whale touches the target, but Tillikum doesn’t, no bridging signal and no fish.  It doesn’t take long for the experienced whales to figure out that when Tillikum is around, fish are scarce.  This frustrated the experienced whales.  What happens when learners are frustrated is that they are much more likely to be aggressive.  In whales, the females are usually in charge of how things happen, and the males live on the outside of the pod and do what the females want them to do.  In an ocean with lots of room, if a female whale asks a male whale to get out of her way, there is a lot of space for him to politely do that.  In SeaLand, there was not much space, and thus, when Tillikum didn’t get out of the way, the female whales started to gang up on him and harass him and rake him with their teeth and ultimately, injure him on a regular basis. 

What the trainers seemed to hope was that by rewarding the whole group for satisfactory performances, Tillikum would fall into line and do what the other whales were doing in order to get fish rewards.  This is where the whole thing falls apart.  For the most part training happens on an individual basis.  In order for groups of animals to learn to work together, each individual must first learn his part.  Working groups of animals mostly relies upon each animal learning his part in the event separately and then integrating that into the whole.  Tillikum had no idea what the trainers wanted so he could not get the right answer.  When he didn’t get the right answer, the other whales got penalized by not getting the reward in spite of their best efforts.  Further to this, the whales were being food deprived for part of the day in order to get them into their overnight holding tank.  This meant for grouchy whales, and what resulted was two adult female whales ganging up on an adolescent male whale.  This was not an environment that would foster learning and this was not a training plan that would result in specific behaviours increasing.

I often see owners of multiple dogs making similar mistakes.  If one dog misbehaves, all the dogs in the household miss out on an opportunity.  One case I particularly remember was a woman who would only exercise her dogs if all the dogs had behaved well all day.  If at 4pm any of the dogs had misbehaved over the course of the day, she just wouldn’t take any of her four dogs for a walk.  Interestingly, most of the days that walks happened were days when she had been out of the house for a significant portion of the day so that she could not record any transgressions.  What this resulted in was a houseful of dogs who all needed more exercise than they were getting and who were thus always anxious, frustrated and getting into mischief.

If the documentary is correct in how they describe Tillikum’s training, then whomever was in charge of his training plan didn’t understand the mechanics of teaching groups of animals to perform together.  There are some situations where animals can be hooked together and forced to do the same thing at the same time and the novice animal will learn from the experienced animal.  This happens for instance when a naive or novice horse is hitched into a couple with an experienced horse in order to learn to pull a cart.  The first horse learns what to do because doing what is incorrect is uncomfortable.  Negative reinforcement is the tool that the horse experiences that makes him fall into line and pull the cart.  Quite simply, if the horse doesn’t pull in tandem with the other horse, his harness will pinch and pull and be uncomfortable.  He is most comfortable by matching what the other horse is doing.  The same is true of sled dogs who pull in concert with one another, and hounds that run in braces; the more experienced dog sets the pace and it is uncomfortable for the naive dog to resist.

Most of the time, when you are teaching a skill, each individual must learn that skill to fluency before you can put the animal into a group to perform in tandem.  This is especially true when you are using positive reinforcement techniques.  You can sometimes effectively teach animals to work together using negative reinforcement if making a mistake is uncomfortable for the naive learner, but this can create problems too; frustration on the part of the experienced animal can lead to aggression between the animals or towards the handler.  Image credit: anilah / 123RF Stock Photo

So why didn’t Tillikum just figure out what the trainers wanted him to learn and then all the whales would have gotten their fish?  The first part of the problem lies in the set up of the training scenario.  If there were a way to set up negative reinforcement so that the whale was more comfortable and less annoyed by choosing the right behaviour, then he would have learned quickly and efficiently how to behave so that he got what he wanted.  Instead what they did was withhold the reinforcement from the whale who got the right answer, and give Tillikum no feedback about what was desired.  It didn’t take long for the other whale to notice that the best predictor of a lack of reward was Tillikum.  At that point, the experienced whale started to harass Tillikum, further decreasing his likelihood of getting the right answer.  In short order the older more experienced whales were routinely harassing the source of their problems; Tillikum. Training became little more than a part of an elaborate hazing of the  youngest and least experienced whale.

I have seen this exact scenario play out when a family gets a second dog.  The first dog knows that he has to sit before his meals.  If the second dog doesn’t know how to sit, and the family decides that the rule is the rule, then the dog who doesn’t know the drill doesn’t get any feedback about what is expected and the dog who does know the drill gets to wait a long time before the end of the ritual comes along.  It doesn’t take long before the first dog realizes that there is a much longer delay to dinner when the second dog is around and then I get a call telling me that although the first dog accepted the second dog in the beginning, there are now problems.  

This leads of course to the question, “How should be train two animals?”  The answer is deceptively simple.  Teach each dog the target behaviour individually.  When each animal can perform the target behaviour fluently, get the two animals to perform the behaviour together.  Make all reinforcers contingent on correct performance of the behaviour to begin with and then increase the difficulty until the contingency is performing the behaviour correctly on cue at the same time as the other individual.  In real life, this might look like this.  You teach the old dog to sit for his dinner.  You get a new dog.  In a separate room or at a separate time, you teach the new dog to sit for his dinner.  Then you bring both dogs into the room and ask them both to sit for their dinners and you feed them as they are successful (it is best to do this with a second handler to help at this stage).  When both dogs are quickly sitting when asked, then you can put the food down only when both dogs are sitting.  Easy peasy.  Frustration is minimal, and success is fast and reliable by not jumping ahead and skipping steps.

The place where I see this fall apart the most often for people with two dogs is leash walking.  If you are going to leash walk one dog, why not two at once?  Often the first problem is that the older dog doesn’t have good leash skills to start with.  If the experienced dog isn’t very good at the behaviour, the naive dog isn’t going to magically figure out the drill.  The second problem is that giving feedback in the dynamic environment of the average neighbourhood street is really difficult and rarely successful.  Trying to do this with two dogs is just an exercise in frustration!

The next part of Tillikkum’s story happens when he kills a trainer in Vancouver.  Tragically a young woman who was working with him was dragged into the pool and drowned.  Information about how this happened is sketchy at best, but the reality is that when you have an animal who is being kept in inadequate facilities (three whales in a small pool and even smaller holding tank for 30% of their lives is not enough space for these large seafaring mammals), who is being food deprived part of the day (the only way that they were successful in getting the three whales into the holding tank was to withhold food for many hours before they were put in for the night), and who is being mentally frustrated, aggression is not an uncommon result.

Neither one of these dogs has good leash skills, and walking them together is not going to help either of them learn to walk nicely on leash.  Notice that the Boston Terrier in front is showing whale eye and a pinned ear and a tightened mouth; he is frustrated and if the other dog harasses him, it would not be a surprise if there was an aggressive incident between the two dogs.  Image credit: digitalduck / 123RF Stock Photo

I have seen this in family pets too.  When the dog isn’t getting sufficient exercise, when he is crated for too many hours a day, when he is harassed by the other dogs in the home, when training is inefficient and unfair, then dogs can become aggressive and hard to handle.  Often this starts out with the first dog harassing the second dog, but this can evolve over time into the second dog becoming aggressive to family members.  This is one of the reasons that I always recommend that families bring both their dogs to an obedience class when they are in the process of adding a second dog to the home; this forces you to spend individual time with your dog, and avoids a lot of big problems.  When you are adding a dog to the home, each dogs will need to have their needs met, will need individual attention and will need guidance and boundaries to help them be successful when they are learning the new ropes of having both dogs in the home.


  1. This post is really incredible, one of the most helpful I have ever read, indeed. Professional Pet Grooming

  2. All excellent points, and good comparisons. So sad about the Tillikkum; I hate that this practice continues.