Border Collie? Malinois? I Don’t Need One!! (Or Do I?)

(This is the next blog post on interesting things I learned from the recent PABA Conference at the University of Guelph. See Part 1 to learn about it plus Dr. William “Deak” Helton.)

Before I begin let me confess two things: One, I absolutely love watching Schutzhund. I say watch because 1) I’ve never trained bitework or tracking 2) I don’t really own a dog that’s “suitable” for it. I think I like it for the same reason I like participating in triathlons. I just think it’s amazing for a single athlete to excel in multiple disciplines.

Two: While I am training my Beagle-mix Petey for Rally Obedience, I do have an eye to the future, so it’s no surprise to my friends I one day want a Malinois or Tervuren. Objectively they just look cool (see Hannah Branigan and Gambit):

But besides looking cool… why do I want one? Why do I need one? Dr. Helton’s lecture on working dogs also included QUITE a bit of interesting information on Breed Intelligence/Trainability and the overall Selection Template for working dogs. Maybe I don’t need a Belgian Shepherd after all?

Stanley Coren’s Rankings (1994)

His book, “The Intelligence of Dogs” suggested that Border Collies, Poodles, German Shepherds, Golden Retrievers, and Dobermans were the most intelligent and…

Afghans, Basenjis, Bulldogs, and Chow-Chows were the least intelligent.

Most dog enthusiasts have heard of the book, but what you may not know is how the rankings were achieved. It was from surveying AKC and CKC judges for their personal opinions about dog breed intelligence. Not exactly objective or quantifiable data.

Dog Sports Data Holds Answers

Helton went to the world of dog sports to seek out data. Competitive sports of any sort is useful for scientists as a) everything is measured and data is accurately tracked and b) scoring and timing is consistent since that’s required for fairness and honesty in competition.

Going to agility, he found two interesting facts:

  1. Precision (frequency of faults) was equal across all breeds, assuming level of practice and experience of dog and handler were equal
  2. Speed (time through course) was better for “elite breeds”

No mystery there – Border Collies are designed to move fast. But, all dogs can be trained to have reliable contacts or enter weave poles accurately.

Scott and Fuller – A Better (and Older) Study:

So let’s forget about Coren. Instead let’s look to Scott and Fuller:

“We can conclude that all breeds show about the same average level of performance in problem solving, provided they can be adequately motivated, provided physical differences and handicaps do not affect the tests..” Scott and Fuller (1965, p.258)

Physical Characteristics Are Key:

So, if all breeds are generally equal in intelligence, then it’s the physical traits of the dogs we should be focused on.

If you’re looking for speed, we’d want a dog like a Greyhound. They are built for running.

If you’re looking for a fighter, we’d want a dog like a Pitbull. They are built for fighting.

Here are a few other considerations:

Size and Thermoregulation – dogs that are too large will overheat; dogs that are too small will freeze. That’s why sled dogs are pretty small – they are designed to dissipate heat while sledding long distances.

Ocular Overlap – Dogs with eyes spread far apart (sighthounds) have less visual acuity in the centre, but are designed to see the horizon. Dogs with more centre-set eyes have more acuity in the centre and see less on the peripheral. These dogs can see what’s in front, and as a result, are more focused on what’s in front of them, and less distracted by things happening on the horizon – obviously making them less distracted during training.

Built for Speed vs Built for Biting: the broader, wider mouths of dogs like the Pitbull are designed to cause maximum damage during biting, whereas the mouth of a Greyhound would deliver far less bite strength.

The Goldilocks Rule

But, historically, dogs serve many purposes, and the most useful ones do multiple tasks well (going back to Schutzhund – obedience, tracking, and protection). So trainers have a bias towards dogs that aren’t too big, aren’t too small, are fast enough, but can kick ass enough. (German Shepherd!) Helton suggests its no wonder why all the dogs favoured by the AKC and CKC judges are all roughly the same size, shape, and even generally have similar shaped heads.

Perhaps the most important consideration for breed selection is simply convenience of the size and shape to the trainer. Let me tell you, Petey is a little dog, and it is a pain to feed him during heelwork training. A dog whose mouth is exactly where your hand is during heelwork is a lot easier to train. When I was working with Chase, the Jack Russell, his physical size required me to get down on the floor and use props to elevate him:

Since these “Not too big, not too small, just right” sized dogs are easier to observe (and hence reinforce accurately), and comfortable to train (no bending, crouching, or gymnastics required), that would bias trainers towards them. If I had two dogs – a Daschund, and a Labrador Retriever, I would most likely spend more time training the Lab, since I would get tired of having to hunch over the Doxie for extended periods of time!

Innate Skill or Born With It?

Helton suggested that practice is the most reliable predictor of success.

However, he shared a really interesting study done by Slabbert and Rasa (1997) – Observational learning of an acquired maternal behaviour pattern by working dog pup: an alternative training method?

Puppies that were allowed to stay with their mother and simply OBSERVE their mother perform work were significantly more likely to become skilled at that work later in life.

Helton went on to explain that this may explain why there is a tendency in human family trees to see multiple generations pursue the same craft – grandfather was a doctor, dad was a doctor, and the son is a doctor. Is there a doctor gene? A more likely explanation is early role models greatly affect early socialization and gives the child a huge head start in getting practice in the skill. The child of a professional hockey coach is much more likely to be dragged around to hockey rinks before they can even walk, and have a huge advantage to his peers when he finally hits the ice.

Perhaps keeping your border collie puppy with mom for a bit longer, and letting him watch mom do weave entries and contacts could become part of raising the perfect agility dog?

Love Your Little Dog

Maybe you’re like me and have a little dog. Fine, maybe Petey doesn’t look as cool as Gambit because he’s tiny, but that’s OK – I’ve gotten used to having to crouch to feed in position.

There are many jobs out there where little dogs SHOULD be the ideal candidate.

Like for Search and Rescue. Why would we want a big heavy dog like a German Shepherd, that occupies more space on transport, requires more food and water, and can’t crawl into small spaces, when we could have more Momo the Long-Haired Chihuahuas searching for victims?

Or more seriously why would I want a heavy dog that will set off mines doing mine detection, when a Beagle or a Jack Russell would be light enough NOT to cause us to get blown up?

Mine Detection Dog in Bosnia

Schutzhund For All?

And maybe I could try bitework with Petey. After all, he is part Jack Russell, just like Claudia Romard’s Mr. Murphy. After all, he’s just as intelligent as a German Shepherd.

Why Can’t My Dog Stay on Task? (The Vigilance Decrement)

I just came back from the third and final day of the Professional Animal Behavior Associates symposium for 2011 – this year, the presentation was focused on working dogs (search and rescue, detection, etc.) I’ll let you read about the lineup here. The PABA conference draws a lot of folks from across Canada, including my friend Katherine Mutzke from Clever K9 Sports in Squamish, BC, Alice Fisher of Dogsmart from Vancouver, Carolyn Clark, Karen Pryor Academy Faculty, and Nicky Barnham of Carolark in Ottawa.

This year I knew I had to go, because I knew absolutely nothing about the topics being presented, and only knew of one of the presenters speaking (Steve White). Having no expectations going in, I went in and right off the bat with the first speaker I was delighted with what I was learning!

The headlining speaker that I found brought the most new information to the table for me was Dr. William “Deak” Helton, the author of “Canine Ergonomics – The Science of Working Dogs“. He had a ton to say, and MOST of what he said I had never heard of before – enough to fill multiple blog posts for weeks… so in the interests of brevity I am going to write about his discussion of the Viligence Decrement. I believe it has tremendous application to us in pet dog training, because it can explain why our dogs fatigue and stop working in class or in dog sport. In class I often point out to students that their dogs “look fried” or “check out”. Here, a scientist was actually going to explain what might be going on in our dogs when this happens!

(Before I begin, you have to understand a bit about my background – I stopped taking sciences in Grade 11 – public school, I focused on Liberal Arts, and in university, I majored in business. So go easy on me if I get anything wrong)

Vigilance Decrement – What is it?

Vigilance decrement is defined as “deterioration in the ability to remain vigilant for critical signals with time, as indicated by a decline in the rate of the correct detection of signals”

In real life, it is as simple as an airport security officer missing contraband items hidden in baggage under x-ray, or a working dog failing to detect and indicate the presence of a landmine. For the airport security officer, that might be the shape of items on screen that fuzzily match what contraband looks like. For a detection dog, that might be the scent of a landmine. In both cases, shapes and smells are just signals.

In my opinion it’s reasonable to assume that a verbal or visual cue is also just a signal. A cue is a signal to indicate there is an opportunity to perform a behaviour and earn reinforcement. Why would a dog respond to a cue one minute, and assuming motivation is sufficient, and the cue is delivered identically, why would a dog not respond to the cue?

Important to note – we are always assuming that the learner is sufficiently motivated. For training, we would assume the food is appealing enough, or the toy desirable enough to want to continue doing work.

Two Competing Theories of Vigilance Decrement

The first theory is boredom. The task is so easy and repetitive that the subject withdraws effort and stops paying attention – and the subject is then distracted and daydreams.

The second theory is fatigue/resource theory. Mental energy is limited, and task performance draws on mental energy. Unless replenished, mental energy is depleted.

Helton does not believe the boredom model holds true. If it held true, you should get greater vigilance decrements if you gave people easier tasks vs. harder ones. In fact, the opposite was true – in studies with humans (male and female) and rats, the harder the signal detection test, the faster the decrease in performance.

In plain English – if I gave you a book on quantum mechanics and said “hey read this”, you’d probably look very bored by page three. But it’s not because it’s too easy – it’s because it requires a ton of mental energy to follow it.

Applied to dog training? In class, many students (especially newer ones, with greener dogs) often attribute the dog’s unresponsiveness as boredom. If a dog stops responding to a known cue in class, assuming sufficient motivation, it’s NOT because their bored, but in fact, have run out of mental energy. I believe their brain is no longer detecting the signal/discriminative stimulus.

Evolutionary Function of “Shutting Down”

Simple! As Helton explained, the brain is not unlike an engine. Excessive use will cause damage, so by shutting down, it is protecting itself. The brain, literally, heats up from use, which is why brain damage can occur with high fever, and we’re mentally more capable if the brain is kept cold. Helton shared that you can measure the temperature inside the ear canal as an accurate way to monitor when the subject is going to burn out!

Refilling Mental Energy

Through rest! That’s why agility people swear by Crate Games – so mental energy is preserved or restored between performance.

On “Daydreaming”

The likelihood of the subject to start “daydreaming” or focusing on off-task activities increase as resources decrease. Just like in my last post – where Petey only noticed the bowl of food in the training area at the very end of the training session. Ignoring distractions requires frontal lobe capacity – going for the food bowl is a “lizard brain” type decision. Using the frontal lobe draws on mental energy, and once depleted, lizard brain kicked in.

A Model for Predicting Detection Success

Drawing from human-models, Helton shared his model for predicting whether the expert (in this case the detection dog) would correctly respond:

P(A) = sS – eFEF + (ex EX + vV)

Probability of Attending = Salience of of the Olfactory Target – Body Movements + Expectations + Value of Information

Salience = How big/obvious is the target?

Body Movements = How much does the dog have to worry about navigating the environment while working? i.e. searching in dangerous rubble vs. a lab environment

Expectations = If you only train the dog to search the floors, he’ll never think to sniff up high where the item may be hidden

Value of Information = I think this means the importance of the consequence to the dog – What’s in it for me if I respond?

Applying his Model to Scent Detection

Rather than have the dog do the search on-site, Helton suggests the dog could stay inside a controlled laboratory environment, and just be given air samples of the targets to smell inside the lab. For container search, for example, air samples of containers could be captured and then brought to the dog to sniff one by one. The salience increases, and body movements are minimized. Another example would be for detecting breast cancer. Rather than have the dog sniff the actual patient, patient breath samples would be collected, and then sent to the dog.

In Norway, there is a company that is doing this – they call it RASCO – Remote Air Sampling for Canine Olfaction.

Applying this Model to my Training

I have thought about how I might apply this to training (and already do):

Salience = make cues easy to discriminate for the dog. If verbal cues, avoid using similar sounding cues (down and bow). Keep them clean and consistent so mental energy doesn’t have to be wasted thinking “was that really the sit cue? or was it something else?)

Distractions/Interruptions/Noise = I think that anything that causes mental energy to be expended, either consciously or subconsciously, will cause performance to drop quickly. So that’s why we follow good training practices to train with high intensity and make it easy for the dog to focus and preserve mental energy.

Expectations = I think this closely relates to the process of generalization and gradually training in as many environments, with as many variables (over time) as possible.

Value of Information = This is just creating good reinforcers, to make sure that our cues are worth paying attention to and acting upon.

Other Ideas (Buck Rogers Stuff?)

Random musings, but…

Should I put an ice pack on Petey’s head between training sessions? Or right before we do our runs at the CARO trial?

Caffeine reduces vigilance decrement in people AND honeybees (yes, it was in the presentation). Caffeine pills for Petey before training sessions or at least trials?

Fun Fact (PABA Related)

The Canadian Border Services Agency’s detection programs were created with the help of the University of Guelph. Their dogs are trained using positive reinforcement – they work for a ball! Three working dogs were brought for demos over the course of the weekend. This black lab cross was too much for his past owners – not a good house pet for the average dog owner – but perfect for working dog life – and worked like a champ looking for currency in the room (he catches people that lie when they say they aren’t carrying over $10,000 in currency).

Currency Detection Dog
Currency Detection Dog

The next post I’ll do about this topic will delve into Helton’s findings regarding breeds and dog characteristics and their intelligence and suitability for work. As a Beagle owner I loved hearing this. Maybe I don’t need that Malinois after all… ?

Canine Cognition Research – Updates and Some Results

Hello Readers –

Did you know that Krista Macpherson, a researcher from the University of Western Ontario, conducts canine cognition research at When Hounds Fly?  Many of her test subjects are current and past students!  Here is an update she sent out to past test participants, including a deck she presented at a conference with interim findings back in March.

Want to volunteer your dog and your time?  Please email and mention if you are (or aren’t) a student (current or past) at school.  She is always looking for volunteers to help.


From Krista:

Hi all!

First of all, my apologies for not following up with you all sooner…I keep telling myself “I’ll talk about results once I have collected a bit more data”…but of course I’m always collecting a bit more data, so I might as well talk about results now!
Most of the people on this email list have participated in one (or both) of two experiments…the first was a timing experiment, and the second was a counting experiment…I’ll speak to the progress of each study below:
Timing Experiment
In this experiment, we had one or two “Manners Minder” machines spit out a a treat on a set interval (either every 30 seconds, or every 1 minute).   With this study, we were interested in determining whether over time, the dogs would come to anticipate the arrival of food from the machines, and thus approach the machine as the time interval approached.  This ended up being problematic, because most of the dogs easily figured out that the ideal solution was simply to lay beside the machine the entire time and wait for the snacks!  While this was a good strategy on the dogs’ part, it made it impossible to measure timing in terms of proximity to the machine.
Luckily, the psychology department at Western has a great team of woodworkers and electricians to build experimental materials…and they have revamped the Manners Minder machine for me.  It is now bolted to a plank of wood, and has a large button hooked up to it.  So the dogs will still be able to get the food at a set interval…they’ll just have to hit a button first.  This means that instead of measuring timing in terms of the dogs’ proximity to the machine, we can measure timing in terms of the number of button presses that occur as the interval approaches.  I’m going to start pilot testing this new apparatus in the next couple of weeks!
Counting Experiment
With the counting experiment, dogs watched as the experimenter (me) dropped different quantities of food into each of two bowls.  If dogs can count, then they should reliably choose the bowl with the larger quantity of food.  What we found is that when the food ratio involves “food versus no food” (which in this experiment, was done using 3 versus 0 pieces of food , and 1 versus 0 pieces of food) the dogs are extremely good at the task (in fact, they rarely choose the wrong bowl.  In any other combination (e.g. 1vs3, 4vs1, 2vs3, 3vs4, 2vs1)  the dogs really struggle with this task, and to get right answer only about 60-65% of the time (and since there are only two bowls, chance performance is 50% on this task).
It is interesting that dogs struggle on the counting task, because pretty much any other species tested on this task has performed quite well at it.  A couple reasons that they may not be good at this task are as follows:
1.  As we have selectively bred dogs to attend to us (herd our sheep, guard our homes etc.), we might have “bred out” other characteristics that were more important for survival in the wild, like numerical discrimination.
2. Dogs are dumb (I personally am not a fan of this explanation!).
Another possibility is that this procedure, for whatever reason, was not “meaningful” to the dog.  In talking to another researcher who was working on a similar project, is occurred to me that I have always had the experimenter drop the food into the bowl, while the owner held the dog (but did not otherwise interact with the dog).  What if the owner was the one to drop the food into the bowls?  Would the dog pay more attention to the task?
Long story short, I changed the procedure so that the owner was the one dropping the food into the bowls–and preliminary results show that in many cases this made a big difference to the dogs. We need to test this more formally, but these initial findings are really exciting!
I presented the counting data (minus the new “owner dropping the food” version of the experiment) at a conference in Florida in March, and it was very well received.  I have attached a copy of the slides that I presented, for any of you who may be interested.
Upcoming Projects
I am currently looking for some more dogs to do more counting studies with–I need naive dogs, who have never done this task before.  If you have a dog or know someone who has a dog who would like to participate, please let me know!  I will be running sessions at When Hounds Fly on April 30th/May 1st, as well as May 14th/15th, between 2pm and 7pm. It takes about an hour to complete the experiment.
As always, many thanks for all of your help with these projects!

Want to see the slide presentation?  Click on any of the images below: