American Psychological Association
117th Annual Convention
Toronto, August 6-9 2009
Invited Plenary Address: Stanley Coren, Ph.D., FRSC, University of British Columbia.
HOW DOGS THINK
Dr. Coren was introduced as professor emeritus, and Ph.D. from Stanford, best known for his books on dogs but also known for his work in areas of neuropsychology, cognition, and sensorimotor functioning. He lives in Vancouver with 2 dogs, his wife, and her cat.
OK, so where do we begin in looking at how dogs think? This was not to be a "dog & pony show" where animals are merely a source of entertainment, but a loving and empirical look at how dogs actually think, employing a scientific framework not unlike the work of Piaget and others who look at early acquisition of language, "object constancy", and goal-directed behavior. The methodology may vary, but some of the well-known dimensions of human learning, such as learning of words and sequences and quantity, are quite amenable to rigorous study. Some of the results are quite interesting, and Dr. Coren presents an engaging blend of historical and sociological canine functioning, as well as some empirical study of cognitive functioning. Along the way he joked about the way people dearly love their breeds and tend to forget the origins of how that breeding came about. Clearly he loves them all, whether smart or not so smart, and his humor underscored how we have many preconceptions of what "smart" really means - to dogs. Here are some scientific studies of dogs, along with some video clips (in his presentation) and clever, engaging graphics to illustrate some of the talents and foibles of (wo)man's best friend. Here's a taste of this dense-packed, multi-media presentation.
First, Dr. Coren said, "You gotta understand: Our thinking about dogs comes in 2 extremes:
1. "Dogs are 4-footed people in fur coats", versus
2. "Dogs are unthinking concomitants - a programmable chassis...."
Where did dogs come from, anyway? Possibly from wolves, and/or jackals, coyotes, selected fox species, and wild dogs. Their remains from earliest history suggest they've long been "man's best friend", with one curious exception: "Neanderthals never formed an alliance with dogs, which may have contributed to their extinction." Embraced by generations of societies, dogs are arguably "the best example of behavioral engineering we've got." They have long been genetically engineered to produce specific behaviors and body types.
For 14,000 years dogs have been our friends. They were the first domesticated animal, in fact, before sheep, cattle, cats, chickens or horses.
And so, we ask, How intelligent are dogs?
Dr.Coren showed a video clip to illustrate cleverness. (Imagine) The doorbell rings and the human goes to the front door as the dog quickly helps himself/herself to some food on the kitchen counter, and then a drink. Very quickly the dog runs out and fetches the cat by the scruff of its neck and deposits it on the kitchen counter just before the human returns to the kitchen. Clever?
Some have looked at animal intelligence in terms of "cephalization index", or the size of the brain relative to the body.
The top of the list for big brains:
- Homo sapiens
Labrador retrievers do relatively well on the cephalization index; moles and rats do particularly poorly in the brain mass continuum.
Coren sees intelligence as being more than one unidimensional entity, positing an instinctive intelligence, adaptive intelligence, and working/obedience intelligence. Instinctive Intelligence references what the dog was bred to do, and thus this is a hard measure to employ comparatively. Adaptive Intelligence refers to problem-solving ability and adaptation skills. (Even within breeds there can be wide variability.) It is in this domain that much research has been done, including language, quantitative, and problem-solving ability. Lastly there is Working and Obedience Intelligence, which can be operationalized as "how well the dog will learn to perform commands and execute instructions for humans." This is said to be the canine equivalent of "school learning", and tends to be seen as a good general indicator of ability in this domain.
Even after arriving at some broad types of "intelligence" is there not an aspect of subjective judgment? Is retrieving, for example, truly "smarter", say, than herding? One needs to consider, in the case of dogs, what a dog was bred to do - E.g., "herding dogs herd & retrievers retrieve".
There are several standardized tests which have been developed to assess dogs along various dimensions of (canine) intelligence. "Obedience performance" is considered a good measure", one we see at play in contests, for example.
Two hundred eight [dog competition] judges were asked to identify the top breeds for intelligence and the responses were "very consistent". With an N of 99 (of the 208 judges), the most esteemed dog in terms of intellect is not, as some might expect, a poodle. "Yes. The poodle is a retriever. He did not ask for that stupid haircut! Poodles can compete as retrievers." [Still, poodle lovers can take comfort: they are #2 on the list, second only to border collies.]
Beagles, however, "descended from the bottom, so the chair you're sitting on is more trainable than a beagle." Other dogs attributed with great intelligence also may not be as sharp as they look. Dr. Coren wondered aloud, "Why keep an Afghan?"
[Video of an advertisement for long, flowing hair -- pick your choice and visualize for 10 seconds.]
Afghans' breeding makes them particularly good at running after antelope, and they are a beautiful sight to watch, but not too many of us do actual work with antelope. Fine as eye-candy, they are at the bottom of the list, cognitively.
Of course, individual tastes do vary, and social popularity changes too over time. While we know hounds are low on the scale in terms of natural dog-brights, Pharaoh Rameses the Great (circa 1275 BC) buried his hound with him. Actually, an image of the mummified dog reveals it to be a greyhound. [Feel free to speculate what cultural or religious priorities made a greyhound, in particular, so prized.]
But hounds are, practically speaking, not very specialized: "All a hound needs to be able to do is get to the prey and not eat it." Of the bottom 10 (in estimated canine intelligence among trainers and borne out by experimentation), 60% are Hounds, and 100% are "old breeds" that have been around for centuries. Moreover, "hounds were designed to be independent hunters that did not depend upon interaction with humans to do their job."
For the curious (and not representing any personal opinion by this neutral reporter!) here are the top and bottom breeds as reported in terms of their canine cognition rating:
Smartest Dogs (Descending Order)
- Border Collie
- German Shepherd
- Golden Retreiver
- Doberman Pinscher
- Shetland Sheepdog
- Labrador Retriever
- Australian Cattle Dog
The Bottom 10 (Ascending Order)
- 101. Basset Hound
- 102. Mastiff
- 103. Beagle
- 104. Pekingese
- 105. Bloodhound
- 106. Borzoi
- 107. Chow Chow
- 108. Bulldog
- 109. Basenji
- 110. Afghan Hound
Dogs can learn up to 165 words or so (including gestures), and a few "super dogs" can learn around 200 words, similar to the human equivalent at around 2 years. Some of our language comprehension tests shown to be useful with young children have been adapted to use with dogs (e.g., the MacArthur Communicative Development Inventory). Rico the border collie (considered overwhelmingly to be the smartest dog) understands over 200 words.
Research has shown dogs engage in "exclusive thinking", meaning they can be taught to label objects. Shown 4 familiar objects and a 5th unknown one, after identifying the 4 known objects by name, it readily learns that the 5th word being said must be the name of the still-unidentified 5th object rather than any of the others. They hold this learned information for around 4 weeks.
Dogs can demonstrate some remarkable skills with receptive language, as illustrated in some fascinating video clips where dogs follow both explicit and subtle direction. [E.g., David Hartwig and Skidboot.] They also listen carefully to other dogs and can learn specific types of barks. (Wolves too can learn social barks by being around dogs.)
Dogs are less capable with productive language (sounds). The average dog has a repertoire of about 35 different sounds and signals. What's the best we get? [See for example, elRellano.com]
Some retrievers can count to 5. The video shows a dog following the directions: "Paco, fetch it up!" and Paco gets one of five objects. "Paco, fetch it up!" and Paco gets object # 2. Then 3, and 4, and 5. If the human then asks Paco to "fetch it up" again, Paco won't be fooled and knows there are only five and he has fetched them all. Some researchers look at size discrimination among dogs, too (e.g., Norton Milgram, at University of Toronto).
In looking at this level of thinking and learning, Piaget's work was invoked, and his interest in whether (human) infants can add/subtract. Dogs took tests similar to Piaget's object permanency tests, and they generally knew whether to expect an object or not, unless the Experimenter cheats and removes or adds something in which case the dog stops and looks confused, *knowing* that something changed unpredictably. Some of the most sophisticated dog-specific research on object permanency derives from the work of Gagnon and Doré with their "Invisible Placement Test", similar to Piaget's procedure, though modified. One interesting finding is that dogs, whose survival skills need to develop relatively quickly, demonstrate object permanency equivalent to that of an 18-month old child at only 8 weeks.
Geometry? Dogs don't care about triangle formulas, but they can find the shortest distance between 2 places.
Do dogs dream? REM studies suggest yes, the average-sized dog will have a dream (REM) about every 20-25 minutes, each lasting 2-3 minutes. Smaller dogs tend to have shorter but more frequent dreams - a pug, for example, exhibits 1 minute of REM for every 10 minutes asleep. Usually REM surpresses movement but dog owners know that sometimes dogs do appear to be chasing or running in their sleep. As we can't simply ask what their dreams are made of, that remains a realm for speculation.
Dogs have some unique differences from other "smart" primates, like chimps. Chimps can become absorbed by their mirror reflection and realize it is themselves, play around and mug, etc. Dogs don't. Why? Is it that they're not so vane? More likely it's that their keen sense of smell doesn't match the image on the glass and they lose interest or dismiss it as not meaningful. (Beckoff, U of Colorado, did some heroic "yellow snow" research to look at this important factor in dog processing and daily experience: smell).
Then there are other tasks of great intelligence and occasional mystery, like the dogs from the Hospice of St. Bernard. They are trained in groups of 3, where one dog remains in the area of an accident, a second keeps the victim warm and calm, and the 3rd seeks help. What happens is fascinating as dog teams seem to demonstrate preferences and end up choosing whether to become a "goer" or "stayer". Learning happens independent of simple reward and punishment explanations. Social learning plays a key role across many breeds and typical tasks. For example, herding and herd-guarding dogs learn faster if sent out to the flock with experienced dogs. Dogs are also attentive to human gestures and behavior, having been selected and bred to do so for centuries.
We know dogs learn through watching and imitating other dogs. Can they learn from watching humans? Yes. One study in Budapest used a V-shaped barrier where the way to get to the treat was to do the opposite of what was tried, and back off enough to go around the side rather than directly proceed forward. Interestingly, the dogs imitated the human model exactly, and if a hole was cut to provide an easier approach, the dogs stuck with the first solution they learned.
"Dogs have a theory of mind", figuring out intent, for example, when the human points at something. Wolves won't do this, though they will visually track the hand. More: If 2 people have a treat and one person is blindfolded, the dog will beg only to the one who can see. Dogs can not only be practical like this, but clever in other ways, such as *deceiving* another dog so as to get its food (illustrated in a few video clips). [A comparison chart shows that the success rate implementing deception slides, but not badly "relative to people".] Although religious and cultural belief has opposed such notions (as consciousness implies a soul), laboratory experiments, to say nothing of our daily experience, suggest that indeed dogs have conscious intent, knowledge, and experience. Descartes, on the other hand, argued that animals were all merely machines. The era of Behaviorism in psychology (e.g., Watson) continued to supress any real efforts at exploring the inner life (or even overt behavior) of dogs, as pigeons and rats were preferred in the lab. Only Darwin made the case that animals do indeed have consciousness.
Dogs of course do learn human words and their meaning. They seem to readily understand the human's intention too, like "BALL, get the BALL", or more complex behaviors such as "fetch" and "roll over". One tip teaching: the most powerful place to repeat an action word is at the end of a sentence.
Finally this should be mentioned: Dogs can learn undesirable behaviors too! [A clip was presented of a dog smoking a cigarette and blowing smoke.] And dogs do master some rather complex activities. [Clip of a dog pushing itself along on a skateboard and then gliding, doing a few laps across a parking lot. Currently making the rounds on YouTube/Facebook, both a border collie as well as a very athletically talented bulldog.] Another clip shows a border collie, our brightest dog, doing a choreographed dance routine with a Charlie Chaplin character.
And as time ran out, the speaker received a rousing applause.
Disclaimer: I try to ensure accurate reports of study results, names, dates, etc., and use a combination of verbatim notes, presenter materials, Power Point data summaries, and direct follow-ups with presenters. If I have inadvertently misstated or mis-typed any information (names, dates, numbers, etc.) I would be grateful for any corrections and will be sure to update/correct any articles I present pertaining to these presentations.