Peer-Reviewed Papers on Dog Ethology

Just caught wind of this thread - I'm in too!  Will read and post comments as time permits (and right up my alley - gotta put all that post-secondary education to some use!)

I read through once and then, re-read some again, including the conclusion.
 
There is some mention of the dog's physical limitations in pulling a string. For example, a cat's paw, IMHO, is more prehensile, able to curl and grasp better. A dog usually grabs with his teeth. I believe this is why dogs are "mouthy." It's about the only tool they have for anything besides digging. The results show that the dog does not necessarily learn which string to pull unless one pull met with success, which will cause a rapid learning to pull that one string. But that they don't have the cognitive ability to look and think "well, that diagonal string is connect and that's the one I need to pull." They keep pulling strings until they get something, starting with the string nearest them. This test of cognitive ability is one that is normally connected with testing in humans and other primates.
 
Here are some of the implications I can think of, regarding these results. A) humans and dogs think differently and have different mental abilities. In addition, IMO, just because a dog cannot do something that a human can, does not mean the dog lacks a thinking ability or is a successful organism. Is the dog smart like a human? No, but he may be smart like a dog.
 
B) dogs following human signals are more successful in the test. Dogs are social animals. In the results of this test and from other evidence, dogs succeed much more in understanding human signals than do wolves. This has been attributed to domestication and breeding dogs to respond to man. An implication is that a dog is more readily willing to look to a human as a leader or for cues, than a wolf is. This has implications in training. The human must lead because the dog will look to him to lead. In the test, stringent methods were used to prevent a human from accidently signaling a dog which string to pull. Just as there are innate behaviors in a dog, one of them might be an innate behavior to seek the company of humans. They are a form of canid that is friendly to humans. It does not mean they are not wolf-like. Some assumptions cannot be made intuitively.
 
Would the gist of this study say that dogs lack a problem-solving ability normally attributed to primates, including man, means they are stupid or deficient? Not that I can think of. A truly fair test would include a test of wolves, a test of NGSDs, a test of Dingos, and a comparison of all of them to determine if any canid can accomplish the string pull problem solving.
 
Another implication is that humans and dogs are not equal, per se, but complimentary, almost symbiotic. It is theorized that the dog "no longer" has this problem-solving ability because the human provides. I can only accept that in an inconclusive theory because, before dogs were domesticated, I don't think this particular ability (which is only shown lacking in this one test) was necessary in time immemorial. I do think dogs are quick learners. Does this inability to solve a string pulling problem count as an innate behavior? As in, a dog innately cannot solve a means-end problem involving string pulling. Innately trying the closest string and, whether by accident or design, finally pulling the right string for a treat?
 
I'm reminded of a cartoon in a science mag that I used to read. A monkey is placed in a room and the door has a peephole. The scientist looks in, only to see the monkey looking back at him.
 
I also wonder how this test result may differ or coincide with the experience of field dog work, or other dog work that relies on the thinking of a dog. For example, Siberian Huskies are intelligent and independent. In some cases, they had to be. They can hear and smell better than humans can and might detect ahead on the trail a thin spot of ice or a crevass and must ignore the musher's command and stop or divert the team to avoid a catastrophe. Are some purpose-bred breeds better at problem-solving than other breeds? Should this test be expanded to try several different breeds? Is it a result of pure-breeding?
 

Ron, jumping the gun a little bit, aren't you? []

Nonetheless, I guess we may as well discuss while the paper is fresh in some of our minds and people can chime in when they get around to reading it.

I have a slight problem with their methods as I understand them. I wonder if the dogs simply don't bother trying to comprehend means-end because they get the treat eventually regardless of what they try first? I remember having a discussion like this with some people about pigeon navigation. The scientists doing the study were disappointed that homing pigeons don't make their routes more efficient once they learn them. But that's a very human way of looking at it. Humans naturally strive for efficiency and don't have the risks stacked against them that other animals do. Many of us thought that if there was no need to deviate from the known route to make it more efficient, then it would be detrimental to deviate, because what if you got lost, or discovered a sparrowhawk territory, or some other unforseeable danger in deviating from the known route. So just because pigeons didn't economise their route home doesn't necessarily mean they can't economise.

Similarly with the dogs, as far as I could make out, the trial continued until the dog pulled the correct string. I think that really reduces the motivation in the dog to get it right the first time. If they don't get it right, they still have a chance to get the treat by trying something else. I've seen my dog madly run through her repertoire of taught behaviours before I get a chance to give her a command, hoping that she'll hit on the right behaviour and get the treat she knows I've gone and got just for her. I'm wondering if we'd see more thoughtfulness in dogs if the trial ended immediately after the first string pull, regardless of whether it was the right one or not. As long as the dogs were trained that they got nothing and we all leave when they pick the wrong string, maybe they would make less errors. It makes me wonder how they train sniffer dogs, because that's something that they really don't want the dogs to make errors on.

Anyway, maybe it wouldn't make any difference to the performance of the dogs, but I feel like you can't say for certain that dogs don't understand means-end until you're sure they are trying their hardest to get it right the first time. We saw from the first couple of experiments that the dogs went for the correct strings most of the time, but it wasn't as hard in those trials for the dogs to work out which string was the right one. Perhaps it was actually quicker for them to try the obvious string, and then the other one if the obvious one yielded no treats, then it would be to try to work out which string was the right one to pull in the first place.

I have to work all weekend, but I'll try to get it read by Monday evening!
Thanks for taking the time to do this, BTW. []

But that's a very human way of looking at it. Humans naturally strive for efficiency and don't have the risks stacked against them that other animals do

I agree with you there. Hence, just because the dog failed this test may not be conclusive evidence that they can't problem solve.
 
ETA:
Humans have limitations. Evolution-wise, we are incomplete. The South American Mountain Gorilla is more "evolved" than we are in respect to it's environment. They grow a thick pelt, are several times stronger than we are, can run faster and climb trees better. Their physiology is more complete for interaction with their environment. What humans have is a combination of odd talents that other animals don't necessarily have in the same degree. Opposable thumb, large forebrain, omnivorous, higher levels of abstract thought, and an un-paralleled ability to manipulate our own environment. Man is the only species to wear the hide of another species. And we have to wear something in the winter, as we don't grow a double pelt to survive that. I think the jury is still out on whether we have the evolutionary advantage. One odd talent peculiar to the species of Homo Sapien is the ability to destroy an entire planet. In the end, this may make us an evolutionary dead-end.
 
We are also ego-centric. We constantly compare things to ourselves. Has this affected the test and our interpretations of the results? It is the fact that sometimes, the observer affects the observed by observing. An example can be found in Quantum Mechanics. The electrons in orbit around a nucleus have a spin, noted up or down, for clarity's sake. When an instrument is applied to observe the electron, the force field of the device will, by it's nature, align the spin of the observed electron. It is inescapable and, the observed is affected by the observer. This, among other things, is a thorny problem in Poincare's 3 body problem. The observer becomes one of the initial conditions in the anlaysis. A friend with a doctorate in physics did his thesis on this problem in connection with chaos theory. So, do have a similar problem in this test? A test, by it's own admission, was developed to judge developing cognitive ability in primates.
 

The comments made so far have been good!
 
I think the main point the experimenters were trying to make is that, based on their experiment, dogs don't seem to be able to formulate a plan in order to reach a goal, i.e. they don't think about what the "best" and quickest solution to the problem is, they just try the most obvious, simplest solution first (i.e. paw closest to the food), and if that fails, try something else. I think the experimenters do attribute problem solving abilities to the subjects, just not a means-end understanding (see the General Discussion).
One explanation for why dogs (and other animals) don't understand means-end stuff is because animals need a strong conceptual system (that most likely involves language) in order to represent the problem/solution so that they can plan.
 
I think the main problem I see with this experiment, similar to what Ron was saying, is that there is very little evolutionary basis for a dog to be able to do this sort of task at the level the experimenters were hoping. Perhaps if they used a design that was more ethologically relevant, the dogs would have performed better?
I also find their training trials a little bit suspect- partly because they don't comment on how dogs performed in those- maybe they learnt something during the training trials that hindered performance in the tests.
 
 
 

ORIGINAL: Vinia

 I think the experimenters do attribute problem solving abilities to the subjects, just not a means-end understanding (see the General Discussion).
One explanation for why dogs (and other animals) don't understand means-end stuff is because animals need a strong conceptual system (that most likely involves language) in order to represent the problem/solution so that they can plan.

I think the main problem I see with this experiment, similar to what Ron was saying, is that there is very little evolutionary basis for a dog to be able to do this sort of task at the level the experimenters were hoping. Perhaps if they used a design that was more ethologically relevant, the dogs would have performed better?
I also find their training trials a little bit suspect- partly because they don't comment on how dogs performed in those- maybe they learnt something during the training trials that hindered performance in the tests.

Good points. I immediately thought of the evolutionary basis when I started reading. When I was working on small birds, our research group had real problems with a large, predatory songbird called a currawong. Currawongs are a little smaller than crows, but have a long beak and they're spooky smart. They raise their babies on the chicks of other birds and they were upsetting experiments by picking off all our nests. It got so bad we started caging the nests using wire, but the currawongs quickly learnt that they could bend the wire out of shape by charging it repeatedly, then they reached through the wire and grabbed at the nest and they'd just pull little bits of it until they could get to the babies. It's almost a natural string-pulling experiment. I can't see where dogs might encounter something like that, though.

Something else I thought about while running errands today. We went to Canine Comissary after I stopped at the Vitamin Shoppe. At the pet supply store, they have these fresh baked turkey treats. I remember a time before and Shadow and I were in line waiting as they were busy. He knew I had treats in my hand. He ran through all the obedience postures he knew, trying to figure the right one to get a treat. While he didn't exactly plan which order to do the tricks in, would not the independent decision on his part to do obedience constitute a means-end scenario? That is, he wanted the treat and he usually gets treats in obedience training. So, may be he ought to try the sit. No? The down. No? Stand on hind legs and get a treat from the hand without grazing with the teeth. No? The sit and shake hands. No? I was not looking for a particular obedience move and had not commanded him to do one.

Back to human cues. At the register today, the clerk wanted to give him a treat. He commanded a sit. Shadow sat. He commanded a down. Shadow downed. And got his treat and made me proud.

I'm not sure how long range a dog could think. So, I don't know if the whole "let's hang around humans and get food" thing that dogs, as a species, do would be a means-end on a grand scale. I think that is more likely an evolutionary thing. The dogs that were docile enough to hang around humans got food more regularly, lived longer with greater food supply, and learned to pick up cues from humans, as it led to better living. I think man and dog have affected the evolution of each other, but not necessarily in a radially adaptive way. I say that in the sense of how I learned it, in that an animal does not evolve something because it needs it. An animal mutates and either survives better or doesn't. This would explain why humans still have an appendix. Also, it would explain why dogs have carnassials yet their biology is best suited by an omnivorous diet. Man feeds them an omnivorous diet but they haven't needed to evolve different teeth. That is, their survival in eating an omnivorous diet has not been hampered by them not having big, wide grinding molars and lateral mastication.

And I agree with Vinia's point. Perhaps another test that isn't meant for primates. However, how could we define what would be means-end solution in a dog, which is different than a primate?

How often have we heard that a whale is possibly as intelligent as a human yet they wouldn't necessarily pass the string pull. Or, would they? Or, would they grab all the strings in their big mouth and yank the box off of its anchors?
 
ETA: let me hedge my statement a little bit. I'm not exactly sure if man and dog have affected each other's evolution, other than the statement about breeding in the study. That is, what is dog may have simply been a different canid than, say, the gray wolf, and not necessarily descended from the gray wolf, in spite of Wayne's summation, and that this dog canid had, as a freak mutation, or whatever, a sensitivity to man, as dolphins do. I think that makes better sense.
 
I may draw some wrath for not buying into Wayne's and the Smithsonian's opinion that dog descended from the gray wolf. I still think they came from a common canid ancestor, though, by all evidence, the dog and the gray wolf are the most genetically similar, as is man to the chimpanzee, each example having only between 1 to 2 percent difference in genetics. In human evolution, no one has suggested that man evolved from the chimp. It is still considered that we evolved from a common ancestor. I tend to think the same of dogs, in spite of current speculation to the contrary.
 

All good points.  Besides the general question of whether dogs understand a means-ends relationship and can plan based on that understanding, I had some questions about the set-up of the experiment itself.  From what I understand, dogs don't see like humans do, and I believe they have limited color vision.  Could they have had problems actually seeing the strings (they apparently used red and blue strings, although I've heard that dogs see the color yellow much better than blue or red?).  Also, as we know, some dogs just aren't very food motivated and if the motivation isn't there, why try to figure out the fastest way to get the food, when as a dog, you know you'll get it sooner or later?  They also conducted the experiment in some cases in a testing room at the university, but some times at the home of the owner, or the dog club.  I would think that the different environments would effect the results in that some dogs would be stressed in new enviroments and not learn as quickly...
 
And finally, dogs are very "in the moment" creatures.  As others have said, maybe they don't "need" to figure out the most efficient method of string pulling because they are not as goal oriented as humans.  

Excellent points, Janey. Yes dogs not only do not see all the colors that primates do, but they have limited focal ability, especially close in. They can see movement a long way off but they will smell what it is before they can actually focus on it. Dogs see like I do. Not so good close up and a bit of trouble at extreme distance but well enough in between, a general far-sightedness. The canine eye lacks some of the cones that a primate eye has. Plus a string may not look like string to them, close up.
 
Perhaps we are agreeing that we see problems with this test and do not see it as a definitive step in stating that dogs cannot means-end solve a problem. Ask some with an escape artist Husky if their dog can problem solve and the answer will be a resounding yes. Different dogs may have different levels or variations of problem solving ability, I think. Shadow had a rope toy with a big knot in it. He untied the knot and then shredded it (no, he doesn't swallow the string.) It may have to do with motivation. Shadow is big enough and fast enough that he could get out of the yard if he wanted to, though I've seen video of a smaller dog scaling a 6' fence. Anyway, Shadow does not get out. He knows he has everything he needs here. Since being neutered, he won't have nearly the need to mate. So, he doesn't have to necessary motivation to get out of the yard in order to solve the problem of how to get out.
 
So, I also agree that dogs may have different motivations than humans and may be laughing at us with the string pull test.

ORIGINAL: JaneyB

All good points.  Besides the general question of whether dogs understand a means-ends relationship and can plan based on that understanding, I had some questions about the set-up of the experiment itself.  From what I understand, dogs don't see like humans do, and I believe they have limited color vision.  Could they have had problems actually seeing the strings (they apparently used red and blue strings, although I've heard that dogs see the color yellow much better than blue or red?).

I'm not sure, but it seemed like the experimental apparatus had the food kind of inside a wire cage. I'm not sure what effect that would have on the dog's ability to find the string visually, but I think the important thing was that it was a dark string on a white board. Dog's aren't so good at colours, but they are quite good at differences in contrast. It looks in the picture as though the wire is considerably thinner than the string, so should be easy to spot. On top of that, the wire would give the dogs the opportunity to locate the food by smell and maybe from there they could make the connection to the string.

But it does make me wonder if they could have made the string smell yummy during the training periods at least, so the dog makes a better connection between string and food. But that would mean that in the ones with two strings, baited and unbaited, both strings would have to smell yummy.

string smell yummy during the training periods at least, so the dog makes a better connection between string and food.

That's a good one, too. I do know that dogs can't focus well close-up but, as you said, this may not matter as long as they can contrast the box and the string and the floor. But dogs smell better than they see. They are born smelling, then later, hearing, then seeing. A better version of the test might have smell on the string. Work with their strongest, most used sense to more accurately judge problem solving. A suitable conversion would have been to have a human or other primate blindfolded and try to smell the correct string. With humans, our vision is the first sense to make any sense. Later, we understand the noises we hear and the textures we feel.
 

I agree to some extent that the dogs may not be motivated to "plan & problem solve" and choose the correct string the first time. If they get what they want eventually, what difference does it make? I would be really interested in seeing what the learning curve would be if the trial ended every time they pulled the wrong string.
 
Do canines in the wild have any reason to evolve means-end relationship understanding? I would say no - I can't think of why that would benefit them in the way they capture prey and survive. How about primates? I think so - in many cases their food is hidden inside something and they need to use a tool or otherwise make the connection in order to get the food.
 
In regards to human evolution, as a sidenote: I feel that we as a species have effectively negated the effects of natural selection... for example if a diabetic or a person with really poor vision was living "in the wild", they would likely not survive or reproduce. Those ailments would eventually fade out of the population. However we artificially bring everyone to a level playing field with medicine & other tools, and our species' genetic makeup does not change. The fact that we are a species capable of artificially aiding ourselves in this way is quite unique. Does it improve our species as a whole? No. But is it a good thing? I would say yes of course... social Darwinism might say no.

One possible reason why they didnt make the strings smell like food is because smell is a relatively difficult variable to control, and so any variability in smell might have been systematic/biasing, which would have weakened the experiment further...
 
One of my cognition lecturers described a simple means-end experiment once- put a dog on one side of a long (but not closed off) wire-mesh fence, with food/toy/something it wants on the other side. Does the dog try to go around the fence? or does it try to go through it?
 

ORIGINAL: Vinia

One possible reason why they didnt make the strings smell like food is because smell is a relatively difficult variable to control, and so any variability in smell might have been systematic/biasing, which would have weakened the experiment further...

One of my cognition lecturers described a simple means-end experiment once- put a dog on one side of a long (but not closed off) wire-mesh fence, with food/toy/something it wants on the other side. Does the dog try to go around the fence? or does it try to go through it?

I think that whether the dog tries to go around or through is somewhat dependent on breed and problem solving ability, not on species specific behavior.  As an example, if I toss something over the fence, my hound stares and whines at it.  My Aussie (see wheels turning here) figures out that she can't get to it directly, and runs around to the gate.
On the other hand, there are times when she loses sight of her frisbee and, if it is green, like the grass, and isn't moving, it takes her a bit to find it, presumably by scent.  But, here's the thing - she almost seems to be running search patterns to find it. []