Banner
    What Your Dog Is-And Isn't-Trying To Tell You
    By Caitlin Kight | October 2nd 2011 12:32 PM | 20 comments | Print | E-mail | Track Comments
    About Caitlin

    I am a research scientist who dabbles in freelance writing and editing, birding, cooking, indoor gardening, needleworking, various athletics, music...

    View Caitlin's Profile

    Fossil evidence indicates that approximately 30,000 years ago, humans captured, tamed, and bred gray wolves (Canis lupus), ultimately producing Canis familiarus: the domestic dog. Centuries of breeding to reduce aggression and fear have yielded animals that are not only comfortable in the presence of humans, but may also be overtly friendly, eager to please, and even helpful.

    As anyone who has seen a working dog in action already knows, dogs are quite adept at understanding human communication; in fact, they are even better than apes at using human gestures in a cooperative communication context. Likewise, humans can interpret signals from their canine companions, including jumping, running back and forth, gazing, and bark patterns--which may, for instance, be used to indicate the locations of desired objects. One long standing question is whether dogs use these communicative behaviors because they recognize human ignorance about a particular issue--in the above case, where an object is hidden--and wish to provide information about it, or whether they are simply performing begging behaviors in order to achieve an outcome they desire for themselves.

    Although this may seem like a subtle difference to most of us, the ability to recognize others' ignorance has only been demonstrated in primates; further, only humans have been found to voluntarily provide information to others in order to reduce their ignorance. Is it possible that dogs, having been bred so long to possess traits that are valuable to their human owners, might have evolved this mental capacity?

    This question was recently investigated by researchers from the Max-Planck Institute for Evolutionary Anthropology, who performed two experiments in order to evaluate dog-human communication. The first was designed to assess whether dogs indicate the location of a hidden object not just to request it, but to inform others about it. Dogs were tested either with their owner or with a stranger, and went through multiple rounds of testing. In each round, the dogs were exposed one of the following: their favorite toy, which they played with alone; a non-favorite toy, which was handled by both the human and the dog; a hole puncher, which was not handled by the dog but was used by the human companion; and a vase, which was handled by neither. After 60 s of exposure to the item, the companion left the room and a second person came in and hid the object while the dog looked on. The human companion then returned and attempted to use the dog's behavior to locate the item. If the dogs were generally interested in informing their human companions, they should have communicated about even the items in which they were uninterested (the hole puncher and vase); if they were just begging for the object, they should have communicated exclusively about the items that were most relevant to them (the two toys).

    Across all trials, people were more likely to find hidden objects than would have been expected by chance, indicating that canine communication was both produced and received. However, dogs' behaviors were most helpful to their human companions when the hidden object was something the dog had a direct interest in--in other words, one of the toys. They did not communicate as much when the object in question was of interest solely to the human. Overall, this indicates that dogs are begging/requesting, rather than providing information to rectify ignorance. Interestingly, though, dogs were generally more communicative with their owners than with strangers; this led owners to have higher overall success rates at finding objects, regardless of type.

    Even though dogs helped their owners find the "human-interest" objects at a fairly high rate, they did not distinguish between the object that their owners had been interested in (hole punch) and the object that had simply been in the room (vase). The researchers were curious about whether this represented the dogs' inability to understand the difference between objects that were desirable and those that were not, or simply a general interest in helping their owners locate any hidden object regardless of how much the human had previously interacted with it. Thus, the second round of testing, which focused only on dog-owner pairs, involved the presentation of pairs of objects. One object was relevant to the owner--and was actively used in a task--and the other object was not. Each dog was tested with six different pairs of objects, plus a control treatment in which the two objects were both toys--one of which was the dog's favorite toy. In all scenarios, the human companion left the room, the object was hidden by a second experimenter, and then the companion returned to search for the hidden item.

    As before, humans were more likely to find hidden objects than would have been expected by chance, indicating that the dogs were actively communicating. This was true regardless of item type, but the relationship was particularly strong for the control trial; dogs provided significantly more information about an object of interest to them than an object of interest to their human companion. Furthermore, even though they clearly showed more interested in their favorite toy during the control experiment--indicating an ability to assign preference to one object over another--they did not prioritize their owner's item of interest, but instead pointed to both types of hidden object equally.

    Overall, this suggests that dogs, like most other animals, do not voluntarily provide information in order to reduce a human companion's ignorance. Instead, they request items for themselves or engage in general "fetching" behavior. This may be a result of dogs' inability to interpret humans' lack of knowledge, difficulty with connecting past and current events in order to understand when information transfer is necessary/useful, and/or trouble interpreting human behaviors as goal-directed. However, throughout the course of the experimental trials, owners' success rates at finding hidden objects did not decrease over time. This indicates that the dogs remained motivated to offer assistance, even in the case of items in which they were not interested. These results are in contrast to those observed in chimps, which become increasingly less helpful in similar situations; rather, the behaviors are more like those of human children.

    In short, the results of these two experiments indicate that, although we may have bred many human-compatible traits into our canine best friends, we have not made them into mini-people--yet. However, the tests highlight the perpetual attentiveness with which our furry friends are endowed and also indicate the impressive efficacy of cross-species communication.  No wonder dogs can be such excellent sidekicks to people in a variety of professions.

    ---

    Kaminski, J., Neumann, M., Bräuer, J., Call, J., Tomasello, M. 2011. Dogs, Canis familiaris, communicate with humans to request but not to inform. Animal Behaviour 82:651-658.


    Thanks to the following websites for providing the images used in this post:

    http://alovefordogs.com/40/how-to-teach-your-dog-tricks/

    http://www.doginstructor.com/basic-dog-tricks/

    http://www.pawprintsthemagazine.com/?p=6417

    http://www.germanshepherdrescue.co.uk/police-dog-jake.html

    Comments

    rholley
    I’ve just been watching Countryfile (BBC).  In one item, they were testing a small robot plane (drone?) to see if it can herd sheep.

    Result – Rex, you don’t have to worry about early retirement, at least not yet.
    Robert H. Olley / Quondam Physics Department / University of Reading / England
    specialagentCK
    Last week I saw some working dogs in action in the Yorkshire Dales. It was impressive!
    NSF Postdoctoral Researcher, University of Exeter--Tremough Campus, UK. Personal website: http://www.caitlinkight.com
    vongehr
    "dogs provided significantly more information about an object of interest to them than an object of interest to ... we have not made them into mini-people"
    If that is the main "control experiment", it seems a little flawed, no? Why expect dogs to be nicer people than people? Just saying.
    specialagentCK
    I wondered the same thing when I first read the study. I think the main point of this particular control was to make sure that dogs could see two objects and prioritize one over the other. So the controls indicates that they can remember that there were two objects, that they can develop an interest in one of them more so than the other, and they can track the location of that one particularly desired object. This indicates that, theoretically, they do have the mental capacity to volunteer information about the object of particular human interest. However, they don't, indicating that--unlike humans--they don't make the connection about need and ignorance.

    But I agree, no matter how much genetic engineering we've done over the years to make dogs helpful companions, I think it would be pretty surprising for them to have developed this sort of ability. Then again, I haven't read all the canine literature, so what do I know?
    NSF Postdoctoral Researcher, University of Exeter--Tremough Campus, UK. Personal website: http://www.caitlinkight.com
    Bonny Bonobo alias Brat
    I have three dogs who I spend a lot of time with each day often playing with them and their toys and I can't help feeling that the choice of objects in this experiment was probably flawed. Dogs are very aware of smells and their toys are usually chewable and/or transportable and/or smelly for obvious reasons. The first being that the dog toy manufacturers make them smelly, the second being the dogs and myself fling them around and chew them (the dogs not me) and all of these factors also usually make them acquire odours and become smelly.
     
    At any time I can walk into the garden and say 'Get a toy' to any of my dogs and they will go off and find one for me, usually locating it by smell if it is hidden from sight. Its difficult to see how a vase or hole-puncher could ever fit into an object category that would ever interest my dogs, first of all they are not smelly, secondly they are not really throwable and thirdly they are not easily carried by a dog or even chewable. To do this experiment properly the objects maybe should have been made equally smelly, transportable and chewable but one should have been a favourite toy, another should have been just a toy, the others of human interest but not of dog interest could have been something like a lemon or orange, a bar of soap, a rubber hairbrush or even a human toothbrush maybe but not a vase or hole punch, at least in my opinion. 

    My article about researchers identifying a potential blue green algae cause & L-Serine treatment for Lou Gehrig's ALS, MND, Parkinsons & Alzheimers is at http://www.science20.com/forums/medicine
    specialagentCK
    One of the things that perplexed me about the study was why they were changing both type of object and type of handling routine simultaneously. I understood their logic, but I didn't necessarily think it was the soundest experimental technique, since it introduces two variables rather than one.

    I also wondered how they were sure that the dog was really watching the second experimenter hide the object. In the case of the favorite toy, the dogs were probably much more motivated to pay attention to what the person was doing; with the vase, maybe the dogs looked away because the object was boring and they didn't care? The paper didn't say anything about verifying the dog's attention.

    Certainly, the smell and taste and feel of an object are all important, even though the dogs weren't actively fetching the toys themselves (but, rather, providing body language cues for the searching person). If the dogs engaged with the objects with more than one sense, they might be more interested in it, more invested in it, and better able to remember what happened with it.

    I suppose this is a difficult issue to investigate, because you want a scenario in which the dog provides information to the human even when the human is the only one who benefits from the communication. In other words, you have to use an object that the dog doesn't care about, but then you also want to be able to compare this to baseline and maximum levels of helpfulness, for which purposes you want something neutral and something the dog is interested in. Maybe the hidden-object setup is not the optimum circumstance in which to test for communication; is this something canines do much in the wild, anyway? I wonder if there is some sort of teamwork circumstance (e.g., a hunting simulation or some other situation that canines are evolutionarily geared for) that might be more appropriate.
    NSF Postdoctoral Researcher, University of Exeter--Tremough Campus, UK. Personal website: http://www.caitlinkight.com
    Gerhard Adam
    I'm not sure that this was a good test, because dogs are also quite sensitive regarding issues of "ownership" and may behave in a way to specifically not interfere with an object they perceive as belonging to someone else (or the "leader").  We do see dogs, such as retrievers, that are quite capable of locating objects that don't belong to them on behalf of the owner, so this entire test seems a bit too abstract to offer much conclusive information. 
    Mundus vult decipi
    specialagentCK
    On a related note, I was curious about the authors' decision to group together dogs of a variety of different species. Some breeds have deliberately been created as helpers, and so might be expected to pay more attention to human activities and human desires--and, as you pointed out, to go do tasks for us. Other dogs have absolutely no evolutionary experience with this sort of thought process or behavior. It's also true that some breeds are just generally more social and/or more intelligent than others. All of that is a result of weeding out particular genes and emphasizing others, which is something the authors were focused on anyway. So why not do the study on a single breed, or on two contrasting breeds?
    NSF Postdoctoral Researcher, University of Exeter--Tremough Campus, UK. Personal website: http://www.caitlinkight.com
    Gerhard Adam
    In thinking about this some more, I'm not convinced that these kinds of tests are valid in almost any context.  What I find bothersome, is that most of these tests seem to be focused on trying to establish whether animals are interested in us, as humans.  It doesn't actually measure anything to do with intelligence, but rather how well an animal adheres to our social and cultural standards.  It should be obvious that our social interactions are governed by thousands of years of evolving to operate in uniquely human groups, while other animals have their own specific social settings and "rules of behavior". 

    As a result, it seems presumptuous to assume that any animal can behave according to human norms as an indicator of "intelligence", any more than a human can behave according to a particular animal society's social norms. 

    I would expect that we would see significant variability in behavior based on a common "value" system.  In a way, it would be like to trying to measure "ambition" in an individual ant, or the degree of altruism in an adult male grizzly bear.  These are traits that would be out of character for the particular "society" of that animal's existence, and consequently be uninformative regarding any inherent "intelligence" as measured by human behavioral standards.

    Since their assumption was to measure the degree to which a dog would recognize a human's ignorance and provide the missing information, it would seem that the first step in such a process would be to assess why a dog should care about it.
    Mundus vult decipi
    Bonny Bonobo alias Brat
    Since their assumption was to measure the degree to which a dog would recognize a human's ignorance and provide the missing information, it would seem that the first step in such a process would be to assess why a dog should care about it.
    Well maybe because we feed them, after all that's a pretty good reason that Pavlov milked to the nth degree. My dogs are very good at letting me know if I haven't fed them and it is their dinner time or worse it is past their dinner time. In fact they are pretty hilarious and ingenious at thinking up ways to bring this to my attention, like staring at their empty food bowls and then looking intently at me or getting very excited about any movement I make in the general direction towards their food bowls or refusing to fetch a ball and then pointing at their mouths with their paws (only joking) but nothing as ingenious as my cat who used to deliberately trip me up preferably to fall directly onto his food bowl and occasionally brought me a bird in his mouth before then letting it go in front of me!

    My article about researchers identifying a potential blue green algae cause & L-Serine treatment for Lou Gehrig's ALS, MND, Parkinsons & Alzheimers is at http://www.science20.com/forums/medicine
    Gerhard Adam
    That's true, but the point isn't about what the dog is interested in, but rather whether the dog would provide information about something the human was interested in.

    Mundus vult decipi
    Bonny Bonobo alias Brat
    Well isn't that what they are doing when they bark whenever someone arrives on our properties? They are hardly barking hello are they?
    My article about researchers identifying a potential blue green algae cause & L-Serine treatment for Lou Gehrig's ALS, MND, Parkinsons & Alzheimers is at http://www.science20.com/forums/medicine
    Gerhard Adam
    Not really, since most animals have a sense of territory are usually quite vocal about announcing it and issuing warnings.  In addition, this is precisely the same behavior you would expect in a pack setting, so it is a normal trait for this social animal, rather than one that is specific to humans.
    Mundus vult decipi
    Bonny Bonobo alias Brat
    I once read somewhere that dogs and humans have lived together symbiotically for so long that we have actually altered the course of each other's evolution. The article said that if we didn't have dogs with their sensitive hearing warning us about intruders we would have evolved better hearing and also that by utilising our dog's keen sense of smell to sniff out prey and track things we saved ourselves from developing highly sensitive dog like noses for smelling, they also claimed that this allowed us to invest in and evolve other useful attributes like higher intelligence. This psychology today article says that :-
    Relying on dogs to hear the approach of danger and to sniff out the scentof prey animals, our ancestors experienced a decline in these sensory abilities compared to other primates. This conclusion is confirmed by shrinkage of brain regions devoted to these senses (the olfactory bulb and lateral geniculate body).
    During the long period of our association, dogs brains have shrunk by about 20 percent, typical for animals such as sheep and pigs who enjoy our protection. Domesticated animals undergo tissue loss in the cerebral hemispheres critical for learning and cognition. If we relied on dogs to do the hearing and smelling, they evidently relied on us to do some of their thinking.
    My article about researchers identifying a potential blue green algae cause & L-Serine treatment for Lou Gehrig's ALS, MND, Parkinsons & Alzheimers is at http://www.science20.com/forums/medicine
    Gerhard Adam
    Well, we certainly altered the dog's evolution, since there wouldn't be any without domestication.  However, most of the other stuff just seems fanciful conjecture.  In the absence of defects, humans have excellent hearing and it is quite unlikely that any amount of evolution could correct the deficiencies in our sense of smell.  After all, even dogs aren't all equal regarding that sense.  

    The problem with stories like Psychology Today are that they are assuming that a significant change has occurred without quantifying if it actually did.  While many of those regions of the brain may have shrunk, that is certainly no suggestion that the domestication of dogs had anything to do with it.  In fact, I'm sure a much stronger argument could be made for human social groups, rather than dogs.

    It's also unreasonable to make such an assumption without actually testing the ability we do possess.  The reason why I say this, is that we presume that because some of these senses don't result in conscious recognition, that they aren't being processed by the brain.  This simply isn't true.  We can ignore noises, unless we are in a stressed situation where they suddenly become important.  That doesn't mean our hearing improved.  It simply means that we are now more conscious of that sensory input.

    The only question to consider is whether we differ in any significant way from other primates.  If not, then the dogs are irrelevant.
    Mundus vult decipi
    Bonny Bonobo alias Brat
    I've just thought of another example of dogs behaving altruistically, it is a common sight on Australian beaches to see dogs unhappily swimming out through the surf in a desperate attempt to tell their 'owners' that they are worried about their safety and even trying to rescue them from their incomprehensible (to a dog) recreational swim in the surf.
    My article about researchers identifying a potential blue green algae cause & L-Serine treatment for Lou Gehrig's ALS, MND, Parkinsons & Alzheimers is at http://www.science20.com/forums/medicine
    Gerhard Adam
    Let's be careful with terms like "altruistic".  Dogs are social animals and there's no question that a human individual (or family) may act as a surrogate "pack".  So, the question would be whether this is behavior that would occur for a dog, regardless of pack-mates, or whether it is specific to humans.

    Remember, we're not discussing whether dogs are social, or whether they participate in social behaviors.  The point is simply whether a dog is capable of recognizing "ignorance" in another individual (because the dog possesses information that the other doesn't), and whether the dog is capable of recognizing that condition and providing the missing information.

    In other words, if you walked into a room and began looking around, I would assume that you were looking for something and I could ask what it was.  If I knew where it was, I could then tell you.  In that case, I recognized that you were "ignorant" about some object and I might possess the knowledge of where it was.

    Obviously dogs can't converse, and obviously dogs don't have the same sense of ownership, etc. that humans have.  However, the question was whether a dog would be capable of recognizing that a human was looking for an object that the dog saw and to convey that information to the person looking.
    Mundus vult decipi
    specialagentCK
    I got the sense that the authors were not trying to anthropomorphize dogs, but simply to try to assess what they were mentally capable of. They emphasized that dogs' ancestors and closest relatives, wolves, are not capable of the kind of communication that has been observed in dogs. So, clearly, the selective breeding associated with the domestication process has selected for certain cognitive traits. Without saying so specifically, they seemed to be interested in investigating whether, in fact, we have selected to make dogs think, and therefore act, in many ways like humans. This series was a way of testing for that. It isn't so much an issue of WHY they should care about us, as whether we have selected for dogs that DO care about us, and/or are capable of communicating that in some way. Because the authors frequently mentioned cognitive abilities in apes (especially chimps) and humans (especially developing humans), I assumed that their broader purpose was to try to help map the evolutionary trajectories associated with the sort of thinking that we humans are capable of.

    Now, whether or not those are valid and interesting questions, and whether or not their studies accurately assessed those questions, is a whole other issue!
    NSF Postdoctoral Researcher, University of Exeter--Tremough Campus, UK. Personal website: http://www.caitlinkight.com
    Gerhard Adam
    It isn't so much an issue of WHY they should care about us, as whether we have selected for dogs that DO care about us, and/or are capable of communicating that in some way.
    This is a good point and I wanted to clarify my reasoning for the question in the first place.  We already know that any social animal has various "rules" that govern its behavior within a group, so when we domesticate such animals we are exploiting certain intrinsic behavioral traits they already possess.  However, this also presumes that we understand the original dynamic of that social behavior so that we can determine whether the behavior has actually changed, or simply been redirected.  So the question becomes whether a dog sees me as a human or as a "pack leader", or something in-between. 

    Consider a similar (but contrived) situation among humans.  If someone walks into a room appearing to look for something we may well ask what they are looking for and respond if we have any knowledge about.  However, imagine if suddenly the President of the U.S. walked in and behaved in a similar fashion.  You may suddenly keep quiet, simply because you don't know what's going on (and his position may seem intimidating).   My point is not to argue that dogs perceive us as gods, but rather we need to consider how dogs view us, to determine the degree of involvement they consider permissible.  In most cases, they obviously recognize us as "rule-makers", so it is entirely plausible that we aren't getting a completely free response.  That's why I asked the question of establishing WHY they should care.

    I've seen similar contrary reactions in horses, between domesticated and wild.  A domesticated horse can be made to run in a round-pen as a means of achieving control and discipline.  However a wild horse made to run in a round-pen will do that all day long, because it's a natural flight response and will convey nothing about "control and discipline".  We know that there is no material difference in the horse's mental abilities, therefore the difference in reaction is simply due to how we (humans) are viewed and what is considered to be the "natural" or normal response to the horse, itself.
    Mundus vult decipi
    My last canine pal was 3/8 wolf, short stocky and a beautiful dog. She was not a very highly trained dog in any of the usual doggy games or disciplines, but I couldn't help the feeling that she always new what I was talking about and would banter playfully with me using a kind of yelp. I always told my friends that she talks, she just couldn't make those important consonant sounds.