Title: Learning and owner–stranger effects on interspecific communication in domestic dogs (Canis familiaris)
Authors: Angel M. Elgier, Adriana Jakovcevic, Alba E. Mustaca, Mariana Bentosela
Published: December 2008 in Behavioural Processes
These researchers were interested in how a dog’s relationship to its owner (which has been described in previous studies as a bond attachment that is similar to that of infant and parental figure. Specifically, the authors were interested in how the processes of extinction (the disappearance of a learned behavior) and reversal learning (learning to do the opposite of a human social cue) would be affected by an owner’s presence or absence. The 13 dogs that were enrolled in the study were randomly assigned an experiment (extinction or reversal) and whether the experiment would be conducted by the owner or a stranger (owner group or stranger group). Then, the authors trained 13 dogs in their lab to seek a treat that was hidden in one of two identical opaque containers that was pointed at by a human standing in the middle of the two containers. After this “acquisition phase”, the dogs were then given their predetermined experiment. In the extinction experiment, neither container had a treat and a “success” for an extinction experiment achieved when the dog did not seek either container within 15 seconds of the point cue. In the reversal experiment, the treat was in the container that was not being pointed at by the human experimenter and a “success” for this experiment was achieved when the dog immediately chose the container that was not being pointed.
The authors found that owner group dogs were much slower to “learn” in the extinction experiment than the stranger group dogs, and that owner group dogs learned the reversal behavior faster than dogs tested with the stranger group dogs. In explanation for the relative poor performance of the owner group dogs in the extinction experiment, the authors hypothesized that the owner group dogs might be more obedient to owners or likely have experienced intermittent reinforcement to their owner’s social cues (resulting in a much more reliable behavior that took longer to “unlearn”). In regards to the better performance of the owner group dogs in the reversal experiment, the researchers hypothesized that an owner’s presence may facilitate learning on the dog’s behalf due to reduced stress. Obviously, however, none of these hypotheses fully explains the results of this experiment.
Let’s start off with what I thought was well done in this paper. The data presentation was great: there was a table with demographic information about the dogs (although it did not feature spay/neuter status) as well as a table listing which dog was assigned to which treatment groups and the number of trials each dog took to achieve extinction/reversal learning. The authors also supplied a bar chart of results, which neatly described their findings.
One quibble I do have with this paper is that it is probably the most poorly written research article that I have ever read. All of the authors are from organizations in Argentina and I suspect that the paper was originally written in Spanish and translated very poorly. For example, there were several instances of “to verb” rather than just “verb“, as well as some vocabulary that doesn’t make sense to me (specifically, the use of the word “ontogeny”, which refers to the fetal development of an organism). But as I’m not in the field of applied animal behavior, perhaps I just don’t understand the use of the term in this context.
Unfortunately, I also have some major concerns with the actual experiment procedures. First of all, all of the dogs assigned to the extinction got 25-30 correct trials out of 30 possible trials in the acquisition phase, while every single dog in the reversal group got only 4 correct trials in this phase. This hardly seems to be possible by coincidence but the authors don’t offer an explanation of why the dogs were divided this way. (And I can’t think of one either!)
The researchers also stated that at the beginning of all training (both the acquisition phase and the experiment), the person performing the point cue attempted to gain eye contact with the dog and looked at the dog for the entirety of the training. I’m confused why this was done because staring at a dog can be threatening and stressful to the dog, and it’s not outside the realm of possibility that this could impact the stranger/owner effect since a staring stranger may be substantially more scary than a staring owner.
The other significant concern I had was the use of “correction” in the reversal learning experiment. If the dog chose the container towards which the human was pointing, the handler would jerk the dog’s leash and say “No,” and just verbally reprimand the dog after two leash jerk/verbal corrections. This type of correction is positive punishment and it’s a very ineffective and inefficient way to get a dog to learn. In fact, the average number of trials needed by dogs in this reversal group (both the owner group and the stranger group dogs) were higher than the average number of trials needed by dogs in the extinction group. Obviously, this could be a function of the task difficulty (perhaps extinction is easier to “learn” than reversal) but the higher number of trials could also have been the result of this confusing, inefficient training methodology.
Even though I have substantial qualms with the protocols of this research, because all dogs were treated similarly, I do think that this research provides evidence that the bond to an owner influences learning in some way. I thought the authors’ focus on an encompassing explanation of why owner group dogs learned extinction slower but reversal faster was perhaps misguided. It makes a lot of sense to me that dogs would have a very strong behavioral habit of going where their owner pointed due to the intermittent reinforcement previously theorized. I would hypothesize that owner group dogs learned reversal faster because it was a totally novel behavior and the owner/dog attachment facilitated that learning. I would be very interested to see this research question addressed by another group of researchers with more veterinary behaviorist oversight to see if these results can be replicated.