Can dogs feel jealousy?

This was the question posed by a study recently published in PLoS ONE, a peer reviewed, online, scientific journal.  With this focus, aggregator sites quickly summarized the report, and innumerable dog lovers then shared these retellings on line.  The study (Harris & Prouvost, 2014) raises some interesting, though hardly novel, issues for behavior analysts, although I confess it is only part of the reason for my attention to it here.  I have been one of those dog lovers since childhood and am even now waiting for my next Border Collie, still but a twinkle in the eyes of the planned sire and dam.  My affection for dogs is one of the reasons I spent more than ten years directing a major research program on canine olfaction and its use as a technology for detecting explosives and contraband.  More to the present interest, I’ve often used test questions such as this in my radical behaviorism course over the years, my favorite one asking if dogs can feel embarrassed.

Talking about dogs is a great way to confront many conceptual topics because we tend to anthropomorphize with abandon when we’re talking about our beloved pets.  This is enough of a problem by itself, but the human qualities we’re attributing to Fido are usually fictional in the first place.  It also doesn’t help that we talk about dogs in our vernacular dialect, which assures a wealth of still more problems.  Of course, dogs have no verbal repertoire of their own, a fact that is endlessly frustrating to most owners.  This apparent shortcoming at least saves us from falling into the trap of interpreting the meaning of what they might have to say.  All in all, it’s hardly surprising that our everyday convictions about canine behavior are not complicated by scientific findings.  And then when some experimental evidence about why dogs do what they do comes to our attention, it suffers the same interpretive challenges we’re all too familiar with in mainstream psychology.

Methods and conclusions

So, what was this study about?  The background and cited literature place it in the area of animal social cognition.  The authors are interested in whether a primordial form of jealousy is evident in dogs.  The study tested 36 pure bred and mixed breed dogs in their homes following a protocol similar to that used to assess jealousy in infants.

Each dog was tested for one minute under each of three conditions, with a brief break between trials.  In one condition, the owner interacted with a realistic stuffed dog that barked, whined, and wagged its tail while they ignored their dog.  In a second condition, the owners interacted with a jack-o-lantern just as they did with the stuffed dog, and in a third condition, the owners read aloud from a children’s book.  In all three conditions, the owner’s interactions with the objects lasted for about eight seconds.

Two blind observers rated one-minute videos from each condition and coded the dog’s behavior for aggression, attention seeking behavior or behavior that attempted to interrupt the owner’s interaction with the objects, and actions described as looking at or away from the owner or object.  Behavior was also coded for 30 seconds following the eight-second interaction periods.

The data are displayed in bar graphs showing the proportion of dogs exhibiting each behavior under each condition.  Suffice it to say, the authors describe the evidence as showing that the stuffed dog condition generated more behavior that could be interpreted as jealousy than the other conditions.  They conclude that their results support the notion that there is a primordial form of jealousy in dogs.  In general, the study was carefully done and will presumably been seen as a worthy contribution to this literature.

What would B. F. Skinner do?

How might behavior analysts evaluate this study and its conclusions?  Wearing our scientist hats, we should be open to the possibility that jealousy is more than just a verbal response – a tact of certain ways of behaving.  This hypothesis allows that jealousy could indeed be a distinctive physical phenomenon.  This notion is consistent with everyday dialect, which suggests that jealousy is an emotion or feeling.  Following Skinner’s argument that we not deny physiological correlates of behavior just because they often occur inside the skin, Chapter 3 discusses feelings as real physiological goings on that we often learn to tact.  It is more complicated than this, of course, but when we describe ourselves as feeling are angry, anxious, sad, or happy there is often more happening than simply labeling behavior.

Of course, we can’t consider the question of whether dogs – or even humans – experience physical sensations that we call jealousy without confronting the matter of what jealousy is, if it is anything at all.  I have suggested two possibilities that might be interesting to behavior analysts.  One is that there are coordinated physiological events that are part of what we learn to tact as jealousy.  On the other hand, there may be nothing more going on that our learning to tact certain ways of behaving under certain conditions in the absence of any particular physiological sensations.  Determining what is going on when we say that we feel jealous will be difficult unless research takes this distinction into account in its methods and conclusions.

The distinction is important because of where the two alternative possibilities lead.  If jealousy involves no more than tacting behavior as taught by our verbal community, our curiosities would focus on the influences on such tacts.  Why do we describe someone as engaging in any particular behavior?  What contingencies lead us to describe behavior as “drinking coffee” or “walking” versus “running?”  These are relatively straightforward questions about our verbal learning history, though they are not necessarily simply to answer.

If instead jealousy involves a particular combination of physiological events we learn to tact together with related behavioral and environmental elements, other questions arise.  There is still the matter of how we learn such tacts (somewhat complicated because of the private nature of some of the physiological events), but now we must determine the nature of the physiological changes associated with what we call an emotion or feeling.  Furthermore, we can ask whether these physiological correlates of environmental events and behavioral reactions are entirely learned or have a genetic component.  That is, if jealousy involves physiological features, is their role fully endowed by ontological contingencies, or is there a phylogenetic “push” that predisposes us to have this feeling – in the terms of the study, a primordial element?  If the latter were true, we could then wonder if this genetic endowment is unique to humans or can be found in other species – the focus of the canine study.

But let’s not get ahead of ourselves

These are fascinating behavioral questions, but they cannot be satisfactorily answered without navigating around some methodological and conceptual pitfalls.   The reported study embodied both kinds of problems.  For example, although its basic preparation seems to have been relatively sound, questions might be raised about the particular response classes targeted by observation (as well as behavior that was not measured) and how they were defined and scored.  The fact that each dog’s behavior was sampled only once under each condition should lead us to wonder whether the observed behaviors are typical of these conditions and would be reliably observed across additional trials.  Furthermore, reporting data in terms of the proportion of subjects showing target behaviors hides potentially useful information about the extent to which each dog engaged in each behavior.

The study’s conceptual challenges may be more serious because the author’s conventional frame of reference assured opportunities for mentalistic missteps.  They interpreted the observed canine behaviors as similar to those reported in studies of human jealousy and concluded that the dogs were therefore experiencing the same emotion.  The anthropomorphism is straightforward and was, after all, the reason for doing the study.  As is often the case, however, the human quality being attributed to the dogs has not been adequately described and explained, if it even has any physical dimensions that could be scientifically investigated.

It is important to recognize the embedded argument.  The authors assumed that the dog’s behavior was caused by an unobserved phenomenon for which there is insufficient scientifically documented, physical evidence.  The lack of such evidence about the existence of a biological phenomenon that might be called jealousy does not mean there is no such phenomenon, of course.  However, assuming its existence as a basis for explaining behavior is dangerously close to mentalism, although in this instance the hypothesized cause is supposedly physical rather than mental.  This kind of explanation is mischievous because it is so convenient.

An alternative explanation

Let’s say you are not satisfied with these niggling complaints, however.  After all, you’ve seen your dog act just like those in the study when you gave attention to your cat or your spouse.  What else could be going on but your dog feeling jealous, just like you felt in high school when your boyfriend danced with other girls at the prom?  What other explanations could there be for this sort of behavior?

Ever heard of the Columban Simulation Project (Epstein, 1981)?  Seen the movie?  (“Columban” is not a misspelled reference to a South American country, by the way.  We’re talking about pigeons here – Columba livia. )  This fascinating series of studies conducted by Robert Epstein and B. F. Skinner focused on how we learn complex behaviors, especially those that some might attribute to “higher mental processes.”  (Google this project and you can learn more.)  Their studies (also documented in a film) showed the kind of learning history required to do things that naïve observers might say demonstrate a sense of self, reasoning, and other cognitive capabilities.

For instance, one project showed the kind of history that can produce the behavior of responding to oneself in a mirror image.  Another analyzed the history that can generate behavior said to show “insight.”  Both involved separately shaping component behaviors required for a complex performance.  These different behaviors then came together in a coordinated way without additional training when the different skills were collectively required to earn reinforcement.

In one study, pigeons were taught to push a small box around the chamber and, separately, to peck a miniature banana.  When the banana was too high to peck, the pigeons pushed the box under the banana and hopped on it, thereby reaching the banana, pecking it, and earning food.  This combining of separate skills had not been trained.  (Their use of this preparation gently mocked Wolfgang Kohler’s famous studies with chimpanzees said to show insight.)  Anyone viewing the pigeon’s final performance, but lacking a behavior analytic background, would explain the bird’s behavior by referring to cognitive processes such as reasoning.  Viewing the learning history dispenses with the need for such mentalism, however.

Can this project help us understand references to dogs feeling jealous under certain conditions?  When we see our dog engage in what we would label as jealous behavior in humans, what learning history might explain its behavior?  Let’s say, for example, that your cat hops up on your desk and you begin petting and talking to it.  Your faithful dog lying at your feel then gets up and puts its head in your lap, perhaps nuzzling in a pushy way.   What reinforcement history could explain the different components of this performance?

There is the behavior of putting its head in your lap, certainly likely to have a rich reinforcement history.  Nuzzling “insistently” is also likely to be more effective than more passive responding.  But why might your dog do this just when you are petting the cat?  Could it be that the probability of reinforcement for this behavior increases when you interrupt your work to do other things, such as leaning back in your chair for a break, taking a sip of coffee, or petting the cat?  Might these actions on your part have become discriminative stimuli for the dog’s behavior?  How might you apply this kind of analysis if the situation were somewhat different, such as the cat hopping into your lap while you are on the sofa watching TV and your dog then standing up and growling at the cat or even pushing at the cat with its muzzle?  What is the likely reinforcement history for the dog’s behavior of growling at the cat or pushing it away?  When the dog growls at the cat or pushes it under both this and other circumstances, isn’t the cat likely to back off or leave the scene, thereby increasing the likelihood of your petting the dog instead?

This way of thinking about the origins of component behaviors of a complex performance can save lots of needless mentalism, not to mention misplaced anthropomorphism.  We may label such behavior in dogs in whatever way we wish.  It probably makes sense to call it jealous behavior in the vernacular dialect because of how we label apparently similar human behavior, but the usual risks of anthropomorphism apply.  Of course, an explanation of “jealous behavior” in terms of operant learning does not address the question of whether tacts of jealousy involve corollary physiological events.  And if there is a physiological component associated with jealous behavior, it is reasonable to ask if it is also present in dogs or other species.

Even if research shows that physiological events are operating when we say we feel jealous, an operant history is certainly a major aspect of the behavior we – and our dogs – emit.  Do dogs feel jealousy?  I don’t know what dogs feel, nor does any one else.  However, it is unwise to observe “jealous behavior” and assign causation to a hypothetical underlying biological process without considering the more parsimonious explanation involving operant reinforcement history.

Reference

Epstein, R. (1981).  On pigeons and people: A preliminary look at the Columban

simultation project. The behavior analyst, 4, 43-55.

Harris, C.R. & Prouvost, C. (2014). Jealousy in Dogs. PLoS ONE 9(7): e94597.

doi:10.1371/journal.pone.0094597