Conflicting Conclusions

The most exciting phrase to hear in science, the one that heralds new discoveries, is not Eureka! (I found it!) but rather, “hmm…. that’s funny….”

Isaac Asimov

Quiz for the day: To prevent or avoid canine aggression problems, should you:

1. Live in the U.S.; spay or neuter your dog and train him without any aversives under the supervision of a veterinary behaviorist;
   or…
2. Live in Europe; plan ahead to acquire a very large, sexually intact male or spayed female dog; refrain from spoiling your dog; exercise it regularly and train it yourself using effective corrections?

According to a pair of studies published in 2009 — both answers are right.

In the previous post I wrote about Meghan Herron, Frances Shofer, and Ilana Reisner’s (Herron et al.) “Survey of the use and outcome of confrontational and non-confrontational training methods in client-owned dogs showing undesired behaviors“. This study has been quoted extensively as proving that “confrontational training methods” provoke aggressive responses in dogs.

During the same time period that Herron et al’s study was published, Joaquin Perez-Guisado and Andres Munoz-Serrano (PG-MS) published an article in The Journal of Animal and Veterinary Advances stating that owner behavior is the primary factor influencing dominance aggression in dogs.  From the abstract:

Modifiable and non-modifiable factors that are associated with higher levels of dominance aggression and depend on the owner include first time ownership, a lack of obedience training, the owner not being the main obedience trainer, spoiling the dog, not using physical punishment, acquisition as a present, as a pet, impulsively, or to guard and spaying female dogs. Modifiable factors have the greatest influence on dominance aggression in dogs. Dog-dependent factors (gender, breed, age, size and coat color) are fewer than owner-dependent factors.  There was an association between certain dog behavior patterns and higher level of dominance aggression.

The sample for this study included a very number of dogs trained using “confrontational methods”. Dogs that should have reacted aggressively to their owners based on the results reported by Herron et al. Contrary to Herron et al’s finding, PG-MS’s data indicate that the use of physical corrections was inversely correlated to dominance aggression. In other words, their data suggest that the use of aversive training methods prevents aggressive behavior.

How can two groups of scientists study aggression in dogs during the same year and come up with conflicting findings? Tabulated data weren’t published in the PG-MS study, so this review will be more limited than I’d like. Still, a closer look at these apparently contradictory studies may provide some valuable insights.

First there were significant differences in the populations sampled. Herron et al.’s respondents were Americans getting advice from a veterinary behavior clinic on existing behavior problems. PG-MS’s were Spaniards taking their dogs for a walk. Herron et al. surveyed 140 dog owners; PG-MS interviewed 711.

In the previous post I said that I believed that there was a significant bias problem in Herron et al’s sample because it only included owners who were actively seeking help from a veterinary behavior clinic. Given this limitation, all dog owners who found that aversive methods worked effectively for them were removed from Herron et al’s sample. I stated that I thought this was the most significant problem with the study – and PG-MS’s data appear to corroborate my suspicion. In a much larger population without that bias, the correlation between aggressive behavior and aversive training methods not only disappeared – it appears to have been reversed.

How could this happen?

First of all we’re comparing two different varieties of apples here. Herron et al stated that their goal was to assess whether there was a connection between the use of aversive techniques and aggressive responses. PG-MS focused on the importance of nature versus nurture in the development of dominance aggression.

Next, I believe that there were significant differences in how the two groups defined aversiveness.  Herron et al. did it subjectively and included what I thought were a rather odd range of “interventions” in both the aversive and non-aversive groups. They evaluated “hitting and kicking” as a training method. Three different types of verbal corrections were considered separately – but the wide array of techniques used in clicker training were lumped together under a single heading. Two different types of leash corrections are included in the aversive group and avoiding any exposure to triggers is considered training in the “neutral” category. This bizarre mish-mash of categories appears to have been put together by someone with little hands-on experience with dogs.

Instead of using objective criteria (like a measure of the dogs’ responses) to evaluate aversiveness, Herron et al. used subjective value judgments to determine which methods were aversive, neutral and non-aversive. On the other hand, PG-MS don’t provide any information on how they assessed the relative aversiveness of training methods, so there may be problems with their assessment too.

One of the most interesting things I discovered in comparing these studies was that it appears that we’re all so sure we know what defines a method as positive or aversive, that scientists don’t even feel the need to clarify their definitions. And that doesn’t make any sense. Rewards and punishments are, by definition, subjective and personal measures of discomfort and pleasure. If trainers and scientists can’t find something other than subjective measures of “niceness” or “not-niceness” to define training methods – we’ll never agree on how to use them.

Moving beyond the problem of definitions, I think it’s important to ask whether the relative aversiveness or positiveness of a method is the most important factor in dog training.

In an effort to figure out how the dogs involved felt, I took data from tables published in Herron et al. and, ignoring  subjective measures of positive or negativeness, ranked them by percent of positive responses as evaluated by the dogs’ owners. Then I plotted that data against the percent of aggressive responses observed. Here’s the chart (click for big).

There’s an inverse relationship between positive effect and aggressive response. In other words, methods that worked well rarely provoked aggressive responses. You can also see that as positive response decreased, the degree of correlation became much more erratic. How can we explain this?

If you look at the interventions that elicited the highest percentage of aggressive responses (hit or kick, growl at, pulling an item out of a dog’s mouth, alpha-rolling, holding a dog in a submissive position on the ground and scruffing it) – you’ll see that these are the kinds of things people are most likely to do when they’re frustrated with their dogs. As I noted yesterday, I believe it is likely that Herron et al’s sample included an anomalously high number of frustrated dog owners.

PG-MS’s stated that being a first time pet owner, acquiring a pet on impulse, spending less time with your dog, a lack of exercise or obedience training, having someone else train your dog for you and spoiling it were positively correlated with aggressive behavior. In my experience as a dog trainer, these are all good indicators of a potentially frustrated dog owner.

In Herron et al’s study, the interventions that elicited the highest percentage of effective responses (using food to trade for an item, making a dog sit for everything, using food rewards, increasing exercise, using an attention command, clicker training, leash corrections and use of food-stuffed toys) with few exceptions* were methods that require a reasonably calm mind and some forethought. PG-MS correlated aggressive behavior to a lack of training, lack of exercise and excess of pampering. These factors don’t generally indicate mindful behavior.

Going back to the quiz we started with; I don’t believe that the answer to preventing and curing dog aggression will be found in subjective, abstract concepts tossed around by people who are more interested in parsing out behavior than handling and interacting with dogs. I’m convinced that the key to preventing and curing canine aggression is proactive, mindful behavior on the part of dog owners.

In science it often happens that scientists say, “You know that’s a really good argument; my position is mistaken,” and then they actually change their minds and you never hear that old view from them again.  They really do it.  It doesn’t happen as often as it should, because scientists are human and change is sometimes painful.  But it happens every day.  I cannot recall the last time something like that happened in politics or religion.

Carl Sagan

---

* Avoiding the problem – which I refuse to accept as training; and the “force down” – which elicited a noticeably higher percentage of aggressive responses than other similarly effective methods.

Note: Based on the recommendations of both studies – Audie should morph into a vicious killing machine any day now. I used aversives to train him and he meets all the criteria for “pampered” according to PG-MS. Stay tuned for further developments…