Wednesday, June 9, 2010

Thinking About Pit Bulls

Last fall, on a clicker-training forum, we got into a lively, at times impassioned, discussion about pit bulls. Some forum members characterized them as "weapon dogs" whose vicious aggression is essentially determined by their genes--dogs trapped in their genetic histories. Some forum members seemed to believe that the dog aggression that was part of their history would inevitably bleed over into aggression against humans. Many forum members, however,  defended pit bulls, with some advocates arguing that their dogs are no different than other dogs. Other pit bull lovers argued that their dogs are indeed fundamentally different than other dogs, and their dog aggression could never be bred out of them without destroying what was essential to their other good qualities: intensity, determination, human- friendliness, toughness, comic good humor. Finally, a group of pit bull lovers argued (or at least we hoped) that it might be possible, with selective breeding,  to maintain what's wonderful about pits while also decreasing their dog aggression. Dogs don't stay fixed throughout their evolutionary history. But selective breeding has a troubled history--often, in getting what you wished for, you also get surprising and undesirable traits as well.
Aggression is extraordinarily complex in evolutionary terms, and it doesn't make much sense to speak of it as one trait. Conspecific aggression (fighting) is different than predation, and in turn both are different than aggression against humans, which in turn is different than prey-aggression (which really shouldn't even be called aggression); they don't appear to be part of one continuum.
Aggressive traits tend to be polygenic (controlled by more than one gene, just like hip dysplasia), and many the genes that affect various kinds of aggression are also likely to affect many other characteristics (when one gene controls several traits, it is called pleiotropy). Epigenetic influences in turn affect gene expression, and hormones affect it all. What this ends up meaning is that it wouldn't be difficult to breed for very low human aggression and high fight aggression in the same individual, and this historically is supposedly what happened with game-bred pits in the late 19th century. It probably wouldn't be impossible to breed for low human aggression, high gameness, and low fight aggression as well. But because these traits are polygenic and each of those genes are pleiotropic, it's also quite easy to start messing with dogs in deleterious ways (see-- golden retrievers: hip dysplasia; pointers: nervousness; etc)
Play behavior in many young predator species is an interesting combination of prey behavior ( practice for hunting) and ritualized fighting (conspecific aggression). Because of their breeding, many pitties seem to slip more quickly from play into aggressive or prey behavior than some other dog breeds, but for all dogs, play behavior has close links to both hunting and conspecific aggression. All dogs have the potential to get overaroused when playing and slide into fight behavior or prey behavior. Certain breeds, of course, are more likely to do this.
Many pit bull breeders argue that "gameness" in pits is not just about fighting--it's terrier tenacity, and it can be turned to fighting conspecifics, or it can be turned to farm work, or it can be turned to agility purposes. Today, on the fighting websites, I think it is indeed used as a euphemism for "fighting". But gameness is a real quality, and my impression is that it's closely linked in pre-1950s pits with the legendary pit gentleness toward children and strangers.This is just an impression (although many pit breeders argue that it's true, I don't know of any data that have been gathered to support the hypothesis. It could be tested, however)

AGGRESSION: what is 'normal'??
Group-living species such as canids evolve ritualistic displays of aggression that substitute for actual fights that might result in fatal injuries--and in most non-Molosser breeds, most of the time, conspecific aggression is a noisy threat display that rarely leads to severe wounds. Yet pits can be the exception. Their conspecific aggression is not all about ritualized displays; rips and tears and blood are ofteninvolved.
Alexandra Semyonova (2002) has argued that 'normal' dogs don't exhibit wounding aggression toward conspecifics, so if pits do this, by definition they aren't normal. There are several problems with this logic. First, it's circular. Semyonova start by arguing that only dogs who don't fight to kill are normal, and therefore normal dogs don't fight to kill. Bad logic, bad science. You can't exclude a breed that now makes up almost a quarter of domestic dogs in many areas in order to define what "normal" dogs do. This doesn't make any more sense than arguing that normal dogs don't have hip dysplasia, so if labs and goldens do, they aren't normal dogs.
Aside from the circular reasoning, the deeper evolutionary point is this: conspecific aggression that leads to fatal wounds is part of the evolutionary heritage of all dogs, just as prey aggression is. Genes, however, are very labile things, and it's possible to select against certain traits and bring about changes in just a few generations, even in wild species (see the silver fox studies). But genes are also very complicated things, and selecting against certain traits we like or don't like can often lead to sobering and unintended consequences (see English springer show gaits and aggression).
Semyonova (2002) argues that 'normal' dogs have evolved to display only ritualized threats, and that pits are 'abnormal' dogs because their threats sometimes turn into actual wounding fights (and therefore, she continues, all pitbulls and related breeds, including boxers, amstaffs, etc, must be eliminated, with all individuals containing any pit bull blood immediately euthanized). But ritualized threat displays can only be an effective strategy, in evolutionary terms, if a certain percentage of threat displays are backed up by actual, costly, violence. In the species that have been studied, such ritualistic displays continue to persist in a population only when they are "backed up" by a certain frequency of actual, violent, damaging aggression. To understand this, think about painted cattle-guards versus actual metal cattle-guards as a useful metaphor. In ranch country, ranchers know they can have a certain percentage of "fake" cattle guards that are just painted onto the road--ritualistic displays of cattle-guardedness. But a certain percentage of cattle guards need to be actual grates that hurt cows' feet when they test them, or else the painted cattle guards lose their effectiveness. Ritualistic displays--roaring, snarling, mock-fights--similarly lose their communicative power (over generations) when they never result in actual bloodshed.
This doesn't mean that all dogs will fight to kill; what it does mean is that violent aggression is not abnormal in dogs. It may be rare (thankfully), but it's normal. We can't single out a large group of dogs that do it and say they're abnormal for doing it, so that we can persist in our fond hope that dog's aren't really canids with big teeth and the potential for violence. We may not like that violent behavior in pits, but the only reason it could be selected for by breeders is because the genetic potential exists for it in the larger dog gene-pool. You can't breed a dog who can fly, but you can breed a dog who can fight, because fighting played an important role in their evolutionary histories.
Conspecific aggression that leads to fatalities is actually surprisingly common in many species, including wild canids (and it's also observed in feral dogs and village dogs.) In fact, it's quite common in the evolutionary history of many group-living organisms. We would like to hope that such aggression is rarely fatal, but it's actually quite often fatal.
Wild wolves do sometimes kill other wolves. Recent and very interesting research supports the (depressing) hypothesis that the altruistic behavior in social mammals seems able to evolve only when there's a high level of fatal conspecific aggression. This may seem contradictory, but when you think about it, it makes (depressing) sense. In small groups, wild canids and people will act for the benefit of the others in the group even when they don't share genetic material. This was long a theoretical puzzle for evolutionary biologists, but evidence (both theoretical modelling studies and empirical evidence) supports the argument that altruism evolves when high levels of fatal conspecific aggression exists between groups. Kind behavior, in other words, toward the same species seems to evolve because of (not in spite of) nasty violence toward the same species.

Pit bull histories
Dr Edmund Russell at the University of Virginia is working on a fascinating book that, in part, explores the co-evolution of pit bulls in England and people. His research suggests how very labile (in evolutionary terms) this breed has been.
The evolution of the pit bull-type dog included farm dogs (probably mastiff-type dogs) selected to bait bulls, which was important order to tenderize the meat for market (not just to protect the farmer from angry bulls). Farmers selected dogs with high prey drive, rather than high fight drive. Dogs that retained the instinct to prey on animals larger than themselves were the ones who were chosen to bait bulls. So pit bulls who fight other dogs in a particularly deadly way are likely to be showing prey behavior, rather than conspecific fight behavior.
People started placing bets on bull-baiting for all the reasons that people like to bet on such things. Queen Elizabeth was particularly fond of the spectacle.
Around 1800, bull baiting became so popular that bettors needed to standardize one way to settle bets. Two rules had been dominant: the dog who attacks for the longest period, and the dog who runs the most number of times at a bull, would win. Certain smaller, faster baiting dogs, however, were good a grabbing the bull's nose and hanging on. Gamblers called this trait "tenacity" and created a new standard rule for placing bets by throwing them into the ring, where the winner would be determined by the dog that "pinned" with a bite to the nose. Breeders quickly began selecting for the trait of grabbing only the bull's nose and lips, not the other parts of its body. Breeders would set a litter of pups near a bull, would watch to see which pups ran at the bulls head, and keep only those pups--very strong selection for tenacious dogs that gripped and held the nose, but didn't shake or rip or grab the flanks. The longer the dog immobilized the bull, the better. Dogs who ripped and tore weren't good; dogs who held on were favored.
Dog-fighting overtook bull-baiting in the early 19th century (partly because of urbanization--bull baiting was closely connected to an agrarian England and outlawed in 1835). New rules developed for betting, and these rules favored the dogs who disabled, rather than immobilized, their opponent. Grabbing and holding on (ie, tenacity) was no longer favored; instead, "ferocity", or repeating biting and tearing was. Breeders crossed the bulldogs with terriers to get this tearing, shaking, behavior (what a terrier does with a rat--that prey-shake, which wolves don't do with their larger prey). 
What happened when pits came to America? Prof. Russell doesn't examine this in his work, but here are some preliminary thoughts, which are not yet backed up with good sources:
When the Irish immigration to the US intensified, many Irish families brought their pits with them, and fighting dogs were so valuable to an Irish family that the dogs lived inside with the family (something that was very rare indeed in Europe for a larger breed), and strong selection for gentleness with children might have begun at that point.
In the early 20th century, pit bulls were one of  the most popular breeds in America--renowned for their prowess in dog-fighting, of course, but almost famous for their courage, sweetness with children, and stable temperament. The list of famous pit bull-type dogs can go on and on: President Teddy Roosevelt's pit bull, Helen Keller's dog, the pup in Little Rascals, Nipper the Victrola Dog, the dog on the military poster, the mascots for football teams.
One trainer wondered if all these famously stable pet pits suggested that lines of pit bulls being bred for calm pets had already diverged from fighting lines early in the 20th century. My sense is that this isn't the case: game-bred pits seemed to have been popular as pets, and it seems that they may not have often fought other dogs in normal conditions, only in actual fighting pits. It may have been too risky to an important family investment to have a good fighting dog that would pick fights with stranger dogs on the street. The old story about game pits was that they wouldn't start a fight, but they also wouldn't end one. Is this true? Who knows.
In the 1980s, pit bulls became popular with inner-city black adolescents, and their reputation spiraled downwards among the white middle class. Some of this was probably connected with racism and media bias: white people became freaked out over pits because they associated pits with a certain kind of African-American culture. The media's job was to sell papers and magazines, and scary pit stories sold. But it wasn't all racism. The pits themselves seem to have changed, as backyard and inner-city breeders began breeding for large heads, for much larger sizes (instead of the 40 to 45 pounds of the Irish pits, we now see 95 lb dogs). Some people began intentionally trying to breed pits for human aggression, and some human aggression probably slipped in as an unintended consequence of breeding for large heads and large body size, which meant crossing pits with guarding breeds.

Pit bulls and Human Aggression
Many websites argue old pit yardmen and fighters used to cull out any dog that showed any degree of human aggression, making pits especially human-friendly. Is this true? None of these studies are published or peer-reviewed, so it's hard to know. But it would be possible to find out, using archival data from the 18th and 19th centuries, and both archival sources and interviews from the 20th C.
Anti-pit bull advocates cite the CDC figures from the 1990s that appeared to show pits and rotties were overrepresented in the breeds responsible for fatal attacks, but those data were never peer-reviewed, and the CDC is the first to argue that they do not actually tell us anything about which breeds were indeed responsible for attacks. There are NO peer-reviewed published statistics on rates of human aggression by dogs of any particular breed or mix.
Similarly, pro-pit bull advocates like to cite Karen Delise's statistics showing that only about 20% of fatal attacks on people were by pits or pit-mixes, which is likely lower than the percentage of pits in the general pet population (suggesting that pits are LESS likely than other breeds to attack people). Again, as someone who has pit bulls I might like to believe that argument, but Delise has not published her work in a peer-reviewed journal, and I believe that until she does, we can't draw any conclusions from it.
I do think that pitties are impulsive, easily-aroused, high-intensity dogs with an enormous love for people. Typically, they don't have an ounce of human aggression toward adults. 
-If a pit is abused, it can often be quite  resilient. But increasingly, pit bulls seem to show fear aggression.
-If a pit mix was badly bred by being crossed with guarding breeds, it can also show human aggression.
-If a pit bull gets over-excited and too pushy with children, I think hyper-arousal can sometimes tip over into prey-drive, if the dog lacks training in impulse-control.
 The risk of pit bulls inflicting fatal wounds on children is vanishingly small, and if we're going to bring human aggression into the argument, we need to recognize that pits were indeed long bred for low rates of human aggression, just as they were long bred for high rates of conspecific aggression. I personally certainly hope the two aren't connected, (so that we could breed for continued human-sociability while breeding against fighting) but I'm not sure there's any published evidence that shows us one way or the other.

Can you breed a game pit bull to be dog friendly? Is gameness in pits--a quality we love--inevitably tied to scary dog aggression? Can a breeder do this without messing up the breed?
Since 2005, the dog genome has been mapped. That's a very big deal. Molecular biologists now know quite a bit about different alleles (forms of genes) at a particular locus correlated with the development of various forms of aggression; about the SNPs (single nucleotide polymorphisms--ie, the chunks of genes that vary most between individuals, where the really interesting stuff is happening); correlated with various types of aggression about various SNPs correlated with idiopathic aggression in those much-studied golden retrievers; about the genetic changes in silver fox genomes that correlated with a strong reduction in impulsive aggression.
Gene tests that tell you if your dog is going to be aggressive aren't available yet, but there's no real reason that they couldn't be relatively soon. Good breeders are taking advantage of canine genomics to reduce the rates of certain diseases in specific breeds, so why not to reduce the rates of certain kinds of aggression? Genes do not directly control aggressive behavior--they affect the behavioral regulators known as neurotransmitters, that in turn may influence the probability of certain kinds of aggressive behavior. Every year, more  is becoming known about the various SNPs that contribute to increased risk of various kinds of aggression in canines. The silver fox research group does it for their silver foxes. If breeders wanted these test, they could be on the market in not very many years, for a reasonable price (look at bovine genomics, for example: it's a huge field, because there's a demand for it).
If I were a canine geneticist, the first place I'd look would be at the MAOA-L locus, which has been correlated with a strong increase in impulsive aggression and increased reactivity in all the species examined (including foxes, I think). MAOA is an enzyme that regulates the breakdown of the neurotransmitter 5-HT or serotonin, and one little mutation in MAOA leads to all sorts of changes in serotonin, which in turn leads to an increase in reactivity and impulsive aggression. (Specifically, the genetic mutation leads to an increase in circulating serotonin in early development, which then creates a cascade of epigenetic effects that lead to lowered serotonin levels and all sorts of havoc later in life. This is partly why nasty fight breeders give their dogs tons of steroids early in life--it works with the genetic templates to create radical changes in behavior).
In foxes, it's possible to selectively breed for changes in MAOA and serotonin uptake, which then modulate aggressive behavior. I believe it's not hard to map these MAOA variations in a DNA sample.
Serotonin gene systems are actually more complicated than first suspected, involving an extra two genes, and changes in those help regulate territorial aggression versus intermale aggression. Tiny changes in these 2 genes (replacing one base pair) actually change the THRESHOLD of the aggressive response--ie, which foxes are hot tempered and which are not. Again, I believe it would be possible to sequence these genetic variations in a DNA sample, giving you some sense of the dog's later aggression threshold.
What's really nifty is that this "hot-tempered" behavior does not actually correlate, genetically, with intensity of fighting--ie, gameness; they're controlled by different genetic mechanisms, which means that a good breeder might indeed be able to separate them out, especially with the help of good genetic tests.
When you breed foxes for lack of aggression, you get changes in neurotransmitter 5-HT metabolic pathways, one of which involves the inhibition of fear-induce defensive aggression. (Which is why giving prozac to fear-aggressive pups may not always a great idea, since it can remove the inhibitory factors keeping it from being expressed. Prozac-treated rats are less likely to fight each other, but more likely to bite people in fear). This finding suggests a possible explanation for why traditionally game-bred pit bulls had a low threshold for fighting and low impulse control but were so friendly to people--lots of inhibition, genetically, of fear-aggression.
But the fox studies also suggest that it's possible to breed for low aggression and high sociability with people.
The review article that goes over all this is "Canid genomics: mapping genes for behavior in the silver fox," Tyrone C Spady and Elaine Ostrander, 2007, Genomics 17: 259-263.
So if I were a breeder, I'd want those genomic studies of aggression, and I'd want to know what those SNPs of my potential breeding dogs looked like. The cow folks are already doing this; why not the dog breeders? And while some pit bull lovers don't want to see dog aggression bred out of pits, plenty of others disagree. In my opinion, the breed has been so badly abused by terrible backyard breeding in the past 2 decades that we're nowhere near some mythical past standard. Anyway, breeds are constantly changing. If we can use genetic tests to reduce terrible diseases in many dog breeds, why not use them to alter aggression thresholds, while keeping careful watch on possible fear-aggression changes.


  1. Thanks for a informative post! Well said!

  2. Thanks for shedding light on this misunderstood breed! Fozzie the foster pittie lover-man concurs.