Lunch Free at Stockdog Trials?

There is no free lunch in animal training.  Healthy animals won’t do something for nothing.  Some people are very disappointed by this, or engage in unhelpful denial contortions, most often when the species happens to be dogs. 

So what about behaviors that go and go like perpetual motion machines?  Understanding what’s actually going on in such scenarios requires us to examine the too-sharp distinction between behaviors and reinforcers, which, while useful in a functional analysis, misses the Premackian point: what we call reinforcers are in fact just preferred behaviors. 

Sitting and staying still and coming when called are all behaviors, but so are eating and drinking and playing.  Preferred behaviors have the sensational ability to boost less preferred behaviors provided the contingency is made apparent to the animal:  “You must do X units of less preferred behavior in order to have the opportunity to do Y units of preferred behavior.”  As b-mod shorthand, we call the most preferred behaviors “reinforcers” (for example, we say “food” rather than “eating”) and we call making the contingency clear “training.”  A student of mine a long time ago said, “so what you’re saying is it’s all Premack.”  Yes.  It’s all Premack. 

Three behaviors have immutable high preference among living animals: eating, drinking and sex.  (You could also say, as many trainers do, that they are “self-reinforcing” behaviors.)   They all have refractory periods, i.e. they wax and wane depending on how recently the animal was able to do them, but they are uniformly motivating.  Animals not motivated by food or water die within days (water) or weeks (food).  The genes associated with animals unmotivated by sex tend to not get themselves passed on very well.

Other strong preferences vary individual by individual (and in dogs, breed by breed).  For instance, for some dogs, fetch is a preferred behavior.  They do it “for free.”  Border collies herd sheep “for free.”  Terriers play tug “for free.”  And so on.  I read books about the English monarchy and watch Yankees games “for free” too.  And I would pay to do these behaviors with non-preferred behavior.  I know many people who would need payment in the form of one of their own preferred behaviors in order for them to watch a unit of baseball.  This doesn’t make them – or me – morally inferior or superior, or harder to train.  However it would be idiotic to try motivating any of us with a non-preferred behavior.  Or to wait for non-preferred behavior freebies.

Voodoo trainers mangle all this.  They feel entitled to a free – or deeply discounted – lunch in the form of massive amounts of non-preferred behavior, and mistake dogs doing certain preferred behaviors “for free” as evidence the dog is motivated by praise doled out when the dog does the self-reinforcing behavior. 

You won’t come across this error for behaviors like eating and drinking (“He eats because I’m pack leader” or “He drinks water because it pleases me”) but you will see it for behavior that happens to have utilitarian value, such as border collies herding sheep.  Because it’s something we find helpful or useful, we make the logical leap that the dog is doing it so that he can be helpful or useful to us, rather than for his own reasons, usually to do with selective breeding.  Just as I have been (inadvertently) selectively bred to find watching the New York Yankees reinforcing, border collies have been (deliberately) selectively bred to strongly prefer eye-stalking moving objects.  The refinement of herding in border collies – which flanking maneuver to do when, driving stock away from the handler instead of gathering, when to put the brakes on etc. – is achived via a combination of letting the dog continue to eye-stalk if he complies with these cues, and punishment, usually in the form of loud, gruff voices, which for soft dogs like border collies, are a sufficiently aversive P+ to do the trick. 

Preferred behaviors are our most precious dog training resource.  They are our leverage. They motivate dogs to do the things we would like them to do as well as motivate the replacement behaviors for the myriad things we would like them not to do.  As if that weren’t enough, they allow us potent non-violent punishment options in the form of cessation of opportunity to continue them.  

Drivey dogs are the cheapest dates as they can be motivated with any combination of fetch, tug and food.  But non-drivey dogs are also a cinch as, without exception, they will be strongly motivated by food.  One would think there would be universal jubilation that we’ve got this leverage bonanza, and yet so many limp on and on and on in their attempt to duck the lunch check.

Cooked Chicken Also Now Dangerous?

This is from an article on rehabbing dogs with multiple problems, in which the author advises owners to:

“• Develop a positive ‘cult of personality’ with your dog. If she thinks you are an influential rock star, she’ll want to please you. Be confident—even haughty. Stoking her desire to please is the key to good behaviour.
• Distrust the power of treats. Though good for initiating behaviours, they can soon outstrip you as the prime motivating force, and create a pushy cookie monster. Praise, and the cult of personality should become the bulwarks of your authority.”

I don’t know the author from Adam so maybe he really believes that being “haughty” with one’s dog will help stop house-soiling, food-bowl guarding, interest in squirrels and cats, and improve sits, downs, recalls and stays.  However my guess is he knows this is ridiculous, probably uses some sort of collar and leash system and has some inkling that without these owners would fall pretty short of rock star influence no matter how “confident” they acted.  But to say to people, “yeah, you could motivate effectively with food, but I feel more comfortable using pain and fear” doesn’t have the same evangelical charm.

Or how about this one:

“Don’t let inferior dog training techniques destroy your relationship with your dog.

Why risk it? You could be putting your dog in emotional or physical harm. We’ve seen both the cookie-bribery trainers and the dog trainers who use the super-harsh methods and we don’t agree with either. Using either approach can undermine your relationship with your dog and/or cause physical harm.”

I can connect the dots between super-harsh methods and emotional and physical harm and relationship destruction, but I’m stumped about how cookie-bribery does this.  Maybe it’s more tortured logic wherein if the dog finds out how off-the-planet fantastic chicken is, the owner will feel dissed because the dog rarely reacts that way to him.  If this is the case, then “the relationship” is weasel-speak for “my narcissism.”  It also means, incidentally, that your significant other should get no movies, laughter with friends, football, cake, good books, fun time on fast vehicles or whatever their passions are so that you stand a better chance of ascending to rock star status amid the now more barren fun-o-sphere.    

Or this one:

“The food-bribery dog training method is short-term at best and the dog is not taught to respect its human.”

“Respect,” by the way, is code for “fear.”  He continues:

“The dog begins biting people as a direct result of owners trying to train their dogs with food, too often resulting in the dog having to be destroyed. Giving an aggressive or dominant dog food to train is dangerous positive reinforcement and can make many dogs even more aggressive and dominant.”

Good heavens, biting people as a direct result of using food to train?  Dangerous positive reinforcement?  All peer-reviewed research on this actually finds the diametric opposite but I’m betting the positive-reinforcement-is-dangerous crowd doesn’t scour academic journals in an attempt to refine their craft. 

This vilifying of food is interesting stuff.  If I wanted to spike crime rates in humans, I’d probably not start by offering baked goods to juvie hall kids for desired behaviors.  And picture a food-o-phobe telling a zoo trainer that elephants and gibbons should be motivated to comply with veterinary procedures by the keeper cultivating a rock star persona.  “Uh, let’s see you try it first…” 

Putting aside the *actual* motivators these people use – choke collars, prong collars, electric shock, pinning dogs to the ground and other things designed to intimidate animals into compliance – we could build a continuum of reinforcer chastity: 

An item’s proposed location on the continuum is based on: 1) the righteousness (going in) of dogs who find that item motivating, and 2) whether using that item actually corrupts the dog’s goodness.  Praise is the uber winner on both counts.  Dogs who work for praise are the most noble, worthy and admirable, and the least conniving. 

And because of clause #2, the more praise is used on a dog, the purer the dog gets.  If it’s not effective, we invoke clause #1: base, attitudinal dog (in need of flagellation – ta-da!). 

Door-opening, attention, giving the dog a chew toy and the opportunity to sniff on a walk – so-called “life rewards” – are less virtuous than praise but not as erosive of moral fiber as, say, food.  (Eating is apparently divorced from the dog’s “life.”)

Tug is middle-ish but with a bullet.  Whereas twenty years ago we all parroted how playing tug was the gateway to anarchy, now most trainers extoll its virtues as a convenient, recyclable motivator and energy-burner for drivey dogs, and one that has the fabulous asset of not being food. 

Which brings us to food and the mercenary, profligate dogs who work for it.  Notice nobody ever talks derisively about dogs “begging for walks” or “becoming dependent on having the leash removed at the dog park” or the atrocity of “praise bribery.”   

The crushing irony is, of course, that for all living dogs, food is insanely powerful, insanely convenient (ever try bringing door-opening along on walks?) and carries the spectacular side effect of the dog liking the training a bit more every time it’s dispensed, liking the trainer and the trainer’s hands a bit more, and liking anything else that is a reasonable predictor of it.  This Pavlovian conditioning side effect is squandered on food bowls and 5:00 p.m.  I wonder if trainers who hate food feel a pang of jealousy when they are forced – by law – to put some of it down on a regular enough basis to stave off their dogs’ starvation.  Or perhaps free food is exempt.  I’m not clear on this. 

I’ve long thought it is food’s very potency that freaks some people out.  It makes dogs look like crackheads, which makes the coercion crowd feel dissed or invisible.  A client could get perilously close to an excellent outcome with desirable side-effects using food, but (drum roll) “he wouldn’t be doing it for you, he’d be doing it because of the food.”    

Plenty of well-meaning owners, who work hard for a living, are made to pay money for this drivel.  And non-aversive trainers actually sometimes kowtow to it by getting all apologetic, rushing to assure owners that this execrable thing, this FOOD, can be phased out once behavior is learned (which would be extinction, which would be bad) or replaced by wholesome, upright motivators such as door-opening, attention, walks and so on.

Of course it’s good practice to employ intermittent schedules to increase resilience to extinction, and of course it’s crafty to diversify the motivator base but some light needs to be shed on this tacit consensus that The Less Food the Better.  Do we think owners will fall to pieces if we said:

Yeah, food is an AWESOME motivator with transcendent side effects so let’s figure out how to get you in the habit of having high value food available when you need it so we can train up strong behavior in a wide variety of contexts.  He’ll never do it because you’re a rock star, and I’m actually pretty incensed that good people like you have been sold the lie that dogs will do it just for you or just for praise.  Trainers who would have you believe this are as dependent on their choking and shocking and pinning and scary stuff as we will be on our food, but that’s how it is with living things.  No free lunch.    

A student of mine pointed out that it’s actually kind of condescending to assume owners can’t handle this truth.  They’re competent adults and won’t tumble into existential obliteration if you level with them about the economics of behavior. 

It also got me thinking of an Emperor’s New Clothes challenge where trainers who pour scorn on food are invited to see if they can train up a single behavior on a single green dog, using their haughtiness, desire to please stoking and personality cults but without any choking or pinning or shocking or bullying or yanking on leashes.  We all know the result they’d get.  Because guess what: dogs aren’t doing it for them either.

The Weasel Files Part One

I did this piece a while back (it may still be floating around on an unauthorized website) but I’d like to post it here, re-edited, because the topic, that of re-packaging basic behavioral science – sometimes mangled basic behavioral science – as New & Improved training methods, is as alive as ever.

The subtitle of The Culture Clash has the word “revolutionary” in it.  The book actually makes a case for training using operant and classical conditioning, technologies that have been around for over sixty years.  Marketing gurus have known for a long time that “new and improved” sells stuff.  If I were to be generous I’d say that at the time of that book’s original publication, OC and CC were given short shrift in terms of their importance in applied dog behavior (versus, say, preceding dogs through doorways), which made them “revolutionary.”  If I weren’t being generous, I’d have to admit that there’s been nothing fundamentally new in animal training and behavior change technology since the early decades of the previous century.  This isn’t because the field is stagnant.  It’s because operant and classical conditioning are potent technologies with vast explanatory power.  They’re under-exploited and poorly understood because they’re not easy to get full mastery of and not easy to sell to clients who watch TV trainers talk about energy.  And they’re not new.     

New & Improved sells books and conference tickets.  But because there isn’t actually anything real that’s new in training and behavior modification, people are tempted to make stuff up.  Dog training is a particularly fertile discipline for this, compared to, say, geology.  Equal time isn’t given to the latest ramblings of the Flat Earth Society at geology conferences or in journals, in the name of free speech or kindness.  But in dog training, anybody can say pretty much anything.  Then, if it’s competently critiqued, the quackery purveyor can cry foul in one of three ways:  

  1. Appeal to free speech, open-mindedness or “the more ideas the merrier” – the earth can be flat, round, oval, French fry shaped or tacoid: we need to respect all points of view and teach them at professional conferences.  For instance, Rapture-readiness advocates should get equal time at cosmology conferences.  Right?     
  2. Appeal to experience or credentials: the earth is flat because, hey, look around you.  Or I went to Yale so no matter what garbage comes out of my mouth, you should believe it.
  3. Appeal to loving kindness, which goes something like this:

New Method Promulgator:  This seminar/book/other revenue-generating item is about healing your dog’s brain tissue and limbic system so he is no longer aggressive to strangers. I call it HAT – Histological Alteration Training.   It is an exciting, new way to…

Trainer Versed in Learning Science:  Wait.  I read your stuff and watched the demonstration video and this is basic negative reinforcement with some obfuscating and irrelevant additions.  The dogs are stressed, and there are existing techniques that are better-vetted, lower risk and more humane.  But aside from that, you really ought to use the existing lexicon of applied behavior analysis.  Is it that you don’t know the basics of applied learning science and have just attempted a wheel-reinvention, or that you do know the basics but think you can dupe well-meaning dog trainers?  I’m frankly not sure which one would be worse. 

New Method Promulgator:  Oh wow, you know I worked very hard on this exciting new method and I feel hurt by your unkind attack.  You know, what we need is more benign gentle loving kindness in this field.  You just don’t get it.   

Incorrect.  Plenty of us get it just fine.

Cognitive neuroscientist Steven Pinker makes a fabulous point about this. The greatest strength of science, he says, is that it is a societal realm where some of the usual rules governing interaction don’t get invoked.  Controls must be applied, evidence must be publicly check-able and findings must be replicable.  If not, you’re a quack.  Period.  In our interactions outside of science, we can employ tools like status or appeals to compassion to deflect criticism.  Or free speech.  Free speech is critical to a free society.  And you can say anything in science too, by the way.   It’s just that if you say something that’s palpable nonsense and don’t back it up with evidence, you’ll be laughed out of the room.  Which isn’t very nice but it gets us computers and antibiotics.

I take particular exception to the Po-Mo notion that all truths are equally valid.  If advocates of this philosophy really believe this, they need to put their money where their mouths are.  But they don’t, do they.  They want real surgeons trained in real scientific medicine when their appendix ruptures.  They don’t leap off buildings based on their conviction that gravity is but one narrative.  They communicate with each other about how nothing is real using real cell phones and real computers designed by real scientists.  They don’t get on airplanes their friends conjure up in philosophical tangents in restaurants.  They want real airplanes designed by real aeronautics engineers.  Richard Dawkins once said, “Show me a cultural relativist at 30 000 feet and I’ll show you a hypocrite.”  When the chips are down, the all-ways-of-knowing-truth-are-equally-valid crowd vote with their feet.  

But in realms where there isn’t the instantaneous death of hundreds of people at high velocity per unit failure, and hence regulation of who gets to build airplanes, and where nobody has called you much on your quackery, and where there’s a career to be had broadcasting your half-baked theories to the credulous, you just might get away with it.  At least for a while.