You are currently browsing the tag archive for the ‘positive reinforcement’ tag.
My daughter was invited to the circus last week by a classmate celebrating her 7th birthday. The circus comes regularly to a very nearby venue, but I’d successfully avoided it up to this point. The kids all enjoyed themselves, of course, and were especially excited to see real live elephants and tigers. I remember my own awe as a child being exposed to such creatures.
To say I felt conflicted would be an understatement, especially while watching the lions and tigers perform. It’s hard to know exactly how much of such an act is scripted versus accidental. But a number of those big cats looked genuinely pissed, and the third time one of the females balked at a cue and then ran at the trainer, it did not look choreographed.
Likely all part of the show on some level, but is that even any better?
My three-year-old son pretty much summed up the absurdity of the evening when he exclaimed quizzically, “Look, it’s tigers on stools!”
Why yes, my son. This is how men demonstrate their physical prowess and mental superiority over other living creatures–by making them sit on stools.
Or, in the case of the majestic killer whale, by housing them in fishbowls and training them to splash tourists.
The maiden post to this blog was in response to the death of trainer Dawn Brancheau at Sea World’s Orlando park. It was the third death associated with the bull orca Tilikum, known to tourists as Shamu.
That post was flip in some respects, but the business of keeping killer whales for fun and profit is not. It does speak directly to my problem with gauging the humaneness of a training program purely according to which learning quadrants happen to dominate. And sadly, it speaks to the power of money to distort our judgement.
I think it’s well established that a higher percentage of clicker-trained killer whales actually kill their trainers than do dogs trained by any method. I do not mean that as an indictment of clicker-training. I mean it as a challenge to the dual myths that A) killer whales are a model of reliable behavior compared to dogs, and B) their handling is a model of humane training.
What we do to killer whales is an atrocity without moral justification, in my opinion. They suffer lives of abject deprivation, void of any genuine opportunities to self-reward. They are prisoners of our selfish desire to engage with an intelligent species that wants little to do with us, absent our trapping and keeping them like lab rats.
Positive reinforcement based operant conditioning has proven utility both within and without the confining walls of zoos and amusement parks. I don’t deny that. I deny the legitimacy of extending the analogy between dogs and killer whales to the point of suggesting the best tools for engaging the latter must also be the best choice for training the former. And to the extent I personally find the level of management involved in captive marine mammal training to be abusive, I think there is some danger associated with modeling dog training practices after that example.
© Ruth Crisler and Spot Check, 2012.
[The following was written for Animal Behavior College in response to a request for an article on balanced training.]
I’m not big on labels, especially empty ones. Tags like positive or balanced don’t signify much, in my opinion, beyond the public image a trainer wishes to project. These are marketing terms, not definable training methods. Yet all of dogdom seems to have settled into the opinion that every trainer is either one or the other.
The more serious trainers I get to know from both sides of the supposed divide, the more I reject this idea. But if pressed, I do identify as balanced, and will continue to do so going forward, regardless how the scales of positive and negative or punishment and reinforcement truly fall within my programs. Balanced may not go a long way toward conveying what I do, but it’s not inaccurate, and I like to think it evokes something of what I actually mean when I use it to describe myself and my colleagues.
To my mind, balanced trainers are by and large pragmatic. We approach tools and methods sensibly, unfettered by politics or ideology. We do what works, within an ethical framework involving fair expectations, clear communication, and respect for the dog in front of us. This may mean building a foundation of understanding and enthusiasm via positive reinforcement, then layering instructional corrections overtop to enhance reliability and steadiness under distraction. It may mean choosing negative reinforcement or positive punishment first, if so doing resolves a problem safely and efficiently. In all cases, it means remaining flexible, reading the dog at every turn, and keeping all options on the table.
Balanced training is unapologetically results oriented. Results matter, both to the client rightly expecting some deliverables, and to the dog, whose quality of life may ultimately depend a great deal more on whether his owner may walk him confidently in public or take him hiking off-leash, than on which quadrants of operant conditioning happened to land the dog such opportunities. Good results also represent the most reliable indicator that a given method is sound, which is not to say ends justify means. Means, particularly highly aversive or costly ones, are justified by the knowledge and experience that they represent the optimal path toward a good result, not the mere hope of achieving one. Being results oriented is not about being a cowboy. It is about being open to both new and traditional tools and methods as long as they have practical utility, and being prepared to do some amount of internal calculus before settling on the best approach.
Balanced trainers acknowledge that the deliberate inclusion of aversives within a training program is neither inhumane nor unscientific. Our commitment to canine welfare and fostering healthy relationships between dogs and people does not inhibit us from taking ownership of those aversives we employ. We focus our energies on applying them productively and responsibly, whether via electronic collar or head halter, with maximum efficiency and minimal risk. Denying their legitimate (and largely unavoidable) role in training and behavior modification both constrains trainers unnecessarily and inhibits frank discussion of how more socially acceptable tools and protocols actually work.
There have been some major shifts in dog training culture over the past several decades. On the upside, there’s been a great surge of interest and innovation, along with a new emphasis on ethical standards and humane methods. On the downside, it has become highly politicized, and lousy with specious claims driven by competition over market share. Balanced trainers recognize their work as existing on a continuum with what dog men have been doing for centuries, not as a departure so radical as to deny their influence and contributions to our field. Even if our approach bears little outward resemblance to the training of old, we refuse to reject traditional tools based on popular trends, and balk at the arrogant dismissal of generations of skilled and accomplished trainers as backward thinking relics.
In the end, it’s a matter of devotion to craft ahead of devotion to methodology.
Does every trainer currently advertising himself as balanced conform to my private definition? No, but I think it holds true for the balanced trainers I know best and attempt to model myself after, including a number who do not identify as such.
Whether positive trainers will mainly sympathize with or feel excluded by the above, I cannot guess. But I invite them to rethink the utility of defining ourselves according to terms that are ultimately more divisive than descriptive, and to help move our industry away from empty labels and toward an honest discussion of what we really do and why.
© Ruth Crisler and Spot Check, 2012.
Steve White presented this video at the APDT conference last week. It’s by a UK trainer, Chirag Patel of Domesticated Manners, whose Youtube channel is definitely worth visiting. Here’s a lovely demonstration of how he teaches a Drop cue. It’s very similar in some ways to how I introduce my Leave It command, although those exercises include the additional goal of reorientation and movement away from the item.
I really like how Patel uses his hands to help the dog find the food from the very beginning, and find the narration very thorough and instructive. I also like that he included a handling mistake in the final cut, explaining what he did wrong and demonstrating a corrected approach immediately afterward.
So, I’ve begun working on the first homework assignment from Silvia Trkman’s long-distance tricks course.
And I suspect Atlas is already pining for the days when all he had to do for his dinner was sit. Poor animal surely thinks I’ve lost my mind. But bless him, he keeps a stiff upper lip.
A friend of mine who teaches agility wrote me a few weeks back, and mentioned her enrollment in a pair of long-distance courses taught by Silvia Trkman. I’d heard of Trkman before now, but not paid too close attention. Mainly, I recall her producing a video called Heeling Is Just Another Trick, and that it had inspired an interesting discussion on a trainer list or two.
Well, I visited her website, and decided to take the plunge. So many of my clients have expressed interest in learning tricks over the years, and in preparing their dogs for agility work, I thought her Puppy/Tricks Class would be a sound investment. It starts next week and lasts three months, over which students are encouraged (or possibly required, not really sure) to post videos of their practice sessions for comment and discussion.
There are several samples of such videos on her website, but the one I enjoyed most was this one, depicting puppy class graduation night at her training center in Slovenia. Some of the exercises may be slightly less dignified or practical than what I’m personally accustomed to, but I would certainly Not Be Embarrassed to turn out a puppy class of this caliber.
I am taking a break this morning from bitching about irresponsible rhetoric in dog training, in order to point anyone in a more conciliatory frame of mind to a handful of blog posts that argue for integrity, clear-headedness, and professionalism when discussing fellow trainers and competing methodologies. I am sure there are many more out there. Feel free to post your own recommendations.
Show Me, Don’t tell Me by Eric Goebelbecker
Integrity: a Response to Drayton Michaels by Ann Withun
One Trainer’s Path to Change by Jeff Silverman
If your dog was to be subjected to an aversive, would you rather it occurred randomly or control the timing yourself?
I put this question to a rational positive-reinforcement trainer, who responded unhesitatingly that she would prefer to control the timing of the aversive, so as to minimize fallout, and in order to potentially create some practical inhibition.
The logic of her choice hinges on a pair of sensible assumptions. First, that controlling an aversive (even just the timing) naturally lends any competent handler the opportunity to avoid (or at least temper) detrimental associations; second, that the well-timed application of an aversive has potential utility. Of course, she would prefer to avoid aversives altogether, and clearly stated so.
No surprise, given the well-publicized risks. According to the American Veterinary Society of Animal Behavior,
the potential adverse effects of punishment [include] but are not limited to: inhibition of learning, increased fear-related and aggressive behaviors, and injury to animals and people
Moreover, we are warned that risks such as extreme generalized fear and negative associations with the dog’s environment or handler, can occur “regardless of the strength of the punishment.”
This last claim must rest on belief in a dark sort of behavioral homeopathy, whereby the magical effects of punishment  endure despite its infinite dilution.
But there is another problem. If we accept that the experience of even mild punishment carries an arguably prohibitive level of risk, and we acknowledge that the deliberate application of an aversive is nonetheless safer in obvious respects than allowing exposure to randomly occurring ones, how is it that trainers come to fret over distilling off every atom of punishment from their training programs, while blithely acknowledging that naturally occurring aversives are both largely unavoidable and relatively innocuous?
One would think such events as getting stepped on or startled would carry a risk (of potentially extreme and irreversible fallout) equal to that borne by the deliberate application of a comparable aversive. Yet few cautionary tales exist to illustrate these hazards, such as happen to dogs every day of their
Even the authors of some of the most dire warnings regarding the purposeful use of aversives to punish behavior, seem to understand that the bulk of natural or accidentally inflicted aversives are fairly harmless.
I imagine it is intuitively obvious to them, as it is to me and to the dog-owning public, that a dog’s stubbing its toe while chasing a frisbee is unlikely to sour him on the game or ruin his relationship with the person who threw it.
So, what makes the demon punishment so extra-special potent, and its measured application so inescapably treacherous, compared to those unplanned aversives our dogs regularly suffer and gracefully overcome?
Aversives v. Punishment
Karen Pryor explains the critical distinction in a 2007 blog post (emphasis mine):
There’s a difference between aversive events and punishment. Life is full of aversive events—it rains, you stub your toe, the train leaves without you. These things happen to all of us, and to our pets, and we don’t control when or if they occur. Kay Laurence has an amusing paragraph about the aversive events that befall her Gordon setters (all of which they ignore)—falling off the bed, running into door posts, and more (read that article here).
In general, all that we learn from the inevitable aversives in daily life is to avoid them if we can.
On the other hand, a punishment is something aversive that you do on purpose. It may be contingent on a behavior, and it may stop or interrupt that behavior—which reinforces YOU for punishing, so watch out for that.
I find this explanation notable for several reasons.
First, it happens to be framed in response (albeit indirect) to the question, “Can you teach everything without punishment?”, yet that question is in no way addressed either by the above or within the remainder of Pryor’s comments.
It does, however, illustrate the tendency to frame discussions on tools and methods in terms of human intent, rather than in terms of the dog’s actual experience.
It’s a common tendency–and problematic, as when assumptions regarding the intention behind either the design or application of a given tool are offered as proxy for an objective analysis of how the tool actually operates or is actually applied.
Consider the myth, held true by many and even promoted by such authorities as Dr. Karen Overall, that head-halters are non-aversive. It’s an error that persists despite the reality that dogs do not casually accept wearing them, nor reliably tolerate being steered or restrained with their assistance.
It’s surprising that a phenomenon so widely observed and even scientifically documented  would be so widely ignored. But if we accept that our intentions are directly relevant to any and all contemplations of tools and methods, it’s only a small leap to imagine they may represent an acceptable standard of measurement.
And if we buy that, head-halters clearly rate as non-aversive by virtue of their gentle intention (indicated right there on the package), whereas prong and electronic collars may fairly be judged inhumane by virtue of being, as Dr. Overall put it in a 2007 editorial, “rooted in an adversarial, confrontational interaction with the dog.”
Why would anyone invest in a scheme so clearly divorced from objective analysis?
For starters, it allows one to rationalize bypassing the complicated business of assessing how a given dog experiences a given tool wielded by a given trainer under given circumstances, instead suggesting a far easier equation, according to which one need only infer a tool’s intention in order to gauge its virtue.
This represents a boon, of course, for the purveyors of tools designed more for the purpose of persuading us of their kindness than actually facilitating it, as well as for anyone in the business of evoking faith in good intentions above promoting trust in skill or effectiveness. Moreover, substituting cursory judgements for true investigation is a real time-saver, freeing one up to concentrate one’s efforts on cementing the stigma attached to those intentions deemed impure, or on promoting the prohibition of those tools and methods associated with them.
But most importantly, it diverts attention from the fact that to a dog, an aversive is just an aversive, whether willfully administered or the result of mere clumsiness, a point that–if fully appreciated–would stand to undermine the endowment of punishment with extra-normal danger and potency.
To be clear, I’m not arguing for or against specific equipment or methods. I’m suggesting good intentions wield little to no dependable influence over how much a dog gains or suffers. And until we make a practice as an industry of evaluating the effect of our actions independently from the righteousness of our intentions, we may remain blind to those cases where to two are in conflict.
Suzanne Clothier lately posted some thoughts on punishment under the title “I had to…”. On her blog, she takes positive trainers to task for dodging responsibility in instances where they’ve made the choice to punish. She offers examples of what she evidently considers lame excuses, like “the client was frustrated,” or “I had tried everything else.” And she challenges trainers to do better:
Replacing the phrase “I had to. . .” with “I chose to. . .” puts the responsibility where it belongs: on the trainer who made the choice to use techniques or equipment. It helps us all remember that in making that choice, by definition we excluded other possibilities. When using force, we need to be very clear that in discarding other options, other possible solutions, we may also be choosing to limit what is possible when we push ourselves.
For the record, I agree force is often used too casually, without due consideration of alternate strategies, and that acting out of mere convenience or fustration should be roundly discouraged. I also believe in the importance of accountability in dog training across the board. However, I was struck reading Clothier’s article by what seemed a misplaced focus on the moral peril (for lack of a better term) associated with use of force, rather than on any harm–real or presumed–that might be dealt the dog as a result.
She details an event involving a young Labrador who’d just head-butted her very hard for the second time, and describes the moment in which she considered her options:
I began to think, “One good correction might get through this dog’s thick skull.” I surprised myself by thinking that, but then I further shocked myself (and some of the audience) when I asked the handler explicity for permission to use a physical correction on her dog. She agreed, trusting me as a trainer to do right by her dog.
In that moment when she trustingly agreed to let me use force on her dog, I found something in myself that surprised me further: a little voice that challenged me to push myself further, to help this dog without force. It was like having a gauntlet thrown down at my feet. Do it without force, without ego, without justifying force.
Compelling words. But what does Clothier’s internal struggle have to do with the needs of this somewhat thick-headed young dog?
We are meant to assume he benefited from Clothier’s suppression of her ego, to understand that what he needed most in that pivotal moment, was not “one good correction,” but rather for Clothier to “take up the gauntlet” and turn the other cheek.
But it’s impossible to deduce that from Clothier’s narrative, because it has nothing to do with the dog’s experience.
Instead, she gives us a parable about overcoming temptation and perfecting one’s intention. Good stuff from a personal improvement standpoint, but no substitute for a reasoned consideration of whether a correction might have been productive. Granted, not the point. But what is??
That we are accountable for our choices to use force, yes. That one should not act out of ego or vengeance, clearly. But was that the temptation Clothier resisted? Remember, she didn’t just refrain from lashing out in anger. She suppressed the instinct to consider punishment as an option.
Despite Clothier’s drawing the familiar analogy between the application of a training correction and the specter of wife-beating, this is ultimately not a lesson in tempering one’s anger or shoring up one’s patience. It is a lesson in training one’s inner voice to distrust one’s rational mind.
Clothier equates the use of aversives with the use of force, and equates force with violence. She frames its contemplation as a sign of moral weakness, and the decision to use “force” as a failure by definition:
Whatever the answer, the solution is to recognize where I went wrong.
Dog training is many things, including a lesson in kindness and patience. But it should not be exploited as a proving ground for fringe notions of moral perfection.
If “I had to…” is a cop-out, then so is “I can’t…” After all, in making that choice, aren’t we also “choosing to limit what is possible”?
Bible and Hatchet
Meanwhile, a generation of trainers is being bullied into signing blood oaths constraining them from ever practicing the productive application of aversives.
Jean Donaldson, Karen Pryor, and Victoria Stillwell all require pledges from their disciples, while selling the public on the idea that hobbling oneself with a vow of irrational temperance is a mark of enlightenment.
The result is a murky and oppressive climate, often dominated by vitriol and intolerance, as in Dr. Karen Overall’s unsubtle insinuation that owning a choke, prong, or electronic collar may lead to child and spousal abuse:
Without exception, such devices will make my anxious patients worse and allow the anger level of my clients to reach levels that are not helpful and may be dangerous. The link between dog abuse and spousal/child abuse is now well-established (Ascione and Arkow, 1999; Lockwood and Ascione, 1998).
Like Pryor’s warning to beware the utility of punishment, lest one’s urge to punish be strengthened, Overall here concerns herself with the threat punishment poses to us. It’s a clumsy argument at best, and less than cleanly scientific. But it succeeds in promoting the point that punishment is poisonous and intoxicating, while skirting the question of what that has to do with training a dog.
Child abuse is real. Animal abuse is real. Drunkenness is real. It’s a fact there are cretins and criminals within our ranks.
Likewise, there’s a history of countering such abuses with fear-mongering, misinformation, and hyperbole. And science, or some fractured fairy tale version of it, has been drafted into these campaigns before.
These tactics are effective, which I’ve heard is reinforcing. But they are a rejection of reason, and an abuse of the influence their authors wield. It’s as old school as tent revivals and temperance unions, and as backward as beating a dog.
There are solid arguments for taking care in applying aversives. But there is no credible foundation, scientific or ethical, for the wholesale exclusion of aversives from a training program, except if one accepts the idea that the very willingness to punish is perverse, and so fit to be stigmatized and suppressed.
Take away that belief, and the dragon vanishes. One is left with a serviceable tool and a solvable problem. The dog doesn’t know you are putting your soul at risk. He doesn’t even need to know you did it on purpose.
It’s not rocket science. It’s not alchemy.
It’s just good bar tending.
4. Overall, K.L., 2007. Considerations for shock and ‘training’ collars: Concerns from and for the working dog community. Journal of Veterinary Behavior: Clinical Applications and Research. Res.2, page 106.
© Ruth Crisler and Spot Check, 2012.
I really like a dog that can keep his act together outside the local coffee shop.
Needless to say, I never stray far from the door, just in case, you know, that dude with the wheelchair and the Chihuahua for sale comes back around.
By the way, if you like looking at dogs tethered to stuff, check out this blog.
On a semi-serious note, I actually rely on an exercise out of Pat Miller’s The Power of Positive Dog Training to combat don’t-you-dare-leave-me-tied-here-while-you-order-a-freaking-frappuccino syndrome.
It’s a brilliant use of negative punishment, and one of the only exercises I know that is capable of eliciting tears from the faint of heart. Of course, on those occasions it does, I whisper softly, “Don’t worry, I got this exercise from a positive reinforcement trainer, so it’s guaranteed humane.”
© Ruth Crisler and Spot Check, 2010.
Rosie came to me last month for a 1-week training program, while her owner, a recent transplant from England, recovered from the metatarsal fracture she’d suffered while walking Rosie on-leash. There was mention of improper footwear, but the real issue seemed to be that at 1-yr-old Rosie had no leash manners whatever. Fact was that back in England, where walks were staged in open fields and marshes and leashes weren’t a big part of life, she hadn’t needed any.
She also hadn’t needed to go up and down stairs, as it turns out.
I’ve taught a few dogs to do stairs over the years, figured it couldn’t be too challenging to accomplish with a young and biddable Lab.
Well, Rosie had the most severe stairs phobia I’d ever seen. No amount of cheese, sausage, or coaxing would get her up any farther than her front end could easily reach while keeping her hind feet safely planted on terra firma.
Gentle “suggestions” that she ought to consider bringing up her rear were met by sheer panic, including throwing her full weight backward and away from the steps. For the record, I did try different types and quantities of steps, and even a low table, always with the same result. I also saw no signs of physical pain or disability.
I’m very reluctant to use force or correction in addressing a phobia, but several things occurred to me in this case. First, that running-backward-at-full-tilt nonsense was going to get someone’s neck broken one day if allowed to persist. Second, it might easily take months and untold quantities of sausage to get this dog up a flight of stairs if the choice was left up to her. Third, this was not a fearful dog generally, despite her paranoia of stairs, so it seemed entirely likely that once she discovered she was in fact perfectly capable of moving up and down stairs without the earth opening up and swallowing her, she might actually enjoy doing so.
Thus, with some reluctance, I popped on her prong collar, and allowed her to discover the downside of running backward away from stairs. It wasn’t pretty, but it worked like gangbusters, let me tell you. With that option off the table, she quickly found the courage to attack the stairs (and the sausage). By the following day, she was happily trotting up and down unbidden, and looking Very Pleased With Herself.
Of course, I was concerned that she might fail to generalize, and continue to balk at the stairs in her house. But her owners reported that as soon as they walked through the front door, she proceeded to show off her new-found confidence. Here is the note I got yesterday along with the above video:
Thought you might like to see Rosie starring in her very own movie.
Before you had her the stair phobia was so bad she hated to even walk by our staircases.
We can’t believe you did so much with her in just one week.
Her leash manners are now so good a neighbor yesterday didn’t recognize her.
Thank you so much.
With all good wishes
Mary and Richard
I gambled a bit on this one, not usually my style, but there you have it. Maybe I was channelling my inner Cesar Millan (a strangely repelling thought). In truth, this was one of those rare training sessions that would have made for excellent television. And, like Cesar, I don’t recommend that anyone try this at home.
© Ruth Crisler and Spot Check, 2010.
I like to watch Top Chef.
Not for the brilliantly embedded advertisements for Whole Paycheck, nor the towering asymmetrical bowls, nor the ever popular Pollock-inspired platings.
But because I like to cook. And I like to train dogs. And the two activities seem to me to have a great deal in common.
Professional chefs remind me of professional dog trainers. Like dog trainers, chefs come to their field from every walk of life. They knowingly follow in the footsteps of masters, tend to respect the need for knowledge of basic principles and a familiarity with a wide range of techniques, develop their skill primarily through years of practice, and finally bring their own individual styles to the table.
You also won’t find too many good examples of either that didn’t do their fair share of grunt work early on in their careers.
Cooking is in fact an interesting metaphor for dog training. It has a long history, is commonly practiced by those with no formal training, and has a collection of perfectly sound rules, theories, and traditions (as well as a handful of ridiculous myths), that were derived empirically by those at work out in the field. Tastes and techniques may differ, new schools of thought may evolve, but in the end, the strength of any given method may generally be appraised by the quality of the results.
Over the past decade or so, a style of cooking called “molecular gastronomy” has developed, begging the question of whether the many styles of cooking that came before were somehow unmolecular. The Wikipedia entry for molecular gastronomy begins,
Molecular gastronomy is a scientific discipline that studies the physical and chemical processes that occur while cooking. Molecular gastronomy seeks to investigate and explain the chemical reasons behind the transformation of ingredients
Practitioners of molecular gastronomy are dedicated to taking a more scientific approach to their craft. They have introduced a number of interesting new methods and tools to the kitchen (like liquid nitrogen, yum). They have also debunked a few myths.
Here’s the thing. The food turned out by this new breed of chef is not obviously or unilaterally superior to what your mom made for dinner, not to mention the dishes turned out by, say, Julia Child.
And while no one has so far suggested, at least to my knowledge, that the art of cooking generally would benefit from banning traditional tools and methods in favor of an all-molecular approach, God knows what the future has in store.
Perhaps one day soon, it will be widely believed that molecular scientists are the natural authorities on what tastes best. Perhaps one day a PhD in chemistry will be considered de rigueur before breaking an egg. Perhaps future seasons of Top Chef will involve judging cheftestants on method alone, while host Padma Lakshmi politely demurs from taking even a bare amuse-bouche between her lips. Perhaps not.
In the meantime, I keep my mind open to both received wisdom and modern science.
Food for thought.
© Ruth Crisler and Spot Check, 2010.