Wednesday, June 27, 2012

The downside of luring a fearful dog

Here's another counter-intuitive idea, one I've taken a long time and many mistakes (some quite recent) to appreciate. I'm generally a big fan of "free shaping," a training method that marks and rewards successive approximations of a desired behavior (or, at its freest, a behavior that the dog invents). Free shaping requires patience, but it can really "light up" a dog, as it engages her initiative, intelligence, and pleasure in learning. That said, I haven't been a purist in this regard; I'll often resort (and encourage my clients and students to resort) to the use of food lures to jump start behaviors when time and/or patience are at a premium. (To begin to teach "down" from a sit, for instance, it's common to take a good, smelly treat and use it to draw a dog's nose down, then out along the floor in an L-motion.)

I'd like to take these shortcuts less often, but I haven't known them to be harmful or problematic in most run-of-the-mill training situations, as long as the lure is faded quickly and the dog kept mentally active. (Too much luring can lead to a kind of "passenger seat" phenomenon - have you ever been driven somewhere three or more times with nothing on your mind but the passing scenery and the fun you'll have when you arrive at your destination, then gotten lost when you had to drive there yourself? The same thing often happens with a dog whose focus is totally consumed by a moving treat.)

As harmless as luring generally is with a confident dog learning basic behaviors, it can backfire badly when used to encourage a fearful dog to "overcome" its fear, and I've become convinced that we should avoid it whenever possible.

The reasoning behind luring a fearful dog collapses the logic of classical conditioning (developing habits of association) with the logic of operant conditioning (developing habits of behavior). If I use the scent and sight of something delicious to draw a fearful dog toward me, my hope is that the dog will not only learn to approach what it fears (me), but also that I'll become less scary as I become more strongly associated with good stuff. In theory, this approach seems like it could be doubly powerful, but in practice it can create serious, even dangerous difficulties.

When luring a confident and physically sound dog in order to sketch out a new behavior, I'm not asking the dog to overcome anything but his ignorance of what he needs to do to get the treat he wants. Imagine instead that the dog were badly arthritic and I tried to lure him to do something physically painful. It wouldn't be surprising if the dog simply balked. Imagine that I offered a series of progressively more tempting treats, until I found the one that "worked" (a bit of steak, say), that is, the one that allowed me to override the dog's strong instinct toward self-preservation. I would not have used "force" as it's commonly understood, but it's doubtful that the dog would now be more likely to perform the desired behavior without the strong, direct temptation of the lure. And multiple repetitions of this "bargain" could well create an association between steak and pain, almost literally poisoning that source of reinforcement. This analogy also helps explain why a fearful dog, when lured, may snatch the treat, then bite the hand that fed him.

A dog riddled by pain and a dog riddled by fear share an especially strong need for autonomy; their great vulnerability naturally heightens their desire to be in control of themselves and their immediate surroundings. For a fearful dog, distance (between the dog and the object of her fear) is often the most critical variable determining her relative sense of comfort, and giving her control of that variable is often the most effective way to boost her confidence, especially if control itself is paired with other good stuff, and the feared object is the source (or the predictor) of that good stuff.

How to do this? One of the simplest and most effective methods I've found for ameliorating a dog's fear of humans is the "Treat and Retreat" exercise developed by Suzanne Clothier and Ian Dunbar. I should say "exercises," since trainers have invented many variations on the theme, but the principle remains the same: rather than luring the dog toward you (by offering the treat ahead of time and using it to draw the dog in), you reward any independent choice by the dog to approach (even the tiniest and most hesitant) with both a food treat and increased distance. My favorite way of doing this is to mark (with a quiet "yes") a dog's approach (or a lean forward, or even an inkling of interest) and toss the treat past her (gently and underhand). She has already established the distance at which she feels just safe enough; when she goes for the treat, the pressure of my frightening presence is relieved (though she may be reluctant to turn her back on me). Having moved further away, she is then set up to move back toward me to her original safe distance, a choice I'll mark and reward again in the same way.

As the dog comes to understand that she's in full control of the distance between us, and as she comes to think of me as a treat (and security!) dispensing machine, she'll draw closer at her own pace. And I will never have put her in the bind of choosing between nourishment and safety.

Grisha Stewart's BAT (Behavior Adjustment Training) extends this general approach to a variety of situations, with a strong emphasis on the "functional" reward of distance/security over the "bonus" reward of food. Its advantage over many other forms of counterconditioning (for the dog's well-being and building confidence) rests in its support of the dog's autonomy and self-control. Though for safety reasons BAT is practiced primarily on-leash, it still effectively allows the dog to choose the working distance: one of the handler's primary responsibilities is to discover and honor the dog's dynamic "zone of minimal discomfort" (as I would put it), that distance at which the dog alerts to a threat but remains capable of self-calming and peaceable choices. I highly recommend her book and videos to anyone who's not yet familiar with her approach.

Tuesday, June 5, 2012

Just extremely cool

Last night I attended a talk by Dr. Diana Reiss, a professor of biopsychology at Hunter College who has been studying animal cognition for roughly thirty years (dolphin and elephant cognition in particular). The event - sponsored by the Oregon Museum of Science and Industry as part of their "Science Pub" series - reminded me of one reason I love living in Portland: it's full to bursting with fellow (and fella) geeks! Arriving at the cavernous Bagdad Theater all of fifteen minutes ahead of time, I foolishly imagined that I'd join a warm huddle of fifty or maybe a hundred other psychology/ neuroscience buffs and dolphin lovers, to hear Dr. Reiss discuss the fine points of the mirror self-recognition test. Instead, I found myself clambering into one of the few remaining seats near the front of the theater, squeezing carefully between an assortment of knees and a line of skinny tables loaded with half-drunk pints of beer and and half-eaten slices of pizza, feeling like I'd crashed a party that was too big for anyone to mind much.

There was a lot to enjoy and a lot to learn from Dr. Reiss's mostly informal account of her research into cetacean thinking (she was there to promote her new book, Dolphin in the Mirror). She told a great anecdote (one I could swear I've heard from another trainer) about having inadvertently taught one of her subjects how to administer a time out: when she failed one day to cut the spines from the tail section of the mackerel she was using as a reinforcement, the dolphin retreated to the opposite side of the pool and waited there (watching her) for fifteen seconds, until he judged that she had registered the negative punishment. And she had - she was doubly careful thereafter to prepare the fish properly, though she did test the recalcitrant dolphin on a handful of occasions by intentionally leaving the spines on, and was given a time out on every occasion.

I also loved hearing about some of the wonderfully open-ended research Reiss conducted back in the early nineties with dolphins at Marine World Africa USA. Like many of her colleagues, she'd been continually frustrated in her attempts to "crack the code" of dolphin communication, but she knew that they were terrific vocal mimics. She decided to create a small, novel, mutually intelligible vocabulary by pairing some easily imitable but previously meaningless whistle patterns with meaningful objects: prized toys. One thing that was especially striking about her approach (and doubtless contributed to the success of the research) was her prioritization of choice and control for her subjects. She created an oversized keyboard with symbols that she'd already determined the dolphins could easily distinguish. When a dolphin pressed a key, it elicited a specific whistle and a specific toy: the dolphins chose their reinforcers, and each reinforcer became associated with the sound as well as with the symbol. As she had hoped, the dolphins started using the "words" for their preferred objects, often repeating them while at play. More remarkably, they invented compound terms ("ball-hoop") that they would use only when they had both objects in their possession. Reiss had built a pause into the keyboard, so that even when keys were pressed in quick succession, the two whistles never overlapped; thus the dolphins' neologisms went beyond mere imitation and association.

More wonderful than any of this, however, were Reiss's video clips of dolphins in a state of narcissistic bliss. I've talked in earlier posts about the mirror self-recognition test and the stubborn skepticism in some corners about whether it demonstrates true self-awareness. Can anyone watch this enchanting video and remain in any doubt?