Poisoning the Process

Clicker training is a great innovation, and lots of fun for trainers, learners, and audiences alike. I can get my dog to sit quietly on her mat in the vet's waiting room, focus on me, and ignore the other dogs—and CATS!—nearby. To the astonishment of my colleagues in traditional education, I can walk into a classroom of hellions, most of whom have done time in the principal's office, click for quiet, sitting in one's chair, and raising a hand to be called on, and have order, turn taking, and good manners almost immediately.

I love all the positive energy at clicker training conferences, and attend them as often as I can. I've noticed something insidious, though: despite the best of intentions, an occasional negative sneaks in, often under the radar of even very skilled positive trainers. This can happen in a number of ways when shaping a behavior, introducing a cue, and pacing sessions. Karen Pryor first drew my attention to this problem in her article "The Poisoned Cue" which appeared in the August/September 2002 issue of Teaching Dogs.

Pryor describes how a trainer can poison the process of learning by including correction for failure as well as positive reinforcement for success: the cues become threats as well as promises, evoking ambivalence instead of certainty. This inclusion of correction causes behavior to deteriorate and reluctance to replace motivation. The cue becomes poisoned because "it is no longer safe…. The shift becomes visible in the learner's attitude, which switches from attentive eagerness to reluctance, often with visible manifestations of stress. Even though successful response to a given discriminative stimulus is still followed by reward, if failure is now followed by punishment, you have made that discriminative stimulus ambiguous in terms of predictable outcome…. You have poisoned your cue."

A trainer can poison the process of learning without poisoning individual cues (that is, despite using positive reinforcement exclusively). This comes about unwittingly—and ironically—because of the trainer's expertise, focus, and purposefulness. The trainer may require the learner to repeat a behavior many times hoping for greater progress or the solidification of gains. The trainer runs the risk of becoming focused on these aims and underestimating or missing the learner's signs of fatigue or waning enthusiasm. The trainer is an expert and the learner a beginner. The trainer's vision, attention span, and buy-in are far greater than the learner's. And the learner performing the behavior over and over is also expending more energy than the trainer. The trainer's lack of punitive intent is irrelevant to the learner who now realizes that training sessions can be exhausting, even stressful and discouraging despite the many clicks and treats.

Trainers of both human and non-human animals run the risk of poisoning the process of learning through excessive repetition of any behavior that requires both mental and physical attentiveness. Here is an example of poisoning the process from ice skating, a sport that I coach:

Landing an axel is a benchmark for ice skaters. Unlike all the other jumps, the axel takes off forwards and involves one and a half turns in the air before landing backwards (in international competition, skaters are doing triple axels—3.5 turns in the air!). Things can go wrong in so many places—you can hesitate slightly at the take off, lean a little to one side or the other, drop a shoulder, kick the free leg imprecisely, give insufficient pull to your arms, look down instead of up and into the jump, and on and on.

As a clicker trainer, I know to work on one element at a time, and to relax the standard for that element when I switch to the next one. Students really want to land an axel and tend to work diligently toward that goal. And then, lo and behold, like magic, after months of work they land an axel. Landing your first axel is cause for celebration, announcements on the rink's public address system, photos on camera phones, web pages, and phone calls to grandparents— but not for repetition. Why not ask the skater to repeat it immediately to solidify the skill? Because the skater will not be able to land another axel right away. As part of the normal vacillation, the ups and downs of acquiring skill on each component, everything serendipitously came together before the skater actually had sufficient technique and control to execute the maneuver.

The skater will not be able to repeat the axel without more practice on each of its components. If asked to do it again, the skater will execute a poor encore and get discouraged. The coach, wary of ending on such a dismal note, will have no choice but to ask the skater to repeat the axel again and again until another is landed. It will not be as good as the first, frustration and fatigue having taken their toll, and the skater will leave discouraged at the ephemeral success that was eclipsed by another 25 or 30 failures.

Not landing an axel if you've never landed one is not a failure. It is just practicing. If you have landed an axel, not landing the next one is a failure. Thus, the overly purposeful coach has managed to grab failure from the jaws of success and wave it around the arena by having the skater try to repeat the triumph.

Why does this happen? Because the trainer has a longer attention span, greater stamina (and is usually exerting less physical energy), and a more linear perspective than the learner. The first time the learner lands the axel, plays the piece with the tricky fingering correctly on the piano, or walks in synch on a loose leash with the dog, the learner is overachieving, performing beyond his/her ability. The success is a fluke. But what a great feeling! If the trainer ends the practice of that skill with that triumph, and adds a bit of hoopla for effect, the learner leaves, chest swelled with pride at this new ability to do the behavior. The learner will come back to the next session confident and eager, filled with the memory and the feeling of the success, and willing to do many repetitions to hit the target again.

If, instead, the trainer asks the learner to repeat the behavior right away, the learner will try, and will do a poor job. The trainer and now-discouraged learner will have to dig in and repeat the behavior enough times for another success to occur. The learner leaves demoralized, knowing he/she did two out of perhaps 50 or more repetitions correctly, a 4% success rate. And, the learner will be ambivalent at the next session. Even though the trainer has clicked and treated exclusively, never correcting, the trainer has nevertheless poisoned the process by going beyond the learner's likely success. The learner leaves frustrated and fatigued, and estranged from the trainer whose vision of success and determination unwittingly poisoned the learner's experience.

Learning curves look very different to the trainer and the learner. From a larger perspective, the trainer sees a fairly linear path toward the goal. The learner, on the other hand, experiences a roller coaster of vacillations, and may not appreciate small incremental steps toward the goal that has not yet been reached. Landing an axel is a clear triumph, and noteworthy even to the learner who cannot see the topography of the forest, for all the trees along the way. If the trainer stops the session (or at least stops working on that specific skill) at the top of each hill, the learner will end each session on a high note despite the learner's perception of a bumpy and unclear journey.

As a positive trainer, how do you avoid poisoning the process? Overcome your eagerness to have the learner repeat behaviors done well the first time. STOP as soon as the learner gets it right. This doesn't necessarily mean you stop the training session, which might feel like punishment to the eager learner, but stop working on that particular skill. Celebrate the achievement, and go to another activity that is very different. A well-paced session will therefore cover more behaviors and fewer repetitions of each behavior. At the end of each session, the learner can bask in the glow of many successes in many different behaviors and unambivalently look forward to the next session with the purely positive trainer.

Many clicker trainers teach groups as well as individuals. Consider the impact of poisoning one learner's process on the other members of the group. Observers find it fascinating to watch a skilled trainer build a behavior. People pay rapt attention and see myriad events to click that they might have missed on their own. The audience actively participates in the process and often bursts into celebratory applause when the learner gets the behavior right.

Then, suppose the trainer says to the learner, "Do it again." Predictably, the learner complies, does his or her best, and executes a poor imitation of the previous triumph. The audience is immediately disheartened and feels for the learner. The learner tries again and does slightly better. The audience can see the painfully long road ahead, and becomes uncomfortable observing the protracted debacle. Unwilling voyeurs, people begin to disengage, turning away, looking for distractions, talking among themselves, checking their schedules and the like—calming signals in effect. Sympathy for the learner's arduous trek develops into wariness of the trainer. Perniciously, the audience's individual and collective relationship with the trainer has also been poisoned. People will be self-protective in the future, reluctant to participate and unwilling to volunteer to demonstrate a new behavior because the trainer has poisoned the process.