Shaping behaviour is a two-way conversation, in which we ask a question, receive an answer, listen and acknowledge, back and forth. When the animal tells us they’re ready we respond by changing the question slightly or asking for a little more. It is an art-form akin to a dance; we might take two steps forward, one step back, three steps forward and in this way we dance our way ever closer towards our goal. You may have heard the terms "splitting" vs "lumping". Or the phrase "rewarding the smallest try". I’m going to add another word to the mix: criteria. Criteria is one corner of the golden triangle of training, to which Bob Bailey asserts we can attribute the vast majority of training problems. It is crucial that we don't always set our criteria too high or else the animal won't be successful and our rate of reinforcement will drop too low. (Cue frustration, shut down or disinterest from the animal). I'll say that again with different words: when we ask for more than the animal is capable of understanding or delivering in that moment what we get is hesitation, wrong guesses, or our animal gives up, shuts down or otherwise "fails" and doesn't get rewarded. Doesn't feel good for them, or for us. Unfortunately, it is human nature to think fast, to set goals for the future and want to achieve them yesterday. We also tend to have unrealistic views (from our comfortable perch atop the mountain of prior knowledge and human agenda) of how “obvious” the answer is, how quickly our animal should be able to figure it out; in short we assume the job of learning is easier than it actually is. In fact, what you think about it all is irrelevant - your animal tells you whether your criteria is small enough or not, by their success rate (and therefore by your rate of reinforcement, i.e. the frequency of your rewards). Setting our expectations and our criteria too high and trying to move on too fast before the animal is ready (aka "lumping") is a pretty well-known training issue. It's safe to say that in almost every scenario it wouldn't hurt to split or “thin-slice” behaviours more than we already are. However, something I see quite a lot but doesn't often get a lot of air-time is the opposite - the tendency to spend too much time drilling at a stage the animal already knows well. This usually happens due to a lack of trainer confidence and the sense of success and safety that comes from the animal getting it right every time. Sometimes we actually sabotage our own progress because we like to stay in the comfort zone where the animal, and therefore the trainer, is successful. It also happens because of a lack of a clear plan as to next steps. "Some of the most annoying hours in my school career were spent listening to an instructor drone interminably through explanations of a concept I already understood. Remaining at a low level of criteria or performance can be intensely frustrating to a learner. In "Don't Shoot the Dog", Karen Pryor points out that failure to move ahead when the student is ready can be just as aversive as any punishment." So once our animal is getting it "right" about 80% of the time at our current criteria, we NEED to move on up and ask for more! More duration, more precision, more distance, more expression, more speed, more softness, more distractions, or whatever it might be. We just gradually paint those layers on, one at a time. (As we shift the focus to another criteria we may see a drop in quality of the other aspects we’ve been working on. Don’t worry. It’ll all even out in the end). Alexandra Kurland talks a lot about loopy training, which is an analogy or framework of sorts for this exact scenario. She describes it as follows: “In a clean loop, the trainee performs all the elements within the loop smoothly and without hesitation, and no unwanted behaviors creep in (as might happen if the emotional balance is off). Once the loop is clean [before and after the click], it’s time to move on to the next criterion. In fact, when a loop is clean, the trainer should move on!". For more about loopy training read this blog. Good training is rhythmic, almost meditative. (Here's a nice example - do you see the clean loops of behaviour?). I find it puts me and my learner into a flow state where the world disappears and we're completely in the moment with each other. The reinforcement process (click, reach for food, deliver food, return to neutral ready for the next rep) is a cohesive part of the “loop” and there’s no hesitations or sticky bits at any point and so it flows. No training session is going to be entirely like this, but this is always my goal. You can see how any messy moments will disrupt your flow - these might include confusion or frustration behaviours from the animal, or your equipment malfunctioning so you're having to reset it all the time, or if there's anxiety or foraging/mugging behaviours going on between reinforcers, etc. By setting criteria well, cleaning up your food delivery and fine-tuning your environment/props/set up you can eliminate many of these messy moments and create smooth clean rhythmic loops of behaviour that spiral steadily outwards towards your goal. Shaping behaviour is a continuous dance in which we step forward, step back, pause, skip three steps forward and so on... if you stand in one place too long or repeat the same steps endlessly when dancing, your dance partner would likely get frustrated or bored. Be brave enough to ask the questions; your animal will probably surprise you. Keep the flow going and enjoy the process as you dance your way toward your goals.
0 Comments
Your comment will be posted after it is approved.
Leave a Reply. |
Subscribe now!Stay in the loop - blog posts, latest news and educational opportunities! Thank you!You have successfully joined our subscriber list. AuthorBex Tasker Archives
March 2020
Categories
All
|