Clean Run Magazine - April 2007 - 37

a conditioned reinforcer’s value links directly to how close it is to reinforcement. While research shows plainly that a conditioned reinforcer need not end behavior, in the studies I reviewed, the conditioned reinforcers that occur throughout a chain of behaviors never take the same form. In other words, the animal doesn’t hear a click, respond on a schedule, hear another click, respond on a different schedule, hear another click, and so forth. Experimenters always use distinct stimuli to provide unambiguous information to the animal. So here’s what I take away from this: if you want your dog to understand clearly the meaning of the click, it is critical to be consistent. So whether the click ends the behavior or not doesn’t matter, the important thing is that it always means the same thing. If a click means “continue, you’re on the right track and you’re getting closer to reinforcement” (also known as an “intermediate bridge” or “keep going” signal) then you should use a distinctly different stimulus to signal that the behavior chain is complete and reward has been earned. In agility training, I use Yes to indicate that my dog is correct (for instance, he entered the weaves properly, he chose the cued obstacle of a discrimination, or he took the desired direction after a jump) but I don’t intend to cease the run and reward. I use a click when I plan to end a sequence and reward. Supporting Intermittent Reinforcement It’s well established that intermittent reinforcement produces more persistent responding than does continuous. While there are numerous benefits accrued from intermittent reinforcement, the most significant is that the animal learns to keep responding in the face of no reward. Once you start challenging your dog with difficult tasks, it is essential for him to understand that trying eventually pays off. But making the transition from continuous to intermittent reward can be frustrating for the dog and daunting for the trainer. When you fail to react to a correct performance, you are effectively punishing the dog. How does he know that he was correct? In operant studies, the unfortunate rat or pigeon doesn’t have anything else to do anyway so it’s likely to try the behavior every so often until it figures things out. In the real world, the dog has choices so how do you distinguish for him when he’s wrong in his choice of behavior versus when he is correct but now you’re rewarding him on an intermittent basis? Think about this scenario from the dog’s perspective. You ask him to take a jump. He can take the jump in the correct manner, or he can do any of a number of other things, including running by the jump, running under the bar, knocking the bar, or taking another jump altogether. If you are working the dog on an intermittent schedule of reward for jumping, then you would ignore the dog when he jumped correctly but wasn’t due for a reward this time. But suppose you also ignore the dog for responding incorrectly. That can be tough on the dog and on the trainer. While the dog will eventually figure it out, it’ll take longer than if you help the dog discriminate when he did the jump correctly but he’s not being rewarded this particular time from when he performed an incorrect behavior. 37 conditioned reinforcement as a “keep going” signal focuses on chain schedules of reinforcement. For instance, in the presence of a red key, the pigeon pecks after 60 seconds and the key turns to blue. In the presence of blue, the pigeon has to peck the key 50 times to turn the key to green. In the presence of green, the pigeon has to wait 2 minutes, peck once and the food hopper opens. The first light, the red, functions as a discriminative stimulus (just like a cue, such as a verbal command or an arm signal) that sets the occasion for the pigeon to begin working. The blue light has two purposes. It is a discriminative stimulus that tells the pigeon about the change in response requirement. It is also conditioned reinforcement, rewarding the pigeon for what it has already done and informing it that the food is a bit closer to happening. Likewise, the green key is both a discriminative stimulus for the next response requirement and a conditioned reinforcer for completing the previous schedule. The green light is a more powerful conditioned reinforcer than is the blue light, simply because it is closer in time to the food. In general, April 07 | Clean Run

Clean Run Magazine - April 2007

Table of Contents for the Digital Edition of Clean Run Magazine - April 2007