A few years ago I worked with a 1400 pound Warmblood horse named Feifel. (To get a sense of proportion of how big he is, here is a video or Feifel learning targeting https://www.youtube.com/watch?v=mHY7vF2pc1s ) He had a number of problems but primarily he bucked when going from a trot to a cantor. To get started with him I associated a clicker with carrots. Then I started working to create some behaviors to develop the relationship before I addressed the primary complaint. I first started with having him turn his head to his left. Each turn received one click and one treat. Within about 15 minutes he started turning his whole body. After the second training session he was making a 360 turn to his left. The diameter of the circle was longer than his body length – a large, lazy loop. Over several more sessions something happened that is not recorded in the literature of behavior analysis or modern training. He improved the behavior to reduce the amount of effort it required to get his carrots. As he refined the behavior, the loop got smaller. When the behavior was firmly established, the cue ‘turn around’ started with his nose right in front of my face, about a foot away. As he hit the 180° mark, his tail would brush my chest. Then he would end up with his head in exactly the same place it was when he started. He had learned to “wheel” rather than take a forward turning step to start a circle. This is not an insignificant thing. Trainers love to assume things about the process that aren’t really true. Paramount at that list is a belief that the trainer is completely in control of the shaping. Nope. I provided the exact same pattern for reinforcement throughout the process and did not use “differential reinforcement” to refine his movements. He did it himself. I didn’t shape the wheel, he did.