The term ‘poisoned cue’ is used much more frequently these days. Whilst that is in itself fabulous, unfortunately it’s true meaning seems to have become a little misunderstood along the way. Many individuals are using the term incorrectly to describe any anxiety related behaviour the horse displays irrespective of how it was trained.
Technically speaking, a poisoned cue is actually what happens when a behaviour has been initially trained using purely positive reinforcement. It is then put ‘on cue’ and at some point down the line the behaviour is punished. This leads to a previously positively trained ‘cue’, which has always resulted in reinforcement, becoming ambigious in its outcome for the horse.
For example, the trainer cues the behaviour, the horse doesn’t respond to the cue for some reason and so the trainer then then applies an aversive as a result of the lack of response. The aversive experience ultimately results in anxiety related behaviour which becomes part of the horse’s response to the cue. So we start to see the horse offering avoidance behaviours on presentation of that cue, broken behaviour chains etc.
The poisioned cue therefore is not what we are seeing if the behaviour has been trained from the outset using aversives (negative reinforcement) such as escalating leg pressure to teach moving forwards. This is a command.
If however, we had free shaped the behaviour of moving forwards using only positive reinforcement, added the cue of a light leg touch to that behaviour, and then someone else rides the horse who escalates that leg aid to a kick in order to get canter, it is highly likely that the light leg touch cue would become poisoned as a result.
Karen Pryor explains this in more depth in her article. To read more, click on the link below:
This entry was posted in Clicker Training, Q & A Section, Stories & Thoughts and tagged aversive experience, Clicker Training, Clicker Training Horses, Karen Pyor, Poisoned Cues, positive reinforcement, punishment, what is a poisoned cue.
Bookmark the permalink.