25 Jul Combining positive and negative reinforcement does not cause poisoned cues
There is a common misconception among animal trainers and clicker trainers in particular that combining positive and negative reinforcement is somehow unethical and will lead to a posioned cue.
This is mostly the due to the results from the paper by
Nicole A Murray on “The Effects of Combining Positive Reinforcement and Negative Reinforcement”.
This paper was supervised by Rosales-Ruiz and has been extensively used and cited as evidence that combining positive and negative reinforcement results in poisoned cues.
There are however several flaws in this paper and I would like to see this study conducted again for the following reasons:
– This was a masters dissertation not a phd and as such had some unsurprising pitfalls.
– N= 1 (Sample size of only one dog was used in the whole study) How was this paper passed with a sample size of 1?
– What was the dogs previous association with the collar and leash?
– What was the dogs previous association with the trainer?
– Had the dog been fed prior to training commencing?
– It was not so much Negative Reinforcement as Positive Punishment. It is unacceptable to use ‘leash tugs’ as the aversive if they wished to achieve a fair comparison and they wished to test Negative Reinforcement. .
– I would like to see a super light minimum pressure applied and then the instant the dog even leaned toward the trainer they click/release/treat. Then do more repetitions, then shape the behaviour gradually. The description “leash tugs” sounds more like a lot of pressure was applied. Was the pressure increased if so how quickly? What was the exact timing of the release?
– Her use of the term ‘aversive consequences’ is a misunderstanding of Negative Reinforcement. Negative Reinforcement has reinforcing properties as it is defined as strengthening a behaviour not decreasing it.
– Throughout the introduction and text, she constantly misunderstands Negative Reinforcement and refers only to the reinforcer of Positive Reinforcement but omits the reinforcer of Negative Reinforcement (removal). She even muddles up her knowledge in the introduction when she says that “little is known, experimentally, about stimuli followed by a reinforcing or an aversive event”. Here she is clearly admitting that it is punishment not Negative Reinforcement that she is testing or else she doesn’t know the difference.
– This is in no way a comprehensive study and should not have been so highly quoted by so many in the industry without actually analysing the paper.
– I would also like to mention that I actually do use capturing, luring, targeting and free shaping for many different behaviours.
– Trainers who suggest that they train only with positive reinforcement are either uneducated or they are not being realistic and honest.
Negative reinforcement can occur when simply walking into a yard with a horse and taking a step away. The slightest movement, retreat or change in a persons posture or touch could be negative reinforcement for any particular behaviour a horse was doing at the time. Any trainer who ever uses a halter and lead rope, bridle or they ride a horse will at some stage be using negative reinforcement even if they are not aware of it.
The majority of the horse training world sadly still uses pressure in extreme degrees. With hardly any understanding of negative reinforcement let alone positive reinforcement.
It is possible to train horses to respond to the lightest possible cues and we CAN make this a pleasant and rewarding experience for the horse.
Contrary to popular opinion combining negative reinforcement with positive reinforcement does NOT cause poisoned cues in fact the opposite occurs. The cue becomes a predictor to earn positive reinforcement. My horse is super keen if he feels the piaffe cue…in anticipation of food.
By teaching Rumba to do the grand prix dressage under saddle and also at liberty I hope to be able to open more peoples mind to the possibility of using clicker training.
**I wrote this post in response to the many abusive messages I have received from other clicker trainers about my training videos of my horse Rumba. I am not disputing that poisoned cues occur, I just think they occur as a result of combining positive reinforcement with punishment, not with negative reinforcement. I have enough statistics of my own to form an opinion that this particular study needs to be repeated. In the last 20 years I have taught 217 Clicker Training clinics, each with an average of 8 horses per clinic. This is a total of 1736 horses that I have had the opportunity to teach and learn from in group clinics. Then there are the thousands of private clicker training lessons I have done as well as horses I have taken for training and started under saddle. I fully support any trainer who wishes to attempt to train purely with positive reinforcement. As an industry we need to come together and support other clicker trainers. We are still the minority in the horse industry!