Thoughts From a Life Time of Animal Training
The recent Zoospenseful blog by Steve Martin on the top ten qualities of an expert trainer is I think a wonderful stimulating question (I love top ten lists- haha). As part of ATR’s commitment to Transformational Learning Theory (learning through open discussion) perhaps we could make this into an interesting discussion piece? We should all probably ask ourselves this question- what is an “expert”? I agree with many thoughts on Steve’s list but here are some important additions from my view:
A demonstrated ability to reliably produce quality behavior in difficult situations would be the most objective and quantifiable measure and ranks highly for me, since many people can talk a good talk but actually putting principle into practice is direct evidence. Ask what has been trained and done by the so-called expert and how advanced and challenging is this behavior?
Further, I feel that the animals “vote” on the subject must be also ranked highly (which I presume is at the heart of many of Steve’s excellent points). How are we to measure the animals’ view or vote on the training system and its practices? For me the willing and eager participation in free space is a good indicator (ie: the extent to which they do not have to be “caught” or controlled). This operational measure allows for the respect of different training styles and choices but with the allegiances to the relationship and welfare issues. Putting the results and the animals’ behavioral indicators together, allow us to measure or operationalize the principles rather than merely citing the dogma or doctrine. This allows and respects the art of training, which is fundamentally individual based (as Susan points out).
Which brings me to my third point, I feel an expert trainer will advocate for and train for the well-being and necessary care of the animal before (or at least in parallel with) any human objectives, not merely in rhetoric (words) but in action and resources dedicated. That is to say that we train towards the obvious necessary care such as basic comfort and desensitization, medical procedures, crating and transport and shifting before we train for shows, riding, interactions or other financial and human objectives. This can be very difficult in many situations to convince management but is the modern and best way to have the best outcomes for both human and animals, and it demonstrates our sacred responsibility to those we care for.
Another issue for me is that the trainer treats humans the way they claim to treat animals. Positive reinforcement trainers should be careful to avoid shaming, coercing or otherwise punishing other trainers into submission and agreement. The community should encourage and support while discussing and debating issues with respect and understanding. Walk the walk. Here I think of the relationship between Desmond Tutu and the Dali Lama- different doctrines still permits mutual respect.
Finally, while I understand Steve’s reasoning and valid concerns regarding backing up each and every bridge stimulus with reinforcement, I should let it be known that my lab has just completed the first piece of a several month long experiment to quantify some of this issue. We will offer a paper on this hopefully by the end of the year. Preview: we have been able to demonstrate perfect bridge maintenance at levels of as low as 75% (with a stimulus that was initially paired at 100%). Again our measure is the animal’s response to the bridge stimulus, SDs and participation in training session was maintained at above 98% for a collection of 5 pinnipeds and two birds over thousands of trials.
We now offer this question to our community: What do you think makes an expert trainer?
There is no one right training system or simplified cookbook of techniques that will provide the best possible outcome given all the different factors of each species and the human objectives involved. Instead, knowing which techniques to use and when to use them is a matter of degrees and variables and greatly depends on the context of the situation. Training and teaching is always individual: a study of one (as Dr. Susan Friedman puts it). Ultimately, it is the goal of any managed care (captive) environment to provide optimal welfare and happiness for the animals while living within the constraints of the situation. This will mean choosing a holistic approach to care that balances the needs of the animal and the trainer. Technique choice, handling methods, environment, medical care and training objectives are all examples of choices that must be taken into account when determining the overall well-being of the animal. So how do we know if we are doing a good job? How do we measure if the balance has been well struck?
At ATR Int’l, we judge our techniques and choices on the measurable and definable net effect on our animals’ relationships with the trainers and the system. There are several behavioral indicators we look for in our animals that let us know our relationship bank account is high and the animals are happy and in good welfare.
- Is the animal engaged and interested in people and interactions?
- Does the animal’s behavior indicate that it wants to participate by approaching eagerly without any coercion?
- Does the animal choose to cooperate in free space with no control (such as leashes, halters, jesses, etc.)?
- Does the animal’s behavior and body language indicate that it is calm, comfortable, and relaxed during training and outside of a session?
So now a question for discussion: what specific characteristics or specific behaviors do YOU look for in your animals (please state the species) that you use that indicate your choices are/have been effective in managing this best outcomes balance?
This controversial conditioned stimulus tells the animal that the training session is over and that their attention is released. Effectively, this is the training equivalent of “goodbye,” and in fact I recommend that exact word as an EOSS (it puts the trainer in a friendlier frame of mind than the more negative connotations of other options such as “no more” or “all gone”).
However, since many people include all or nearly all the reinforcement within the training session, this signal can come to mean to the animal that the opportunity for reinforcement is gone (or a stimulus delta in the psychological sense). In that case, it is a type of secondary negative punisher and may therefore lead to frustration, anxiety, and in the worst case scenario, aggression. For this reason, this stimulus is hotly debated. Whether or not the stimulus comes to represent a negative punisher depends on whether or not it does in fact predict no reinforcement. It is possible to vary the session times randomly and provide post-session reinforcement delivery that will contradict the conditioning of this stimulus as a stimulus delta.
One solution is to routinely provide post-session reinforcement to the animal after the trainer’s departure and the EOSS. Leaving food in the enclosure would be an example. Another solution is to also frequently return repeatedly after varying lengths of time (a technique I call ‘multiple sessions’) to disassociate the EOSS with any predictability of a lengthy absence or loss of reinforcement opportunity. It should be pointed out that without a specifically conditioned stimulus to say such a thing, the animal will naturally associate the various activities that the trainer does in preparation and associated with leaving, as an EOSS that they come to recognize from an associative learning process. So, in some sense, whether or not you deliberately condition this stimulus, the observant animal will generally learn one or several activities of the trainer that are effectively a type of EOSS. This truth makes a heated debate over the evils of the EOSS rather silly in my opinion, since in all likelihood there exists some form of EOSS in any case. Ultimately, the real evil is found in the predictable association to a period of no reinforcement that should, in either case, be mitigated by prudent counter measures like those described above or other means to lessen the negative implications of departure (assuming your animal doesn’t want to see you leave since you have a positive relationship bank).
I have been of late studying a great deal about mindfulness and meditation. And I’ve been curious about how this topic, which is increasingly valued in human medical and psychological circles, might cross over in some of the work that we do with animals? I wonder what your thoughts might be on this?
I noticed that one of the things people remark on about my animals is their composure and calmness. I see this as a kind of mindfulness and, practically speaking, I believe it comes from teaching the animals how to perform extended behavior calmly, particularly targeting or attention and orienting, helping form behavior that can be several minutes in duration and calmly exhibited. Being effective in this requires a basic behavior conditioning approach of extending each second of the behavior until a fuller duration is accomplished.
But I believe that to set the stage for this much more is needed, in particular the development of trust and understanding about their environment, their relationships, and their caregivers. From a place of no tricks, deception, and limited pressure comes the ability to feel confident and trusting. This supplies the foundation from which behavior extension and calmness are formed.
People are not the only ones who need to learn how to relax. Training animals to relax is an important part of promoting their healthy, well adjusted, and peaceful lives. Most relaxation training starts with conditioning a particular posture or behavior that represents a relaxed state in that species. This can be done using any of the six basic operant techniques but is very successful in a free shaping, scan and capture scenario. Initially the smallest unit of relaxation is conditioned and then this is incrementally extended for longer and longer periods of time.
With time, animals who originally may have resisted, will learn to appreciate relaxation for its own sake much like the process of learning to meditate for human beings.
“Train as fast as you can and as slow as necessary.” This adage is a reminder not to plateau an animal by unnecessarily repeating too many of the same approximations before the animal arrives at the goal behavior. Another way I like to think about this idea is to say that a trainer should train FAST TO THE THRESHOLD of the behavior (the edge of the confidence of the animal) and SLOW AT THE THRESHOLD. This allows training to progress quickly through the confident approximations, but suggests a trainer should take more, smaller approximations at the edge of an animal’s confidence. This finessed approach enables the animal to develop confidence in a supported way through more information, rehearsal, and opportunity to eliminate any possible fears at the point where this may be needed.
Most training and learning, indeed any acquired skill, occurs in a series of steps or approximations. This is true regardless of the technique or type of learning that is taking place. So for example, in scan and capture the trainer is looking for the smallest unit of behavior that leads from where the animal currently is to one unit closer to the behavior you are looking for. In baiting, this would be using some form of draw to lure the animal a step closer to your goal and in negative reinforcement it would be to teach the animal to yield to the smallest unit of pressure in the right direction of the goal behavior. Much depends on the sensitivity of the trainer to choose the appropriate size or unit of approximation and to decide expertly when to move forward and when to repeat and rehearse. If fear is a component of the equation, then the trainer should opt for more rehearsals at the same approximation until smooth and eager behavior is demonstrated which indicates the readiness to move on.
If you request something from an animal who fails to correctly perform the request, there is a low chance that requesting the same behavior in the same way in the same location will result in any different outcome. Whatever the antecedents were that lined up against the correct presentation are still in operation even if the consequence is not reinforcement. Immediate rehearsal history will also encourage the incorrect response. The only situation in which asking again in the same manner will result in a different response is if the animal simply did not hear and understand you, but even in that case the odds may be stacked against a correct response due to rehearsal and the antecedent conditions. Better options include: changing the environment, adding more information, changing the position of the animal and the trainer, and changing the behavior being requested to something easier and more likely. These changes can begin to set the animal on the path to reinforcement which will lead to further cooperation and the ability to revisit the problem behavior with a fresh mind. “The definition of insanity is repeating the same thing over and over while expecting a different result.”
Everyone has their ups and downs. People often forget that animals are the same way. Before beginning any training session, take time to connect and assess where the animal is at mentally before making any demands. Are they distracted or engaged? Are they moving easily or with difficulty and hesitation? Taking the time to be as mindful as possible about your partner will help you understand and anticipate their reactions as well as make good choices about what activities or requests they might be able to manage. It is true that you, as the trainer, may set forth the agenda, but the animal will always set the start point.
Excellent behavior modification involves consistency in communication but variability in motivation. Clear explanations provide confidence and a sense of control brought about from predictable understanding. In contrast, diversity in reinforcement brings renewed interest, stimulation and curiosity which inspires engagement.