Written By Susan G. Friedman, PH.D.
Susan G. Friedman, Ph.D. is a professor emeritus in the Department of Psychology at Utah State University. Susan has co-authored chapters on behavior change in five veterinary texts, and her popular articles have been translated into 14 languages. She teaches a well-attended course on animal learning online (Living & Learning With Animals) and seminars and consultations with zoos and animal organizations around the world. Susan was appointed to the F&WS Condor Recovery Team from 2002 – 2010, after which time the team was retired due to the success of the birds in the wild. She is the Chairperson of the Scientific Advisory Committee of American Humane Association (AHA) Film and TV Unit, and a member in good standing of ABAI, ABMA, IAATE and IAABC. See behaviorworks.org and facebook.com/behaviorworks
When Peter asked me to submit a blog for Zoospensefull, I decided to write about the benefits to our profession of adopting the least intrusive principle. Four thousand three hundred ninety-two words and 23 citations later, we agreed to a blog about the blog that had grown into a full-blown article, which can be found here.
A typical debate about negative reinforcement is something like a train wreck – you don’t want to stare but you can’t look away. Some trainers argue negative reinforcement has a place in animal training; others say it does not. It’s a good time to check our understanding of why animals need trainers who adhere to the least intrusive principle. This principle inspired development of procedural hierarchies in our field (see Friedman, 2008) and others (see Byskov, 2019 for an intervention ladder for infectious disease control). This provides a framework for discussing negative reinforcement in the larger, applied context.
The Merriam-Webster dictionary states, “ethics tends to suggest aspects of universal fairness and the question of whether or not an action is responsible.” Ethical considerations are about value judgments; science is not. Einstein (1941) said, “For science can only ascertain what is, but not what should be, and outside of its domain value judgments of all kinds remain necessary.” We have a full toolbox of effective training procedures, but from an ethical perspective, effectiveness is not enough.
Least Intrusive Principle
Every profession has ethical standards followed by members in good standing and many of them (e.g., medicine, special education, mental health, applied behavior analysis, bioethics and law) include the least intrusive principle (aka least invasive principle, least restrictive alternative, least intrusive intervention/mandate/doctrine/means). Across the board, intrusiveness refers to the degree of counter-control, choice and consent – a concept as relevant to the welfare of non- human animals as to people. In fact, if we were to change the name of the least-to-most intrusive hierarchy to the most-to-least control hierarchy, it would be the same concept. It is a pressing issue when the least intrusive principle is not applied with strength and consistency in any of these professions. That said, this problem won’t be solved by having no guidance at all. Professional standards are generally protective of both the provider and the receiver of professional expertise. We know what a lack of professional accountability looks like and it is horrific.
I first advocated for the least intrusive principle regarding the selection of training procedures in the early 2000s. I wrote the article, “What’s Wrong with This Picture: Effectiveness is Not Enough” in 2008, with the proposed hierarchy graphic (see Figure 1 below for a recent version). Readers are encouraged to read this article for behavioral examples at each level on the hierarchy at my website www.behaviorworks.org (direct link to article https://tinyurl.com/y7fst3vf).
As explained in the article, the hierarchy was an expanded conceptualization of a four-category hierarchy of behavior-reduction procedures first described by Alberto and Troutman (2006). The authors also set forth these important implementation details:
- If a less intrusive procedure will accomplish the desired behavior change, it is neither necessary nor ethical to use a more intrusive procedure.
- If the choice is between a less intrusive but ineffective procedure and a more aversive but effective procedure, then the effective procedure should be selected.
- Before a more intrusive procedure is employed, data should be collected to substantiate the ineffectiveness of the less intrusive procedure.
Thus, the least intrusive principle doesn’t preclude the use of negative reinforcement (or punishment for that matter) per se; rather, it limits the use of negative reinforcement when it isn’t necessary. Data collection doesn’t need to be exhaustive or burdensome. Just a few baseline repetitions, or a retrospective report from a knowledgeable observer, may be all that is needed to justify moving along the intrusiveness continuum.
Control and Animal Welfare
It is a small step from understanding that behavior is an evolved tool to achieve functional outcomes to realizing that control over outcomes matters in the lives of all animals. There is no dearth of scientific evidence across several different levels of analysis (e.g., behavioral, neuro, ethological) revealing the direct correlation that exists between control and animal welfare (for more examples, see Friedman, 2005; and, Leotti, Iyengar, and Ochsner, 2010). It’s sensible to add control to the list of so-called primary reinforcers. We are born to control outcomes – indeed, that is the adaptive function of behavior; we need to control outcomes to survive and thrive (for the effects of lack of control, i.e., learned helplessness, see for example, Overmier & Seligman, 1967).
The commitment to using the least intrusive training procedures to provide animals more control has also led to some of our most innovative training advances to date, for example, replacing the traditional command-style monologue training with dialogue between trainer and learner by teaching “ready” signals (see Bertilsson and Johnson Vegh, 2020).
The Hierarchy and Trainers’ Skills
Aversive procedures are, for many people, the default solution to behavior challenges. It will always be easier (requiring little or no skill) and sometimes faster to spray a troop of monkeys with a hose to shift them from one enclosure to another; to push an owl onto the glove with pressure on its abdomen; and to keep a dog in the heel position with a choke collar. Alternatively, it takes skill – the result of knowledge and experience – to use positive reinforcement to shape new behaviors. To adhere to the least intrusive principle, trainers must have expertise identifying an unwell animal, and using antecedent arrangement and positive reinforcement, prompting and fading, shaping new behaviors, strengthening appropriate alternative behaviors, and functional assessment. This is where contemporary trainers excel; this is the deep expertise offered by our profession. We are expert at changing behavior by changing conditions with a minimum of force and coercion.
When the Function Is Escape
What about solving behavior problems? In applied behavior analysis, the first step is assessing why the behavior occurs by asking, WTF? (What’s the function?) What does the animal get, or get away from, by behaving this way in these conditions? There is inherent value in solving a behavior problem by providing the same functional outcomes produced by the problem behavior with a more acceptable alternative behavior and teaching new skills. Freedom is about big skill repertoires (WISABA, 2016).
For example, a parrot who lunges to remove the caregiver’s hand can be taught that the hand will be removed contingent (dependent) on the parrot leaning back instead (an acceptable alternative behavior to escape the hand). At the same time, or soon after, a positive reinforcement shaping program can be implemented to teach enthusiastic stepping onto the hand (building new skills).
A different approach would be to use a stimulus-stimulus procedure (S-S) in which the food immediately follows the presentation of the hand independent of a behavioral criterion (non- contingently). However, food is an arbitrary consequence in this case because the bird is lunging to remove the hand, not to get food. As a result, food consequences can result in weaker or slower progress. As negative reinforcement is more closely aligned with the escape function of the original problem behavior, it may, in the case of the lunging parrot, be the least intrusive, effective intervention.
We should also consider variations of a negative reinforcement procedure for escape-maintained behavior that sets the occasion for animals to escape to something (appetitive) rather than only away from something (aversive). Ramirez (2017) tested this idea when he and his team taught a beluga whale that it could say no to a cue by touching a nearby target for same fish instead.
And yet, it bears repeating that even in the case where the functional reinforcer for the problem behavior is escape, data should be collected to substantiate the need to move along the hierarchy to negative reinforcement. The question to consider is, “Is it necessary?”
When Positive Reinforcement Is Coercive
There are approaches to positive reinforcement that would not fit the spirit of the least intrusive principle and can produce problematic results. One case is the use of a severe motivating operation to increase the value of the reinforcers. Depriving animals of food or social contact with conspecifics and people with whom they have relationships are examples of unnecessary and unacceptable strategies, all of which we know are in use by some “trainers.”
Another consideration is the superimposition of consequences. For example, consider the case where a trainer jerks the dog’s leash to punish running out of the heel position, and then immediately follows the leash pop with food, intending to reinforce coming back into position. This stimulus-stimulus pairing of the leash pop and food may result in the dog working for leash pops, as they predict food (see for example, the seminal work of Holz & Azrin, 1961, where pigeons learn to work for shock).
The superimposition of consequences works in reverse as well. For example, feeding an animal before or while an injection is administered may result in it refusing that food. See Sdao (2016) for a more in-depth discussion of these consequence traps in her article, “What Not to Pair: The Consequence of Mixing Consequences.” Gaining a better understanding of these intricacies helps us make better training decisions and meet our goal of using least intrusive, effective procedures.
Diving Into the Deep End of the Discussion: Acknowledging Dissent
In the 20 years or so that I have been disseminating the least intrusive principle to the animal training community, I have heard dissent – sometimes absurd, but most often thought provoking. So, I share those opinions and my responses here as well.
- “The hierarchy is not supported by science.” Incorrect, see citations above (for more examples, see Friedman, 2005). In addition, as noted, ethics is not the domain of science and so this principle is one case in which science and ethics converge on the same conclusion.
- “Given my experience training this species, I know where to start.” This is really profiling what an animal will do in the future based on what other animals have done in the past. The risk of the self-fulfilling prophecy is too great (for a discussion of how others’ expectations of an individual affect that individual’s performance, see Rosenthal, 1998). We know well that behavior is always a study of one. The detrimental fallout of aversive procedures is too costly to rely on fortune-telling (see Sidman, 1989).
- “Common applications of negative reinforcement are so aversive (e.g., shock collars and ear pinches) that the procedure should be on a level of its own, closer to positive punishment.” Indeed, the continuum of aversiveness is long, ranging from weak to strong. One argument for leaving the hierarchy as it stands is that before a highly aversive event functions as a negative reinforcer, it likely punishes the preceding behavior; positive punishment is generally the most intrusive training procedure.
Conclusion: Asking the Right Questions
There is abundant evidence that animals are learners who share with humans the fundamental principles of behavior change; they are neither commodities nor stimulus-response reflex machines. This understanding has blown the lid off the relevance of training to animal welfare. Still, there is so much more to discover about the procedures we use to teach our learners. As I wrote in the original 2008 article, the commitment to using the least intrusive, effective intervention encourages us to think before we act, so that we make choices about the means by which we accomplish our behavior goals and not just the outcomes. When more intrusive procedures are not necessary, we should not use them, regardless of rationales based on personal recipes or canned protocols.
Ultimately, applying the least intrusive principle to animal training requires a good understanding of how the fundamental principles of learning work (the world as it is) and how we choose to use them (the world as we want it to be); that is, both science and ethics. A profession-wide standard based on the least intrusive principle moves us closer to that goal.
Alberto, P. A. & Troutman, A. C. (2006). Applied behavior analysis for teachers (7th ed.). Englewood Cliffs, NJ: Merrill.
Bertilsson E. & Johnson Vegh, E. (2020). Animals in control – Start button case studies. https://video.clickertraining.com/programs/animals-in-control-start-button-case-studies.
Byskov, M.F. (2019, January 18). Qualitative and quantitative interpretations of the least restrictive means. Bioethics, 33(4), 511-521. https://doi.org/10.1111/bioe.12548
Einstein, A. (1941). Science, philosophy and religion, A symposium, The Conference on Science, Philosophy and Religion in Their Relation to the Democratic Way of Life, Inc., New York
Friedman, S.G., (2005). He said, she said, science says. Good Bird Magazine, 1(1), 10-14. Friedman, S.G., (2008). What’s wrong with this picture? Effectiveness is not enough. Good Bird
Holz, W. & Azrin. N. (1961). Discriminative properties of punishment. Journal of Experimental
Analysis of Behavior. July; 4(3): 225–232.
Lalli, J. & Casey, (1996). Treatment of multiply controlled problem behavior. Journal of Applied
Behavior Analysis of Behavior, 29:3, 391-395.
Merriam-Webster (n.d.). “Code of ethics.” Merriam-Webster.com Dictionary, https://www.merriam-webster.com/dictionary/ethics
Morlino, R., Mauzy, C., Salim, J., Turpin, J., Charlton, E., Zabala, K., Zawoyski, A., & Ringdahl, J. (2020, May). The use of combined contingencies in the treatment of targeted and nontargeted problem behavior. Poster presented at Association of Behavior Analysis Conference.
Overmier, J. B. & Seligman, M. E. P. (1967). Effects of inescapable shock upon subsequent escape and avoidance responding. Journal of Comparative and Physiological Psychology, 63, 28-33.
Ramirez, K. (2017). “Dr. No: How Teaching an animal to Say ‘No’ Can Be the Right Prescription,” [Video file.] https://video.clickertraining.com/programs/dr-no-how- teaching-an-animal-to-say-no-can-be-the-right-prescription
Rosenthal, R. (1998). Covert communication in classrooms, clinics, and courtrooms. Eye on Psi Chi, 3:18-22. http://www.psichi.org/pubs/articles/article_121.asp
Sdao, K. (Sdao, K. (2016). What not to pair: The consequence of mixing consequences. www.kathysdao.com/articles/what-not-to-pair/
Sidman, M. (1989). Coercion and its fallout. Boston, MA: Authors Cooperative.