28/04/2020
Fantastic piece from Hear Your Horse Whisper
"Studies into animal behavior have not always had a positive influence on animal training. Early studies into animal behavior tended to concentrate on how animals organize their social systems and many early papers reported that species lived in hierarchies that were maintained by aggression (Allee, 1938). These studies showed that the dominant individual gained control of the others in the group through the emission of aggressive behavior toward subordinates (i.e., threats and attacks). Similarly, some studies by anthropologists and ethologists seemed to indicate that some pastoral peoples in Africa controlled their cattle by assuming the role of the alpha individual (Lott & Hart 1977,1979). Thus, the scene was set for animal trainers to apply such techniques with their animals (e.g., Grandin, 1981, with farm animals); this type of training, although it was principally punishment, was justified on the basis that it mimicked natural processes and therefore was not considered to cause animal welfare problems. Already, by this point in time, research into animal punishment had shown it to be largely ineffective as a method of training animals, as the undesired behaviors tended to reappear (see Chance, 1998). Perhaps the most well known example of using punishment during animal training was with pet dogs: Owners were recommended to dominate the dog as an alpha individual would do in the wild. Often dog trainers instructed owners in the use of the alpha roll; this was the turning of the dog onto its back and pinning it down by the throat (see, for example Monks of New Skete Staff, 2002). These domination techniques sometimes produced short-term improvements in training but most commonly resulted in an extreme aggressive response from the dog towards the owner. It was not just dogs that suffered. This line of aggressive animal training was also used with elephants (see Stevenson, 2002, for a review of literature) and has been suggested as a method of gaining cooperation from farm housed pigs (Grandin, 1981).
These methods were unsuccessful because they were based on punishment, and because many species of animals do not always respond to humans as if they were conspecifics (see Hemsworth & Barnett, 1987; Hemsworth & Gonyou, 1997). Thus, the reaction of the animal to human training was one of interspecies interaction and not intraspecies; this result would not have been a surprise to comparative psychologists who knew that animals could classify objects.
(...) An animal living in an environment where access to resources are controlled is living in an environment full of opportunities to receive positive reinforcement (i.e., access to resources). Evolutionarily, it appears easier for punishment to evolve as a method for dominant animals to control subordinates. However, theoretical and experimental studies into animal behavior all demonstrate that animals try to avoid aggressive encounters due to their high costs in terms of injuries and energy expenditure (Krebs & Davis, 1987). It is difficult to imagine how a system of intraspecific interactions could have evolved using positive or negative reinforcement because animals do not carry reinforcers with them. Or do they? Many social species of birds and mammals undertake social grooming whose function is not only to clean skin and feathers but also to increase the strength of social relationships (Dunbar, 1991). Grooming sessions are known to be pleasurable to animals, for example, horses and primates groom each other bodies in places where neuronal responses produce positive sensations (Feh & Mazieres, 1993;
Keverne et al., 1989). Thus, the use of intraspecies positive reinforcement in the
animal kingdom might be more widely spread than we have appreciated. Perhaps herein lies the reason why positive reinforcement is the most efficient way to train animals: it is the most natural method.
(...) Punishment, therefore, is not recommended in animal training programs for four reasons: (1) its effects are short-lived; (2) animals can respond very aggressively to punishment; (3) animal welfare concerns (the application of punishment would be illegal in some countries); and (4) animals learn more effectively from positive reinforcement, therefore the use of punishment is a false economy.
(...) Factors influencing animal learning are varied but one factor shown to disrupt learning and memory is stress (Sapolsky, 1996). Studies on laboratory animals exposed to high levels of stress show that their performance in tests of learning and memory decrease dramatically. For example, domestic pigs exposed to stress find it difficult to relocate previously visited food sources (Mendl et al., 1997).
(...) The implication for animal trainers is that animals should only be trained under nonstressful conditions, if a high rate of success is to be achieved. Intuitively, most good animal trainers know this and ensure that the training environment is as stress free as possible (...)"
Robert J. Young (2002). Uncloaking the Magician: Contributions of Comparative Psychology to Understanding Animal Training.International Journal of Comparative Psychology,15, 174-185.
