S-R Contiguity, or the feed-forward approach as I like to call it, is the theory that learning occurs with associations that are formed when a response is paired with a stimulus that occur close together. This theory claims that learned behaviors are evoked by the stimulus, so the stimulus (S) evokes an internal or external response (R). I'm not going to explain why this is the more correct way to explain learning when compared to operant and classical conditioning, but I am going to offer my research paper as an introduction to this feed-forward approach. By "feed-forward," I mean that a stimulus evokes a behavior. "Feed-backwards" would be the theory that a consequence "feeds backwards" to strengthen or weaken the preceding behavior, which is the way learning is currently explained in the dog training field by many trainers.
I got a 95% on this paper due to some citation errors, but my argument wasn't downgraded at all so I made the corrections and now I present to you what I hope to be a valuable introduction to this approach. It is about non-aversive vs aversive dog training methods, and I explain both from a feed-forward perspective. Enjoy!
Aversive or
Non-Aversive Methods: The feed-forward perspective
In
the dog training field, there are different methods that trainers stick to in
order to modify dog behavior, but there is an ongoing argument on which
technique is best in order to treat all behavior problems in dogs. To see which
training methods work best is impossible due to the numerous amounts of
training techniques in existence, but there are two methods that can easily be
compared and contrasted upon. The two that will be compared here will be
classified as aversive, being methods that induce pain or discomfort, or
non-aversive, or methods that do not induce pain or discomfort to the dog. It
is important to look at these two methods because different stimuli evoke different
responses in animals. From a feed-forward perspective, these two methods will
be analyzed in order to reveal why aversive methods work for some behavior
problems in dogs, and why non-aversive methods work for other behavior problems
in dogs.
Non-aversive
methods include positive reinforcement, such as giving a dog a food or toy
reward, in order to strengthen a behavior. By partaking in positive
reinforcement, the goal is to strengthen a behavior. Other forms of positive
reinforcement can include tactile stimulation that the dog finds enjoyable, such
as a side rub or ear massage. Aversive
methods are used on dogs by the means of positive punishment, such as applying
a leash correction. By partaking in positive punishment, the goal is to weaken
or eliminate a behavior. Other aversive methods include the “alpha roll,” which
is when a dog is forced onto its side or back, grabbing the dog’s scruff and
shaking them, and some even go as far as hitting or kicking the dog(Herron,
Shofer, & Reisner, 2009). Aversive methods can also be done by the means of
negative punishment, which is taking something valuable away to weaken or eliminate
a behavior.
In
2009, a study analyzing the differences between these two methods was
published, revealing that some of the ways people typically deal with undesirable
behavior in dogs evokes an aggressive response in them. People seeking help for
their dog’s unwanted behavior at the University of Pennsylvania, Philadelphia,
had to fill out a behavioral questionnaire as a part of their dog’s treatment,
which included the survey used in the study. They were asked “yes or no”
questions before and after their treatment regarding what they did when their
dog engaged in an undesirable behavior and whether or not it evoked a
“growl/bare teeth,” “snap/lunge,” or “bite” response from their dog. The most
common behavioral problems that they owners wished to solve were aggression to
familiar or unfamiliar people, separation anxiety, a more specific fear, or
other behavior problems such as aggression to cats or barking (Herron, Shofer,
& Reisner, 2009).
Their
results showed that the most popular form of positive punishment, the leash
correction, evoked the least amount of bites. The forms of positive punishment
that did evoke the most aggressive responses in the dogs included the alpha
roll, at 31%, “stare downs”, at 30%, and hitting or kicking the dog, at 43%.
The dogs that were aggressive to people were the most likely to show an
aggressive response after they were “alpha rolled” or hit or kicked. The dogs
that were aggressive to familiar people also were the dogs that were aggressive
after they would be told “no,” in a loud or stern voice (Herron, Shofer, &
Reisner, 2009).
The
forms of non-aversive methods used by the clients in the study included using
food rewards and clicker training. The instances where aggressive responses
were evoked were minimal, with 6% of the dogs engaging in an aggressive
behavior when food was being used to trade for an object that the dog had,
which was the treatment that evoked the most bites in this category. Other
forms of non-aversive methods used by the participants in this study included asking
the dog to perform a focusing behavior where the dog would be asked to look at
the owner, or asking the dog to sit in order to earn something. Everything on
the list of non-aversive treatments evoked the least amount of aggressive
responses, which suggests that aversive methods evoke the most aggressive
responses from dogs, and the non-aversive method evoke the least amount of
aggressive responses (Herron, Shofer, & Reisner, 2009).
Another
study in 2008 found correlations between training methods and behavioral
problems in the UK. People walking their dogs or visiting veterinary hospitals
throughout the UK were given questionnaires asking questions about the training
methods they used, and what behavioral problems their dog had. The results were
put under the behavior problem categories, “control problem score,” which involved
goal seeking behavior, “aggression score,” such as lunging and biting,
“fear/avoidance score,” or the “attention seeking score.” The results showed that the dogs subjected to
non-aversive methods had the lowest scores in every behavior problem category,
and the aversive methods had high scores in each of the categories. The
surprising finding was that dogs that were subjected to both positive
reinforcement and positive punishment had the biggest aggression score
although, when looking at just aversive or just non-aversive methods, aversive
methods were correlated to the most undesirable behaviors such as aggression or
engaging in fearful behaviors (Rooney, Twells, Seawright, & Casey, 2008).
Although correlation cannot always translate to causation, stimuli evoke
behaviors in animals, and the treatment that animals are subjected to are all
forms of stimuli. It was previously shown that aversive stimuli evoke
aggressive responses (Herron, Shofer, & Reisner, 2009)., so it is not a
surprise that the dogs that were subjected to aversive stimuli had a higher
aggression score.
Although
that study did not observe what behaviors were evoked by certain training
methods, a study done in 2011 found that the participants who used primarily
punishment-based methods to train more tasks tended to have dogs that did not
interact with a stranger within their home when compared to participants who used
more reward-based methods (Rooney & Cowan, 2011). A way of explaining this
is that using punishment can cause a dog to have a negative association with
people. This would be because aversive techniques evoke behaviors that are
incompatible with approaching and interacting with people, which would manifest
itself in avoidance-like behaviors or calming signals. Calming signals are
behaviors that are done by dogs to either calm themselves down, or calm
another dog down and includes behaviors such as yawning, looking away,
or sniffing the ground suddenly (Rugaas, 2005). From my experience, the best
way to identify calming signals are looking for behaviors that almost makes it
seem as if the dog is ignoring another dog or human. Ignoring a human is
incompatible with interacting with them, so a training method that evokes
calming signals that make the dog “ignore” the handler is not going to make
them focus on the task at hand. Non-aversive methods evoke behaviors that are
more compatible with interacting with people. Food causes dogs to approach the
food, and play causes dogs to engage in playful behaviors that are incompatible
with fearful behaviors. If a person uses these things with a dog, the dog is
more likely to engage in pre-feeding or playful behaviors with that person.
Belgian
military dog-handler teams were analyzed to see how training methods affected
the teams’ performance during their normal standardized evaluations. In these
evaluations, the teams would go through obedience and protection work exercises
and each team did the evaluation twice with 20 days in between each evaluation
to ensure the reliability of the observations. They looked at the dog’s
behavior by analyzing their body postures, “training-related behaviors,” and by
determining how distracted the dogs were. Training-related behaviors were
looked at during the obedience exercises, and they looked at mouth-licking, if
they had their tongue out, yawning, lifting a front paw, replacement behaviors
(such as shaking as if they had just been wet), jumping, and if the dog’s mouth
was opening and closing quickly. The experimenters considered a dog distracted
if their head or body was orientated too far away from their handler or the
direction in which they were being guided through the exercise (Haverbeke,
Laporte, Depiereux, Giffory, & Diederich, 2008). Looking at these behaviors
makes sense because several of those behaviors indicate stress in the dogs,
such as mouth-licking, yawning, lifting the front paw, displaying replacement
behaviors, and if they were opening and closing the mouth quickly (Aloff, 2010).
It
was found that handlers oftentimes punished dogs for not doing an exercise correctly
by pulling on the leash, hanging the dog by its collar, verbally scolding the
dog, or by hitting the dog. The maximum punishment the dogs received was 80% of
the time for incorrect responses for protection work, and 60% during the
obedience exercises. The handlers reward the dogs for doing exercises correctly
67% of the time for both obedience and protection work exercises. The observed
rewards given to the dogs were stroking the dog, verbally praising the dog, or
giving the dog a toy or treat. When looking at the teams’ performance, the dogs
that received the most aversive stimulus had the lowest scores, and the dogs
that received the most non-aversive stimulus had the highest scores. None of
the dogs received exclusively aversive or non-aversive treatment, but the dogs
in the high scoring groups were subjected to half of the aversive stimulus that
the low scoring groups were subjected to. Dogs that experienced more aversive
treatment also were found to high higher levels of distraction, and displayed more
of the training related behaviors that are associated with stress (Haverbeke et
al, 2008). Because these are stress signals that the dogs were engaging themselves in after
punishment occurred, and the dogs were more distracted as well, it means that punishment
evokes stress behaviors and makes the dogs more distracted.
Because
the dogs subjected to the most aversive stimuli had the worst scores, it could
mean that the exercises evoked anticipatory errors of the aversive stimulus.
The evaluations and training exercises have been practiced over the course of
the dog’s lives while being in the military (Haverbeke et al, 2008), which
means that there was time for these anticipatory behaviors to form, which would
have interfered with the target behaviors in some of the exercises. In a
feed-backwards perspective, the punished dogs should have done better because
punishment after an incorrect behavior was more frequent than the amount of
rewards given after a correct behavior, so the incorrect and punished behaviors
should have weakened or extinguished, resulting in a higher score for those
teams. Instead, the aversive stimulus evoked behaviors that were incompatible
with the obedience and protection work exercises, which resulted in the dogs
not performing the targeted exercises as well as the dogs that were subjected
to less aversive stimulus.
Aversive
stimulus appears to make dogs engage in behaviors that are incompatible with
things that people want dogs to do, such as the military dogs in Belgium, but it
can be used for protecting livestock from dogs. According to the Companion
Animal Behaviour Study Group, shock collars are appropriate for this purpose
only, and it has been determined that it is an effective way to get dogs to
leave livestock alone even if they were reintroduced to livestock a year after
being shocked for going after them (Christiansen, Bakken, & Braastad,
2001). There does not appear to be much more evidence in favor of the
usefulness of aversive methods except for during these types of purposes where
one is trying to stop a dog from chasing livestock by using a shock collar.
This would be because shocking the dogs evoked behaviors that are incompatible
with chasing, which included jumping and yelping. Certain breeds of dogs are
genetically predisposed to chasing, such as herding dogs (Sherman, Reisner,
Taliaferro, & Houpt, 1996), so these breeds may or may not respond to this
type of training, just like how some dogs will go through the shock of an
invisible fence in order to chase a rabbit out of the yard. Otherwise, aversive
stimulus evokes the most aggressive behaviors, and has even been linked to
disobedience and an increase in anxious and fearful behaviors so it is not
considered the safest or most effective training method when the target
behaviors are increased confidence and responsiveness to obedience commands (Christiansen,
Bakken, & Braastad, 2001).
Non-aversive
methods have overwhelming evidence that suggests that it evokes behaviors that
are compatible with what is typically considered “more desirable” behavior,
such as the absence of aggressive behavior. For example, it has been shown that
the more a dog was put in obedience classes the less likely they were to be
aggressive (Jagoe & Serpell, 1996). Some traditional trainers would argue
this is because the handler has asserted its dominance over the dog, but a more
valid explanation is that training a dog makes the dog engage in behaviors that
are incompatible with aggression, so the likelihood of aggression developing
decreases as the alternative behaviors increase that are incompatible with
aggression. It is a popular belief that dominance is a key motivator of
behavior in dogs, especially with aggressive dogs, but aggression has been
found to be caused by anxiety or from the dog trying to defend themselves from
something that they perceive to be as dangerous (Herron, Shofer, & Reisner,
2009). This is why, from a feed-forward perspective, non-aversive methods are
so effective in the treatment of aggressive dogs. Feeding and play, the two
most popular forms of positive reinforcement, are incompatible with fearful
behaviors, and the fearful stimulus that previously evoked aggression
eventually evokes pre-feeding or playful behaviors when the dog is subjected to
a systematic desensitization protocol. Non-aversive methods have also been
implemented in treating dogs with phobias, and was found to be 100% effective
when tested on 247 dogs (Rogerson, 1997), which makes perfect sense when seen from
a feed-forward perspective.
Dog
trainers have been arguing about the best way to modify dog behavior. Just like
how behaviorists desire a parsimonious explanation for learning, dog trainers
want to know the dog training method that is going to work for every behavior
problem. The research outlined here suggests that aversive methods do have some
practical uses that are effective, but non-aversive methods are overwhelmingly
effective in the treatment of aggression and phobias in dogs, and it
contributes to a better performance in working dogs. By comparing these two
methods, it has also been found that non-aversive methods result in the less
human-directed aggression when trying to treat these problems, making
non-aversive methods safer to use with aggressive dogs. This makes non-aversive
dog training methods the most effective and safe training method for dog
behavior modifications when done correctly.
References
Aloff, B. (2010) Canine Body Language: A Photographic Guide. Washington: Dogwise Publishing.
Christiansen, F.O., Bakken, M.,
& Braastad, B.O. (2001). Behavioural changes and aversive conditioning in
hunting dogs by the second-year confrontation with domestic sheep. Applied Animal Behaviour Science, vol 72,
131-143.
Haverbeke, A., Laporte, B.,
Depiereux, E., Giffory, J.M., Diederich, C. (2008). Training methods of
military dog handlers and their effects on the team’s performance. Applied Animal Behaviour Science, vol 113,
110-122.
Herron, M.E., Shofer, F.S, Reisner,
I.R. (2009). Survey of the use and outcome of confrontational and
non-confrontational traning methods in client-owned dogs showing undesired
behaviors. Applied Animal Behaviour
Science, vol 117, 47-54.
Jagoe, A., Serpell, J. (1996).
Owner characteristics and interactions and the prevalence of canine behavior
problems. Applied Animal Behaviour
Science, vol 47, 31-42.
Rogerson, J. (1997). Canine fears
and phobias; a regime for treatment without recourse to drugs. Applied Animal Behaviour Science, vol
52, 291-297.
Rooney, N.J, Cowan, S. (2011).
Training methods and owner-dog interactions: Links with dog behavior and
learning ability. Applied Animal
Behaviour Science, vol 132, 169-177.
Rugass, T. (2005). On Talking Terms With Dogs: Calming Signals.
Washington: Dogwise Publishing.
Sherman, C.K., Reisner, I.R.,
Taliaferro, L.A., Houpt, K.A. (1996). Characteristics, treatment, and outcome
of 99 cases of aggression between dogs. Applied
Animal Behaviour Science, vol 47, 91-108.