Could the widely held assumptions about the advantages of clicker training be wrong?
For years I've been trying to convince my clicker training colleagues that a) there is no scientific evidence supporting the assumption that dogs learn faster with a clicker than with verbal markers or food alone; and that b) the use of the clicker needlessly complicates the training process for clients who are already frustrated with their dogs' behaviors, so adding clicker-skills into our training protocols does more harm than good.
Their counter to argument "a" is that scientific studies have shown that there is a highly efficacious relationship between the clicking sound and the amygdala, a structure in the limbic system of the brain. "Science has proved that dogs trained with clickers learn faster" is the common refrain. Yet when I ask where I can find these studies, I am presented with studies about dogs in Skinner boxes (observation chambers where the handlers are not visible to the animals), or horses being led into a trailer, or dolphins being trained in captivity.
Their counter to argument "b" is often condescending and dismissive, e.g. "Find clients who really care about their dogs and they will be willing to learn how to use clickers." That response misses the point. If the clicker is not any better than using a simple verbal marker, why use it? In my view a verbal marker is far superior, because it encourages the dog to watch and listen to ME, instead of to a neutral sound. After all, isn't the whole point of dog training to enhance communication between the dog and the owner?
The debate usually stalls at this point. I'm not a scientist, I'm just a guy who trains dogs. Perhaps they regard me as a hopeless relic from the past, unwilling or unable to modernize my approach.
But wait - it seems that there are recent scientific studies which may support my position:
The following is reprinted with permission from the SPRING 2020 issue of the Chronicle of the Dog, the official publication for the Association
of Professional Dog Trainers:
Some of you may have been at the 2019 APDT conference and attended
a talk by Dr. Clive Wynne discussing recent research on
dogs. One of the studies he mentioned suggested that training
using a clicker showed no benefit over training without a clicker..
Many trainers in the
dog training community have strong feelings about clickers, so a study that
cast doubt on the clicker's
effectiveness generated intense reactions. For many, the idea that clickers
might not be as useful as previously thought was very
difficult to swallow. It may be tempting to simply discount the study since it
doesn't fit with what we "already know." But this was
research done under Dr. Wynne's guidance, and Dr. Wynne is a fine scientist who
knows how to design an experiment, so ignoring his
research is not a practical choice.
The results of the clicker
study from Dr. Wynne's lab were surprising to both of us, and we wound up
discussing Dr. Wynne's talk-
and this research, of course-quite a bit while still at the conference. We came
away from the conference committed to finding out
more. Unfortunately, Dr. Wynne's study is not yet published, but fortunately,
Dr. Spaulding is great at hunting up research on specific
topics, so she was able to come up with several other research papers that
described similar results. We've also learned of at least one
more study addressing the same topic that is pending publication and also shows
the same type of results.
Why ask
"obvious" research questions?
Recent studies
on clicker training are an example of scientists
in labs studying things we already "know" to be true based
on observation. For those of us working with dogs on a daily
basis, it may sometimes seem silly that researchers devote time
and resources to such obvious issues. And indeed, much of the
research that comes out supports what we are already doing.
So, why bother asking research questions about well-established
practices and beliefs?
There
are a few reasons. First, it's beneficial to have solid empirical
data behind the decisions and recommendations we are making.
This allows us to be more confident in our choices, as well as to get
buy in from clients and potential referral sources. And that, in turn,
allows us to be more successful at our jobs and help more people
and
dogs. Second, what we think we know is not always correct.
Given all of the above, we like to ask a few questions:
1. What
do recent studies in dogs tell us about the use of the
clicker and other markers?
2. What
about research in other species?
3. What
do these results mean to us as dog trainers?
Whatever you feel about the
results, these recent studies on the
use of what we will call "deliberate markers" are a great example
of how important it can be to ask "obvious" research questions.
We humans are very good at "knowing" things without having
any evidence to support our assumptions. One of the purposes
of science is to keep us honest, as it were-to help us make sure
we are making choices based on actual data, not gut feelings.
Sometimes it is hard to accept scientific results that don't mesh
with our belief system, but in the end, the data are what we really
need to rely on. So please don't just turn the page and move on
to the next article in this publication if you are uncomfortable
with this topic! Take the time to explore what's going on before
making any judgments.
The willingness to change our opinion is exactly what makes
science so powerful. Author Marilynne Robinson put it very
eloquently in a recent interview on the podcast On Being:
"[Science 's] genius is self-criticism. When found that the
universe
is accelerating and accelerating in its rate of acceleration. . .
this is
not supposed to be true and the moment [scientists]}ind out that
all
major assumptions have been overthrown, there's rejoicing in the
scientific community. . .And that is the authority of science for
me. "
What do the studies on
the use of the clicker actually say? Today,
we will take a look at two studies on clicker training that came
out in the last few years. They both looked directly at the impacts
of using a clicker compared to using no marker (food only), and
in one of the two studies they also looked at the use of a verbal
marker. Neither study found any benefit to using a clicker.
The more recent study was
conducted by Feng and colleagues
(2018)' They had a sample size of 45 dogs, which is a good-sized
lab study when it comes to dogs. Dogs were recruited from the
local area and included various breeds and a relatively even
balance of males and females. None of the dogs had experience
with clicker training or extensive training of any kind. The
beauty of this study is they recruited people interested in a free
introductory dog training course. Although this still creates
some bias (compared to the general dog-owning population), the
sample is likely much less biased than in studies that recruit
dogs
from an already registered list of study volunteers.
The more recent study was conducted by Feng and colleagues
(2018)' They had a sample size of 45 dogs, which is a good-sized
lab study when it comes to dogs. Dogs were recruited from the
local area and included various breeds and a relatively even
balance of males and females. None of the dogs had experience
with clicker training or extensive training of any kind. The
beauty of this study is they recruited people interested in a free
introductory dog training course. Although this still creates
some bias (compared to the general dog-owning population), the
sample is likely much less biased than in studies that recruit dogs
from an already registered list of study volunteers.
",..there is no
empirical evidence to
date that clicker training facilitates
learning in dogs, So does this mean
we should all throw out our clickers?
Probably not. All the research really
tells us is that training with food
alone and training with a clicker both
work about equally well. Neither
technique has an advantage, based
on the research so far. It's probably
still early to make dramatic changes
in how we are already training. So
if you use a clicker, fee! free to keep
using it! If you don't use a clicker, feel
free to stick with food only!"
The dogs were divided into three groups: clicker + food, food
only, and a waiting list control. This means the control group
was told that they were on a "waiting list" and were not
officially
involved in the experiment yet. The humans went through a
reaction time test in the lab prior to beginning training. The
dogs
went through some baseline testing to help establish a starting
point for their responsiveness to commands and their impulsivity.
After that, training
sessions were conducted at each owner's
home. The training sessions were conducted by the study's first
author (Feng). The dogs were trained to perform a series of
tricks:
nose target to a hand, nose target to an object, spin in a circle,
chin rest on the ground, play dead, and station on a mat.
The study found no significant differences between the two
training groups in a variety of measures (including performance
Neither technique has an advantage, based on
the research so far.
doesn't seem likely that it would be a factor in the results as they It's probably still early to make
dramatic changes in how we are
relate to efficacy of the clicker.
already training. So if you use a clicker, feel free to keep using it!
If
you
don't use a clicker, feel free to stick with food only!
The Chiandetti et al. (2016) article was similar to the Feng study
in that it looked at the difference between groups trained with But what about
the idea that the clicker hooks directly
a clicker +
food as well as food only. However, they
also added into the
amygdala?
a third experimental group - dogs tested
with a verbal marker Some
of you may be aware of a hypothesis that proposes that the
("bravo") + food. Again, the dogs had no history of
clicker training clicker is more
effective than a verbal marker because it reaches
or extensive training with any other method. They were pet dogs the amygdala faster than many other
types of stimuli. While this
recruited specifically for participation in this study. hypothesis may
be correct, it is only an idea, and we were not
able
to locate any evidence in support of this hypothesis. Karen
The dogs learned a novel behavior (opening a bread box-type Pryor, one of the two
developers of this hypothesis, has made
apparatus) and the training was done by one of two trainers with the following statement about it:
"That is the hypothesis, based
experience in using primary and secondary reinforcers. Training on various previously unconnected
bodies of research; it is not
was done at the dogs' homes. This research also looked at the data or evidence."* Until
we have more evidence on this point,
ability of the dogs to generalize the learned behavior to a similar, we cannot be sure whether or not the
hypothesis is valid. *If
but different, behavior (opening different types of boxes). you would like to see Karen Pryor's
quote in context, please visit
https://www.clickertraining.com/node/226 .
The researchers measured how long it took the dog to produce the
first behavior, and how many attempts it took to reach the first It's also important to note that a lot
of behavior research takes
behavior. They also measured the length of time and number of place in the lab, using operant
conditioning chambers (a.k.a.,
attempts to reach criterion.* Finally, they recorded how quickly "Skinner boxes," though
that is not the preferred term). In an
the dog learned to generalize to open two other types of boxes. operant conditioning chamber, the
handler is usually not visible
Once again, this study found no difference between the groups. *
to the animal, but in the
recent research on dogs, the dogs
Note the researchers are not very clear on what "first behavior"
and could see the handler. This
raises the question of whether the
"reach criterion" mean, so we cannot explain exactly what they
had dogs being studied are picking
up on other human behaviors.
If the dogs are
picking up on something that the handlers are doing
and using that as a marker, that could make a huge difference to
the salience of the clicker or verbal marker. Putting this another
way, the research may simply indicate that the dogs are picking up
on non-deliberate/unintentional/"natural" markers, and therefore
not giving attention to the deliberate/intentional markers used in
the training (the clicker or verbal marker). Research in which the
trainer is not visible to the dog could help sort that point out. One
can also imagine research in which the two conditions-trainer
visible and trainer hidden-are tested. It's also worth noting that
in one study on horses (McCall & Burgin, 2002) the handler was
invisible to the horse, and they still found no benefit to using a
secondary reinforcer.
However, if it turns out
that clickers are more effective if the trainer
is extremely careful not to give any additional markers, we need
to ask ourselves how valuable this information is to those of us
working with pet dogs. Is it reasonable to expect the average pet
owner to avoid any unintentional markers? Is it even possible for
professionals or sports dog trainers to do this? Think of Clever
Hans, whose handler was completely unaware of the signals his
horse was noticing. It's a very interesting research question and
probably one that's worth exploring.
There is also one other, very important point. "Clicker
training"
means much more than just using a
clicker. Many people who
call themselves "clicker trainers" rarely use a clicker (though
they generally use other types of deliberate markers). What all
clicker trainers have in common, though, is a dedication to a few
basic principles. These include setting the animal up for success,
using positive reinforcement, shaping behavior using successive
approximations, splitting behavior into fine "slices" (rather than
lumping), and adjusting training based on feedback from the
animal ("the dog is always right"). All of these practices are
important; the clicker itself is just
one small part of the picture'
perhaps it is one,or several-of these practices that creates an
advantage over other forms of positive reinforcement training All
of these things are testable!
More research is needed!
As they say,
there is no need to throw out the baby with the
bathwater! More research is needed to sort out what exactly is going
on, including what specific aspects of the overall clicker training
approach may be beneficial. For now, keep an open mind and
o serve your own clients and training. Is the clicker benefiting you'
or just the philosophy that goes with
using the clicker? We don't
knw the answer to this question, but we're interested in learning
more. And in the meantime, we're keeping an eye out for more
studies on this topic! As we accumulate more data in this area, we
will have a much better indication of whether or not we need to
change our current practices.
Regardless of where the evidence takes us, it's almost certain
that we will end up in a place where we know more about how
dogs learn and which exact
practices are most beneficial when
it comes to training new behaviors. This is why research on
"known" questions is so important - behavior is almost always
more complicated than we think! By testing the methods we
are currently using in the field, we will learn more about how
dogs learn and how we can integrate different methods into our
training in a way that maximizes quality of life for both the dogs
and their people!
Kristina Spaulding has a
doctorate in
biopsychology and is a Certijied Applied
Am'mal Behaviorist. She currently
conducts
private behavior work in Upstate New York.
Dr. Span|ding is a member of the Fear Free
Speakers Bureau. She teaches on|inc
classes on
Learning Theory through the Association
of
Professional Dog Trainers and is a member of the APDT Education
Committee.
Irith Bloom, CPDT-KSA, CDBC, KPA
CTP, VSPDT, CBATI, is a member of the
faculty at Victoria Stilwell Dog Traim'ng
Academy and DogBiz's Dog Walking Academy
and speaks at conferences and seminars
regularly. She has been published both
online
and in print. She volunteers with NESR,
Annenberg PetSpace, and the
LA County 2020 HPHF Coah'tion.
Irith is the owner of Ibe Sophisticated Dog, LLC, apet traim'ng
company based in Los Angeles.
References
Feng, L.C., Hodgens, N.H., Woodhead, J.K., Howell, T.J.,
and P'C'
Benett (2018) Is clicker training (Clicker + food)
better than food-only
training for novice companion dogs and their owners? Applied Animal
Behaviour Science, 204,
81-93.
McCall, C.A. and S.E. Burgin (2002) Equine utilization of
secondary
reinforcement during response extinction and acquisition' APPlied
Animal Behaviour Science, 78(2-4), 253-562'
Smith, S.M. and ES. Davies (2008) Clicker increases resistance to
extinction but does no decrease training time of a simple operant task
in domestic dogs (Canis familiaris). Applies Animal Behaviour Science'
110(3-4), 318-329.
Tippett, K. (host).
(2019, November 21 ) On Being with Krista
Ti ett [audio podcast]. Retrieved from https://onbeing.org/Programs/
marilynne-robinson-marcelo-gleiser-the-mystery-we-are/
Williams, J.L., Friend, T.H., Nevin, C.H. and G. Archer (2004) The
efficacy of a secondary reinforcer (clicker) during acquisition and
extinction of an operant task in horses. Applied Animal Behaviour
Science, 88(3-4), 331-341.ver training without a clicker'