LEARNING BEHAVIOR DURING TARGET TRAINING MARES
To download the paper (and be able to see the graphs/tables)
Click on the link to the left.
Comparing learning behavior during target training mares of different
temperaments, with rewards between cookies with release of pressure vs. releasing
pressure for ten seconds
Deanna Beals and
Rachel Barnes
Department of Animal Science, University of California,
Davis
ABSTRACT
The
Equine Industry bases its success from the efficacy of training horses. Rewarding
methods used during training must take into consideration the natural behavior
of horses in order to be effective. For this experiment, six UC Davis broodmares
were used for two different rewarding treatments during target training;
“COOKIE” using release of pressure and cookies as a reward, versus “RELEASE” using
a ten second release of pressure as a reward. The mares initially were
temperament tested, using vocalization and movement to score “cold blooded”,
“warm blooded”, or “hot blooded”, and split between treatments. For three
experiment days, each horse was led into a solid sided round pen with a ball
hanging from the wall. For five minutes, the trainer would teach the horse to
touch the ball using target training. Data was recorded for the horse’s spatial
relationship to the ball, latency to touch the ball, and number of times the
horse touched the ball either on it’s own (unassisted touch), or assisted by the
trainer (assisted touch). Data was statistically analyzed to find significant
differences on day three between the treatments for number of assisted touches,
with COOKIE having less assisted touches. Also on day three, the cool/warm
blooded Quarter Horses showed more unassisted touches than the Thoroughbred and
Arabian.
Results conclude that when
target training mares, using multiple rewarding methods increases bouts of
learning behavior, and horses of cooler temperaments show more learning
behavior.
Key words; horse, behavior, learning, training, rewards
INTRODUCTION
The horse industry is a viable
market, grossing top ten in the United States (American Horse Council, 2005).
Training is essential for the health of the horse industry, not only aiding in
providing horses with safe behavior that act reliably, but also helping keep
the competitive nature of the equine industry through winnings due a large part
to the training of the champion horse (McGreevy et al., 2009).
Instinctually, horses move into
pressure to try and find relief from the pressure. However, when we train
horses we want to teach them to move off of pressure so we can direct their
movements when riding (Waren et al., 2002). Training a horse to move off of
pressure can be achieved by rewarding the horse via a release, or relief of
pressure, and even with rewards.
The temperament of the horse must
also be taken into consideration when training horses (Waren et al., 2002). A horse’s
temperament refers to their personality, which is partly genetically determined
(Waren et al., 2002) (Grandin, 1998). Certain breeds of horses have been
selected for either very hot temperament such as fine boned Arabians, as well
as cool temperaments such as the thick boned draft breeds (Grandin, 1998).
In
general, a “hot blooded” horse gets excited very easily and is more likely to
run when frightened (Grandin, 1998). A “cool blooded” horse has a calmer
disposition and is less likely to run, and a “warm blooded” horse is even tempered
(Grandin, 1998).
When training, a hot blooded horse may be more difficult to
work with for low energy tasks such as target training because of their
excitable nature (Grandin, 1998). Equivocally, a cold blooded horse may be more
difficult to work with for high energy tasks such as racing because of their
quieter nature (Grandin, 1998).
Target training is teaching a horse
to physically touch an object that the trainer decides will be a target.
Targets include tennis balls, soccer balls, the trainer’s hand, a construction
cone, or even trailers. In this experiment we target train mares to touch a
rubber ball with their face. We expect that we will need to physically show
them how to touch the ball with their face and reward them once they do. Then
we will observe the number of times the horses touch the target on their own as
a measurement of the learning behavior for this task.
When training horses, the reward is
important in encouraging learning behavior. There are various rewarding
methods, including releasing pressure, giving a food treat, a vocal reward such
as “good boy”, and petting/itching the horse. Often times several reward
methods are used at once to further reinforce a good behavior.
Releasing
pressure is one of the most commonly used rewards for a majority of training
for riding (Waren et al., 2002). Food rewards are a popular option for target
training horses. This experiment will compare the learning behavior for target
training between a cookie food reward that is combined with a release, and a reward
that is a release of pressure for ten seconds.
We hypothesize that horses will
show more learning behaviors during training when rewarded with more than one
highly valued action such as a food reward doubled with a release of pressure.
We also hypothesize that horses of cooler temperaments will show more learning
behavior than horses of hot temperaments.
Our objective is to determine the
most effective way to train a horse to perform a task. We predict that mares
when given a food reward combined with a release during target training will show
more learning behaviors than mares rewarded with a ten second release. We also
predict that the hot blooded Thoroughbred and Arabian will show less learning
behavior than the cooler blooded Quarter Horses.
MATERIALS AND METHODS
This experiment took place in April
2012 at the UC Davis Horse Barn, in Davis California. The experiment consisted
of an initial temperament testing experiment lasting 5 minutes, followed by three
days of 5-minute experiments per mare to address our hypothesis.
Six UC Davis broodmares were used
for the experiment, consisting of four Quarter horses, one Thoroughbred (Horse
# 27), and one Arabian (Horse # 33) (Table 2). The mares were all in good body
condition. The mares were used frequently for various classes at UC Davis where
a number of individuals would handle them, and sometimes take them into the
round pen. Previous history of the mares is otherwise unknown. The mares had
home housing consisting of 1/8 acre dry lot pens with shade structures, feed
and water troughs. The mares were housed individually, in pairs, triplets and
quadruplets, and sometimes moved between pens or put onto a ½ acre pasture. All
the horses were fed twice a day a diet of alfalfa hay. Horses received 5-10% of
their body weight in hay at approximately 7:00 AM and 6:00 PM.
The experiment site consisted of a
42’ solid sided round pen with 7’ high walls, and an entrance on the East side
(Figure 1). A Jolly Ball was tied
on the West wall, hanging 4’ high. The round pen was visually partitioned into
4 different zones by drawing a line in the dirt with a broom handle. Zones 1,
2, and 3 separated the round pen into three 14’ sections, with zone 3 on the
East side with the entrance, and zone 1 including the ball (Figure 1). Zone 0
represented the ball itself.
The trainer and data collector used
for this experiment were novel individuals to these horses. Horses were
individually removed from their home housing for the initial 5 minute long temperament
test and following experiments, using a 1” nylon flat webbing halter and 10’
nylon lead rope. The trainer was the only person to handle each horse in
transport to the experiment site, during the experiment, and back to the home
housing for the initial temperament testing and the three experiment days. The
horse was led using two hands, so that the trainer walked on the left side of
the horse with her right hand holding the lead rope 6” from the halter, and her
left hand holding the excess rope.
Before conducting the main
experiment, we first conducted a temperament testing to determine the horse’s
personality so each treatment had balanced temperaments. We tested each mare
individually, with the trainer leading her into the middle of the experimental round
pen, taking off the lead rope, and having standing passively in the middle for
five minutes. We let the horse have the unrestricted movement and behavior, and
recorded the number of vocalizations (Table 1), and instantaneously sampled
every 30 seconds if the horse was moving or standing still (Table 1). Horses
that moved more and had more vocalizations were given a “hot blooded”
temperament score, and horses that stood still more and vocalized less were
given “warm blooded” or “cold blooded” temperament score (Table 2).
Our experiment had two treatments
with different rewarding systems. Our rewarding protocol for the cookie plus
release of pressure treatment group, COOKIE, was once a horse reached a goal,
the pressure was immediately released and one cookie was fed as a reward. We
fed Manna Pro “Apple Wafer Horse Treats” from the same 25 lb bag (Model #
0093006125, purchased from Sheldon Feed in Sheldon, CA). The cookie was fed as
soon as possible after the horse reached a goal. If the cookie dropped, it was
retrieved and fed to the horse. If the horse refused the cookie, it was still
offered anyways. As soon as the horse had taken and chewed the cookie about
three or four chews, then the horse was asked to reach another goal, and the
pressure was put back on the lead rope. If the horse tried to sniff or bite at
the trainer for more cookies, the trainer would apply the pressure and ask the
horse to reach another goal.
For release of pressure treatment,
RELEASE, once a horse reached a goal the pressure was immediately released and
the trainer counted to ten seconds and did not apply any pressure on the horse
(unless it was to protect the trainer from being injured by the horse, or to
keep the horse within a safe position next to the trainer). Since handling a
horse takes two hands and a third person was not available to time ten second
intervals, the trainer counted to ten seconds using a “one-one-thousand,
two-two-thousand, three-three-thousand…” counting method.
For the training protocol, each
horse was individually led through the round pen door, turned to the left to
make a circle so they were facing the ball to the West, and began in zone 3
(Figure 1). When the starting time was called, the trainer would begin walking
towards the ball (Figure 1). The trainer would apply pressure by pulling the lead
rope in direction of the ball, only releasing the pressure and rewarding the
behavior once the horse has reached a goal of entering a lower numbered zone. The
trainer would reward with treatment method, of either release with cookies or
ten seconds of release, then continue to apply pressure until another goal is
reached. The horse was allowed as many attempts as possible within the given 5
minute experiment.
Data was collected using instantaneous
and continuous sampling methods, and analyzed using Microsoft Excel with
averages, pivot tables, and t-tests between two paired samples. For the initial
temperament testing we collected data on number of vocalizations (Table 1)
using continuous sampling and tally methods, and movement (Table 1) using a 0/1
(yes/no) instantaneous sampling every 30 seconds. For the following three days
of experiments we recorded which zone the horse is located in (Figure 1) using
instantaneous sampling every 30 seconds, latency to touch (Table 1) using
continuous sampling, how many unassisted touches (Table 1), assisted touches
(Table 1), and total touches (Table 1) using continuous sampling and a tally
method.
RESULTS
The
horses displayed a gradient of temperaments, and all horses were analyzed to
find differences between their temperaments. During temperament testing, Horse
# 14 did not make any vocalizations, whereas Horse # 33 made 26 within the
five-minute period (Table 2). Similarly, horse # 10 only spent 33% of her time moving,
whereas Horse # 33 spent 100% of her time moving (Table 2). Horses that
vocalized and moved less were scored as “cooler” horses. The “hot blooded”
score that both Horse # 27 and Horse # 33 received from our temperament test
(Table 2), support the concept that certain breeds of horses are more
genetically “hot” than other breeds (Grandin, 1998). “Hot blooded” breeds
include the Arabian (Horse # 33) and Thoroughbred
(Horse # 27) (Grandin, 1998).
There
were no statistical differences between the two treatments in latency to touch
the ball over the 3 days (Table 3).
Comparing cookies and release latency to approach showed a trend to increase
over the 3 days (Figure 2). Day 2 showed a trend for cool/warm horses to
approach the ball at a slower rate then the hot horses (Table 4).
Assisted
touches (Table 2) were counted on the second and third day of experiments.
There was a statistically significant difference found in the number of
assisted touches between the cookie and release treatments. The treatment being
given cookies as a reward needed less assisted touches than the treatment being
given ten seconds of release as a reward (Figure 4).
Unassisted
touches (Table 2) were counted on the second and third day of experiments.
There was not a statistically significant difference found in the number of
unassisted touches between the cookie and release treatments (Figure 5). There
was a statistically significant difference on day 3 between cool/warm and hot
horses for number of unassisted touches (Figure 3).
Total
touches (Table 2) were counted all three days of the experiment. There were no
differences between the two treatments in the number of total touches (Table 3).
DISCUSSION
We
found that the hotter tempered horses became distracted more easily and would
show more attempts at being disrespectful and or ignoring the trainer. These
types of actions showed a lack of focus on learning and we think this might be
why the horses received lower scores.
Temperament itself is probably the
biggest reason why the more hot-blooded horses seemed to learn slower, not
because they are stupid, but because they are more concerned with being
isolated and in danger of predation, and thus are not in a learning frame of
mind. Horses are a prey species that are gregarious in nature. They live in
herds to protect themselves from predators, using power in numbers to survive.
Because of their herding behavior, it can be more difficult to teach a horse
something when they are isolated. If a horse feels that it’s life is at threat,
they are unlikely to show learning behaviors when their mind is in “fight or
flight” mode. Supporting this theory, the cool tempered horses learned
statistically equally as fast between the two treatments, showing that there is
an element of temperament to consider when training horses.
When we began the experiment we had
not planned on counting the number of touches, since we thought the horses
would be more frightened of the ball and thus spend more time zones 2 and 3.
Because of this, for the first horse on the first day we did not count any
touches. Then we decided to tally any type of touching so the rest of the
horses on day 1 have total touches (Table 2) tallied. We decided then that it would
be best to be able to see which touches the horses did on their own (signifying
that they are learning to touch the ball) versus the number of assisted touches
(Table 2). So day 2 and 3 we also recorded unassisted verses assisted touches
(Table 2).
On the third day, the “warm” and
“hot” tempered mares spat out their cookies. We think this is because the
novelty of the food reward had been lost, and so had their interest in eating
the cookie. This is a common training problem when using a food rewards.
Another common problem with food rewards is the horse aggressively acting
towards the trainer to get a treat (Hockenhull). In our experiment, we remedied
this inappropriate begging by re-directing the horse to try and achieve the
goal of touching the ball again.
In conclusion we think that using cookies
combined with a release is more effective to target train a horse than only a
release of pressure. Furthermore, horses with cooler temperaments will learn
faster than horses with hot temperaments. However, this does not mean that hot
horses are not as smart, just that they have a stronger sense of
self-preservation to work against.
FIGURES AND TABLES
(note: if you would like to see the figures and tables, download the link from above)
Figure 1. Round Pen
Zones
This figure shows the 4 zones (0, 1, 2, and 3) associated with the 42’ solid
sided round pen used for the experiment housing. The horse is considered to be
in zone 0 when it’s face (in front of the ears) is touching the ball. Zone 1 is
the area from the ball to 14’ away when walking a straight line to the ball.
Zone 2 is from 14’ away to 28’ away, and zone 3 is from 28’ away to 42’ away.
The dashed line represents the door on the round pen that opens inward.
Figure 2. Latency to
touch the Ball
This figure shows the average time of latency to touch the
ball between the COOKIE and RELEASE treatments on each day of the experiment.
Both treatments show a trend of increasing latency to touch the ball.
Figure 3. Average
Unassisted Touches Cool/Warm and Hot
This figure shows the average unassisted touches comparing
between the cool/warm and hot temperament horses. Day 3 showed a statistically
significant difference that cool/warm horses had more unassisted touches then
the hot horses.
Figure 4. Number of
Unassisted Touches Between Treatments
This figure shows the average tallied number of unassisted
touches between treatment COOKIE and RELEASE on day 2 and 3 of the experiment.
This graph shows that over time a trend that it took less unassisted touches
for the COOKIE treatment than RELEASE.
Figure 5. Number of
Assisted Touches Between Treatments
This figure shows the average number of tallied assisted
touches between treatment COOKIE and RELEASE on day 2 and 3 of the experiment.
This graph shows that the COOKIE group had significantly more assisted touches
than RELEASE treatment.
Table 1. Term
Definitions
Vocalization - When the
horse whinnies or neighs by producing a call, usually directed towards another
conspecific. One bout of vocalization is separated from another bout by a
lack of vocalization for at least one second. Vocalization does not include a
horse blowing air through it’s nose in a sneeze.
Moving - When a
horse is in motion foreword, backward, sideways, or upwards. At least one
hoof must be in motion and includes rearing and bucking. Does not include
pawing or lying down.
Standing
Still - When a
horse is not moving. All four feet of the horse motionless, or if the horse
is rolling.
Face - The
horse’s face is designated as the area in front of the ears, including
forehead, nose, chin, and jaws.
Latency to
Touch - How long
it takes for the horse to touch the ball with it’s nose (entering zone 0) for
the first time since the call of start.
Unassisted
Touch - When the
horse touches the ball with it’s face (entering zone 0) without any type of
cue or pressure given from the trainer.
Assisted
Touch - When the
horse touches the ball with its face (entering zone 0) by being physically
pressured to do so by the trainer.
Total
Touch - Number of
times a horse touches the ball (entering zone 0) either by assisted or
unassisted touches.
Table 2. Mares
Temperament Scores
This table shows the results of the temperament testing
experiment, showing a total number of vocalization and the percentage of time
of the 9 different 30 second intervals that the horse was moving. Temperament
score was based heavily off of the number of vocalizations. Breed was not
considered when giving a temperament score, however the Thoroughbred and Arabian
are considered hot blooded horses compared to quarter horses.
Table 3. Average ±
standard deviation of the two treatments (cookies and release) and p-values for
comparison between cookie and release by day.
This table shows that average and standard deviations of
each variable, separated by day. Latency to approach the ball, average total
number of touches, total unassisted touches, and total unassisted touches
variables were measured.
Table 4. Average ±
standard deviation of cool/warm and hot treatments and p-values for comparison
between cool/warm and hot horses.
The latency to approach the ball in seconds was recorded over three
days.
REFERENCES
American Horse
Council. 2005 “National Economic Impact of the U.S. Horse Industry”.
Grandin, Temple,
and Mark J. Deesing. “Genetics and Behavior during Handling, Restraint, and
Herding”.
Grandin, Temple.
1998. “The Way I see it, Dangers of Trait Over-Selection”. Western Horseman,
August, pp. 120-124.
Hockenhull, J.,
& Crighton E. “Survery of hand feeding cookies and unwanted oral investigative
behavior in horses”
McGreevy P.D., Oddie
C., Burton F.L., & McLean A.N. 2009. “The horse-human dyad: Can we align
horse training and handling activities with the equid social ethogram?”. The
Veterinary Journal. 181:12-18
Sankey C.,
Richard-Yris M., Leroy H., Henry S., & Hausberger M. 2010. “Positive
interactions lead to lasting positive memories in horses, Equus caballus. Animal Behaviour. 79:869-875
Waran N.,
McGreevy P., & Casey R. A. 2002. “Training Methods and Horse welfare”.
The
Welfare of Horses. Chapter 7:151-180
Warren N. P., McGregory,
R. & Casey. 2007. “The Welfare of Horses”. Journal of Animal Welfare.
3:151-180
