The performance of your horse is dependent upon both their physical capabilities and their temperament. Performance is optimised when the right match is found between the horse and rider. With 1.3 million regular riders in the U.K., understanding a horse’s temperament is becoming ever more important (British Horse Society, 2015). A staggering 66% of 3,000 non-racing horses were euthanized in France as a result of inappropriate behaviour (Warren-Smith and McGreevy, 2008). Thus, being able to test for temperament from an early stage is valuable, as it will mitigate behavioural problems that develop as a result of a poor fit between a horse’s temperament, environment, and the expectations of their owner, minimise wastage, and maximise welfare (Waran, 2007; Munsters et al., 2013; Dai et al., 2015; Pierard et al., 2017).

The horse is a prey animal and is naturally predisposed to have a flight response for survival when exposed to unexpected stimuli (Goodwin, 2010); the intensity with which the horse reacts can have detrimental effects on the safety of both the horse and rider during training and daily care. A decrease in this escape response improves manageability, consequently enabling successful training, and reducing the risk of any serious accidents and injuries (Christensen et al., 2010; Borstel et al., 2012).

In the United States there are on average 2,300 reports of horse-related injuries to humans annually, many of which occur as a result of unexpected fear reactions. Of these injured people, 48% suffered head injuries and 45% required surgical intervention (Lesimple et al., 2011; Dai et al., 2015). Despite the majority of horse owners recognising the importance of managing a horse’s fear reactions to minimize injury risk, there is currently a scarce amount of research identifying an objective method to quantify a horse’s temperament (Visser et al., 2001; Visser et al., 2002; Gorecka et al., 2007).

Temperament is defined by an individual’s behaviour characteristics and their approach to coping with continuing changes and challenges in their environment (Scolan et al., 1997; Visser et al., 2001; Visser et al., 2002). Temperament can then be further broken down into behavioural traits which are used for assessment, such as fearfulness, emotionality, sociability, anxiousness, dominance, inquisitiveness, and gregariousness. These individual differences have already been demonstrated in other species, including rodents, sheep, dogs, and cattle (Scolan et al., 1997; Wolff et al., 1997; Visser et al., 2001; Lansade et al., 2008). An individual’s temperament will be exhibited in situations when subjected to emotional stress; therefore, it is fundamental to evoke a stress response to reflect an individual’s innate emotional reactivity (Visser et al., 2002; Valenchon et al., 2013).

Past research has shown that the emotionality aspect of a horse’s temperament correlates with learning ability, thus corresponding with trainability, which can further be related to performance. Responses to temperament tests could be used to predict a horse’s future response towards a particular environment or activity, subsequently improving horse selection, breeding selection, and training (Scolan et al., 1997; Visser et al., 2003; Lloyd et al., 2007; Valenchon et al., 2013). Various authors have emphasized the need to incorporate temperament tests in horse selection and breeding programmes; however, research to validate a replicable test that is feasible in field conditions is still in progress despite public opinion showing a high demand for objective criteria to standardise temperament testing (Visser et al., 2010; Borstel et al., 2011; Gorecka-Bruzda et al., 2011; Graf et al., 2013).

Horse breeds are divided into two groups: hot- and cold-blooded. This distinction comes from their temperament and the types of tasks the breed was developed to accomplish. Hot-blooded horse breeds include Thoroughbreds and Arabs, bred for agility and speed. Cold bloods were bred for tasks like agricultural labour and include types such as cobs, shires, and draft breeds (Flanders, 2011). It is typically thought that hot bloods are more sensitive and reactive, thus making them more difficult to handle and less trainable.

The aim of my study was to identify whether equine type determines reactivity to a novel startle test and whether this reflects a precursor to certain temperament traits, which could effectively determine an appropriate lifestyle for each horse. In summary my objectives were to:

  • Evaluate whether equine type affects the heart rate (HR) and behaviour parameters in a novel startle test.
  • Determine whether repeating the novel startle test can result in habituation.
  • Identify whether age is a contributing factor to reactivity to the novel startle test.
Screen shot of experimental design

My experimental design


I used 26 clinically healthy horses from Writtle University College and Mashbury Hall Farm in Chelmsford ranging from 3 to 23 years old (Juvenile N = 9, Mature N = 17). Half the horses were hot-blooded, half were cold-blooded. I determined the groups using the prominent breed in the horse’s bloodline—a horse who is ¾ Thoroughbred, for example, was classed as a hot blood (HB). I decided to look at horses under the age of 5 separately, to eliminate the variables of age and training.

I placed a feed bowl with balancer pellets on the point of X in the testing space, an indoor ring, in order to keep the horses occupied and still. Twenty minutes before each trial, I attached an HR monitor to the horse’s girth area and took a baseline reading. The horses were then taken individually to the testing space, which was familiar to the horses, and performed the startle test: I placed each horse in front of the feed bowl as shown in the image below and released them, in order to remove the potential risks and allow a free reaction. I then proceeded to pop and hold the black umbrella directly in front of the horse from a distance of 2 meters.

The horses’ HR was continuously recorded, and their behaviour was captured on video by the second handler using an iPad. Once the HR returned to within a range of 20 bpm of the original baseline HR and the horses’ behaviour suggested that they had lost interest (or the horse touched the umbrella), the test was repeated. This was done three times for each horse.


Using a Polar Flowsync tool, I was able to determine what the HR was for each horse at the moment of startle, by chronologically correlating it with the video footage. The HR at each moment of startle, the difference between the baseline and maximum HR, and the average HR were noted. I reviewed the video and noted the behaviours associated with startling in an ethogram (see Table 1), then scored them on a Likert-type scale (Table 2).

For each of the three behaviours, I took a total from each startle for each horse to compare the severity of the different behaviours between the groups. I then took a total score from each startle from each horse to identify habituation. I also took an average behaviour score, calculated using the total of each individual behaviour from each horse divided by the number of behaviours measured. The time taken to approach the feed bucket was also scored according to minute intervals (i.e., if the horse returned within one minute they scored a one, and if they didn’t return until the fifth minute, they scored a five).

Table 1 - Ethogram of scored behaviour observed during the trial.

Table 1 – Ethogram of scored behaviour observed during the trial.

Table 2 - The scoring categories for each behaviour (NO=Novel Object, in this case an umbrella).

Table 2 – The scoring categories for each behaviour (NO=Novel Object, in this case an umbrella).


The startle test revealed a significant difference between the mean HR in the hot bloods and the cold bloods in the second startle test (P=0.03, f=0.031, df=24). There was no significant difference found between the juvenile (0 to 5 years) and the mature horses (6 to 23 years), as seen below.

Average heart rate from startles 1, 2, and 3 for both types of equine and age groups.

Average heart rate from startles 1, 2, and 3 for both types of equine and age groups.

As well, when comparing the group of horses as a whole there was a significant difference between the mean HR in startle 1 and startle 3 (P=0.02, t=2.496, df = 25), and a highly significant difference between the mean HR in startle 2 and startle 3 (P=0.008, t=2.864, df=25).

(* P<0.05 ** P<0.005) Mean HR from all horses from startle 1, 2, and 3.

(* P<0.05 ** P<0.005) Mean HR from all horses from startle 1, 2, and 3.

Furthermore, the average behaviour scores from all three startles revealed a significant difference in behaviour response between the hot bloods and cold bloods (P=0.019, f=0.591, df=24), but again no significant difference in the behaviour responses between the juvenile and mature horses (see the image below). Additionally, five horses presented snorting behaviours, and four of those did not return to the feed bucket, so they received the maximum score for time taken to approach the feed bucket. The remaining horses all successfully returned to the feed bucket, including six who went further to investigate the umbrella. Six horses also performed rolling behaviours, two before the test began and the remaining four after the test was completed. However, there was no significant difference in behaviour seen in the same horse throughout the startles.

(* P<0.05). Average behaviour scores from all three startles for both types, juvenile and mature horses.

(* P<0.05). Average behaviour scores from all three startles for both types, juvenile and mature horses.

A Pearson’s correlation indicated there was a significantly positive correlation between the severity of spook and time taken to approach feed bucket (r=0.868, P=-.0005, N=26), suggesting those who showed a more severe initial spook also took the longest to calm down.

The correlation between spook behaviours and the time taken to return to the feed bucket.

The correlation between spook behaviours and the time taken to return to the feed bucket.


What did the startle test reveal?

The startle test revealed differences between the two types of equine. Across all three startles, the hot bloods’ average behaviour score was significantly higher than the cold bloods’. Plus, the average HR significantly differed between the types in the second startle. As seen in Fig 3.1.1 the HRs did not differ much in the first startle, but that significant difference in the second startle reflects the cold bloods becoming less reactive and calming down much more quickly than the hot bloods. This is supported by the lack of significant difference in the spooking behaviour. Spooking is an innate response in a fight-or-flight prey animal. All of the subject horses exhibited it to different degrees.; it was the aftermath that differed between the types.

Even though there were no statistically significant differences in behaviour between the startles for each individual, there was a difference in HR, which was exaggerated between the different types. This was probably a result of the test only being repeated three times. It is highly likely more of a habituation response would have been seen if the test had included additional repetitions of the startle stimulus. However, the severity of the initial reaction to a novel stimulus is an effective precursor of how many training sessions will be required to habituate a horse to a fear-inducing stimulus. Christensen et al. (2010) found a strong correlation between HR from the first training session and the total number of sessions required to habituate horses to a white nylon bag.

How might these insights be useful?

Studies have shown that reactions to novel objects can be generalised to other stimuli, more specifically to their response to performing in unfamiliar, stressful conditions (Christensen et al., 2012).

Research has also suggested that more reactive and emotional individuals require more training sessions, as well as more sensitive training (Borstel et al., 2012). The behaviour responses of the hot blood horses suggest they were significantly more reactive, which is closely related to emotionality. Increased emotionality can interfere with learning (Lansade and Simon, 2010; Christensen et al., 2012)—highly fearful or emotional horses may not make the best eventers, for example, since novel situations are a common occurrence and high levels of concentration are required to come out on top. Visser et al. (2003) were able to successfully predict the quality of the show-jumping performance of a horse later in life by conducting temperament tests at a very young age, which excludes variables like training history.

Although my test confirmed the stereotypes about hot- and cold-blooded breeds, nevertheless these tests could put a stop to all horses of the same type being tarred with the same brush. Although differences were found between the different types of equines in my study, there was variation between individuals—some hot bloods did not react as strongly as others.

If a battery of tests, fine-tuned to be as effective as possible and including a startle test, were used to test different aspects of an individual horse’s personality, they could offer a real insight into whether that horse is going to be a good fit for the lifestyle and environment a buyer can provide them.

Implementing the results of these tests into breeding selection could help create lines that were suitable for what modern people want and use horses for—pleasure riding, equestrian sports, racing, and so on. This could further reduce the risk of failed horse-human partnerships, wastage, and injuries.

Further work needs to focus on fine-tuning these tests further, ensuring they are replicable and consistent.

Concluding points

  • Despite age being taken into consideration, significant differences were still found between the two types of equine.
  • Hot bloods showed a significantly increased HR in the second startle compared to the cold bloods, as well as a significantly higher average behaviour score, suggesting they took longer to habituate and calm down.
  • There is a relationship between severity of response and time taken to return to baseline, which can then further be related to trainability.
  • By implementing effective temperament tests into breeding programmes and horse selection, it is possible we could reduce horse wastage and injuries.



von Borstel, U.K., Pasing, S., and Gauly, M. (2011). Towards a more objective assessment of equine personality using behavioural and physiological observations from performance test training. Applied Animal Behaviour Science 135(4), pp.277-85.

von Borstel, U.K., et. al, (2012). Repeatability and reliability of scores from ridden temperament tests conducted during performance tests. Applied Animal Behaviour Science 139(3-4), pp,251-63.

British Horse Society (2015). Equestrian statistics. Available at: (Accessed on 19 January 2017).

Christensen, J.W., Rundgren, M., and Olsson, K., (2010). Training methods for horses: Habituation to a frightening stimulus. Equine Veterinary Journal 38(5), pp.439-43.

Christensen, J.W., et. al, (2012). Does learning performance in horses relate to fearfulness, baseline stress hormone, and social rank? Applied Animal Behaviour Science 140(1-2), pp.44-52.

Dai, F., et. al, (2015). Validation of a fear test in sport horses using infrared thermography. Journal of Veterinary Behavior: Clinical Applications and Research 10(2), pp.128-36.

Evans, P., (2010). Equine vision and its effect on behaviour. Utah State University 1-3.

Flanders, F. (2011) Exploring Animal Science. NY: Delmar, Cengage Learning

Goodwin, D. (2010). The importance of ethology in understanding the behaviour of the horse. Equine Veterinary Journal 31(S28), pp.15-19.

Graf, P., von Borstel, U.K., and Gauly, M. (2013). Importance of personality traits in horses to breeders and riders. Journal of Veterinary Behavior: Clinical Applications and Research 8(5), pp.316-25.

Górecka-Bruzda, A., et. al, (2011). Reactivity to humans and fearfulness tests: Field validation in Polish cold blood horses. Applied Animal Behaviour Science 133(3-4), pp.207-15.

Koenen, E.P.C., Aldridge, L.I., and Philipsson, J., (2004). An overview of breeding objectives for warmblood sport horses. Livestock Production Science 88(1-2), pp.77-84.

Lansade, L., Pichard, G., and Leconte, M., (2008). Sensory sensitivities: Components of a horse’s temperament dimension. Applied Animal Behaviour Science 114(3-4), pp.534-53.

Lansade, L., and Simon, F., (2010). Horses’ learning performances are under the influence of several temperamental dimensions. Applied Animal Behaviour Science 125(1-2), pp.30-37.

Lesimple, C. et. al, (2011). Housing conditions and breed are associated with emotionality and cognitive abilities in riding school horses. Applied Animal Behaviour Science 129(2-4), pp.92-99.

Lloyd, A.S., (2007). Evaluation of a novel method of horse personality assessment: Rater-agreement and links to behaviour. Applied Animal Behaviour Science 105(1-3), pp.205-22.

Munsters, C.B.M. et. al, (2013). The effects of transport, riot control training and night patrols on the workload and stress of mounted police horses. Applied Animal Behaviour Science 143(1), pp.52-60.

Pierard, M., Mcgreevy, M., and Geers, R. (2017). Developing behavioral tests to support selection of police horses. Journal of Veterinary Behavior: Clinical Applications and Research 19, pp.7-13.

Le Scolan, N., Hausberger, M., and Wolff, A., (1997). Stability over situations in temperamental traits of horses as revealed by experimental and scoring approaches. Behavioural Processes 41(3), pp.257-66.

Seaman, S.C., Davidson, J.P.B., and Waran, N.K., (2002). How reliable Is temperament assessment in the domestic horse (Equus caballus)? Applied Animal Behaviour Science 78(2-4), pp.175-91.

Valenchon, M., et. al, (2013). Stress modulates instrumental learning performances in horses (Equus caballus) in interaction with temperament. PLoS ONE 8(4), pp.1-10.

Visser, E.K. et. al, (2001). Quantifying aspects of young horses’ temperament: Consistency of behavioural variables. Applied Animal Behaviour Science 74(4), pp.241-58.

Visser, E.K. et. al, (2002). Heart rate and heart rate variability during a novel object test and a handling test in young horses. Physiology & Behavior 76(2), pp.289-96.

Visser, E.K. et. al, (2003). The association between performance in show-jumping and personality traits earlier in life. Applied Animal Behaviour Science 82(4), pp.279-95.

Visser, E.K. et. al, (2008). Does horse temperament influence horse–rider cooperation? Journal of Applied Animal Welfare Science 11(3), pp.267-84.

Visser, E.K., (2010). Experts’ assessment of temperament in sport horses. Journal of Veterinary Behavior: Clinical Applications and Research 5(4), pp. 214-15.

Waran, Natalie (2007). The Welfare of Horses. Dordrecht, Springer.

Warren-Smith, A.K., and Mcgreevy, P.D., (2008). Equestrian coaches’ understanding and application of learning theory in horse training. Anthrozoos: A Multidisciplinary Journal of The Interactions of People & Animals 21(2), pp.153-62.

Wolff, A., Hausberger, M., and Le Scolan, N. (1997). Experimental tests to assess emotionality in horses. Behavioural Processes 40(3), pp.209-21.