Hartpury Student Research Journal

Home » Does learning ability differ with personality in dogs?

Does learning ability differ with personality in dogs?

Author Names: Lauren Branfield (BSc (Hons) Animal Behaviour and Welfare) and Dr Carrie Ijichi



Understanding dog personality and learning ability can  improve training and welfare. Training can become more effective and aligned to an individual’s needs after understanding the subject’s personality and performance, which in turn can improve the welfare of dogs during training. The aim of this study was to understand how personality affects a dog’s cognitive performance. Personality was assessed using the validated Monash Canine Personality Questionnaire. An objective approach was utilised to measure cognitive performance through two learning ability tasks, the V-shaped Fence task and the Spin the Bottle task. Three measurements were used for each task; performance, engagement and the subject’s ability to learn through sets and attempts at each task. These performance factors were then compared to the individual’s results from the personality questionnaire. It was hypothesised that personality does affect learning ability and more specifically that an individual with a neurotic state would perform less well than a less neurotic subject. After analysis the hypotheses were accepted as significant findings were found for the personality trait, Motivation, and the average level of engagement with the learning ability task Spin the Bottle (P= 0.046, R=.362, N=31). PCA analysis also found correlations between neuroticism and learning ability. However, this was not reflected in the Post Hoc analysis which indicates the need for further research and a larger sample size to validate results.


1.0 Introduction

Personality is often defined as individual characteristics which form a distinctive combination of qualities (Ley et al, 2009; Gosling et al, 2003). Cognition can be defined as the mental action of acquiring knowledge and understanding it through experience and senses (Broom et al, 2007). Advances in understanding dog personality and learning ability can help match dogs to an owner or handler thus providing a suitable match between owner and dog (Coppinger et al, 2004). Improved knowledge about an individual’s personality can also promote better training and welfare (Svartberg, 2002).

There are a number of reasons why developing knowledge about dog personality and learning ability is important. Firstly, animal cognition is a developing area with regards to assessing an individual’s welfare (Boissy et al, 2007; Wemelsfelder et al, 2001). Welfare is not just negative stimuli absence but also involves the presence of positive stimuli (Boissy et al, 2007; Broom et al, 2007; Spruijt et al, 2001). Stafford et al (2012) suggest that dog welfare depends on what the dog experiences and that there are numerous factors affecting experiences which can compromise welfare. Identifying how an individual learns could help minimise their exposure to negative stimuli. For example a negative stimulus could be a cognitive task that a dog cannot perform, and instead providing it with a task it can successfully complete, therefore may improve its experiences and welfare.

Secondly, further research into dog personality and learning ability can improve our understanding of the factors that affect cognitive performance which is much more complicated than simply intelligence. Understanding if personality is a factor in cognitive performance can help assess whether a particular individual is capable of a task or activity. Lack of understanding can potentially cause dogs frustration and poor welfare states. Frustration is a key factor that can impact negatively upon welfare (Latham & Mason, 2010). The inability to perform a task or activity may lead a dog to experience high levels of frustration, often caused by a reward withdrawal (Bentosela et al, 2008). This reward withdrawal can be driven by lack of success and not reaching the reward in a cognitive task (Whitehead, 2012; Rolls, 2007). Understanding individual cognitive performance can encourage positive welfare as training can be altered to be maximally effective by enabling dogs to be treated as individuals and tailoring sessions to their abilities. Owners or handlers can then set an appropriate learning task which can be successfully completed by the dog.

Thirdly, new found knowledge on personality and learning ability can be used to assess the suitability of dogs, especially puppies, for certain tasks. These assessments could be particularly useful for working or assistance dogs such as Guide Dogs, Dogs for the Disabled, Police dogs and gun dogs. Assessing dogs from an earlier age can save time, money and also prevent stress on the puppy during training for the task (Batt et al, 2008; Serpell & Hsu, 2001; Goddard & Beilharz, 1983). Training dogs for particular purposes from an early age is essential for two further reasons. Firstly, puppies go through a “sensitive period”, previously known as the critical period, which is a time in an animal’s life when experiences have large effects on later behaviour (Webster, 1997; Scott et al, 1965). This means that if puppies are assessed as suitable for the personality and learning ability requirements for a particular task, training can begin as early as possible so that these experiences can be learnt in the sensitive period. Learning these training experiences in the sensitive period can enable the puppy to learn tasks more easily and to be able to cope with new experiences in later life (Webster, 1997). Secondly, if a puppy is assessed for certain personality traits and learning abilities it can help improve the likelihood of the dog succeeding in that particular task, therefore reducing potentially unsuccessful dogs from undergoing unnecessary training and stress. For example, a dog showing separation related behaviours is more likely to show negative emotional states which correlates with a neurotic cognitive bias (Mendl et al, 2010). Certain behaviours, such as separation anxiety, may affect a dog’s ability to carry out a task (Mendl et al, 2010;Svartberg, 2002). In contrast, a dog which is showing a confident, relaxed emotional state may perform better than a neurotic individual (Carter et al, 2012; Turcsán et al, 2011; Mendl et al, 2010).

To utilise personality effectively, it must be reliably assessed. Questionnaire based approaches are a useful, valid tool to assess personality in animals (Serpell et al, 2005; Momozawa et al, 2003; Gosling et al, 2002). The Monash Canine Personality Questionnaire (MCPQ) was developed to measure personality in dogs (Ley et al, 2009). The test was designed so that pet owners or handlers and their dogs could be matched based on their personalities. The questionnaire is a 26-item, adjective-based, owner-administered questionnaire that measures dog personality along five dimensions: Extraversion, Motivation, Training Focus, Amicability and Neuroticism (Ley et al, 2009). The study examined 65 couples using the Monash Canine Personality Questionnaire and used Inter-rater reliability and test re-test measures. Inter-rater reliability tests the degree to which the rater gives consistent estimates of the same behaviour and test re-test is measured by carrying out the test twice. As the questionnaire is deemed reliable, the assessment will be used in the current study to assess dog personality as measured by their owners. Although subjective assessments have been proved reliable, previous studies indicate caution should be taken as the results are only as good as the person filling it in (Funder, 1995).

Objective assessments have been used to measure cognitive performance in animals. For example, Pongracz et al (2001) examined how a human demonstrator affected a dog’s performance when the dog was presented with a novel task. The study found after behavioural assessments that the dogs performed better at reaching the target (food) behind a V-shaped fence after demonstration by a human. This test of performance was assessed by how quickly the dogs reached the target. The V-shaped fence tested the subject’s ability to understand “object permanence” – the understanding that the ball is still there even though it is out of sight (Zucca et al, 2007; Gagnon et al, 1982). Svartberg et al (2002) investigated the relationship between personality and performance in working dogs, finding relationships between “shyness” and lower performing dogs. Within this study, learning ability was assessed through performance in working dog trials, and was concluded to be a successful measure to evaluate a dog’s learning ability (Svartberg, 2002). This study was carried out on working dogs of two breeds, German Shepherds and Belgian Tervuren. Although the study demonstrates that personality and performance do indicate a relationship, it is not a good representation of pet dogs and a mix of breeds. Therefore, this shows the importance of further research into personality and learning ability in dogs.

Carter et al (2011) used both subjective and objective approaches and found that they correlate positively. Subjective questionnaires provide multiple and predefined contexts and objective behavioural assessments provide relative terms but in limited contexts (Cater et al, 2011). Using both approaches in the current research will encourage an accurate and valid conclusion about how personality affects learning abilities in dogs.

The overall aim of this piece of research is to understand how personality affects a dog’s cognitive performance. To this end, dog personality will be assessed by a subjective questionnaire and compared to cognitive performance during two objective learning ability tasks, the V-shaped fence and Spin the Bottle tasks. In the current research the V-shaped fence task will be utilised as per Pongracz et al (2001) through assessing the time taken by the individual to reach the target, however without a human demonstrator in this study. Pongracz et al (2001) used a food reward behind the V-shaped fence, however the current study will be using a toy reward. The second task, the Spin the Bottle task, was designed by the researcher to provide a food reward motivator (Figure 2). This is a novel test and was designed for the purpose of this study by the researcher. Using novel tasks as a means to measure personality is a validated method (Exnerová et al, 2010; Pongracz et al, 2001). Assessing subjects in new conditions can encourage direct measures of performance as affected by their personalities.  It is hypothesised that dog personality affects learning ability and more specifically that a dog with a neurotic state may perform less well than an individual which is not neurotic.


2.0 Methodology

2.1 Subjects

A total of 31 dogs were used comprising 18 males and 13 females, aged 8 months-12 years (mean = 55 months. Standard deviation = 33.61 months). The sample of dogs were comprised of a wide range of breeds. Subjects were selected via opportunistic sampling through clients from a Canine Behaviourist in Worcestershire. The selection criteria were that the subjects have no known lameness or illness, reliably responded to the release command “OK” and were not known to suffer from behaviour issues likely to cause unnecessary stress during the testing in order to reduce ethical concerns.


2.2 Personality assessment

Dog personalities, often described as psychological tendencies, were determined using the Monash Canine Personality Questionnaire  Ley et al, 2009; McCrae et al, 2002) following owner consent. The owner with the strongest relationship with the subject completed the questionnaire. The owner was aware of the purpose of the study but was blind to their dog’s performance in the cognitive test at the time they completed the questionnaire. Therefore data collected from the questionnaire was not influenced by the subject’s performance from the cognitive tests.


2.2.1 Monash Canine Personality Questionnaire

The questionnaire rated personality traits over five dimensions: Extraversion, Motivation, Training focus, Amicability and Neuroticism. Adjectives were used to describe each trait within these dimensions (Table 1). The owner rated the adjectives by a scoring assessment on a six point scale. 1 = really does not describe my dog, 6 = really describes my dog. All adjectives were randomised so that they did not fall into their dimension groupings in order to reduce proximity error. Each dimension was then expressed as a percentage score.


Table 1: Shows the Monash Canine Personality Questionnaire dimensions and adjectives to describe them.

Dimension Adjectives
Extraversion Active, energetic, excitable, hyperactivity, lively, and restless.
Motivation Assertive, determined, independent, persevering, tenacious.
Training focus Attentive, biddable, intelligent, obedient, reliable and trainable.
Amicability Easy going, friendly, non-aggressive, relaxed, and sociable.
Neuroticism Fearful, nervous, submissive and timid.


2.3 Learning ability tasks

Learning ability was measured through two novel behaviour assessments. One task used a ball as a reward, whereas the other task used cheese. As dogs may have different reward preferences it is important not to assume that every subject will be highly motivated by one reward. Therefore, two rewards were used to counteract different motivational preferences. The tasks took place in a secure, quiet room at Beech Behaviour Centre with the consent of its owner, Sandra Raw. Owners were not present which prevented them subliminally cueing the subject, avoiding the “Clever Hans Effect”. All subjects were handled by the researcher (LB) to ensure that all attempts were identical and not influenced by different handlers. Subjects were tested in random order where possible. Owners could remove their dog from testing at any time. The experimenter was blind to the subject’s personality scores at the time of testing and throughout behavioural analysis. Therefore the experimenter was not influenced by the results from the personality questionnaire.

Each task was completed within a set of three attempts but task order was randomised. Upon completion of an attempt within a set, subjects were sent back to the markers and released again. Time taken to complete each attempt was measured using a stop watch during testing. Dogs that did not complete the task were given a maximum time score of 120 seconds for the V-shaped fence and 180 seconds for the Spin the Bottle task. Time taken to complete the attempt was subtracted from the maximum available time to give a score for “performance”. Therefore, a high score indicated a short latency to solve the task and a good cognitive performance. A low score indicated a long latency and a poor cognitive performance.

Engagement with the task was measured retrospectively from video recordings by measuring how long the subject spent engaging with the task as a percentage of the total time taken to complete the task. A dog was recorded as being engaged or not, based on the behavioural indicators seen in Table 2.


Table 2: Shows behaviours seen if engaged or not engaged with the tasks.

Engaged with the task Not engaged with the task
Touching the task including: sniffing, licking, pawing, nudging it with its nose. Facing away from test area including: laying down not looking at the task, looking at the experimenter, focused on other areas of the test area and not looking at the task.
Looking at the task including: walking around it, laying down and sitting down. Stops to groom itself including: licking body and itching.
Vocalisations including: barking and whining at the task. Vocalisations including: barking or whining to get out the door or the test area.


Latency to complete the task across multiple attempts within the set was measured to assess how readily the subjects learnt the tasks and improved with repeat performance. Improvement in performance across attempts within a set was determined by the following formulas:

Improvement 1 = latency in attempt 1 – latency attempt 2

Improvement 2 = latency attempt 2 – latency attempt 3

Overall Improvement = latency attempt 1 – latency attempt 3


2.3.1 V-shaped fence

The researcher placed a ball behind a solid fence and the subject was allowed to search for it (Pongracz et al, 2001). This tested the subject’s ability to understand “object permanence” – the understanding that the ball is still there even though it is out of sight (Zucca et al, 2007; Gagnon et al, 1992). This task was completed three times to assess improved performance as a proxy for learning.

The V-Shaped fence was approximately 3.5’ x 2’ x 2’ in dimensions and was constructed out of two 3.5’ x 2’ boards, arranged at right angles to one another. These were joined via a common spine along the centre of the “V”. The structure was secured by two bracing pieces, attached at the top and at the bottom of the structure. Both the central spine and bracing pieces were made from ½’ x ½’ wooden battens. All components were secured using a high number of small nails which were hammered flush to avoid any sharp edges which could be a risk to any subject using the equipment (Figure 1).

Branfield figure 1

Figure 1: Shows the V-shaped fence learning ability task.

Subjects were brought to a marker 10ft from the object and given a command to sit. The researcher (LB) then released the leash and, when ready, gave the release command “OK”. No further interaction or demonstration was given. The time began from the release command and ended when either the subject successfully retrieved the ball or the maximum time was reached. The maximum time allowed to complete this task was two minutes as pilot testing revealed that if a subject had not completed the task within two minutes, it was unlikely to do so. Subjects that did not complete the task were given a score of 120 seconds.


2.3.2 Spin the Bottle

The Spin the Bottle object was approximately 1’ x 1’ x 1.5’ in dimensions. A bottle was suspended on an axle where it was able to swing freely. The axle was supported by a strong structure around it. The frame was made from various cut lengths of ½” x 1” wood, the axle was a length of ¾” wooden batten and the plastic bottle was a recycled soft drink container which had been washed and dried. All components were secured using a number of medium-heavy duty small nails which were all hammered flush to avoid sharp edges which could have been a risk to any subjects using the equipment. Also, all edges were sanded down to avoid splinters and any unnecessary rough edges (Figure 2).

Branfield figure 2

Figure 2: Shows the Spin the Bottle learning ability task.

The researcher put one piece of cheese inside a bottle which was connected to a wooden frame. A pilot study showed that more than one piece of cheese required high levels of impulse control; impulse control being the ability to inhibit a response (Bouwknecht et al, 2001). As the subjects ranged in training status, multiple pieces increased motivation and normally resulted in the subject leaving the marker before the release command. The task required the subject to knock or paw at the bottle until the bottle span around, releasing the food reward. This task was completed three times to assess learning via reduced time to complete the task.

Again, subjects were brought to a marker 10ft from the object and given a command to sit. The handler then released the leash and, when ready, gave the release command “OK”. No further interaction or demonstration was given after the release command. Subjects were given three minutes to complete the task because pilot testing revealed that if subjects had not successfully released the food item after three minutes, they were unlikely to be successful. The time began when the release command was given and ended when the subject either released and collected the food reward, or the maximum time was reached. Subjects that did not successfully complete the task were given a score of 180 seconds.


2.4 Statistical testing

A Kolmogorov-Smirnov test was used to test for normality of all variables. Data was non-parametric, and was transformed to meet parametric criteria using a Log Transformation test so further tests could be completed.

Principal component analysis (PCA) was used to reduce the large number of variables and look for inter-correlations. PCA transforms a number of possible correlated variables into a smaller number (n=53) so variance and inter-correlations can clearly be seen. Strong loadings were selected if above +0.6, and those at 0.3 and below were regarded as weak loadings but still of interest (Frey and Pimental, 1978). However, it has been suggested that PCA is a guide and further analysis should be continued after the test (Wold, Esbensen, & Geladi, 1987). As PCA gives a data matrix it extracts strong patterns producing loadings and factors. However, these loadings do not show whether or how each individual loading interacts with one another (A interacts with B, B interacts with C, but A may not interact with C). Therefore, data was clarified by using Pearson’s Product-Moment Correlation Coefficient test to assess linear findings. Multiple correlations were corrected for by a False Discovery Rate Test (Benjamini et al, 1995). This test helped control error rate with mass testing and so all true data were collected. Paired t-tests were also conducted to investigate the level of engagement between the V-shaped fence and Spin the Bottle task. All analyses was conducted using SPSS software.


3.0 Results

After the Log Transformation test was completed a PCA (Principal Component Analysis) for personality and learning ability tests was used to simplify the data. PCA extracted six factors accounting for a total of 81.368% variance (Table 3).

Table 3: Shows PCA factors for personality and learning ability tests in a component matrix (N=31). Figures in bold are main loadings.

1 2 3 4 5 6
-Average Performance, V-shaped Fence .903 .300 -.164 .169
-V-shaped Fence attempt 1 .827 .333 -.167 -.208 -.117
-Average Level of Engagement, V-shaped Fence -.820 -.192 .268 -.123 -.302
-V-shaped Fence attempt 2 .817 .192 -.114 -.192 .220 -.190
-V-shaped Fence attempt 3 .816 .168 -.275 .161 .334
-Average Level of Engagement, Spin the Bottle -.597 .527 .191 .206
-Improvement V-shaped Fence attempt 2 .532 .311 .307 .216 .274 -.370
-Overall Time Improvement, V-shaped Fence .461 .447 .119 .266 -.418 -.348
-Spin the Bottle attempt 3 .358 -.853 .115 .309
-Spin the Bottle attempt 2 .162 -.797 -.368 .342 .119
-Overall Time Improvement, Spin the Bottle -.485 .690 -.445 -.200 .126
-Average Performance, Spin the Bottle .181 -.672 .296 .210 .416
-Improvement Spin the Bottle attempt 1 -.154 .603 .502 -.375 .319
-Improvement Spin the Bottle attempt 2 -.468 .242 -.689   .280
-Training Focus -.106 .231 -.275 .640 -.238
-Neuroticism .162 -.345 .155 -.619 .112
-Motivation -.102 .455 .228 .538 .480 .128
Extraversion .186 .563 .109 .598 -.112
Improvement V-shaped Fence attempt 1 .423 .433 .305 .168 -.468
Amicability .175   .453 -.311 .635

In the first factor, accounting for 22.456% variance, all demonstrated personality traits were non-significant therefore this factor was not used for further analysis as it does not relate to the hypothesis. However, there were notable loadings from the PCA test in this component in terms of the learning ability task variables. Within Component 1 all learning ability variables apart from Spin the Bottle attempt 2 (0.162), Average Performance on Spin the Bottle (0.181) and Improvement in Spin the Bottle 1 (-0.154) showed correlations.

The second factor, accounting for 19.990% variance, loaded weak and positive on Motivation (0.455) and weak and negative on Neuroticism (-0.345). There were strong loadings of Spin the Bottle attempt 2 (-0.797) and 3 (-0.853), Spin the Bottle Improvement 1 (0.603), Overall Time Improvement Spin the Bottle (0.69) and Average Performance Spin the Bottle (-0.672). This factor was labelled as “Component 2”. Another loading of interest developed in Component 2. In the personality traits, Motivation (0.455) and a negative Neuroticism (-0.345) correlated with the majority of the Spin the Bottle tasks.

The third factor, accounting for 11.635% variance, approached a significant and positive loading on Extraversion (0.563). There were strong, negative loadings of Spin the Bottle Improvement 2 (-.0.689). This factor was labelled “Component 3”. A notable loading within Component 3 was also established. This loading included personality trait Extraversion (0.563) and learning ability task Spin the Bottle attempt 2 (-0.689).

The fourth factor, accounting for 10.255% variance, loaded strong and negative on Neuroticism (-0.619), weak and positive on Motivation (0.538) and weak and negative on Amicability (0.453). There were no strong loadings on learning ability variables however PCA found weak and positive loadings on Spin the Bottle attempt 2 (0.342) and 3 (0.309) and Improvement 1 Spin the Bottle (-0.375). This factor was labelled as “Component 4”.

The fifth factor, accounting for 9.374% variance, loaded strong and positive on Extraversion (0.598), weak and positive on Motivation (0.48) and weak and negative on Amicability (-0.311). There were no strong loadings in the learning ability variables, however PCA found weak loadings on Overall Time Improvement V-shaped Fence (-0.418), Improvement 2 Spin the Bottle (0.28) and Improvement 1 V-shaped Fence (-0.468). This factor was labelled as “Component 5”.

And finally the sixth factor, accounting for 7.658% variance, loaded strong and positive on Amicability (0.635). There were no strong loadings in the learning ability variables, however PCA found weak and negative loadings on Average Level of Engagement V-shaped Fence (-0.302), Improvement 2 V-shaped Fence (-0.37), and Overall Time Improvement V-shaped Fence (-0.348), and weak, positive loadings in V-shaped Fence attempt 3 (0.334), Average Performance Spin the Bottle (0.416), Improvement 1 Spin the Bottle (0.319). This factor was labelled as “Component 6”.

After the PCA, test factors loading +/- 0.3 in each factor were then analysed through a Pearson’s Product-Moment Correlation Coefficient test. It revealed that there was no statistically significant correlation between strong loadings from Component 2, however there was a significant correlation between Motivation and Average Level of Engagement in Spin the Bottle (P= 0.046, R=.362, N=31; Table 4).  No other significant data were found in Component 3, Component 4, Component 5 and Component 6.


Table 4: Shows Pearson’s Correlation results in component 2, (N=31). Figures in bold are significant data.

Component 2
Motivation Neuroticism
-Average Performance, V-shaped Fence 0.753 0.956
-V-shaped Fence attempt 1 0.41 0.428
-Average Level of Engagement, S-the Bottle 0.046 0.104
-Improvement V-shaped Fence attempt 2 0.056 0.411
-Overall Time Improvement, V-shaped Fence 0.641 0.511
-Spin the Bottle attempt 2 0.214 0.407
-Spin the Bottle attempt 3 0.189 0.574
Overall Time Improvement, Spin the Bottle 0.229 0.176
-Average Performance, Spin the Bottle 0.618 0.354
-Improvement Spin the Bottle attempt 1 0.203 0.812
-Improvement V-shaped Fence attempt 1 0.513 0.274


Paired T-tests showed that there was a significant difference in overall level of engagement between the V-shaped fence and Spin the Bottle task (N=31, R=.321, P=0.00). The V-shaped fence task showed higher levels of engagement (mean=3.9960) than the Spin the Bottle task (mean=2.5076).


4.0 Discussion

The importance of dogs within today’s society provides a niche area to explore personalities and how they impact learning ability. Understanding dog personality and learning ability benefits dogs and their owners. Firstly, understanding how personality affects dog performance can alter how people train their animals and allow the provision of appropriate goals which dogs can achieve (Marshall-Pescini et al, 2008; Lindsay, 2001). Secondly, providing appropriate training goals can drastically improve welfare as standards and expectations can be set specifically for that individual (Ley, McGreevy, & Bennett, 2009; Bennett & Rohlf, 2007). In the current study, it was hypothesised that personality affects learning ability and, more specifically, that individuals scoring more highly for neuroticism would perform less well in cognitive tasks. This was assessed by comparing performance in Spin the Bottle and the V-shaped fence tasks against scores for personality, measured by a subjective questionnaire (Ley, Bennett, & Coleman, 2009; Pongrácz et al, 2001). The data indicates correlations between certain aspects of personality and performance.

PCA Component 1 loaded all learning ability variables apart from Spin the Bottle attempt 2, Average Performance Spin the Bottle and Improvement Spin the Bottle 1. Personality traits were not loaded in this component which means that component 1 does not support the hypothesis. The PCA table indicated that most learning ability variables correlated and showed how a dog’s performance in multiple facets of the cognitive tests interacted. Post Hoc tests did not support the PCA results and showed no significant data in this component. This suggests that while multiple variables may be interacting, there is no clear linear relationship between any given pair.

Component 2 loaded negatively but weakly in Neuroticism and positively but weakly in Motivation. This suggests that there is a tendency for dogs scoring lower for Neuroticism to be more motivated. Post Hoc analysis revealed that Motivation correlated significantly with Average Level of Engagement in Spin the Bottle. This Post Hoc analysis correlation also loaded in the PCA test suggesting a meaningful link between these variables. However, a significant correlation between Neuroticism and Motivation was not reflected in the Post Hoc test. This suggests that a high score for subjectively assessed Motivation could predict a higher level of engagement in this particular task. Increased interaction rate stems from an improved problem solving ability and learning speed (Marshall-Pescini et al, 2008; Panksepp, 1998). Therefore, it may be that Motivation was related more to the Spin the Bottle task instead of the V-shaped fence task as it may have required a higher level of problem solving ability. Understanding that Motivation can predict engagement level is useful, as training goals can be set appropriately for that individual regarding its motivational state.

Component 3 loaded positively on Extraversion, however, Post Hoc analysis did not support this. In Component 3, there was a trend between all but one Spin the Bottle variables (Spin the Bottle attempt 3). This suggests that an individual with a high Extraversion score was overall unable to carry out the Spin the Bottle task effectively. This relationship within this component was not expected but is of possible interest (Mendl et al, 2010). Extraversion may impact learning ability negatively as these individuals are described as friendly, enthusiastic and sociable as measured in the questionnaire (Ley, Bennett, et al, 2009; Canli et al, 2002). Therefore, individuals scoring highly for this personality factor may be more inclined to interact with the researcher than the cognitive task. This was supported by the researcher’s informal observations during testing. Subjects with this personality trait would jump up, bark and look up at the researcher, a possible indicator of attention seeking (Fisher, 2012; Rooney et al, 2002).

Component 4 loaded approaching positive significance on Motivation, positively but weak on Amicability and negatively on Neuroticism. Within this component, three learning ability variables were found as well as these personality variables. These learning ability variables included positive but weak loading for Spin the Bottle attempt 3, positive but weak loading for Spin the Bottle attempt 2 and negative but weak loading for Improvement Spin the Bottle attempt 1. These relationships were not confirmed by Post Hoc analysis suggesting that, while they may interact together, there is no clear linear relationship between any given pair of variables. Having a tendency for increased Motivation and Amicability with a low Neurotic personality may explain the correlation with the Spin the Bottle tasks. A dog which is motivated and social, could appear to be less fearful within the testing area according to the emotion framework (Mendl et al, 2010; Figure 3). Therefore, being in a calm, confident and motivated state may improve performance in the Spin the Bottle task through increased focus and interaction (Whitehead, 2012; Rolls, 2007).


Branfield figure 3

Figure 3; shows the movement through animal emotions and moods when faced with a negative or positive valence (Mendl et al, 2010).

Component 5 loaded positively with a strong tendency on Extraversion, positively but weak on Motivation and negatively but weak on Amicability. In addition, three learning ability variables loaded on this component. These variables include loading negatively but weak on Overall Time Improvement V-shaped Fence, positively but weak on Improvement Spin the Bottle attempt 2, and negatively with a significant tendency in Improvement V-shaped Fence attempt 1. However, Post Hoc analysis did not support linear relationships between these pairs of variables. Personality loadings in Component 5 suggest that the dog was confident, moderately motivated and possibly less human focused. Although in Component 3 a high score in Extraversion was evaluated to result in being more human focused, the contradiction of a weak score on Amicability in this component may have decreased human focus. Less human focus may be the result of negative results in two out of three learning ability variables against a positive, strong Extraversion and negative, weak Amicability score. When comparing the personality and learning ability, results can indicate a positive but weak tendency of Amicability and negative but strong tendancy in Extraversion. This suggests subjects in this component may be less human focused as scoring could show a dog which is more focused on the task than the researcher. This could mean that dogs with this personality could be quite engaged and motivated in the task which did not require them to focus or look back at the researcher for “help”. It has been suggested that dogs can recognise and use cues from humans, including eye gazing and hand gestures, to assist during tasks (Hare et al, 2002). Together this could mean a dog that shows little Amicability does not use human cues as much as a dog with a higher Amicability personality.

Component 6 loaded positively and strongly in Amicability in the PCA test. Learning variables included negative but weak loading on Average Level of Engagement V-shaped Fence, positive but weak tendency in V-shaped Fence attempt 3, negative but weak in Improvement V-shaped Fence attempt 2, negative but weak in Overall Time Improvement V-shaped Fence, positive but weak in Average Performance Spin the Bottle and positive but weak in Improvement Spin the Bottle attempt 1. Again, Post Hoc analysis did not confirm these results. PCA loadings may be explained such that amicable dogs are less interested or motivated to complete the task, instead focusing on the novelty of a researcher in the testing area. This increased focus on the researcher may be driven from an urge to socialise, or could be a predisposition of separation related behaviours as owners were out of the room (DeMartini, 2014).

Separation anxiety in dogs is also associated with dogs that are very dependent on their owners (Serpell, 1996). Dogs who were unable to cope during testing when owners left may have gone through separation anxiety and would have showed a negative valence and a possible high arousal state (Mendl et al, 2010; Serpell, 1996; Figure 3). This type of behaviour may have been seen in dogs with an amicable personality. Dogs were then expected to carry out novel tasks when they were in an anxious or fearful state which therefore could have influenced their performance. In contrast, if the dogs were interested in socialising with the researcher they may have had a positive valence with a medium to high arousal state, resulting in a happy or excited dog (Figure 3). However, even if the dog was in a positive state they may have been more concerned and interested in the researcher rather than solving the novel task. Together, this may indicate why only small improvements and averages were found in the learning ability variables in this component.


4.1 Limitations and future research

The sample size, while adequate for this field of research, may not have been sufficient for PCA analysis. PCA requires an optimal sample of five times the number of variables used in statistical analysis (Bryant & Yarnold, 1995). The sample size did not meet optimal numbers to account for the number of variables included (Osborne et al, 2004; Bryant and Yarnold, 1995; Hatcher 1994). This may explain why findings in the PCA were not supported by those from Pearson’s Product-Moment Correlation Coefficient test. In addition, personality requires large variance in order to yield significant results. Insufficient variance will lead to smaller effect sizes which requires larger sample sizes to detect. However, personality variables appeared to have adequate variance so this is not likely to explain limited significant findings.

Subjective assessment can be only as good as the person filling it in (Funder, 1995). Funder (1995) discussed how certain judgements may be viewed as accurate if it leads to successful social interactions. These judgements of dog personality may have resulted in the owner being subjectively biased and inaccurate regarding their dog’s actual personality. Although the questionnaire used has been validated (Ley et al, 2009) the subjective findings must be used with caution. Momozawa et al (2003) concluded that questionnaires are as reliable as behavioural assessment methods and provide valid evidence. To ensure that the owner is providing accurate data they were advised to be as truthful as possible during the questionnaire. The current study did not identify key information about the owners. Key information that may be relevant for future research includes; age, gender, education, length of acquaintance with the dog, and experience with a variety of individuals in the species. In personality research as measured by owners, distinctions between individuals and breeds as a whole will not arise if the rater is only familiar with a small or homogeneous number of individuals (Gosling, 2001). If an owner is only familiar with one breed type, they may have biased views towards the personality of that particular dog within the breed. This can be a limiting factor when scoring an individual’s personality during questionnaires.

Whilst the subjective personality assessment is well validated, the validity of the Spin the Bottle task is uncertain. As the task was designed by the researcher, it was novel and had not been previously validated. To validate this task’s measure of performance further studies should be performed to replicate it. According to Panksepp (2005) variation in motivation within the tasks could have resulted from the value of the reward. The amount and quality of the reward has direct effects on the behaviour; animals work harder for larger or tastier rewards (Panksepp, 2005; Reid, 1996). Burch et al (1999) also mentions how motivation is at the heart of reinforcement. The current study only used one piece of cheese in the Spin the Bottle task. The pilot study found that multiple pieces of cheese caused subjects to move off the starting point before the “OK” cue was given. This meant that multiple pieces would have been too motivating which would have impacted upon the repeatability of the study and the accuracy of readings obtained  when timing the subjects. The value of the reward also depends on the individual. As different dogs are motivated by different rewards two tasks were used. Each task resulted in different motivation preferences, food and play. Using different rewards was an attempt to counteract these differences in play versus food motivation.

Location was another possible factor affecting dog behaviour during testing. Although a quiet, low stimulating room was used, it was impossible to control all noises from outside. Noises such as passing traffic and neighbours leaving or coming home could have been a distraction or induced a fear response which may have affected dog performance. Distance of tasks between each other may also have contributed to inaccurate results. Experimental equipment was set up throughout testing so dogs would not get distracted by the movement of equipment between tasks. Moving the equipment out of the testing area could have created a distracting or over-stimulating environment. Therefore it was decided that equipment would be set down for the entire testing time to avoid affecting data validity as performance may have been influenced by excitement levels if equipment was moved. Although distraction was minimised by keeping equipment in one place, dogs could have got confused to which tasks they were required to engage in. One particular dog went to the wrong task when the “OK” cue was given. If the study was repeated, after each task the subject could be taken out of the testing area which would allow the researcher to alternate the equipment in the absence of the subject.

In addition, it was not possible to control when the dog had been fed and exercised. This lack of control may have affected the motivational state of the dogs. According to Pryor (2002) how much a dog has eaten can affect its motivational outcome during training. Therefore this could have affected data validity regarding how motivated and engaged the dog was. A novel task to solve to reach the reward may have looked less appealing to a dog which had just been fed a large bowl of food and exercised heavily compared to a dog which had not (Pryor 2002; Burch et al, 1999). Excitement levels from a dog which has previously been on a long walk compared to a dog which had not may also differ in performance outcomes.


5.0 Conclusion

Although only one significant piece of data was found this key piece of data could be baseline knowledge on which to develop future research. Understanding that a dog’s motivational state may impact their learning ability may encourage owners to provide them with an appropriate task or training goal. Understanding that individuals learn and perform differently is vital to improve welfare so animals remain in a balanced emotional state.

The study found correlations between personality and performance in the Spin the Bottle task. With further validation, the Spin the Bottle task has the potential to be a tool for future research to measure performance in dogs. There is also the potential to improve the knowledge that owners have regarding their dogs performance levels, as each task can be easily designed and implemented for cognitive tasks to stimulate their dog’s minds. However, it is important to remember that the learning ability task Spin the Bottle was a novel task to scientific research. As it was not previously validated it may have not been an appropriate task to measure performance. Notwithstanding this potential limitation, the knowledge gained about personality and learning ability has the potential to improve dog welfare. Understanding how an individual performs and how its personality affects this performance can alter owners or handlers’ methods of training. From this research a conclusion can be drawn that a dog with a motivational personality can potentially perform better at the Spin the Bottle task than a dog which is in a neurotic state.



Batt, L. S. Batt, M. S. Baguley, J. A. McGreevy, P. D. (2008). Factors associated with success in guide dog training. Journal of Veterinary Behavior: Clinical Applications and Research. 3(4), 143–151.

Broom, D. and Fraser, A. (2007). Domestic animal behaviour and welfare. Wallingford, UK: CABI.

Boissy A. Manteuffel, G. Jensen, M. Oppermann Moe, R. Spruiji, B. Keeling, L. Winckler, C. Forkman, B. Dimitrov, I. Langbein, J. Bakken, M. Veissier, I. Aubert, A. (2007). Assessment of positive emotions in animals to improve their welfare. Physiology and Behaviour. 92, 375–397.

Bennett, P. C. Rohlf, V. I. (2007). Owner-companion dog interactions: Relationships between demographic variables, potentially problematic behaviours, training engagement and shared activities. Applied Animal Behaviour Science. 102(1-2), 65–84.

Bentosela, M. Barrera, G. Jakovcevic, A. (2008). Effect of reinforcement, reinforcer omission and extinction on a communicative response in domestic dogs (Canis Familiaris). Behavioural Processes. 78 (3), 464-9.

Burch, M. R. Bailey, J.S. (1999) How Dogs Learn. New York: Howell

Bouwknecht, J. A. Hijzen, T. H. van der Gugten, J. Maes, R. A. . Hen, R. Olivier, B. (2001). Absence of 5-HT1B receptors is associated with impaired impulse control in male 5-HT1B knockout mice. Biological Psychiatry. 49(7), 557–568.

Bryant, F. B. Yarnold, P. R. (1995). Principal-components analysis and exploratory and confirmatory factor analysis. American Psychological Association. 99–136.

Benjamini, Y. Hochberg, Y. (1995). Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society. 57(1), 289-300

Canli, T. Sivers, H. Whitfield, S. L. Gotlib, I. H. & Gabrieli, J. D. E. (2002). Amygdala response to happy faces as a function of extraversion. Science (New York, N.Y.). 296(5576), 2191.

Carter, A. Marshall, H. Heinsohn, R. Cowlishaw, G. (2012). How not to measure boldness: novel object and antipredator responses are not the same in wild baboons. Animal Behaviour. 84(3), 603–609.

Carter, A. Marshall, H. Heinsohn, R. Cowlishaw, G. (2011). Evaluating animal personalities: do observer assessments and experimental tests measure the same thing? Behavioural Ecology and Sociobiology. 66(1), 153-160.

Coppinger, R. Coppinger, L. (2004). Dogs: a new understanding of canine origin, behaviour and evolution. USA: Scribner.

DeMartini-Price, M. (2014). Treating Separation Anxiety in Dogs. USA: Dogwise Publishing.

Exnerová, A. Svádová, K. Fučíková, E. Drent, P. Štys, P. (2010). Personality matters: individual variation in reactions of naive bird predators to aposematic prey. Proceedings of the Royal Society. B, Biological sciences. 277(1682), 723-728.

Frey, D. Pimental, R. (1978). Principal components analysis and factor analysis P. Cogan (Ed.), Quantitative Ethology (1st ed.), John Wiley & Sons, New York, pp. 219–245.

Funder, D. C. (1995). On the accuracy of personality judgment: A realistic approach. American Psychological Association. 102(4), 652–670.

Fisher, J. (2012). Think Dog, the bestselling guide to canine psychology. Revised edition. UK: Octopus Publishing Group Ltd.

Gagnon, S. Doré, F. Y. (1992). Search behavior in various breeds of adult dogs (Canis familiaris): Object permanence and olfactory cues. Journal of Comparative Psychology. 106(1).

Goddard, M. E. Beilharz, R. G. (1983). Genetics of traits which determine the suitability of dogs as guide-dogs for the blind. Applied Animal Ethology, 9(3-4), 299–315.

Gosling, S. D. (2001). From mice to men: What can we learn about personality from animal research? Psychological Bulletin. 127(1), 45–86.

Gosling, S. Kwan, V. John, O. (2003). A dog’s got personality: a cross-species comparative approach to personality judgments in dogs and humans. Journal of Personality and Social Psychology. 85(6), 1161-1169.

Gosling, S. Vazire, S. (2002). Are we barking up the right tree? Evaluating a comparative approach to personality. Journal of Research in Personality. 36(6), 607–614.

Hare, B. Brown, M. Williamson, B. Tomasello, M. (2002). The Domestication of Social Cognition in Dogs. Science. 298(5598), 1634-1636.

Hatcher, L. (1994). A step-by-step Approach to Using the SAS system for Factor Analysis and Structural Equation Modelling. Cray, N.C.: SAS Institute, Inc.

Ley, J. Bennett, P. Coleman, G. (2009). A refinement and validation of the Monash Canine Personality Questionnaire (MCPQ). Applied Animal Behaviour Science. 116(2-4), 220-227.

Ley, J. Bennett, P. Coleman, G. (2009). Inter-rater and test–retest reliability of the Monash Canine Personality Questionnaire-Revised (MCPQ-R). Applied Animal Behaviour Science. 119(1–2), 85–90.

Latham, A. Mason, G. (2010). Frustration and perseveration in stereotypic captive animals: is a taste of enrichment worse than none at all? Behavioural Brain Research. 211 (1), 96-104.

Lindsay, S.R. (2001). Handbook of Applied Dog Behavior and Training, volume. 2: Etiology and Assessment of Behavioural Problems. Iowa State University Press, Ames

McCrae, R. Costa, P. Fritz, O. Angleitner, A. Hřebíčková, M. Avia, M. Sanz, J. Sánchez-Bernardos, M. Kusdil, M. Woodfield, R. Saunders, P. Smith, P. (2002). Nature over nurture: Temperament, personality, and life span development. Journal of personality and social psychology. 78(1), 173-186.

Marshall-Pescini, S. Valsecchi, P. Petak, I. Accorsi, P. A. & Previde, E. P. (2008). Does training make you smarter? The effects of training on dogs’ performance (Canis familiaris) in a problem solving task. Behavioural Processes. 78(3), 449–54.

Mendl, M. Brooks, J. Basse, C. Burman, O. Paul, E. Blackwell, E. Casey, R. (2010). Dogs showing separation-related behaviour exhibit a ‘pessimistic’ cognitive bias. Current Biology. 20(19), R839-R840.

Mendl, M. Burman, O. Paul, E. (2010). An integrative and functional framework for the study of animal emotion and mood. Proceedings of the Royal Society. B, Biological sciences. 227(1696), 2895-2904.

Momozawa, Y. Ono, T. Sato, F. Kikusui, T. Takeuchi, Y. Mori, Y. Kusunose, R. (2003). Assessment of equine temperament by a questionnaire survey to caretakers and evaluation of its reliability by simultaneous behavior test. Applied Animal Behaviour Science. 84(2), 127–138.

Osborne, J.W. Costello, A,B. (2004). Sample size and subject to item ratio in principle component analysis. Practical Assessment, Research and Evaluation. 9(11).

Pongracz, P. Miklosi, A. Kubinyi, E. Gurobi, K. Topal, J. Csanyi, V. (2001). Social learning in dogs: the effect of a human demonstrator on the performance of dogs in a detour task. Animal Behaviour. 62(6), 1109-1117.

Panksepp, J. (1998). Affective neuroscience. The foundations of human and animal emotions. Oxford: Oxford University Press Inc.

Pryor, K (2002). Don’t Shoot the Dog, the new art of teaching and training. Surrey: Ringpress Books Ltd. Revised edition.

Rolls, E.T (2007). Emotion Explained. United States: Oxford University Press.

Rooney, N, J. Bradshaw, J, W.S. (2002). An experimental study of the effects of play upon the dog-human relationship. Applied Animal Behaviour Science. 75(2), 161-176

Scott, J. Fuller, J. (1965). Genetics and the social behaviour of the dog, the classic study. USA: The University of Chicago.

Serpell, J. A. (1996). Evidence for an association between pet behavior and owner attachment levels. Applied Animal Behaviour Science, 47(1-2), 49–60.

Serpell, J. Hsu, Y. (2001). Development and validation of a novel method for evaluating behavior and temperament in guide dogs. Applied Animal Behaviour Science. 72(4), 347–364.

Serpell, J. Hsu, Y. (2005). Effects of breed, sex, and neuter status on trainability in dogs. Anthrozoos: A Multidisciplinary Journal of The Interactions of People & Animals. 18(4), 196-207.

Spruiji, M. van den Bos, R. Pijlman, F. (2001). A concept of welfare based on reward evaluating mechanism inS the brain: anticipatory behaviour as an indicator for the state of reward systems. Applied Animal Behaviour Science. 72, 145-171.

Svartberg, K. (2002). Shyness-boldness predicts performance in working dogs. Applied animal behaviour and science. 79(2), 157-174.

Turcsán, B. Kubinyi, E. Virányi, Z. Range, F. (2011). Personality matching in owner-dog dyads. Journal of Veterinary Behaviour: Clinical Applications and Research. 6(1), 77.

Wold, S. Esbensen, K. Geladi, P. (1987). Principal component analysis. Chemometrics and Intelligent Laboratory Systems, 2(1-3), 37–52.

Wemelsfelder, F. Hunter, T. Mendl, M. (2001). Assessing the “whole animal”: a free choice profiling approach. Animal Behaviour, 62(2), 209-220.

Whitehead, S (2012). The secrets your dog wants you to know, Clever Dog. London: Harper Collins publishers.

Webster, 1997 Webster, S.D. Being sensitive to the sensitive period;

D.S. Mills, S.E. Heath, L.J. Harrington (1997). Proceedings of the First International Conference on Veterinary Behavioural Medicine, Birmingham, UK, April 1(2), 20–2.

Zucca, P. Milos, N. Vallortigara, G. (2007). Piagetian object permanence and its development in Eurasian jays (Garrulus glandarius). Animal Cognition. 10(2), 243–58.