Mike Ko Personal Portfolio


Home-School Education
Hong Kong

University of Durham
Bachelor of Science
United Kingdom

University of Sussex

Master of Arts
United Kingdom

Academic Works
                                Year 3
: Literature Review

What are the Evidences for Personalities as Repeatable Traits in Vertebrates?

Table of Contents:

    1. Summary
    2. Introduction
    3. Defining and Measuring Personalities
    4. The General Evidence for Repeatability
    5. Factors that Influence Personalities and its Repeatability                
    6. Challenges in Studying and Identifying Personalities
    7. Indirect Evidence: Theoretical Bases for Suboptimal Behavioural Variations
    8. Conclusion



            Interest in the concept of personalities in animal behaviour has grown in recent decades. Although conventional views of behaviour do not place great significance on consistent behavioural variations, evidence suggests that they do exist and thus may require theoretical explanations. A crucial aspect of personality is repeatability, which is mainly estimated through the index R. Many studies have been conducted on the topic, yet no universal set of definitions is in place. There are many supportive findings in the literature for the existence of personalities. Yet misinterpretations of relationships between measured variables and personality traits as well as lax control of confounding variables mean that such evidence must be further scrutinised. Theoretical theories and models show that personalities can evolve as an adaptive trait, and thus can fit into the traditional evolutionary scheme based on optimality. This field is still in its early infancy and requires improvement to its terminology and methodologies for progress to be made.


            Animals exhibit various behaviours in response to different kinds of stimuli from either the environment or from other animals. To understand the underpinnings of such behaviours, studies can treat these behaviours as an optimised adaptation against a given situation (Maynard Smith, 1978). With this assumption, all members of a given animal species can be expected to react in similar ways under the same conditions. Yet various studies have shown that there can actually be consistent variations in behavioural responses between individuals over time. For example, some individuals may be more inclined to actively explore novel surroundings than others. Such persistent individual-specific differences in behaviour are now often referred to as “personalities” in the literature (Carter et al., 2012).

            In the past, personalities have mostly been regarded as random non-adaptive variations around an adaptive optimum. However, the potential prevalence of distinct personalities suggests that such variations may be under selective pressures (Dall, Houston and McNamara, 2004). Thus they may be something that warrants further investigation with respect to their development and impacts on ecology and evolution. In particular, the existence of personalities can change how we interpret the effects of selection pressures on behaviours. While natural selection has previously been assumed to cause optimal behaviours to develop, new theories may now be required to account for animal personalities.

            While there is now great interest in personality within animal behaviour, it is still very much an emerging field. There is a relative lack of standardisation in both the methods and interpretations for personality studies. Certain researchers may also be hesitant to investigate it in their studies, sometimes due to concerns about the significance of the entire concept of personalities itself (Réale et al., 2007). This review aims to evaluate the evidence in the literature for the existence of personalities, and thus highlight whether or not animal personalities will be a significant topic for study. The background for personality and related concepts are explained, followed by a review of the basic evidence for personalities in vertebrates within the literature. Next, potential issues in study methods and definitions that that may confound these otherwise supportive findings are stressed. The theoretical explanations for personalities are also described as an indirect support for the concept. Due to scope limitations, this review will only focus on personality in vertebrates.

Defining and Measuring Personalities

          The literature have proposed and indeed shown common use of the term “personality” to describe behavioural variations between individuals that are consistent over time (Dall, Houston and McNamara, 2004; Réale et al., 2007). Yet different terms have been used to describe behavioural variations in the past. In addition to “personality”, there have been some instances of where “temperament” (Réale et al., 2007), “behavioural syndromes” (Conrad et al., 2011; Sih, Bell and Johnson, 2004) and “coping strategies” (Koolhas et al., 1999) have also been used, although such alternatives can also refer to slightly different or derived concepts.

            While personality refers to all consistent behavioural variations in general, in most cases such variations are studied with regard to a specific aspect of behaviour such as boldness. In this context, such aspects are sometimes called “personality traits” (Carter et al., 2012). Interestingly, personality traits can be both defined and tested in various ways, depending on the study. For example, boldness has been defined through the behavioural response of individuals when exposed to predation risk (Brown, Jones and Braithwaite, 2004), novel environments (Fraser et al., 2001) or objects (Wright et al., 2006), respectively. Thus caution must be exercised when interpreting personality traits, as they may not necessarily refer to the same variable.

            Although there is no universal classification system for personality traits for the field of study, suggestions have been made to define them more clearly. In particular, it has been proposed that there should be a tentative framework to sort all traits into five general types as shown in Table 1 (Réale et al., 2007). Tests of different behaviours can be counted as a measure for one or more of these personality traits. This scheme covers many of the traits studied in the literature so far, although there are also other behaviours that are omitted by this categorisation. Yet there is currently insufficient data on the subject to fully and clearly differentiate between different traits, which will be required to make an unambiguous classification system.

Table 1. Classification Categories for Personality Traits as proposed by Réale et al., (2007).

Personality Traits:

Behaviour described:


“An individual’s reaction to any risky situation, but not new situations.”


“An individual’s reaction to a new situation.”


“The general level of activity of an individual.”


“An individual’s agonistic reaction towards conspecifics.”


“An individual’s reaction to the presence or absence of conspecifics (excluding aggressive behaviour).”


            Regardless of the actual trait, any behavioural variations between individuals can only be defined as a personality if tests show that they satisfy certain criteria. Given the definition of personality, it follows that an important criterion is the repeatability of the behavioural variation in a given individual. A commonly used measure of repeatability is the intraclass correlation coefficient (Hayes and Jenkins, 1997; Bell, Hankison and Laskowski, 2009), sometimes referred to as “narrow sense repeatability” or R (Biro and Stamps, 2015). R is calculated from the variances of measurements both among and within individuals. The larger the value of R, the more repeatable a behaviour within an individual is (Hayes and Jenkins, 1997). Various studies have utilised R or other alternative methods to estimate the repeatability of personality traits, either as a general study or in response to changes in an independent variable. Thus there is ample material in the literature that assesses the evidence for personalities in vertebrates.

The General Evidence for Repeatability      

            The idea that behavioural traits can be repeatable over time has been suggested several decades ago, most notably by the study of the three-spined stickleback fish, Gasterosteus aculeatus (Huntingford, 1976). The study found that consistent inter-individual behavioural variations in with respect to their aggressiveness against territory intruders of different species. Recent studies in the literature have since further supported the notion of personality in various vertebrate species. For example, Pleizier et al. (2015) found that checkered pufferfish (Sphoeroides testudineus) showed repeatable inter-individual variation in traits such as puff performance, level of activity and the duration of startled behaviour in response to stress by cortisol injections. In male guppies (Poecilia reticulata), individuals demonstrated consistent differences in mating and exploratory behaviour (Kelley, Phillips and Evans, 2013). Specifically, different individuals consistently exhibited particular frequencies of courtship displays and approaches towards new objects, the latter of which was taken as an indicator of boldness. Such trends persisted even when light levels were varied during the study. In juvenile lemon damselfishes (Pomacentrus moluccensis), certain individuals showed consistently higher levels of activity, boldness and aggression at given temperatures (Biro, Beckmann and Stamps, 2010). Somewhat unusually, there is evidence that behavioural predictability can be a repeatable trait. Biro and Adriaenssens (2012) found that certain mosquitofishes (Gambusia holbrooki) repeatedly showed greater variability in activity levels than others over time.

            In birds, multiple studies have focused on the great tit Parus major to evaluate the repeatability of behavioural traits, for which there seems to be some degree of evidence. As part of a study on four European P. major populations, it was shown that there were consistent differences between individuals’ exploration behaviour (Dingemanse et al., 2012). Exploration was measured through an integrated score calculated from the number of hops and flights exhibited within the first two minutes of confinement in a cage. Another study of P. major similarly showed that there were consistent differences in the speed score for exhibited exploratory behaviour between two genetically distinct bird lines (Carere et al., 2005). Here the scores were dependent on the time it took for the birds to explore a number of specific objects in the confinement area of the study. Based on the scores obtained, the two lines were accordingly identified to have either “slow” or “fast” exploratory behaviours. However, the results of the study also indicated that differences between the lines in some aspects of exploratory behaviour could decrease with age. Furthermore, variation of social behavioural traits such as gregariousness (from the number of nearby conspecifics during foraging), sociability (interaction rate with conspecifics) and “betweenness” (centrality of an individual’s position relative to conspecifics) in P. major were shown to be repeatable over three years (Aplin et al., 2015).

            While studies on reptiles were relatively uncommon in recent literature, work has indicated that individuals of the common lizard Zootoca vivipara can show repeatable differences in measures of exploration behaviour and activity for time scales of between a week and a year (Le Galliard et al., 2012). Another study arrived at the same conclusion for boldness in hatchling keelback snakes (Tropidonophis mairii) over a timescale of days. Boldness was measured by the time of emergence from shelter by individual hatchlings (Mayer, Shine and Brown, 2016).

            Mammalian studies further support the notion of repeatable inter-individual differences in behaviour. This was the case for traits including playfulness, chase-proneness, sociability, curiosity and aggressiveness in the domestic dog, Canis familiaris (Svartberg et al., 2005). Similarly, the little brown bat, Myotis lucifugus, also show consistent behavioural differences between individuals with respect to activity, exploration and relatively specific stress responses, although only over a 24-hour period (Menzies et al., 2013). In the meerkat Suricata suricatta, the extent of social behaviours such as babysitting and provisioning for the young were repeatable within individuals over a decade (English, Nakagawa and Clutton-Brock, 2010). However, babysitting was comparatively less repeatable than provisioning and can show greater variation in consistency with age.

            Overall, one meta-analysis of 114 studies on 98 species of invertebrates and vertebrates found that certain traits were quite repeatable, including aggression, mating and habitat selection (Bell, Hankison and Laskowski, 2014). However, the analysis also found that certain traits such as migration, activity and mate preferences were relatively unrepeatable. Regardless, there is a good amount of supporting evidence that there are indeed repeatable behavioural traits and hence personalities do exists within individuals of a population.

Factors that Influence Personalities and its Repeatability

            Despite the numerous studies reporting distinct personalities in vertebrates, it should be noted that not all personality traits are as repeatable as others. For example, the meta-analysis found traits such as migration, activity and mate preferences were relatively less repeatable than other persona (Bell, Hankison and Laskowski, 2014). In certain instances, whether personalities in a specific behavioural trait actually exist in the first place may be questionable. When testing for personalities, it is important to bear in mind the results can be influenced by many underlying factors.

            Age is an important variable that can lead to differences in repeatable behaviour within individuals. As individuals develop from a juvenile into an adult, the personality that they adopt may change. In the Little Brown Bat M. lucifugus, personalities of pre-weaning young individuals showed lower activity compared to relatively more mature individuals (Menzies et al, 2013). This is not an unreasonable result, as personalities that may be beneficial in juvenile stages may be suboptimal in adult conditions. Thus repeatability of behaviours may vary depending on a study’s observation interval and the age of target individuals. Even beyond the developmental stages, it is possible for the personality of individuals to change as a fully developed adult. Carere et al. (2005) had found exploration behaviours to be consistent in two genetically distinct lines of P. major, with one line being quicker in measures of exploration. Yet over the study interval the comparatively slow line eventually became quicker. These results suggest that personalities are not fixed within individuals. In effect, there may be differing behavioural plasticity in different life stages despite the potential to have personalities. The same can potentially apply to plasticity in different seasons or time in general.

            Behavioural plasticity is an important concept related to personalities. Plasticity refers to the ability to alter a trait (in this case behaviour) in response to changing situations or contexts. In contrast, personalities are mostly maintained regardless of the context. At first glance, both concepts appear to be mutually exclusive, however it may be possible these two aspects actually operate in concert. However, this review found no definitive solution to this apparent dilemma literature.

            Repeatability of intra-individual behaviours can be dependent on whether or not it is part of a “behavioural syndrome” (Sih, Bell and Johnson, 2004), a distinct non-synonymous term from personality. All behaviours have a genetic basis, thus pleiotropy with respect to multiple behaviours and hence personalities can occur. These personalities are likely be correlated, thus creating a behavioural syndrome. Depending on their physiological basis, a personality that is part of a behavioural syndrome may become more repeatable.

            In addition, abiotic factors such as temperature have been found to affect repeatability. Biro, Beckmann, and Stamps, (2010) discovered that small temperature fluctuations of about 3°C led to increases in measures for boldness, aggressiveness and activity for juvenile individuals of two species of coral reef fish (Pomacentrus moluccensis and Pomacentrus bankanensis). This study highlights the importance of controlling test conditions in testing personalities in individuals. In this case, individual fishes tested at colder conditions may appear more bold, aggressive or active than it would have and vice-versa. Such discrepancies will give data sets that are mistakenly regarded as comparable and give incorrect results for behavioural repeatability.

            Of course, these are undoubtedly only some of the many variables that can influence the repeatability of behavioural variations. Depending on how studies are conducted, positive findings for personalities may turn out to be unfounded as behavioural variations may actually be non-repeatable. Thus it is vital that studies of personalities and their interpretations take into account such potential confounding factors.

Challenges in Studying and Identifying Personalities

            While a variety of factors can affect the identification of personalities, the method for personality studies can also cast doubt upon the validity of their findings. As mentioned before, this relatively new field of behaviour lacks a standardised system of terminology, in particular a formal set of terms and definitions for personality traits. Due to this ambiguity, the methods for studying personalities have become similarly inconsistent. Past studies have used differing measures of behaviour to investigate given personality traits, however they ultimately may not be measuring the same trait (Carter et al., 2012). As previously mentioned, some studies have studied boldness through measuring responses to novel objects, environment or predation risks. Yet other studies found no correlation between some of these measures. In one case the propensity of the mountain chickadee Poecile gambeli to explore novel environments was not correlated to their propensity to touch a novel object (Fox et al., 2009). Thus the two variables may not actually be interchangeable indicators of boldness. A further compounding factor may be that different species or taxa may respond to specific tests in different ways. This potential mismatch between personality traits and measured indicative behaviours is a serious issue, as strictly speaking findings of studies may either be invalid or mutually incomparable.

            The possibility of researchers misattributing the connection between an indicator variable and a personality trait can be problematic. Sometimes it is not immediately apparent what personality trait a given variable actually reflects. One study repeatedly exposed juvenile lemon shark (Negaprion brevirostris) individuals to a novel environment in the form of open fields (Finger et al., 2016). While individuals showed consistent variation in their movement rates in the fields, the exact personality trait it reflects can either by activity or merely the reaction to novelty. An additional test was required to distinguish between the two, through testing for the habituation of movement rates. Ultimately it was found that the sharks’ movement rate did decrease with repeated exposure to the fields, thus suggesting that first test actually was an indicator of reaction to novelty. Thus it is very important that researchers ensure that the variables measured are actually linked to the given personality traits.

            The confusion caused by the lack of formal defined terms is also partly due to the complicated relationship between measured variables and the personality traits they reflect. By some definitions for personality traits, it is possible for a particular variable to serve as a measure for different traits. This causes uncertainty over what trait the variable actually reflects. Conversely, there can also be situations when multiple variables actually serve as an indicator for the same personality trait. Researchers who are working concurrently may inadvertently use different labels for the same trait, hence further exacerbating the potential confusion. These two scenarios are referred to as the “jingle-jangle fallacy” in psychology (Carter et al., 2012).

            There are also multiple issues with how repeatability is currently defined. As it stands, there is no consensus on the time period over which intra-individual behaviours must be consistent to be defined as personalities. Menzies et al, (2013) found that relatively younger juvenile individuals of M. lucifugus appear to demonstrate consistently lower activity over a 24-hour observation period. Whether due to sampling bias or other factors, the repeatability of behaviours may change if the observation period was extended.

            Additionally, the use of R as a measure of repeatability can introduce statistical inaccuracies. The problem is that R values are calculated with the assumption that behaviours do not vary with time, which is possible. If the mean for a measured behaviour (distance travelled in a novel environment, for instance) varies over time for an individual, then the derived values for R will become incorrect. This can be partially mitigated by including a time variable into calculations for R. However, there are also other factors that can affect R, such as changes in conditions (Biro and Stamps, 2015).

            Another concern for personality studies is how accurately does measured behaviour in laboratory conditions reflect actual personality traits in the wild. Because it is often difficult if not impossible to conduct controlled and repeated behavioural measurements, many observations in personalities studies are made on subjects in captivity. Yet findings from laboratory conditions do not always translate into the wild. A study compared the repeatability of behavioural measures for shyness, activity and exploration in the cricket Gryllus campestris, both in the wild and in captivity (Fisher et al., 2015). While findings for activity and exploration were found to be consistent between the two groups, shyness was not. However, just how well results from the laboratory and the wild corroborate with one another for personalities is generally unknown. Regardless, the above study illustrates the point that findings in the laboratory must ultimately be checked against actual behaviours in the wild, whether for vertebrates or otherwise.

            Given these sources of potential issues, the positive findings in identifying personalities in vertebrates must be regarded with caution. The ambiguities that currently exist in the field mean that current findings may eventually have to be re-examined or even discarded.

Indirect Evidence: Theoretical Bases for Suboptimal Behavioural Variations

            The ultimately debatable validity of the evidence for personalities in vertebrates notwithstanding, the fundamental question of why personalities should exist must be accounted for as well. Regardless of quality, current studies have encouraged more work in accounting for the underlying causes for behavioural variations. Provided that these theoretical accounts or models are plausible, they can serve as indirect supporting evidence for the notion of personalities.

            Animal behaviours are often interpreted as being evolved adaptive optimal states. In contrast, the concept of personalities means that there will be suboptimal variations around the optimal behaviour, hence going against the assumption of optimality. Regardless of whether or not personalities can be demonstrated empirically, the theoretical question of how such ‘sub-optimal’ variations persists in a population remains. Generally speaking, there will always be natural variation in a given trait (either on the genetic or the phenotypic level) due to random mutations. However, this notion may be more difficult to apply to behaviours, as non-adaptive behavioural variations will probably infer costs without benefits and thus should be selected against. While there is no definitive answer to this issue yet, different hypothesis and models have been put forth that can give personalities an adaptive significance.

            For example, Johnstone (2001) modified the conventional Hawk-Dove game model to including  “eavesdroppers”, players who can gain information of the last strategy employed by its future opponent beforehand. The result from this model is that eavesdroppers may know the outcome of upcoming contests and choose the most beneficial strategy. In effect, an aggressive Hawk who knows the opponent will be a docile Dove will maintain its strategy, while it may switch strategies if a fellow Hawk will be the opponent. Under certain conditions, the modified model will result in Evolutionary Stable States where all individuals adopt a single strategy (J. Thus this theoretical model constructed conditions that allowed the persistence of a specific behavioural variation, in effect personalities.

            Another theoretical model on dynamic foraging game for two players shows that an adaptive behavioural variation can be encouraged in one of the participants (Rands et al., 2003). In the model, two players are in shelters that they must leave to forage when food stores drop below a threshold. The model includes predation risk during foraging, thus probability of survival is higher when both players forage at the same time. The model predicts that given a random level of food stores and constant foraging success, the player with a lower food store should always adopt a risk-taking strategy and head out to forage first. Again, the conditions of a theoretical model shows that it is possible to fix different behavioural variations and create personalities that are also adaptive.

            It has been further argued that possessing a flexible behaviour can be maladaptive in certain circumstances. In a constantly changing environment, the costs associated with having a flexible but slowly responsive behaviour that could adapt inappropriately may be greater than a single-strategy one (Tufto, 1999).

            From a theoretical standpoint, it is possible for personalities to be adaptive, upon which they can be acted upon be selective pressures and maintained within the population. Yet these theories and models must ultimately be supported by empirical findings that support the existence of personalities.


            The literature contains numerous studies showing that inter-individual behavioural variations are indeed repeatable and thus personalities do exist in vertebrates. Certain theoretical models also support this notion in principal. Yet many factors exists that affect the quality of the studies and their findings. In particular, the lack of universal definitions and terminology for the field may very well stifle future progress and generate confusion. Existing methodologies with regard to repeatability of behaviours and the control of test variables have also been relatively lax, allowing confounding factors such as temperature and time to affect the assessment of intra-individual behavioural repeatability.

            In the future, researchers must set clear definitions for personality traits that currently exists as well as those that will undoubtedly be put forth in time. Although difficult, a framework for linking behavioural variables and the reflected personality trait will have to be made to avoid so-called “Jingle-Jangle Fallacies”. Standards of studies must also be raised so that potentially confounding variables are properly controlled. Only then will researchers be equipped to answer the question of whether personalities exists in vertebrates and other animals.



Back to Top