An atlas of personality, emotion and behaviour

Anthony E. D. Mobbs, Conceptualization , Data curation , Formal analysis , Investigation , Methodology , Project administration , Resources , Software , Validation , Visualization , Writing – original draft , Writing – review & editing *

Anthony E. D. Mobbs

Department of Psychology, Faculty of Human Sciences, Macquarie University, Sydney, New South Wales, Australia

Department of Psychology, Faculty of Human Sciences, Macquarie University, Sydney, New South Wales, Australia

Competing Interests: Anthony E.D. Mobbs has submitted Patent Cooperation Treaty Application Number PCT/AU2019/051233 titled ‘An Improved Psychometric Testing System’. This does not alter our adherence to PLOS ONE policies on sharing data and materials.

This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Associated Data

S1 Table: Atlas of personality, emotion and behaviour poster. Expanded version of Table 4 with 25 words for each combination of affiliation, dominance and domain.

S2 Table: Example questionnaires with 48 questions. Two 48 Question tests. The first test uses antonymic adjectival descriptors of personality traits. The second using antonymic abstract noun descriptors of emotion.

S1 File: Antonymic word pairs. Pool of antonymic word pairs. Groups of 12 may be selected to form a questionnaire based on the atlas.

All Python code and datasets are available at doi.org/10.6084/m9.figshare.c.4792323, with the exception of data obtained under licence from Oxford and Merriam-Webster Dictionary. The data underlying the results presented in the study are available from Oxford Dictionary and Merriam-Webster Dictionary. Subsequent to obtaining licence(s) from Oxford and Merriam-Webster Dictionary, the Python code automatically downloads the licenced data enabling the Python code to fully operate.

Abstract

A novel two-dimensional matrix taxonomy, or atlas, of personality, emotion and behaviour is presented. The two dimensions of the atlas, affiliation and dominance, are demonstrated to have theoretical foundations in neurobiology and social psychology. Both dimensions are divided into five ordinal categories, creating a square matrix of 25 cells. A new catalogue of 20,669 English words descriptive of personality, emotion, behaviour, and power is also presented. The catalogue is more comprehensive than previous catalogues, and is novel in its inclusion of intrapersonal, group, and societal behaviours. All words in the catalogue were scored according to the atlas, facilitating visualisation in two dimensions. This enabled a contiguous and novel comparison of existing psychological taxonomies, as well as broader societal concepts such as leadership, ethics, and crime. Using the atlas, a novel psychological test is developed with improved sensitivity and specificity.

Introduction

Since antiquity, humans have sought to identify a framework for understanding the whole person, encompassing personality, emotion and behaviour[1–3]. Raymond Cattell, who introduced factor analysis to personality research, originally trained as a chemist and stated that his lifelong ambition was to identify a model of personality with similar explanatory power as the periodic table of elements[4]. Taxonomies are systems of measuring or classifying phenomena that facilitate precise communication and common understanding. The law of parsimony states that ‘the simplest explanation of an event or observation is the preferred explanation’[5], or in other words, ‘Everything should be made as simple as possible, but not simpler.’[6]. Taxonomies may be said to be parsimonious if they precisely describe a broad range of phenomena with the minimum number of independent variables[7]. The characteristics of a parsimonious model of personality are well understood[3,8,9], yet no ‘periodic table’ or grand theory of the whole person currently exists[3,8]. There remains disagreement about the number or nature of personality factors[10] giving rise to the diverse variety of personality constructs[8]. Identifying a grand theory has been deemed one of the most important goals of personality research[11] with impact for both diagnostic and therapeutic understanding. Psychological connections have been observed between the concepts of emotion, behaviour and personality[12–19]. It was therefore hypothesised that a taxonomy may encompass all three concepts of emotion, behaviour and personality. A unifying taxonomy encompassing personality, emotions and behaviour would be more parsimonious than three separate taxonomies. It was hypothesised that the lexical approach could be utilised to identify an overarching taxonomy of personality, emotion and behaviour.

Lexical analysis

The lexical hypothesis states that, ‘All aspects of human personality, which are or have been of importance, interest or utility, have already become recorded in the substance of language’[20] and, ‘When an idea is important, people are likely to have a word for it … the more important something is, … the more words there are likely to be’[21]. Lexical analysis is typically performed in two phases. Firstly, the words relevant to a topic are catalogued. Secondly, the catalogue is analysed to identify a parsimonious taxonomy that reduces complexity, simplifies categorisation and enhances communication. The completeness of the catalogue is desirable to ensure optimal selection of the form and parameterisation of the taxonomy. Conversely, an incomplete catalogue may lead to inappropriate selection of the taxonomic form or incorrect parameterisation.

Previous taxonomies of personality derived from lexical analysis, such as 16PF[22], HEXACO[23] and five-factor models[24] have largely focussed upon adjectival descriptors of personality. Verbs and nouns have largely been excluded[25], with notable exceptions[26,27], therefore leading to incomplete catalogues. In the context of lexical analysis, there are a range of verbs and nouns that are related to adjectival descriptors of personality traits. For example, verbs are used to describe interpersonal interactions (e.g. hit, hug, and harmonize) whereas abstract nouns may be used to describe emotions (e.g. hate, happiness and helplessness) and other nouns may be used to describe power (e.g. celebrity, chief, rich and poor).

Interpersonal circumplex

For thousands of years, circles have been used in various ways to map the breadth of the human experience[28]. For instance, ancient Greek astrology divided the sky into twelve equal portions of a circle, and from that derived the star-signs which are still discussed in popular culture today. The Interpersonal Circumplex was born out of this tradition. In the early-mid twentieth century, the American psychoanalyst, Harry Stack Sullivan, began mapping theories of interpersonal dynamics in circular forms. One sketch from 1948 was similar to the subsequent Interpersonal Circumplex; it depicts two individuals in an interpersonal interaction, connected by one arc representing a disaffiliative force, and two arcs representing affiliative forces. After Sullivan’s death, the Kaiser Foundation Group, which included Timothy Leary, operationalised Sullivan’s concepts and were credited with discovering the circumplex. Leary continued developing the Interpersonal Circumplex, which gained international recognition through his seminal text, ‘The Interpersonal Diagnosis of Personality’[1]. The Interpersonal Circumplex became a watershed theory in personality psychology, and although it is not often used in contemporary application, it remains a foundational influence.

Interpersonal Circumplex taxonomies[1,29,30] are characterised by radially divided concentric circles superimposed over two orthogonal axes ( Fig 1A ). When used as a taxonomy of personality, affiliation and dominance have commonly been used as the orthogonal dimensions[1,31–34]. Other researchers have used a range of synonymic terms for affiliation and dominance including: agency/communion[35], getting-ahead/getting along[36], ambitious/agreeable[37], assertiveness/compassion[38], dominant/friendly[29,39] and domineering/nurturant[40]. The superimposed concentric circles measure intensity, with the least intensity at the origin and gaining intensity in proportion to the radius. The most extreme behaviours and personality descriptors are located on the perimeter of the outermost concentric circle. Many psychological constructs have previously been mapped onto Interpersonal Circumplex models; for example, Fig 1B shows the components of the Dark Triad mapped onto the Circumplex[41]. The circumplex has also been used as a taxonomy of emotion in which case the valence/affect and intensity/activation are often used as the orthogonal dimensions[42–44].

(A) Interpersonal Circumplex used as a taxonomy of personality[1,30,45] with radially divided concentric circles superimposed over the two orthogonal dimensions of affiliation and dominance. (B) The Interpersonal Circumplex used to represent the Dark Triad[41]. (C) The Interpersonal Circumplex used as a taxonomy of emotion.

Criticisms of the circumplex approach have included the apparent subjectivity of item placement[46], difficulty in operationalising due to overly numerical application[47], the superior performance of alternative taxonomies[48] and the inability to place personality disorders[49]. Proponents of the Interpersonal Circumplex have used methodological devices such as rotations in an apparent effort to overcome its inherent limitations[50]. In addition to these criticisms observed by others, three additional deficiencies of the Interpersonal Circumplex were identified in the present research. Firstly, that the constraints imposed by the use of concentric circles to measure intensity, impose correlation between otherwise orthogonal axes. Secondly, the Interpersonal Circumplex does not make allowance for behaviours and traits of neutral dominance ( Fig 2A ) or neutral affiliation ( Fig 2B ). Finally, Interpersonal Circumplex taxonomies are ambiguous as to the placement of extreme behaviours. For example, a behaviour that is both maximally dominant and maximally disaffiliative, such as killing, is unable to be mapped onto the Interpersonal Circumplex models with precision due to the concentricity constraint ( Fig 2C ).

(A) Adominant behaviours (neutral dominance) not specifically identified on previous versions of the Interpersonal Circumplex, for example: ‘separate’, ‘gamble’ and ‘cooperate’. (B) Intrapersonal behaviours (neutral affiliation) not described on previous versions of the Interpersonal Circumplex[51] ‘innovate’, ‘learn’ and ‘stagnate’. (C) Maximally dominant and disaffiliative behaviours, such as killing, fall beyond the limits of the outermost concentric circle of the Interpersonal Circumplex.

The Abridged Big Five-Dimensional Circumplex (AB5C)[52,53] advances the Interpersonal Circumplex and measures many additional traits, thus overcoming some of the identified deficiencies. Although the Abridged Big Five-Dimensional Circumplex incorporates behaviours and traits of neutral affiliation and dominance, it retains the requirement for intensity to be measured by concentric circles, perpetuating the imposed correlation between axes and the ambiguous placement of extreme behaviours.

Proposing a unifying taxonomy

The characteristics of a parsimonious taxonomy of personality have previously been described[3,8,9], four of which are:

Synthetic: Integrating knowledge of the various components of personality within a single coherent framework[3].

Specific: The dimensions of the taxonomy should be orthogonal and divisible into non-overlapping categories so that phenomena may be uniquely placed within the taxonomy[9].

No current taxonomy appears to satisfy all these criteria. Narrow taxonomies, such as the Dark Triad[41] are neither comprehensive nor synthetic, as they specifically limit their scope to a particular subset of behaviours (e.g. Psychopathy, Narcissism, and Machiavellianism). Factor taxonomies, such as 16PF[22], HEXACO[23] and five-factor models[24] are derived using dimensional reduction techniques such as principal components analysis or factorisation. Although there is clear evidence of heritability of personality characteristics[54], there is currently no accepted theory as to the neurological mechanisms that support the dimensions of factor models[55]. Other criticisms of five-factor models include the unexplained correlations between dimensions[24,56–60] and that factor models have been assessed as being insufficiently comprehensive[61]. For these reasons, both narrow and factor taxonomies were excluded from consideration as the foundation of a parsimonious taxonomy.

In contrast to five-factor models of personality, the dimensions of the Interpersonal Circumplex, affiliation and dominance, have strong biological support. Functional neuroimaging has identified independent neural pathways for affiliation and dominance (see Fig 3 )[62]. These neural pathways have been identified in non-mammalian vertebrates[63] and are evident across five major vertebrate lineages of mammals, birds, reptiles, amphibians, and teleost fish[64]. The psychoactive hormones of oxytocin and testosterone have also been correlated with affiliation and dominance[65,66]. Additionally, the evolutionary bases of affiliation and dominance have been extensively investigated and established[67–69]. Given the comprehensive usage and biological basis for the selection of affiliation and dominance, we conclude that affiliation and dominance satisfy the criteria for being mechanistic and are therefore suitable candidates for the dimensions of a parsimonious taxonomy of personality.

(A) The dorsolateral prefrontal cortex is related to dominance and submissiveness. (B) The Putamen is related to affiliation.

To avoid ambiguity of trait placement and achieve the requisite level of specificity required for parsimony, two structural modifications to the Interpersonal Circumplex are proposed. Firstly, to overcome the imposed correlation between dimensions, it was proposed to remove the concentric circles. Therefore, behaviours and personality traits may be measured independently by reference to each orthogonal axis. Secondly, to allow for the categorisation of behaviours and traits of neutral dominance and affiliation, one of the ordinal categories for each dimension must be created to specifically allow for neutral behaviours. In keeping with the concept of neutrality, this category will be assigned the value of 0.

In order to categorise phenomena according to the taxonomy with specificity, each dimension must be divided into non-overlapping categories[9]. Ordinal categories are proposed that measure the enduring effect of behaviour on observers ( Table 1 ). Behaviours that have negligible effect on observers were scored towards the centre of the scale. Behaviours that have an enduring impact on the observer were scored towards the extremity of the scale. Affiliation was scored as positive and disaffiliative behaviours as negative. Dominant behaviours were scored positive and submissive behaviours, negative.

Table 1

An actor may attempt to assert, proffer, evoke or induce states of affiliation and/or dominance with respect to an observer; however, the response of the observer is inherently influenced by their perception of the actor’s behaviour/emotion. These perceptions may be dependent upon many factors, such as individual neurobiological variation in personality, cognitive states including attentional networks, power status, context, and the cultural milieu. For these reasons, the proposed classification scale does not specify or infer causal relations, rather, it identifies correlations rated typical of general cases.

Representative examples of the application of this table include:

Behaviours that are maximally disaffiliative and maximally dominant (-2,2) include ‘maim’, ‘attack’, as well as the absolute behaviour, ‘kill’. Observers of these behaviours may label the actor as ‘cruel’, ‘violent’ or ‘criminal’. Antecedent emotions to such behaviours and traits include ‘rage’, ‘anger’ or ‘wrath’.

Behaviours that are maximally disaffiliative and maximally submissive (-2,-2) include ‘self-harm’ and the absolute behaviour, ‘suicide’. Observers of these behaviours may label the actor as ‘dejected’, ‘morose’ or ‘melancholic’. Antecedent emotions to such behaviours and traits include ‘despair’, ‘emptiness’ or ‘futility’.

Behaviours that are maximally affiliative and maximally submissive (2,-2) include ‘worship’, ‘honour’ and ‘adore’, and the absolute behaviour, ‘martyrdom’. Observers of these behaviours may label the actor as ‘reverent’ or ‘devoted’. Antecedent emotions to such behaviours and traits include ‘love’, ‘veneration’ and ‘devotion’.

Behaviours that are maximally affiliative and maximally dominant (2,2) include ‘charm’, ‘excite’ and ‘inspire’. Observers of these behaviours may label the actor as ‘exuberant’, ‘dynamic’ or ‘charismatic’, and the absolute behaviour, ‘perfection’. Antecedent emotions to such behaviours and traits include ‘ecstasy’, ‘passion’ or ‘triumph’.

Behaviours that are of neutral dominance and affiliation (0,0) will be rarely noticed by either the actor or observer(s). These neutral emotions and behaviours include the awareness of our basic senses and involuntary behaviours such as ‘digestion’ and ‘respiration’.

The atlas taxonomy addresses the deficiencies of the Interpersonal Circumplex, thus forming a parsimonious taxonomy of human personality. By removing the radial and concentric constraints and dividing each axis into ordinal divisions, a matrix structure is created, see Fig 4 . The matrix’s standard taxonomic form[9] resembles other parsimonious taxonomies, such as the periodic table of elements and cartographic maps. This enables the full spectrum of interpersonal, intrapersonal, dominant, submissive and adominant behaviours to be precisely measured.

Matrix taxonomy of personality created by ordinal division of the affiliation and dominance dimensions. Neutrally affiliative and dominant behaviours and traits are shown in yellow. Behaviours may be independently classified by reference to each orthogonal axis.

Studies 1 to 4 in the present research seek to confirm that the proposed atlas taxonomy is both comprehensive and synthetic. If so confirmed, the atlas taxonomy will be demonstrated to be comprehensive, synthetic, mechanistic and specific, and will therefore satisfy the criterion for being a parsimonious taxonomy of personality. Study 5 in the present research develops a pilot psychological test founded upon the atlas that may be used as the basis for future research using the atlas.

Study 1

The objective of this Study was to catalogue all English language words descriptive of human interaction and emotion.

Method

WordNet[70] was identified as a reputable lexical database for the English language developed within the Princeton University Department of Psychology. Multiple scans of the WordNet repository cataloguing all words descriptive of emotions, behaviour, personality and power. The Oxford and Merriam Webster thesauri[71,72] were used to identify synonyms and antonyms for all catalogued words. The synonyms and antonyms descriptive of emotions, behaviour, personality and power were added to the catalogue. This process was repeated until no further words were identified. The Oxford English Dictionary[71] was used to classify the part of speech for each word.

To achieve consistent categorisation, the four concepts were defined as follows: behaviour is an observable action (typically verbs), emotion is the perception of a neurological impulse that initiates behaviour (typically abstract nouns)[73], personality traits are descriptors of characteristic behaviours (typically adjectives), and power is the capacity to influence another, or the capacity to avoid being influenced by another (typically nouns)[74–77]. The concepts of power and the trait of dominance are often conflated, yet others have sought to differentiate these concepts[78]. Power has been observed to have a significant moderating effect on personality and emotion. Given the moderating effect of power on personality and emotion, the present research distinguished power from other concepts where possible.

Results

20,669 words were identified as being descriptive of personality, emotion, behaviour and power. A summary of the words is shown in Table 2 .

Table 2

Examples of the 1,970 words descriptive of power, included ‘rich’, ‘poor’, ‘skilled’, ‘unskilled’, ‘employed’, ‘unemployed’, ‘king’ and ‘servant’.

Discussion

Previous catalogues of adjectives are in the range of 1,710 to 4,500 words[27,79–81]. The limited size of previous catalogues casts doubt on the completeness of taxonomies derived from them. Reconciliations of the new catalogue were performed with previous catalogues where available. The reconciliations showed that previous catalogues included archaic words that are uncommon in modern dictionaries, such as ‘indeliberate’, ‘granousier’, ‘eremitic’ and ‘scientistic’. The reconciliation showed that currently popular words, such as ‘adaptable’, ‘charismatic’, ‘perfectionist’ and, ‘withdrawn’, had been omitted from earlier catalogues. This is demonstrative of the ability of the catalogue to be culturally sensitive. The existence of modern online word catalogues, dictionaries and thesauri greatly assisted the compilation of the catalogue, and thus formed the most comprehensive catalogue of English-language words in personality research to date. The procedures adopted, as well as the absolute number of words identified, formed a catalogue that was considered to be unbiased and sufficient for the purposes of identifying a comprehensive taxonomy.

Despite the comprehensiveness of the catalogue, without replication, it is possible that important descriptors of personality, emotion or behaviour may have been overlooked.

Study 2

The objective of this study was: firstly, to confirm that the words in the catalogue can be classified according to the atlas matrix taxonomy, and secondly, to score a subset of up to 20% of the catalogued words.

Method

A Delphi approach was utilised to obtain a reliable consensus of opinions of a small group of health professionals[82]. The Delphi approach requires a group of experts to independently record their professional opinion and then achieve consensus through discussion. A cognitive and behavioural neurologist and three registered clinical psychologists assisted as linguistic judges. It was determined that a relatively small subset of the catalogue could be manually scored and therefore a five phase Delphi approach with sampling was used.

Phase one: To ensure the Delphi process covered all of the 25 cells of the atlas, the author performed an initial scoring of the entire catalogue. From this initial scoring, 10 nouns, 10 verbs and 10 adjectives were selected from each of the 25 cells of the atlas (750 words in total).

Phase two: Without being informed of the scoring performed in Phase one, the judges were asked to independently score each word. The judges then discussed their scores, during which, the judges were encouraged to revise their individual scoring until consensus was achieved.

Phase three: Synonyms of words scored in Phase two were identified from which were selected 35 nouns, 35 verbs and 35 adjectives from each cell (1,925 words in total). These word lists were collectively discussed by the judges. The scoring of each word was reviewed by exception until consensus was achieved. It was observed in this phase that the judges referred to synonyms and antonyms of other words selected to achieve a level of consistency in their scoring. Subsequent to the group process, the author and neurologist rescored the remainder of the catalogue.

Phase four: The author selected 1,620 antonymic word pairs. The word pairs consisted of words made opposite by prefix or suffix, such as ‘observant/unobservant’, ‘engaged/unengaged’ and ‘merciless/merciful’, and word pairs identified as antonyms in either or both of the reference thesauri. If both words had previously been scored as diametrically opposite in earlier phases, the scoring was retained. If both words had not been previously scored as diametrically opposite, a manual revision of their scoring was performed to achieve exact opposite scoring. These word pairs were then independently reviewed by three judges. In cases where at least two of the three judges had assigned identical scoring to word pairs, these word pairs were considered to be archetypal. The remaining words scored by the judges in Phase three were also deemed to be archetypal.

Phase five: The author reviewed the scored archetypal words and made minor amendments to ensure that synonyms, conjugates and inflections were proximately located where appropriate. The judges then performed an overarching review of a summary of 1,800 archetypal words until consensus was achieved by all judges (see S1 Table).

It was identified that there are examples of English language words that can be used in multiple contexts. In these situations, the words are usually classified as different parts of speech. For example, ‘bully’, ‘calm’ or ‘tidy’ all of which can all be descriptive of a behaviour (verb) or personality trait (adjective). In these instances, the words were categorised by the judging panel in the same cell irrespective of the context. An example of a word that can be used in different contexts but would be categorised in different cells is ‘humble’. ‘Humble’ used as an adjective was scored by the reviewers as having affiliation of 1 and dominance of -2. When used as a verb, ‘humble’ was scored as having affiliation of -2 and dominance of 1. When such instances were detected, the word was excluded as being candidate archetypal words. However, due to there being relatively few words in the cell (1,-2) and the word ‘humble’ was retained as an archetypal word for this cell.

Results

Of the 1,620 antonymically opposite word pairs, 27 were rejected by two or more judges leaving 1,593 word pairs remaining. Of the remaining 1,593 word pairs, 150 were opposite due to a prefix or suffix, 711 were identified as antonyms in both reference thesauri and 731 were identified as antonyms in one of the reference thesauri. Consensus amongst the four judges was achieved for 96% of word pairs, with a single dissenting judge on 4% of word pairs. The near complete consensus was viewed as confirming that the catalogued words can be successfully classified using the atlas. Table 3 shows an example of a personality trait, emotion and behaviour for each cell in the atlas taxonomy.

Table 3

Colouration applied to assist interpretation. See S1 Table for an expanded version of this table with 25 words for each combination of affiliation, dominance and domain.

Discussion

The thesauri did not frequently identify synonymic associations between the words descriptive of personality, emotion and behaviour; for example, the words ‘kill’ (behaviour) and ‘murderer’ (personality) were not synonymously related. The reference thesauri did however nominate ‘killer’ (personality) and ‘murderer’ (personality) as synonyms. ‘Kill’ and ‘killer’ can be linked by virtue of having the same linguistic stem. The linking of stem words was performed manually in the present research; however, it could be automated in future. By supplementing the thesauri derived synonyms with manually linked stem words, a robust association between personality, emotions and behaviours was achieved.

Some emotions are known to give rise to physiological changes[83], such as happiness, love, pride, anger, fear, anxiety, shame, sadness, depression, disgust, contempt, and envy. It was noted that these emotions are located on the outer edge of the atlas. Strong emotions promote high levels of arousal, and the associated physiological changes are thought to lead to an evolutionary advantage by creating a state of readiness for action[84,85]. Conversely, emotions and behaviours at the centre of the atlas are largely occult or involuntary without any sense of urgency. Emotional intensity and readiness for action appear to be correlated concepts, and future analysis with the aid of the atlas may establish causality between these concepts.

Historically, it has been acknowledged that relationships exist between personality, emotions and behaviour, although the nature of these relationships is yet unclear. The present research has illustrated the linguistic associations between the catalogued words. Building upon this, future research may use the atlas as a tool to clarify the nature of the causal relationships between personality, emotion and behaviour.

Extensive research has been conducted on the human and non-human ability to recognise emotion through facial expression[18,19,86–88]. Several studies have used the Interpersonal Circumplex as a taxonomy for arranging the biological spectrum of facial expressions[34,89–91]. The atlas’ inclusion of emotions prompted the categorisation of facial expressions according to the atlas. An artist's impression of the emotions represented in Table 3 is shown in Fig 5 . Emoji, which have become ubiquitous forms of electronic communication, have been arranged according to the atlas as shown in Fig 6 . The facial expressions shown in Fig 5 demonstrate smooth gradients of expression according to the orthogonal dimensions of affiliation and dominance. The ability of the atlas to represent facial expressions lends support to efficacy of the two-dimensional model of personality and emotion, as well as the selection of affiliation and dominance as the two orthogonal dimensions.