Infant rhesus macaque (Macaca mulatta) personality and subjective well-being.

Infant temperament is theorized to lay the foundation for adult personality; however, many questions remain regarding personality in infancy, including the number of dimensions, extent to which they are adult-like, and their relation to other outcomes, such as mental and physical health. Here we tested whether adult-like personality dimensions are already present in infancy in a nonhuman primate species. We measured personality and subjective well-being in 7-month-old rhesus macaques (N = 55) using the Hominoid Personality Questionnaire and Subjective Well-Being Questionnaire, both of which were developed for adult primates based on human measures. Multiple human raters, who provided infants with daily care since birth, independently rated each infant. We found high interrater reliability. Results from a parallel analysis and scree plot indicated a five component structure, which, using principal components analysis, we found to be comprised of dimensions relating to Openness (e.g., curiosity, inquisitive, playfulness), Assertiveness (e.g., dominance, bullying, aggressive), Anxiety (e.g., vigilance, fearful), Friendliness (e.g., sociable, affectionate, sympathetic), and Intellect (e.g., organized, not erratic). These components are largely analogous to those in adult macaques, suggesting remarkably stable structural personality components across the lifespan. Infant macaques' subjective well-being positively correlates with Openness and Assertiveness and negatively correlated with Anxiety, similar to findings in adult macaques and other primates. Together, these findings suggest that, in macaques, infant personality dimensions may be conceptually related to adult personality and challenge the view that infant temperament may be disorganized and not as meaningful as adult personality. Further research is necessary to explore the antecedents, predictive validity, and stability of these personality components across situations and with development.

Introduction

Human and nonhuman primate infants display individual differences in various aspects of their psychology and behavior [1-4]. Individual differences in infancy are often described in terms of temperament [2-4]. Temperament—a commonly used term to describe infant personality [5]—refers to biologically based inter-individual differences in behavioral tendencies (e.g., attention, motor behavior, emotions, self-regulation), which constitute stable patterns across contexts and over time [6-8]. A related but distinguishable individual difference in infancy is happiness or subjective well-being, which refers to having high levels of life satisfaction [9], high levels of positive affect, and low levels of negative affect [10]. In humans, infant temperament and well-being are considered the early foundations of adult personality and well-being [11,12]; however, many questions remain about their development. For instance, are adult-like personality characteristics present in infancy? That is, to what extent do individual differences in personality traits and well-being appear early and persist across development?

More is known about personality and well-being in human adults relative to in infants or children [10,13]. Adult personality is most commonly measured as five stable domains or constructs (i.e., the Five-Factor Model or the “Big Five”): Extraversion, Openness, Conscientiousness, Neuroticism, and Agreeableness [14,15]. This Five-Factor Model is generalizable across methods of measurement, gender, age, and culture, with strong test-retest reliability and internal consistency [16,17]. These personality dimensions are heritable [18,19] and largely stable but continue to change with age in adults [20,21]. The five factors are predictive of academic performance [22], career success [23], romantic relationship satisfaction [24], health [25], and subjective well-being [26].

Subjective well-being—a construct typically measured in adults through self-reports—includes individual differences in emotions, such as positive affect and happiness, and cognitive components, such as goal achievement and life satisfaction [27,28]. Subjective well-being is related to, but also distinguishable from, personality [29]. For example, higher levels of Extroversion, Agreeableness, Conscientiousness, and emotional stability (i.e., lower Neuroticism) are associated with greater subjective well-being [13]. Subjective well-being is heritable [30], and, in adults, largely stable over time [31] but also still changing with age [32], and positively associated with mental and physical health [33,34].

Human infant personality and subjective well-being

Given the role of personality and well-being for predicting health and success, it is important to uncover their early roots and how they emerge and develop. While it is theorized that the adult five-factor personality dimensions are largely synonymous with infant temperament [11,35], to date, there have been few tests of this proposal, and therefore, there is rather limited support for this idea. Toddlers’ temperaments predict their five-factor personality scores into later childhood and adolescence [36-39], which suggest some degree of stability in personality. However, little is known about whether these dimensions are present earlier in development, in infancy [36,40,41].

Few studies have examined infant well-being and whether there are stable intra-individual differences. One study found that parent reports of 1-year-old infants’ temperament predicted infants’ later life satisfaction as adults through 29 years of age [10]. This longitudinal study found that infants’ levels of positive affect, in particular, predicted their life satisfaction as adults, whereas infants’ negative affect, in contrast, did not predict any measures of adult well-being. Though limited, there is also some evidence of an association between infant personality and later subjective well-being. For example, one study found that the developmental trajectories of externalizing behavior problems (e.g., aggression, temper loss, noncompliance), from infancy (1.5 years old) to mid-adolescence (14 years old), are associated with well-being in young adults, at 18 years of age [42]. However, to our knowledge, there are no studies that have directly measured both well-being and personality in infancy to explore their development and relation to one another. In sum, we know little about the early emergence of subjective well-being in infancy or how it is related to other dimensions of infant personality. Nor do we understand the extent to which these dimensions may be adult-like, established early in development.

Value of animal models

Studies in animals may shed light on the early development of individual differences in personality and well-being, which are not unique to humans. Indeed, personality in animals is a well-established phenomenon across a wide range of species, including octopi [43], dogs [44], snakes [45], zebra finches [46], bees [47], and whales [48]. While animal studies of personality are interesting in their own right, they also widen our understanding of human personality through enabling approaches to questions that are difficult or impossible to answer with studies in humans [49-51]. Studies of personality in animals are necessary to clarify the phylogenetic history of specific traits, offering insights into their evolutionary origins [52,53]. Furthermore, animals are useful for developmental studies of personality, as many species exhibit more rapid development and shorter lifespans, making it possible to longitudinally measure personality over the lifespan in a shorter period of time and with less attrition [54].

Finally, in animals, there is more experimental control and manipulability, enabling more accurate measures of prenatal and postnatal contributions to personality and well-being, which would be ethically or practically difficult if not impossible in humans [55-59]. For instance, in humans, infant and parent well-being are linked, with infants’ temperaments affecting maternal well-being [60] and maternal well-being, during and after pregnancy, affecting children’s well-being [61]. Determining causal relations among these complex systems is challenging. Animal studies can overcome these limitations by enabling a high degree of experimental control over infants’ environments. By standardizing infants’ environments, for example, this can help disentangle environmental contributions to individual differences in personality or well-being observed in a specific sample. For example, in chimpanzees, infants who were raised in an environment that included less contact with conspecifics, compared to infants reared in more species-typical environments, displayed lower levels of extroversion later in life [62]. At the same time, caution is warranted when generalizing findings from one sample to another, when infants’ early environments vary substantially [63].

The Five-Factor Model has been adapted for a variety of nonhuman primate species, although the number and nature of the factors varies somewhat across species [64-68]. These studies often use an approach similar to that used with human infants, assessing personality and well-being through knowledgeable informants, such as animal care staff [69,70], which reveal strong levels of inter-observer agreement, and predictive validity of behaviors in various real-world contexts [49]. For example, rhesus macaques are reported to have six personality components: Confidence, Friendliness, Dominance (hereafter referred to as Assertiveness to avoid confusion with traditional measures of hierarchical dominance), Anxiety, Openness, and Activity [53]. Much like in humans, these personality dimensions are heritable [71,72] and are associated with specific patterns of behavior. For example, individuals rated higher in sociability (Extraversion), tend to engage in more affiliative interactions, whereas individuals higher in confidence tend to engage in more aggressive behaviors [73]. Furthermore, some of these rhesus macaque personality dimensions are associated with lifetime injury incidence [74] and well-being [53].

Subjective well-being has been reported as a valid measure in a variety of nonhuman primate species [30,75-79]. In studies of nonhuman animal subjective well-being, human raters are asked how often each animal is happy, how satisfied each animal is with their social relationships, how successful each animal is in achieving their goals, and to imagine how happy they would be if they were that animal for a week [80]. Such studies reveal that, in adult macaques, much like in humans and chimpanzees, higher confidence and friendliness, and lower anxiety are associated with higher subjective well-being [53]. However, we know little about the early emergence of subjective well-being in infancy or how it may relate to other dimensions of infant personality. Nor do we understand the extent to which these dimensions may be adult-like, established early in development.

Infant macaque personality

A barrier to understanding personality development in infancy is the lack of well-established measures. One approach involves placing infants into various situations (e.g., novel environment, person, or object), and measuring their behavioral and physiological reactions (e.g., stress-related behaviors, salivary cortisol). An example of this approach is the Brazelton Newborn Behavioral Assessment Scale (NBAS), which is the most common measure of temperament in human newborns and has also been adapted for macaque newborns (Infant Behavioral Assessment Scale, IBAS; [81]). The IBAS focuses primarily on neurological development, including sensory and motor abilities (e.g., reflexes, orienting). Similarly, the Biobehavioral Assessment (BBA) is designed to assess 3- to 4-month-old macaques’ behavioral and physiological responses to a variety of stressors over a 48-hour period [63]. While these approaches offer valuable insight into activity levels, irritability, and stress-related aspects of infants’ development, there remains a need to better capture more positive types of traits, such as infants’ curiosity, playfulness, and sociability. Both the IBAS and the BBA assessments are also costly and labor-intensive, requiring animal handling by trained staff, as well as intensive behavioral scoring by reliable observers.

An alternative method, which may give us a broader view of personality dimensions, is to use caregiver surveys [62]. In humans, parental surveys capitalize on the fact that caregivers have extensive observations of their infants across a wide variety of contexts and are therefore one of the easiest, most reliable, and most predictive measures of temperament [82,83]. Similarly, animal care technicians, who interact with individual animals daily over the course of many months, and sometimes many years, have been shown to provide animal personality ratings that are consistent across raters and over time [53, 67,77,84–87]. Early surveys of macaque infant personality were developed prior to the human Five-Factor Model, which has expanded our understanding of a wider variety of personality domains [87]. Previous studies, therefore, may not have captured all of the dimensions of infant macaque personality [53].

Current study

In the current study, we examined personality in infant rhesus macaque monkeys raised in a well-controlled, standardized, laboratory environment by human caretakers. This unique early environment offered a degree of experimental control and standardization that enabled us to observe natural variation in personality with limited environmental influences, a level of control impossible to achieve in human studies of personality and subjective well-being. We explored whether caretakers—who were intricately familiar with each infant through daily interactions since birth—could reliably rate infant macaque personality and subjective well-being. We also examined whether there were personality dimensions in infant monkeys that resemble those in adult monkeys (i.e., six component adult rhesus macaque structure: Table 1 in [53]). Finally, we tested whether infant personality dimensions are related to their well-being.

Materials and methods

Ethics statement

This study was purely observational. The Eunice Kennedy Shriver National Institute of Child Health and Human Development Animal Care and Use Committee approved all animal procedures. We conducted the study in accordance with the Guide for the Care and Use of Laboratory Animals and complied with the Animal Welfare Act. During the course of this study, infants were fed with Similac® Advance® (Abbott Laboratories) and, starting at 2 weeks old, Purina Monkey chow (#5054). Additional food enrichment, including fruits, seeds, and nuts, was introduced twice daily when infants were 2 months old. Water was available ad libitum. Infants’ housing was enriched by an inanimate surrogate mother covered with fleece fabric as well as blankets and various plastic and rubber toys, which were rotated daily. At the conclusion of data collection for the current study, infants continued to be housed in the nursery as part of ongoing, unrelated research studies until ca. 6–8 months of age, after which they were transferred to large peer groups. See [88] for further details on housing, enrichment, and feeding.

Subjects

We studied 55 healthy, full-term infant rhesus macaques (Macaca mulatta), including 29 females and 26 males born in four cohorts between April of 2013 and July of 2016: N = 18 (7 females) born in 2013, N = 10 (6 females) born in 2014, N = 16 (9 females) born in 2015, and N = 11 (7 females) born in 2016. Infants were housed in the Animal Care Center at the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health. Infants were rated when they were between 6.5 and 7.5 months old (196 to 225 days old; mean (SD) = 211 ± 9 days old). Infants were separated from their mothers on the day of birth (typically by 8am) and reared in a nursery facility by human caretakers for ongoing, unrelated research studies. Infants were housed in adjacent incubators (51 cm × 38 cm × 43 cm) for the first 2 weeks of life and in larger cages (65 cm × 73 cm × 83 cm) thereafter. Human caretakers were present for 13 hours each day and interacted with infants every 2 hours for feeding and cleaning purposes. In both housing arrangements infants could see and hear other infants.

In the first 5 weeks after birth, infants were singly housed and raised identically. Once the youngest infant reached 36 days of age, infants were placed into small, same-aged peer groups. Infants were randomly assigned to one of two rearing conditions for unrelated research studies: low-socialization infants (N = 27) and high-socialization infants (N = 28). Low-socialization infants continued to be individually housed but assigned to playgroups composed of 3 to 4 peers and put together for 2 hours a day, 5 days a week. High-socialization infants were raised in groups with 3 to 4 peers (for details: [88-90]). By 6 months of age, all infants had extensive experience with same-aged conspecifics. Between 6–8 months old, all infants from each year cohort were placed into one large peer group together with one adult male and several mother-reared infants born the same year (for details see [91]). Therefore, rearing experiences converged after this initial period of differential rearing.

We choose to assess personality and well-being in these infants while they were still in the nursery setting, with one-on-one interactions occurring daily between animal care staff and the infants. At the same time, we waited until infants were 7 months of to give the raters time to get to know their individual personalities.

Measures

Each infant monkey’s personality and subjective well-being was rated by two to three of the full-time animal care staff who worked with the animals. These six raters had observed and interacted with the infants since the day they were brought to the nursery. Raters were asked to make their judgments on the basis of their own understanding of each trait and the descriptions of each trait that were provided. They were instructed to use the monkey’s behaviors and interactions with other monkeys to make their ratings, considering their understanding of typical monkey behavior, to decide if a particular monkey is above, below, or average for each trait. In written and verbal instructions, each rater was instructed to keep their ratings private and not discuss their ratings with the other raters. Infant macaques (N = 55) were rated when they reached approximately 7 months of age by at least two raters (mean = 2.8 raters per subject; range 2 to 3 raters per subject). We collected data between November 2013 and February 2017. For the personality ratings, there were no missing items out of 8,316 items ratings; for the subjective well-being ratings there were no missing items out of 616 item ratings.

Personality

We measured personality using the Hominoid Personality Questionnaire (HPQ) [92], which is 54-item questionnaire where each item is made up of an adjective and 1–3 descriptive sentences. As an example, the item ‘gentle’ is presented as, “GENTLE: Subject responds to others in an easy-going, kind, and considerate manner. Subject is not rough or threatening.” Each item is followed by a 7-item Likert scale with answers ranging from 1 “Least: Displays either total absence or negligible amounts of the trait” to 7 “Most: Displays extremely large amounts of the trait.” The HPQ can be downloaded from [93].

Subjective well-being

Each macaque was rated on subjective well-being, a four-item questionnaire based on King and Landau’s questionnaire [80]. Using this questionnaire, raters were asked to answer questions on how often each animal is happy, how successful each animal is in achieving their goals, to imagine how happy they would be if they were that animal for a week, and to estimate how satisfied each animal is with their social relationships. Each question is followed by a 7-item Likert scale with answers ranging from “Displays either total absence or negligible amounts of the trait or state” to “Displays extremely large amounts of the trait.” The subjective well-being questionnaire can be downloaded from [94].

Data analysis

All analyses were conducted using R, version 3.5.1 [95]. Principal components analyses and parallel analysis were conducted using the psych package [96]. The R script is available in supporting materials.

Item interrater reliabilities

We used two intraclass correlations (ICC) to estimate interrater reliabilities among raters [97]. ICC(3,1) measures the reliability of individual ratings whereas ICC(3,k) measures the reliability of average ratings across k raters.

Principal components analyses

After the ICCs were performed all data were averaged across raters resulting in a single score for each animal, and these scores were used for all remaining analyses. To examine the structure of infant macaque personality we used a principal component analysis (PCA). We determined the numbers of components to extract by using a parallel analysis and examining the scree plot [98,99] using the fa.parallel function in the ‘psych’ package [96]. We followed the comparative personality methods outlined by Robinson et al. [100] and discussion by Weiss [101] and calculated two additional structures, a structure with one less component than recommended by the parallel analysis and a structure with one more component. We examined each structure with both varimax and promax rotations of the structure(s); if the correlations in the promax rotation were relatively low (below r = 0.40) then we used the varimax rotation.

Following this analysis, we next computed unit-weighted component scores [102], based on the derived structures and the published adult six component structure [53] where a weight of +1 was assigned to loadings that were greater or equal to .4 and a weight of -1 was assigned to loadings that were equal to or less than -.4; all other loadings were assigned weights of 0. If an item loaded at greater than or equal to |.4| on multiple components, then the item was assigned to the component on which it loaded the highest. We performed a single PCA and followed the statistical procedure to determine the structure derived from the four subjective well-being items.

Component interrater reliabilities

To check interrater reliability at the component level, we created unit-weighted component scores [102] of individual scores based on the results of the PCA again using ICC(3,1) and ICC(3,k).

Pearson’s correlations

We ran two sets of Pearson’s correlations. We first tested for associations between the infant macaque personality component scores and the previously published adult macaque personality component scores [53]. This approach allowed us to determine which structure most closely resembled the adult six component structure (Table 1 in [53]). In the second set of correlations, we tested for associations between the infant macaque personality component scores and the infant macaque subjective well-being component scores.

Results

Interrater reliabilities

Observers were found to agree on all four subjective well-being items and all but two HPQ items (Table 1). For the subjective well-being items the mean ICC(3,1) was 0.45 (SD ± 0.08, range = 0.34 to 0.52) and the ICC(3,k) was 0.69 (SD ± 0.07, range = 0.59 to 0.75). For the HPQ items the mean ICC(3,1) was 0.32 (SD ± 0.17, range -0.14 to 0.66) and the ICC(3,k) was 0.52 (SD ± 0.25, range -0.52 to 0.85). We excluded the HPQ items for which observers were unreliable—unperceptive and imitative—from further analysis.

Table 1

Interrater reliability of subjective well-being and hominoid personality questionnaire items.

Subjective Well-Being Item	ICC(3,1)	ICC(3,k)
Time animal is happy	0.52	0.75
Goal achievement	0.49	0.73
Happiness as animal	0.47	0.71
Social satisfaction	0.34	0.59
Subjective well-being average	0.45	0.69
Personality Item	ICC(3,1)	ICC(3,k)
Dominant	0.66	0.85
Timid	0.62	0.82
Submissive	0.61	0.82
Cautious	0.60	0.81
Aggressive	0.60	0.81
Bullying	0.56	0.78
Curious	0.54	0.77
Fearful	0.49	0.73
Playful	0.44	0.69
Inquisitive	0.44	0.69
Helpful	0.44	0.69
Anxious	0.44	0.68
Independent	0.43	0.68
Autistic	0.42	0.67
Active	0.42	0.67
Thoughtless	0.40	0.66
Stingy	0.40	0.65
Reckless	0.39	0.64
Individualistic	0.39	0.64
Persistent	0.38	0.63
Depressed	0.38	0.63
Affectionate	0.38	0.63
Manipulative	0.37	0.63
Cool	0.36	0.61
Stable	0.36	0.61
Impulsive	0.36	0.61
Solitary	0.33	0.58
Jealous	0.33	0.58
Vulnerable	0.33	0.58
Dependent	0.31	0.56
Sociable	0.31	0.56
Protective	0.29	0.54
Gentle	0.29	0.53
Distractible	0.28	0.53
Sympathetic	0.28	0.52
Friendly	0.27	0.51
Inventive	0.27	0.51
Lazy	0.26	0.50
Disorganized	0.23	0.46
Irritable	0.22	0.45
Decisive	0.21	0.42
Conventional	0.20	0.42
Innovative	0.18	0.38
Defiant	0.17	0.37
Erratic	0.16	0.35
Intelligent	0.15	0.33
Clumsy	0.14	0.32
Excitable	0.14	0.31
Sensitive	0.07	0.18
Unemotional	0.07	0.17
Quitting	0.05	0.13
Predictable	0.01	0.04
Unperceptive	-0.06	-0.18
Imitative	-0.14	-0.52
Hominoid Personality Questionnaire average	0.32	0.52

Intraclass correlations (ICC) were based on 55 rhesus macaques, the number of raters (k) ranged between 2 and 3, k = 2.8.

Principal components analyses

The parallel analysis and scree plot of the personality items suggested a five component structure (see Table 2), which accounted for 62% of total variance. The promax rotation (S1 Table) showed relatively low correlations between components (highest correlation = 0.35; S2 Table) therefore we decided to interpret the varimax rotated five component structure. We report the promax rotated four, five, and six component structures in S1 Table; see S2 Table for the component correlations. We report the varimax rotated four and six component structures in S3 Table. We also ran a factor analysis (see S4 Table) and compared the results to this structure using a congruence test. We found the results of both tests to be virtually identical (congruence = 1.00 across all corresponding components) and therefore continued with the PCA approach to be consistent with the method used in Weiss [53]. For the suggested five component structure the mean ICC(3,1) was 0.49 (SD ± 0.18, range 0.21 to 0.64) and ICC(3,k) was 0.71 (SD ± 0.17, range 0.42 to 0.83) (see S5 Table). The parallel analysis of the subjective well-being items suggested a single component structure (see S6 Table).

Table 2

Varimax rotated infant rhesus macaque structure.

Item	Infant Macaque Components						Corresponding Adult Component
	Openness	Assertiveness	Anxious*	Friendliness	Intellect*	h2
Curious	0.80	0.02	-0.18	0.17	-0.10	0.72	Openness +
Active	0.80	0.09	-0.02	-0.13	-0.09	0.68	Activity +
Inquisitive	0.78	-0.09	-0.23	0.10	-0.13	0.70	Openness +
Lazy	-0.78	-0.01	-0.23	-0.02	-0.04	0.66	Activity -
Playful	0.76	0.20	-0.16	0.17	-0.04	0.67	Activity +
Impulsive	0.67	0.20	0.08	-0.19	-0.41	0.70	Openness +
Depressed	-0.65	-0.08	0.27	-0.29	-0.13	0.60	Friendliness -
Reckless	0.63	0.39	-0.20	-0.02	-0.37	0.73	Dominant +
Distractible	0.62	-0.01	0.06	0.14	-0.51	0.68	Confidence -
Timid	-0.62	-0.26	0.50	-0.10	0.18	0.74	Confidence -
Innovative	0.59	0.19	-0.08	0.01	0.16	0.42	Openness +
Inventive	0.58	0.20	-0.25	0.08	0.06	0.45	Openness +
Thoughtless	0.57	0.00	-0.02	0.16	-0.56	0.67	Openness +
Cautious	-0.56	-0.36	0.36	-0.09	0.33	0.68	Confidence -
Persistent	0.54	0.51	-0.13	-0.16	-0.04	0.60	Friendliness +
Individualistic	0.47	0.22	0.21	-0.32	-0.21	0.47	Dominant +
Sensitive	-0.37	-0.21	0.27	0.13	0.21	0.32	Friendliness +
Aggressive	0.08	0.88	-0.11	-0.05	0.01	0.80	Dominant +
Bullying	0.09	0.87	-0.16	-0.03	0.02	0.79	Dominant +
Dominant	0.25	0.81	-0.23	0.03	0.14	0.78	Dominant +
Gentle	-0.02	-0.75	-0.17	0.27	0.08	0.66	Dominant -
Defiant	0.21	0.72	-0.04	0.11	-0.31	0.68	Dominant +
Stingy	0.16	0.72	-0.01	-0.07	-0.03	0.55	Dominant +
Submissive	-0.20	-0.72	0.40	0.06	-0.15	0.74	Confidence -
Manipulative	0.00	0.67	-0.09	0.20	0.11	0.51	Dominant +
Jealous	0.06	0.64	0.14	0.02	-0.14	0.45	Anxious +
Irritable	-0.48	0.59	0.17	-0.20	-0.22	0.69	Dominant +
Quitting	-0.19	-0.36	0.04	0.07	-0.13	0.19	Anxious +
Conventional	-0.34	-0.36	-0.29	0.30	0.29	0.50	Activity -
Cool	0.18	-0.03	-0.79	0.15	0.22	0.73	Anxious -
Stable	0.25	0.03	-0.77	0.19	0.09	0.70	Confidence +
Unemotional	-0.09	0.01	-0.71	0.01	-0.04	0.51	Anxious -
Anxious	-0.46	-0.12	0.68	-0.24	-0.08	0.75	Anxious +
Fearful	-0.52	-0.14	0.65	-0.13	0.11	0.75	Confidence -
Excitable	0.18	0.06	0.65	0.02	-0.26	0.53	Dominant +
Autistic	-0.13	-0.07	0.60	-0.18	-0.15	0.43	N/A
Vulnerable	-0.09	-0.50	0.59	0.11	-0.10	0.63	Confidence -
Affectionate	0.13	-0.13	-0.15	0.81	-0.09	0.73	Friendliness +
Helpful	0.21	-0.14	-0.36	0.71	-0.13	0.72	Friendliness +
Sympathetic	0.13	-0.27	-0.11	0.70	-0.03	0.60	Friendliness +
Protective	0.00	0.28	-0.20	0.66	-0.01	0.55	Friendliness +
Sociable	0.57	0.04	-0.15	0.64	-0.04	0.75	Friendliness +
Independent	0.29	0.06	-0.50	-0.62	0.03	0.72	Dominant +
Dependent	-0.32	-0.06	0.34	0.59	-0.08	0.58	Confidence -
Solitary	-0.47	-0.26	0.01	-0.54	0.01	0.58	Friendliness -
Friendly	0.39	-0.45	-0.27	0.51	-0.08	0.69	Friendliness +
Decisive	0.02	0.19	-0.22	-0.19	0.73	0.65	Friendliness +
Intelligent	0.24	0.17	-0.17	-0.11	0.68	0.58	Friendliness +
Disorganized	0.46	0.00	0.08	0.12	-0.67	0.69	Confidence -
Clumsy	0.02	0.02	-0.02	0.03	-0.59	0.35	Activity -
Erratic	0.20	0.33	0.19	-0.21	-0.53	0.52	Anxious +
Predictable	-0.23	-0.34	-0.17	0.16	0.45	0.43	Activity -
Proportion of variance	0.19	0.15	-0.11	0.09	0.08

N = 55. Salient loadings are in boldface.

*Indicates a component that has been reflected;

+ indicates a positive loading and

- indicates a negative loading with the corresponding adult personality component (if there is one).

N/A indicates the infant item has no corresponding adult structure component.

For the five component varimax rotated infant structure, the first component was comprised of items relating to curiosity, activity, and innovation such as, inquisitive, playful, and inventive; we named this component Openness. The second component was comprised of items relating to dominance traits such as bullying, aggressive, defiant, and manipulative; we named this component Assertiveness. The third component was comprised of items relating to anxiety and vigilance such as fearful, timid, excitable, vulnerable, and anxious; we named this component Anxiety. The fourth component was comprised of items relating to sociability such as affectionate, sympathetic, helpful, and friendly; we named this component Friendliness. The fifth component was comprised of items relating to decision making behavior such as intelligent, decisive, and predictable; as this dimension appeared to resemble orangutan Intellect (Table 3 in [79]) rather than any one adult rhesus macaque dimension, we named this component Intellect.

The four and six component structures accounted for 57% and 65% of total variance, respectively. These structures resembled that of the five component with the primary exception being the items that loaded onto Intellect. In the four component structure (S3 Table) four of the intellect items (disorganized, excitable, intelligent, erratic) loaded onto the Anxiety component with two items (predictable and clumsy) not loading onto any component. In the six component structure three of the Intellect items (erratic, predictable, and clumsy) loaded onto the fifth component, which we named Predictability. The two remaining Intellect items (decisive and intelligent) loaded onto the sixth component, which we called Intellect. The promax rotated four, five, and six component structures can be found in S1 Table; the varimax rotated four and six component structures can be found in S3 Table.

All four subjective well-being items loaded onto a single component (S4 Table). This structure accounted for 69% of variance. This result matches the structure found in adult rhesus macaques [53].

Pearson’s correlations

We tested for correlations between our infant macaque personality component scores and the component scores based on the published adult macaque personality structure (see Table 3 in [53]). Adult macaque Confidence was significantly correlated (ps < 0.05) with higher infant Openness (rs ≥ 0.65), higher infant Assertiveness (rs = 0.62), and lower infant Anxiety (rs ≥ -0.84) in the infant four, five, and six component structures and with higher infant Intellect (r = 0.72) in the infant six component structure (Table 3). Adult macaque Openness was significantly correlated with higher infant Openness (rs ≥ 0.94) in the four, five, and infant six component structures and with lower infant Intellect (r = -0.50) in the infant five component structure and lower infant Predictability in the infant six component structure (r = -0.47). Adult macaque Assertiveness was correlated with higher infant Openness (rs ≥ 0.52) and higher infant Assertiveness (rs = 0.97) in the infant four, five, and six component structures and higher infant Intellect (r = 0.53) in the infant six component structure. Adult macaque Friendliness was significantly correlated with higher infant Openness (rs ≥ 0.52) and infant Friendliness (rs ≥ 0.80) and lower infant Anxiety (rs ≥ -0.59) in the four, five, and six component structures. Adult macaque Activity was significantly correlated with higher infant Openness (rs ≥ 0.90) and infant Assertiveness (rs = 0.44) in the infant four, five, and six component structures and lower Intellect (r = -0.48) in the infant five component structure and lower Predictability (r = -0.54) in the infant six component structure. Adult macaque Anxiety was significantly correlated with higher infant Anxiety (rs ≥ 0.85) in the infant four, five, and six component structures, lower Intellect (r ≥ -0.45) in the infant five and six component structures.

Table 3

Pearson correlations for infant personality components based on the six component adult personality structure (from [53]).

Infant Structure	Adult Structure
	Confidence		Openness		Assertiveness		Friendliness		Activity		Anxiety
Four Component
Openness	0.67	[0.50,0.80]	0.95	[0.91,0.97]	0.52	[0.30,0.69]	0.52	[0.30,0.69]	0.90	[0.83,0.94]	-0.33	[-0.55,-0.08]
Assertiveness	0.62	[0.42,0.76]	0.23	[-0.04,0.46]	0.97	[0.94,0.98]	0.15	[-0.12,0.40]	0.44	[0.20,0.63]	-0.02	[-0.29,0.24]
Anxiety	-0.84	[-0.90,-0.74]	-0.25	[-0.48,0.02]	-0.30	[-0.52,-0.04]	-0.59	[-0.74,-0.38]	-0.19	[-0.44,0.08]	0.88	[0.80,0.93]
Friendliness	0.02	[-0.25,0.28]	0.22	[-0.04,0.46]	-0.21	[-0.45,0.06]	0.84	[0.74,0.90]	0.02	[-0.24,0.29]	-0.32	[-0.54,-0.06]
Five Component
Openness	0.66	[0.48,0.79]	0.95	[0.91,0.97]	0.53	[0.31,0.70]	0.52	[0.30,0.69]	0.91	[0.84,0.94]	-0.32	[-0.54,-0.06]
Assertiveness	0.62	[0.42,0.76]	0.23	[-0.04,0.46]	0.97	[0.94,0.98]	0.15	[-0.12,0.40]	0.44	[0.20,0.63]	-0.02	[-0.29,0.24]
Anxiety	-0.88	[-0.93,-0.80]	-0.42	[-0.62,-0.17]	-0.33	[-0.55,-0.07]	-0.64	[-0.78,-0.46]	-0.35	[-0.56,-0.09]	0.85	[0.75,0.91]
Friendliness	0.02	[-0.25,0.28]	0.22	[-0.04,0.46]	-0.21	[-0.45,0.06]	0.84	[0.74,0.90]	0.02	[-0.24,0.29]	-0.32	[-0.54,-0.06]
Intellect	0.13	[-0.14,0.38]	-0.50	[-0.68,-0.27]	-0.10	[-0.36,0.17]	0.02	[-0.25,0.28]	-0.48	[-0.66,-0.24]	-0.45	[-0.64,-0.21]
Six Component
Openness	0.65	[0.46,0.78]	0.94	[0.90,0.96]	0.53	[0.31,0.70]	0.54	[0.32,0.70]	0.90	[0.84,0.94]	-0.30	[-0.53,-0.04]
Assertiveness	0.62	[0.42,0.76]	0.23	[-0.04,0.46]	0.97	[0.94,0.98]	0.15	[-0.12,0.40]	0.44	[0.20,0.63]	-0.02	[-0.29,0.24]
Anxiety	-0.88	[-0.93,-0.80]	-0.42	[-0.62,-0.17	-0.33	[-0.55,-0.07]	-0.64	[-0.78,-0.46]	-0.35	[-0.56,-0.09]	0.85	[0.75,0.91]
Friendliness	-0.05	[-0.32,0.21]	0.17	[-0.10,0.42]	-0.30	[-0.52,-0.03]	0.80	[0.67,0.88]	-0.06	[-0.32,0.21]	-0.30	[-0.52,-0.04]
Predictability	-0.12	[-0.38,0.15]	-0.47	[-0.65,-0.23]	-0.32	[-0.54,-0.06]	0.01	[-0.25,0.28]	-0.54	[-0.71,-0.32]	-0.35	[-0.56,-0.09]
Intellect	0.72	[0.57,0.83]	0.33	[0.06,0.54]	0.53	[0.31,0.70]	0.24	[-0.03,0.48]	0.40	[0.15,0.60]	-0.50	[-0.67,-0.27]

Correlated are reported for infant macaque four, five, and six component personality structures. N = 55. Boldface correlations are statistically significant (ps < 0.05), and 95% confidence intervals are in brackets.

The subjective well-being component was significantly correlated with higher infant Openness (rs ≥ 0.71) and Assertiveness (r = 0.44) and lower infant Anxiety (rs ≥ -0.74) in the infant macaque four, five, and six component structures, ps < .05 (Table 4). Infant subjective well-being also positively correlated with adult Confidence (r = 0.78), Openness (r = 0.59), Assertiveness (r = 0.45), Friendliness (r = 0.69), and Activity (r = 0.60), and negatively correlated with adult Anxiety (r = -0.65) on the adult macaque six component structure.

Table 4

Pearson correlation of subjective well-being and infant and adult personality structures.

Structure	SWB component
	SWB	95% CI
Infant
Four Component
Openness	0.72	[0.56,0.83]
Assertiveness	0.44	[0.20,0.63]
Anxiety	-0.74	[-0.84,-0.58]
Friendliness	0.33	[0.08,0.55]
Five Component
Openness	0.71	[0.55,0.82]
Assertiveness	0.44	[0.20,0.63]
Anxiety	-0.80	[-0.88,-0.68]
Friendliness	0.33	[0.08,0.55]
Intellect	0.05	[-0.21,0.31]
Six Component
Openness	0.71	[0.54,0.82]
Assertiveness	0.44	[0.20,0.63]
Anxiety	-0.80	[-0.88,-0.68]
Friendliness	0.26	[-0.00,0.49]
Predictability	-0.01	[-0.28,0.26]
Intellect	0.50	[0.27,0.68]
Adult
Confidence	0.78	[0.72,0.90]
Openness	0.59	[0.39,0.74]
Assertiveness	0.45	[0.22,0.65]
Friendliness	0.69	[0.50,0.80]
Activity	0.60	[0.39,0.74]
Anxiety	-0.65	[-0.80,-0.51]

N = 55. Boldface correlations are significant at p< 0.05; subjective well-being (SWB) 95% confidence intervals (CI) are reported in brackets.

Discussion

We tested whether adult-like personality factors are already present in infancy in rhesus macaque monkeys. We found infant macaques have a five component personality structure, based on caregiver ratings of 52 traits that observers showed agreement on: Openness (e.g., curiosity, inquisitive, playfulness), Assertiveness (e.g., dominance, bullying, aggressive), Anxiety (e.g., vigilance, fearful), Friendliness (e.g., sociable, affectionate, sympathetic), and Intellect (e.g., intelligent, decisive). These components are largely analogous to the six components in adult macaques—Openness, Assertiveness, Anxiety, Friendliness, Confidence, and Activity—although we also found some differences between adult and infant personality. These components in infant macaques are also similar to those reported in human children as young as 2 to 3 years of age, described as the “Little Six”: Openness, Agreeableness, Neuroticism, Extraversion, Conscientiousness, and Activity [36,40,41]. Furthermore, we found links between personality and well-being: Infants’ subjective well-being positively correlated with Openness and Assertiveness and negatively correlated with Anxiety, similar to findings in adult macaques [53]. These findings suggest stable structural personality components within this species.

Interrater reliabilities of personality and subjective well-being

We found that all but two of the 54 HPQ items (unperceptive and imitative) and all four of the subjective well-being items were reliable among raters. Observers were not reliable on ratings of the item unperceptive, similar to results of observer ratings of adult macaques [53]. Previous studies reported that traits related to Extraversion have the highest levels of interrater agreement in both humans and animals, while traits related to Neuroticism have high levels of agreement in animals, but not humans, and traits related to agreeableness have the lowest levels of interrater agreement in both humans and animals [49]. For subjective well-being, our interrater reliabilities were also excellent, and comparable to those reported in adult macaques [53]. Together, these findings suggest that observers agreed on their ratings of infant macaque personality and well-being.

Personality component structures in infant macaques

The five personality constructs we found in infant macaques—Openness, Assertiveness, Anxiety, Friendliness, and Intellect—appear similar to those reported in adult macaques [53], as well as other nonhuman primates [67,69,78,79,103] and human children [36,40,41]. Next, we outline each personality component that we detected in infant macaques and consider the similarities in these components with age and across species.

Infant macaques exhibit a component, which we call Openness, which may be similar to Surgency/Extraversion temperament structure reported in human infants and adult macaques, which refers to infants’ tendency to exhibit energetic activity, positive affect, and high intensity pleasure [50], sometimes referred to as Surgency/Sociability (vs. Shyness/Inhibition) in children [41,104-106]. This component may be similar to the Openness component reported in 2- to 3-year-old children, which includes curiosity and exploring, love of learning, and interest in experiencing new things [36,40,41]. In adult macaques, higher levels of Openness are associated with better cognitive performance [107], so this may be an interesting personality component to study developmentally as it relates to learning.

Infant Assertiveness seems to mirror adult Assertiveness [53]. We decided to name this infant component Assertiveness, rather than Dominance, to avoid confusion with hierarchical dominance. Given the strict dominance hierarchy that rhesus macaques live in [108], it is unsurprising that traits relating to these behaviors would show up in infancy. Similar individual differences in aggressiveness have been reported in human infants and preschoolers, described as being low in Agreeableness, and high levels are sometimes described as having a “difficult temperament” (for a review, see [109]). In macaques, higher levels of assertiveness are associated with social success [110] and visible in facial morphology as a social signal [111], so it may be important to study this personality component in relation to the development of social behaviors and skills. While the infant macaques in the present study were separated from their mothers, disrupting the usual rank inheritance transfer through social experiences [110,112], we still found this to be a distinct component, underscoring its potential importance.

Both infants and adults have a similar Anxiety component [53]. The Anxiety component we found here may be similar to the Negative Affectivity temperament structure reported in human and macaque newborns, which reflects an infant’s tendency to experience negative emotions and distress [50]. Infant macaque Anxiety may also be similar to the Neuroticism component reported in human infants [36,40,41], and similar to Fearfulness in adult macaques [113]. In fact, in macaques, Fearfulness is reported to be one of the most stable personality traits across the first 7 years of life [113]. Higher levels of neuroticism are linked to a range of poor health outcomes, so a better understanding of its developmental origins is of significant clinical relevance (for a review, see [114]).

In macaques, both infants and adults have a component called Friendliness [53]. This personality component appears to be similar to the Sociability dimension reported in adult macaques, associated with being affiliative, warm, and less solitary [115], and the Extraversion component reported in human 2- to 3-year-olds [2,41,104-106]. Sociality is a core individual difference in primates, reported across a wide range of species [53,59,68]. Although we did not test for sex differences in Friendliness in the present study, sex differences in social behaviors are reported in both human and macaques, with females generally showing higher levels of social interest than males already in early infancy [88]. Understanding the causes and consequences of infants with low levels of sociability may help with the development of animal models to study developmental disorders, especially those that disproportionately affect males, such as autism [116,117].

We also found a dimension in infant macaques that was not apparent in adults: Intellect. The infant macaque Intellect component included items related to intelligence, such as being more thoughtful and more decisive, while being less distractible and less clumsy. In adult macaques, the items on this component load across Friendliness, Confidence, Activity, and Anxiety. Instead, this component appears to more closely resemble orangutan Intellect, with which it shares four of its six items [79]. Our findings suggest that this component—Intellect—may not be species-specific, found only in orangutans [80], but may be shared with other primates, at least at some points in development. This Intellect component also shares some similarities with the human toddler dimension Conscientiousness, which includes thoughtfulness, attentiveness, concentration, and planning [41]. Self-regulation, in particular, appears to be a core component of Conscientiousness in human children [118].

In the six component structure we found the items in the Intellect component from the five component structure were split into two components that were comprised of items relating to: (1) a Predictability component (predictable, not erratic, and not clumsy), and (2) an Intellect component (intelligent, organized). While we did not find Predictability in the five component structure, it did appear in the six component structure, suggesting that it may be an emerging component of personality, but may not be as stable as the other components at this age.

The current study offers novel insights into human infant personality and well-being and highlights future directions for research in human infants. Our findings in macaques suggest that there may already be well-established adult-like dimensions of personality detectable in infancy through caregiver report. To our knowledge, no studies, to date, have attempted to measure human adult personality (i.e., the “Big Five”) in human infants, despite reports of similar dimensions (i.e., the “Little Six”) in toddlers and young children (2 to 5 years of age) [36,40,41]. It may therefore be worthwhile to explore whether such personality components may be detected earlier in human infants as well. Such research may help to bridge the gap between studies of infant temperament and studies of adult personality [41], which have historically relied on different instruments.

Subjective well-being in infant macaques

We found that all four subjective well-being items loaded onto a single component, similar to adult macaques [53], orangutans [79], Western lowland gorillas [78], chimpanzees [80], and brown capuchins [76]. We also found that infant macaque subjective well-being positively correlated with Openness, Assertiveness (Dominance), and Friendliness, and negatively correlated with Anxiety. In adult macaques, higher confidence and friendliness, and lower anxiety are associated with higher subjective well-being [53], and similar patterns have been reported in chimpanzees [30], and humans [119]. In human adults, individuals who report higher levels of subjective well-being also tend to be lower in Neuroticism (associated with anxiety) and higher in Extraversion (associated with friendliness) [120]. These results suggest that well-being may be related to personality in similar ways across the lifespan and across primate species.

There are not yet well-established measures of human infant subjective-well being. The measure of subjective well-being used here may be adapted for use with human infants, and offers a number of advantages over previous measures. For example, in addition to including questions about positive emotions, this measure also includes questions about goal-achievement (“Estimate, for your infant, the extent to which he/she is effective or successful in achieving his/her goals or wishes”), infants’ experiences of social interactions (“Estimate the extent to which social interactions with other people are satisfying, enjoyable experiences as opposed to being a source of fright, distress, frustration, or some other negative experience”), and asks raters to imagine themselves as the infant (“Imagine how happy you would be if you were your infant for a week. You would be exactly like your infant. You would behave the same way as your infant, would perceive the world the same way as your infant, and would feel things the same way as your infant.”). This measure of well-being may be adapted for use with human infants and validated through comparisons with other similar measures, such as parent survey-based measures of infant positive affect [10, 11] and by comparing directly with infant behaviors (e.g., smiles, laughter, positive vocalizations). Validating a measure of subjective well-being in human infants could facilitate studies of the early developmental emergence of this construct, its stability across the lifespan, test potential associations between parent and infant well-being before and after birth, and associations between infant subjective well-being and personality. Recent studies on human infant subjective well-being report that parents’ ratings of infant positive, but not negative affect, predict adult life satisfaction [10] as well as cognition in childhood and educational attainment in adulthood [121]. These studies suggest that subjective well-being in infancy may lay the foundation for later success in across numerous domains.

Limitations and future directions

A limitation of the present study is that it included a relatively small sample of infant macaques reared in a neonatal nursery by humans. Future studies are needed to expand this research to larger and more diverse populations, including those socially reared in laboratories, zoos, field stations, the wild, and other contexts, to test the generalizability of these findings. At the same time, this nursery rearing is a strength of the current study because, despite being raised in nearly identical environments, we still found six different personality dimensions and variation among individuals in these dimensions. In addition, the infants in the current study grew up in very different environments relative to those of previous studies in wild populations (Weiss et al., 2011), and yet our findings are largely similar. Together, these findings suggest that individual differences in these personality factors are unlikely to be exclusively due to variation in infants’ postnatal environments, but rather, are more likely due (at least in part) to differences in infants’ prenatal environment and/or genetics. The present study, therefore, offers fundamental insight about personality development, revealing its early ontogenetic roots.

The present study is also limited in that we only focused on one age group using a cross-sectional approach. Infant macaques at 7 months old are approximately equivalent to 2-year-old human infants, given that they are estimated to develop roughly four times faster than human infants, in their cognitive and brain development [122,123]. Given how little is know about infant personality development, both in human and nonhuman primates, future studies are needed in both younger and older infant macaques to determine how personality emerges. Studies in nonhuman primate infants will be instrumental in uncovering how infant temperament interacts with the early environment to shape development over time [124]. In addition, comparative studies exploring personality changes with age will be fruitful. For example, one study found chimpanzees and humans showed remarkable similarities in changes to their personality dimensions across the lifespan, from adolescence into adulthood [84]. Ideally, in future work, researchers could follow infants longitudinally and repeatedly sample their personality as they grow up, into adulthood, to better capture from birth through adulthood, across the lifespan, to further test the developmental stability of these dimensions [113]. A better understanding of infant primate personality may enable us to better identify infants who are outliers in specific traits associated with developmental disabilities (e.g., [125]), which may enable better animal models of human disabilities [117]. Through better understanding the development of personality, we may be able to enhance infants’ development by pre-screening them to identify infants at risk of developing problems and helping them overcome temperament-related challenges, and by training caregivers to better align their responses to fit each child’s characteristics [7,8,126].

Conclusions

Our findings suggest that, in macaques, infant personality dimensions may be conceptually related to adult personality. Given that infant macaque are a popular model of human development, it is critical to understand the ways in which macaque personality may be similar to, or different from, that in humans. Further research is necessary to explore the antecedents, predictive validity, and stability of these personality components across situations and with development. Animal studies of personality can bring unique insights to the biological mechanisms that underlie personality development, including their causes and developmental plasticity. Considering that macaque infants are often studied as a model of human infant development, it is critical to understand the ways in which macaque infants may be similar to, or different from, human infants, in terms of personality. Nonhuman primate models of infant development offer unique insights about the development of personality and subjective well-being, widening our view of individual differences and their early emergence.

Infant macaque code.

R code for statistical analysis that we used in the current study.

(R)

Click here for additional data file.

Infant macaque personality data.

Blinded raw data from the current study.

(XLSX)

Click here for additional data file.

Promax rotated infant rhesus macaque four and six component models.

Salient loadings are in boldface. N = 55. *Indicates a component that has been reflected.

(XLSX)

Click here for additional data file.

Promax-rotated component correlations of four, five, and six component structures.

Correlations from S1 Table promax-rotations.

(XLSX)

Click here for additional data file.

Varimax rotated infant rhesus macaque four and six component models.

Salient loadings are in boldface. N = 55. *Indicates a component that has been reflected.

(XLSX)

Click here for additional data file.

Factor analysis with varimax rotation of infant rhesus macaque structure.

N = 55. Salient loadings are in boldface. *Indicates a component that has been reflected. Proportion of variance = 58%.

(XLSX)

Click here for additional data file.

Interrater reliability of components derived from infant macaque five component structure.

Note. Based on 55 rhesus macaques. k = 2.8.

(XLSX)

Click here for additional data file.

PCA of infant rhesus macaque subjective well-being items.

N = 55. Proportion of variance = 69%.

(XLSX)

Click here for additional data file.

15 Oct 2019

PONE-D-19-26868

Infant rhesus macaque (Macaca mulatta) personality and subjective well-being

PLOS ONE

Dear Dr. Simpson,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

There are several major issues you have to adress, principally:

- Data reduction used: PCA vs. FA

- Reliability of some of the items

- Discussion and the interpretation of the results

We would appreciate receiving your revised manuscript by Nov 29 2019 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Miquel Llorente, PhD

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The overarching focus in this paper is in identifying the personality structure of a set of infant rhesus monkeys that were essentially nursery-reared – some had 2 hrs/day of peer socialization (while living alone) and the others lived in peer groups. Ratings were done when the animals were about 7 months of age. The authors then did a principal components analysis (PCA) to identify a structure to their data, and computed scores on 4, 5, and 6 dimensions. They also computed scores based on a published study of adult monkeys (Weiss et al., 2011), and compared the two sets of scores using Pearson product-moment correlations. The authors’ analysis suggests adult personality has definite antecedents in young animals. (There is a second instrument used in this study, one that assesses subjective well-being; this analysis does not really seem to belong in this paper, and could probably be eliminated.)

There are two major issues that I believe are problematic in this paper.

First, on line 357, the authors state: “We found that all but two of the 54 HPQ items (unperceptive and imitative) and all four of the subjective well-being items were reliable among raters.” Inspection of Table 1, however, indicates that several items on the personality inventory had extremely low reliabilities, the worst being Predictable, whose values are .01 and .04. A psychometrician would hardly call these items “reliable,” and if reliability is a criterion for inclusion in a PCA, then the authors need to justify their decision for including items whose ICC values are extremely low.

Second, the authors use the inappropriate principal components procedure to identify the personality dimensions. PCA is a data reduction technique; the more appropriate technique is factor analysis, which is aimed explicitly at identify the latent variables that explain the observed data. While I understand that there are many papers in the animal personality literature that have used PCA instead of FA, that does not make the practice acceptable. There are many resources in the literature and online describing the differences between PCA and FA, and which technique should be used under which circumstances. In this case, the goal of identifying personality factors (ie, latent traits) makes PCA the wrong technique. The authors might look at a paper by Costello and Osborne, who present a very accessible discussion of this and other relevant issues. (Costello, A.B. and J.W. Osborne. 2005. Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research & Evaluation 10:1–9.)

Two more minor issues are:

References to the supplementary tables are incorrect. In the paragraph starting on line 270, the second line indicates Table S1 is the result of the promax rotation, but the next sentence indicates S1 contains the varimax rotation. Similarly, the correlations of the promax rotated factors is indicated as S3, but I find them in S2 instead. In general, the authors need to insure the references to the tables are correct and accurate.

In line 446ff, the authors suggest “that these six personality differences are unlikely to be due to infants’ postnatal environments, but rather, are more likely due to differences in infants’ prenatal environment and/or genetics.” It’s unclear what they authors are trying to say here. If what they mean is that their six factors were derived from animals with unusual rearing histories, and yet they seem to parallel the structure derived from adult rhesus living in very different circumstances (the Weiss et al paper), then this should be more clearly stated. If, however, what they are suggesting is that postnatal factors do indeed play little role in the development of personality, then I would consider that an overstatement (i.e. I would disagree with the statement “The present study, therefore, offers fundamental insight about personality development, revealing that it does not start after birth, but long before”). I simply don’t see that they have results that support that statement. In fact, they do have animals from two different rearing conditions – Do those two sets of animals differ from each other on these factors? If so, then presumably that is due to the postnatal environment. Moreover, earlier, they indicated their analysis suggested a five-factor solution. Why are they now interpreting the six-factor solution? Finally, their phrase “these six personality differences” seems inaccurate; they are not referring to actual differences, but rather individual difference *factors*.

Reviewer #2: This study tests an impressive number of infant monkeys to assess their personality, using the HPQ, and their wellbeing. However, the methodology suffers from a number of limitations, some of which the authors raise themselves in the Discussion. Given these limitations, these results must be taken with caution, especially their applicability to other populations or monkeys that have been mother reared in typical environments. Indeed, we know that atypical rearing can have long term consequences on the personality expression of primates (e.g., Freeman et al 2016 Developmental Psychobiology), as well as other long-term effects (e.g., Capitanio et al 2006 Nursey Rearing and Biobehavioral Organization), and I think this needs to be better recognized.

Additionally, this study is framed as being novel in testing the personality of young macaques. However, at the California Primate Research Center, such evaluations have been conducted for many years, testing thousands of young macaques (e.g., Capitanio, 2017, Variation in Biobehavioral Organization). This should be more clearly acknowledged.

Finally, and as I note below, it is proposed that macaques offer a good animal model for studying the development of personality. Given this, I would have appreciated greater consideration of how these results related to other aspects of the monkeys’ behavior (to provide a fuller perspective) and how these results relate to what is known about human personality development.

Here are my more specific comments are they arise throughout the article:

Given that your title refers to macaques, when I was reading the opening few lines of your paper it was unclear whether you were referencing literature on human or macaque/primate infants. I suggest you explicitly state that you are referring to human infant research. Furthermore, given this opening on human infant literature, I think your Discussion could also benefit from greater consideration about how these results relate to what we know for human infants and children and whether there are parallels between the species. Given that your proposal that macaques are a good model species for this ontogenetic research, it would be beneficial to hear your conclusions about that in the Discussion.

I do not think the subtitles in the introduction are needed, nor in the Discussion, unless this is a requirement of the journal, I would suggest you omit them.

Line 178 – In your methods, when describing the macaques’ rearing experience, I infer that for the first 5 weeks of their life the macaques are singly housed. Please state this explicitly for clarity.

Line 187 – how many raters total did you include?

Line 215 – why did you rate the monkeys at 7 months old? Is this related to a particular developmental milestone? Please provide more detail about the decision behind this sampling point. This information is key in better understanding the relevance and importance of your results when understanding the development of macaques. Therefore, I think more background information about the ontogeny of macaques in general, and how that maps onto human development, would be helpful given your framing that you are using macaques as a model species to understand human personality development.

Line 215 – please provide the range of the number of raters per subject as well as the mean. I see this is provided with Table 1, but I think it would be helpful to have it presented in the text too.

Do you have any physiological, cognitive or behavioral data that were collected at the time that you could use to further validate these personality measures, especially the measure of wellbeing? For example, were those rate with better wellbeing also those with better body condition, lower cortisol, more responsive to tests of cognition etc.? I think these kind of additional data would really strengthen your data and conclusions. Were those monkeys rated high on the intellect factor also those that performed better in concurrent (or later) cognitive tasks. Please include such meta data if you have it.

Is it possible to get ratings on these monkeys now that they are older to see how these infant ratings translated to when these monkeys were older (sensu Weinstein and Capitanio 2012 J Comp Psychol)?

Line 447 – “to more larger” sounds awkward, I suggest “to larger and more diverse”

Line 454 – how can you conclude “The present study, therefore, offers fundamental insight about personality development, revealing that it does not start after birth, but long before”? You tested monkeys seven months after birth so how do you know that personality development does not start at birth?

I found the raw data set an additional materials. However, I think it would be beneficial if you could also provide your R script along with your data set for full transparency of methods.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

23 Oct 2019

Reviewer #1: The overarching focus in this paper is in identifying the personality structure of a set of infant rhesus monkeys that were essentially nursery-reared – some had 2 hrs/day of peer socialization (while living alone) and the others lived in peer groups. Ratings were done when the animals were about 7 months of age. The authors then did a principal components analysis (PCA) to identify a structure to their data, and computed scores on 4, 5, and 6 dimensions. They also computed scores based on a published study of adult monkeys (Weiss et al., 2011), and compared the two sets of scores using Pearson product-moment correlations. The authors’ analysis suggests adult personality has definite antecedents in young animals. (There is a second instrument used in this study, one that assesses subjective well-being; this analysis does not really seem to belong in this paper, and could probably be eliminated.)

Response: Our analyses are not only focused on the existence of a personality structure in infant macaques, but also on how the personality constructs we discovered relate to subjective well-being. This additional analysis gives us predictive validity of the instrument as used with infant macaques as our findings mirror similar findings in adult macaques. Therefore, we chose to retain the subjective well-being component of this study.

There are two major issues that I believe are problematic in this paper.

1. First, on line 357, the authors state: “We found that all but two of the 54 HPQ items (unperceptive and imitative) and all four of the subjective well-being items were reliable among raters.” Inspection of Table 1, however, indicates that several items on the personality inventory had extremely low reliabilities, the worst being Predictable, whose values are .01 and .04. A psychometrician would hardly call these items “reliable,” and if reliability is a criterion for inclusion in a PCA, then the authors need to justify their decision for including items whose ICC values are extremely low.

Response: We adjusted our wording throughout the manuscript to reflect observer agreement rather than reliability. It is a requirement that observers agree on their ratings, which they do.

A psychometrician was consulted from design through to revision of this study. He said, “They're [the interrater reliabilities] ratios, so [low reliabilities] just means a lot of variance that's not true-score variance compared to the true score variance. That doesn't mean there's no true score variance, though, and you can tell because, if there was no true score variance, the items would not load on factors/components. It would also mean that, when getting ICCs of your scales (the factor scores), it would show no increase, but it does.” See Reise & Henson (2003) for a review of true score variance.

Interrater reliabilities of scale items are typically in line with repeatabilities of behavior. For further information, see papers by Bell et al (2009) and McCrae and Mottus (2019).

Bell, A. M., Hankison, S. J., & Laskowski, K. L. (2009). The repeatability of behaviour: a meta-analysis. Animal Behaviour, 77(4), 771-783.

McCrae, R. R., & Mõttus, R. (2019). A new psychometrics: What personality scales measure, with implications for theory and assessment. Current Directions in Psychological Science.

Reise, S. P., & Henson, J. M. (2003). A discussion of modern versus traditional psychometrics as applied to personality assessment scales. Journal of Personality Assessment, 81(2), 93-103.

Velicer, W. F., & Jackson, D. N. (1990). Component analysis versus common factor analysis: Some issues in selecting an appropriate procedure. Multivariate Behavioral Research, 25(1), 1-28

2. Second, the authors use the inappropriate principal components procedure to identify the personality dimensions. PCA is a data reduction technique; the more appropriate technique is factor analysis, which is aimed explicitly at identify the latent variables that explain the observed data. While I understand that there are many papers in the animal personality literature that have used PCA instead of FA, that does not make the practice acceptable. There are many resources in the literature and online describing the differences between PCA and FA, and which technique should be used under which circumstances. In this case, the goal of identifying personality factors (ie, latent traits) makes PCA the wrong technique. The authors might look at a paper by Costello and Osborne, who present a very accessible discussion of this and other relevant issues. (Costello, A.B. and J.W. Osborne. 2005. Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research & Evaluation 10:1–9.)

Response: Regarding the point about FA versus PCA, we are not in agreement with the reviewer here as the field has not reached agreement on the preferred method, likely because they both give extremely similar results. To quote Velicer and Jackson (1990), “Because factor score estimates are nearly identical to component scores, and are themselves determinate, it is difficult to understand how factor scores can be better and more generalizable.” FA and PCA have both been used in animal personality, even within the same paper. For example, Weiss et al. (2006) reported in Footnote 4 that they performed PCA and FA and found a virtually identical structure.

Given the reviewer’s comments, we ran a FA on the five factor structure and ran a test of congruence comparing it to the structure given using PCA and found the following congruences:

RC1 RC2 RC3 RC4 RC5

MR1 1.00 0.32 0.39 0.20 0.34

MR2 0.32 1.00 0.19 -0.17 0.13

MR3 0.40 0.18 1.00 0.26 -0.20

MR4 0.21 -0.16 0.26 1.00 0.07

MR5 0.37 0.13 -0.21 0.08 1.00

Given this and the fact that we’re comparing our structure with the published adult rhesus macaque structure, which was calculated using PCA, we have chosen to retain the PCA approach and have made the following addition to reflect this: “We also ran a factor analysis (see S4 Table) and compared the results to this structure using a congruence test. We found the results of both tests to be virtually identical (congruence = 1.00 across all corresponding components) and therefore continued with the PCA approach to be consistent with the method used in Weiss [53].” (Lines 352-355)

Supplementary Table 4 includes the results of the FA.

Two more minor issues are:

3. References to the supplementary tables are incorrect. In the paragraph starting on line 270, the second line indicates Table S1 is the result of the promax rotation, but the next sentence indicates S1 contains the varimax rotation. Similarly, the correlations of the promax rotated factors is indicated as S3, but I find them in S2 instead. In general, the authors need to insure the references to the tables are correct and accurate.

Response: Thank you for bringing this to our attention. This has now been corrected in the manuscript.

4. In line 446ff, the authors suggest “that these six personality differences are unlikely to be due to infants’ postnatal environments, but rather, are more likely due to differences in infants’ prenatal environment and/or genetics.” It’s unclear what they authors are trying to say here. If what they mean is that their six factors were derived from animals with unusual rearing histories, and yet they seem to parallel the structure derived from adult rhesus living in very different circumstances (the Weiss et al paper), then this should be more clearly stated. If, however, what they are suggesting is that postnatal factors do indeed play little role in the development of personality, then I would consider that an overstatement (i.e. I would disagree with the statement “The present study, therefore, offers fundamental insight about personality development, revealing that it does not start after birth, but long before”). I simply don’t see that they have results that support that statement. In fact, they do have animals from two different rearing conditions – Do those two sets of animals differ from each other on these factors? If so, then presumably that is due to the postnatal environment. Moreover, earlier, they indicated their analysis suggested a five-factor solution. Why are they now interpreting the six-factor solution? Finally, their phrase “these six personality differences” seems inaccurate; they are not referring to actual differences, but rather individual difference *factors*.

Response: We revised this section for clarity (Lines 584-593). Unfortunately, we do not have a large enough sample size in the current study to examine different rearing groups within our infant sample, but we agree that this is an interesting future direction.

1. Reviewer #2: This study tests an impressive number of infant monkeys to assess their personality, using the HPQ, and their wellbeing. However, the methodology suffers from a number of limitations, some of which the authors raise themselves in the Discussion. Given these limitations, these results must be taken with caution, especially their applicability to other populations or monkeys that have been mother reared in typical environments. Indeed, we know that atypical rearing can have long term consequences on the personality expression of primates (e.g., Freeman et al 2016 Developmental Psychobiology), as well as other long-term effects (e.g., Capitanio et al 2006 Nursey Rearing and Biobehavioral Organization), and I think this needs to be better recognized.

Response: Thank you for drawing our attention to these relevant publications, which we now reference in our introduction, emphasizing the value of animal models for enabling experimental manipulations to infants’ early rearing environments, while also acknowledging such variation is a limitation to generalizability of studies that focus only on one specific type of sample (nursery-reared infants), such as the present study (Lines 147-159). However, given that our findings in human-reared infant monkeys are largely similar to those reported in adult wild populations (Weiss et al., 2011), we think some claims of generalizability of these basic personality factors across different populations with various early environments are not unreasonable.

2. Additionally, this study is framed as being novel in testing the personality of young macaques. However, at the California Primate Research Center, such evaluations have been conducted for many years, testing thousands of young macaques (e.g., Capitanio, 2017, Variation in Biobehavioral Organization). This should be more clearly acknowledged.

Response: We added to our introduction a brief review of previous studies on infant macaque personality and clarify how the present study is novel in its approach to assessing personality exclusively using caretaker surveys developed for adult macaques to assess infants’ personality dimensions (Lines 145-169).

3. Finally, and as I note below, it is proposed that macaques offer a good animal model for studying the development of personality. Given this, I would have appreciated greater consideration of how these results related to other aspects of the monkeys’ behavior (to provide a fuller perspective) and how these results relate to what is known about human personality development.

Response: We agree that an important next step is to consider whether these personality dimensions are related to individual differences in behavior, as we mention in our discussion. Unfortunately, we do not have such data to include in the present study. We added discussion of how these results relate to what is known about human infant development throughout (e.g., Lines 497-499, 509-512, 519-527, 551-573).

Here are my more specific comments are they arise throughout the article:

4. Given that your title refers to macaques, when I was reading the opening few lines of your paper it was unclear whether you were referencing literature on human or macaque/primate infants. I suggest you explicitly state that you are referring to human infant research. Furthermore, given this opening on human infant literature, I think your Discussion could also benefit from greater consideration about how these results relate to what we know for human infants and children and whether there are parallels between the species. Given that your proposal that macaques are a good model species for this ontogenetic research, it would be beneficial to hear your conclusions about that in the Discussion.

Response: Thank you for pointing out these ambiguities. We added “human” and “nonhuman primate” to clarify the species throughout our introduction.

We have also added to our Discussion, as suggested, additional links between the findings of the present study and the study of human infant personality (e.g., Lines 497-499, 509-512, 519-527, 551-573).

5. I do not think the subtitles in the introduction are needed, nor in the Discussion, unless this is a requirement of the journal, I would suggest you omit them.

Response: Some of the subheadings have been removed from the Introduction and Discussion sections, as recommended. We retained some headings as required by the journal.

6. Line 178 – In your methods, when describing the macaques’ rearing experience, I infer that for the first 5 weeks of their life the macaques are singly housed. Please state this explicitly for clarity.

Response: Yes, that is correct. We added that information to the “Subjects” section (Line 222).

7. Line 187 – how many raters total did you include?

Response: We clarify that we had six raters (Line 239).

8. Line 215 – why did you rate the monkeys at 7 months old? Is this related to a particular developmental milestone? Please provide more detail about the decision behind this sampling point. This information is key in better understanding the relevance and importance of your results when understanding the development of macaques. Therefore, I think more background information about the ontogeny of macaques in general, and how that maps onto human development, would be helpful given your framing that you are using macaques as a model species to understand human personality development.

Response: Thank you for pointing out this oversight. We added our justification for why we choose this age group (Lines 232-235) and also discuss what the “equivalent” approximate age is for human infants (Lines 595-597): “Infant macaques at 7 months old are approximately equivalent to 2-year-old human infants, given that they are estimated to develop roughly four times faster than human infants, in their cognitive and brain development [122,123].”

9. Line 215 – please provide the range of the number of raters per subject as well as the mean. I see this is provided with Table 1, but I think it would be helpful to have it presented in the text too.

Response: We now include this information (Line 250).

10. Do you have any physiological, cognitive or behavioral data that were collected at the time that you could use to further validate these personality measures, especially the measure of wellbeing? For example, were those rate with better wellbeing also those with better body condition, lower cortisol, more responsive to tests of cognition etc.? I think these kind of additional data would really strengthen your data and conclusions. Were those monkeys rated high on the intellect factor also those that performed better in concurrent (or later) cognitive tasks. Please include such meta data if you have it.

Response: No, unfortunately additional physiological, cognitive, or behavioral data on these infants are not available. We agree that these are important future directions.

11. Is it possible to get ratings on these monkeys now that they are older to see how these infant ratings translated to when these monkeys were older (sensu Weinstein and Capitanio 2012 J Comp Psychol)?

Response: No, unfortunately we no longer have access to these individuals.

12. Line 447 – “to more larger” sounds awkward, I suggest “to larger and more diverse”

Response: We corrected this typo.

13. Line 454 – how can you conclude “The present study, therefore, offers fundamental insight about personality development, revealing that it does not start after birth, but long before”? You tested monkeys seven months after birth so how do you know that personality development does not start at birth?

Response: We are not claiming that personality development does not start at birth. We reworded this section to avoid confusion (Lines 584-593): “Together, these findings suggest that individual differences in these personality factors are unlikely to be exclusively due to variation in infants’ postnatal environments, but rather, are more likely due (at least in part) to differences in infants’ prenatal environment and/or genetics. The present study, therefore, offers fundamental insight about personality development, revealing its early ontogenetic roots.”

14. I found the raw data set an additional materials. However, I think it would be beneficial if you could also provide your R script along with your data set for full transparency of methods.

Response: We have now included the R script in our supporting materials.

12 Nov 2019

PONE-D-19-26868R1

Infant rhesus macaque (Macaca mulatta) personality and subjective well-being

PLOS ONE

Dear Dr. Simpson,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands yet. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process, specially adressing the comments by reviewer 2.

Although I personally consider that the manuscript improved considerably, issues regarding FA vs. PCA and reliability vs. agreement have to be adressed before our final acceptation.

We would appreciate receiving your revised manuscript by Dec 27 2019 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Miquel Llorente, PhD

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This is a revision of a paper focused on comparing scores on personality ratings of infant monkeys with scores using a component analysis conducted on adults of the same species, with a goal of seeing how the newly-derived infant components compare with the adult components. My previous comments focused on two main issues.

1. Reliability of individual items. The authors responded to my original concern that many items show low reliabilities in two ways.

First, the authors respond that they have altered the wording in multiple places to refer to these values as “agreement” and not “reliability.” This is incorrect. Agreement and reliability are different constructs, and in fact, the authors’ use of ICC(3,1) and ICC(3,k) reflect their focus on *consistency* in reliability, not agreement. Consider an example of three individuals rating five children on a trait. Person 1 gives ratings of 1, 2, 3, 4, 5. Person 2 gives ratings of 2, 3, 4, 5, 6, and Person 3 gives ratings of 3, 4, 5, 6, 7. Absolute agreement in these ratings = 0. The data are, however, quite consistent: for these data, ICC(3,1)=.714, and ICC(3,k)=.882. So to refer to the results of their reliability analysis as reflecting inter-rater agreement is incorrect.

Their second response seems to suggest that as long as an ICC is >0, there is some true score variance present, and they argue that the fact that these items loaded at all onto a factor is proof that, even with low reliabilities, they have value and so are retained. However, it’s generally considered that a measure is reliable if it *mostly* reflects true score variance, and not error variance. In the present case, 1% of the variance in Predictable is true score variance, and 99% is error. But “error” is not the same as “random error.” Using items with low true score variance can lead to spurious relationships owing to some other systematic source of variance embedded within the 99% error component of this item. Unfortunately, there are no hard and fast guidelines for where the line should be drawn so that items above the line are considered “reliable” and those below are considered “unreliable.” I have never seen that line at 1%, however, which is where the current authors are placing it. The authors will need to provide additional justification for their decision.

2. Principal Components Analysis (PCA) vs. Factor Analysis (FA). The idea that there are underlying latent traits that inform the display of behavior is fundamental to the idea of personality. Historically, the way these traits have been identified is through FA, although many in the ethological world use PCA instead. The underlying assumptions of these two procedures is different, as any text on factor analysis will indicate. The authors provide a reference by Velicer and Jackson (1990) as evidence that it really doesn’t make much difference in many cases between PCA and FA. However, the very next paper in that issue, by Gorsuch, does take this view to task. And contrary to the authors’ assertions, in human psychology anyway, FA does indeed seem to be the preferred method. The use of PCA in the animal literature is likely a hold-over historically (PCA is computationally easier than FA, but with the use of computers these days, this issue is moot) and disciplinary (PCA has its roots more in biology where data reduction – the principal reason for PCA – was the goal, not the discovery of latent traits). The authors did do a FA on their data, and a congruence analysis suggested the factor structure was virtually identical. This is reasonable; PCA and FA can, in many cases, lead to similar structures. The results of this analysis are presented in a Supplementary Table. Given that the present authors are trying to replicate a result from an earlier paper by Weiss that used PCA, provision of the FA results in a supplementary table is a sufficient response to my earlier critique.

Reviewer #2: Thank you for responding to all of my comments and suggestions and for providing greater theoretical context for your work in relation to human infant/child personality.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

21 Nov 2019

Reviewer #1

1. Reliability of individual items. The authors responded to my original concern that many items show low reliabilities in two ways.

First, the authors respond that they have altered the wording in multiple places to refer to these values as “agreement” and not “reliability.” This is incorrect. Agreement and reliability are different constructs, and in fact, the authors’ use of ICC(3,1) and ICC(3,k) reflect their focus on *consistency* in reliability, not agreement. Consider an example of three individuals rating five children on a trait. Person 1 gives ratings of 1, 2, 3, 4, 5. Person 2 gives ratings of 2, 3, 4, 5, 6, and Person 3 gives ratings of 3, 4, 5, 6, 7. Absolute agreement in these ratings = 0. The data are, however, quite consistent: for these data, ICC(3,1)=.714, and ICC(3,k)=.882. So to refer to the results of their reliability analysis as reflecting inter-rater agreement is incorrect.

1. Response: Given this feedback, we have adjusted the wording to interrater reliability, where appropriate, as previously written.

2. Their second response seems to suggest that as long as an ICC is >0, there is some true score variance present, and they argue that the fact that these items loaded at all onto a factor is proof that, even with low reliabilities, they have value and so are retained. However, it’s generally considered that a measure is reliable if it *mostly* reflects true score variance, and not error variance. In the present case, 1% of the variance in Predictable is true score variance, and 99% is error. But “error” is not the same as “random error.” Using items with low true score variance can lead to spurious relationships owing to some other systematic source of variance embedded within the 99% error component of this item. Unfortunately, there are no hard and fast guidelines for where the line should be drawn so that items above the line are considered “reliable” and those below are considered “unreliable.” I have never seen that line at 1%, however, which is where the current authors are placing it. The authors will need to provide additional justification for their decision.

2. Response: We thank the reviewer for this thoughtful comment. The reviewer’s comment doesn’t take into account the fact that we aggregate all the ratings that reach 0.01 and above, which makes them more reliable and stable. Rushton, Brenerd, and Pressley speak to this exact point in their 1983 paper, when they state:

“Many important variables in behavioral development are presumed to be unrelated because of repeated failures to obtain substantial correlations. In this article, we explore the possibility that such null findings have often been due to failures to aggregate. The principle of aggregation states that the sum of a set of multiple measurements is a more stable and representative estimator than any single measurement. This greater representation occurs because there is inevitably some error associated with measurement. By combining numerous exemplars, such errors of measurement are averaged out, leaving a clearer view of underlying relationships.”

And also,

“According to the principle of aggregation, the sum of a set of multiple measurements is a more stable and unbiased estimator than any single measurement from the set. One reason is that there is always error associated with measurement. When several measurements are combined, these errors tend to average out, thereby providing a more accurate picture of relationships in the population. Perhaps the most familiar illustration of this effect is the rule in educational and personality testing that the reliability of an instrument increases as the number of items increases (e.g., Gulliksen, 1950; Lord & Novick, 1968).”

Regarding error variance they state,

“A more accurate picture is obtained by using the principle of aggregation and examining the predictability achieved from a number of measures. To reiterate, this effect occurs because the randomness in any one measure (error variance) is averaged out over several measures, leaving a clearer view of what a person's true behavior is like.”

Rushton, J. P., Brainerd, C. J., & Pressley, M. (1983). Behavioral development and construct validity: The principle of aggregation. Psychological Bulletin, 94(1), 18-38.

Regarding the reviewer never having seen the bar set to 0.01, we provide the following examples of papers where the limit is set to 0.01 and above:

Bergvall, U. A., Schäpers, A., Kjellander, P., & Weiss, A. (2011). Personality and foraging decisions in fallow deer, Dama dama. Animal Behaviour, 81(1), 101-112.

"The interrater reliabilities of 47 of the items were greater than 0 and used in the analysis.”

Lee, P. C., & Moss, C. J. (2012). Wild female African elephants (Loxodonta africana) exhibit personality traits of leadership and social integration. Journal of Comparative Psychology, 126(3), 224-232.

“The ICC (3, k) for two adjectives (apprehensive =-5.72 and tense =-2.57) were highly negative and so these were dropped from analyses. ICC for other adjectives ranged from 0.72 (slow, playful) to 0.02 (confident).”

Úbeda, Y., & Llorente, M. (2015). Personality in sanctuary-housed chimpanzees: A comparative approach of psychobiological and penta-factorial human models. Evolutionary Psychology, 13(1), 182-196.

“The ICCs for the single (3, 1) and average (3, k) ratings were generally strong and there were no unreliable coefficients equal to or less than zero to eliminate from the analysis, indicating that raters tended to agree in their judgments about the personality traits of the chimpanzees.”

*The paper we used to make our adult personality comparisons: Weiss, A., Adams, M. J., Widdig, A., & Gerald, M. S. (2011). Rhesus macaques (Macaca mulatta) as living fossils of hominoid personality and subjective well-being. Journal of Comparative Psychology, 125(1), 72.

“We excluded the items autistic and unperceptive from further analysis as unreliable because their interrater reliabilities were less than zero.”

Although these studies include items above 0.0, one paper (Úbeda & Llorente, 2015) reported results that show two personality surveys produce similar results, another paper (Bergvall et al., 2011) showed that ratings correlated with behavioral tests and observations and foraging behavior (e.g., boldness predicted eating novel food), and a third paper (Weiss et al., 2011) demonstrated reliability across individuals and correlations with another measure, subjective well-being. Taken together, these findings suggest that the inclusion of aggregated items, even when reliability is low, do not impede the ability to predict other measures. Furthermore, all the items that we found were above 0.0, with the exception of autistic, were also reported to be reliable in the Weiss et al. study on the same species, with two primary differences being that they studied adult rhesus macaques and had more raters. This again shows that inclusion of these items is valid for comparison with this other study and that they have reliability across studies, even if they were low in ours.

Finally, the interrater reliability of the components, including all items above 0.0, is as follows for the parallel analysis suggested five-component infant structure:

Item icc31 icc3k n_obs n_sub n_rat k

Openness5 0.6393677 0.8323313 154 55 6 2.8

Assertiveness5 0.6412943 0.8334954 154 55 6 2.8

Anxiety5 0.4812548 0.72204 154 55 6 2.8

Friendliness5 0.4920122 0.7305994 154 55 6 2.8

Intellect5 0.2081923 0.4240335 154 55 6 2.8

These results suggest that the inclusion of items 0.0 and above does not hamper the reliability at the component level. We’ve now included this table in the supplementary materials and referenced it in the paper (lines 277-279 and 321-323).

Given that there are no hard and fast guidelines, as this reviewer points out, and given that numerous previous studies in this area have used this approach, we therefore retained our interrater reliability analysis results.

3. Principal Components Analysis (PCA) vs. Factor Analysis (FA). The idea that there are underlying latent traits that inform the display of behavior is fundamental to the idea of personality. Historically, the way these traits have been identified is through FA, although many in the ethological world use PCA instead. The underlying assumptions of these two procedures is different, as any text on factor analysis will indicate. The authors provide a reference by Velicer and Jackson (1990) as evidence that it really doesn’t make much difference in many cases between PCA and FA. However, the very next paper in that issue, by Gorsuch, does take this view to task. And contrary to the authors’ assertions, in human psychology anyway, FA does indeed seem to be the preferred method. The use of PCA in the animal literature is likely a hold-over historically (PCA is computationally easier than FA, but with the use of computers these days, this issue is moot) and disciplinary (PCA has its roots more in biology where data reduction – the principal reason for PCA – was the goal, not the discovery of latent traits). The authors did do a FA on their data, and a congruence analysis suggested the factor structure was virtually identical. This is reasonable; PCA and FA can, in many cases, lead to similar structures. The results of this analysis are presented in a Supplementary Table. Given that the present authors are trying to replicate a result from an earlier paper by Weiss that used PCA, provision of the FA results in a supplementary table is a sufficient response to my earlier critique.

3. Response: We are pleased that the reviewer is happy with our response and accepts our use of PCA in this paper.

Regarding latent variables, Velicer and Jackson also respond to this in their 1990 paper by stating, “If the latent variable approach results in greater generalization under sampling, the factor analyses on the subsample should be closer to the population than the component analysis. Velicer (1972), using several existing well known data sets, found no observable difference.” and “Latent variable procedures (i.e., factor analysis) are asserted to have greater generalizability to the set of unsampled variables and thus represent an advantage for factor analysis. However, empirical studies (Velicer, 1974; Velicer & Fava, 1987, 1990) have generally found no difference between the methods under conditions of variable sampling. This issue does not represent a basis for selecting either method of analysis.”

Reviewer #2

1. Thank you for responding to all of my comments and suggestions and for providing greater theoretical context for your work in relation to human infant/child personality.

1. Response: We thank this reviewer for the positive comments and are happy that the reviewer is satisfied with our changes.

6 Dec 2019

Infant rhesus macaque (Macaca mulatta) personality and subjective well-being

PONE-D-19-26868R2

Dear Dr. Simpson,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Miquel Llorente, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: (No Response)

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: (No Response)

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: (No Response)

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: (No Response)

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The remaining issue pertained to inclusion of individual items with low inter-rater reliability in the authors’ PCA. The authors’ response, namely to invoke the Principle of Aggregation, really is tangential to the issue – yes, aggregating individual items will result in a more reliable measure. The issue is *which* items to aggregate. The authors agree there are no hard and fast rules. I will simply note that if one contrasts the ICC values for the personality items in Table 1, comparing the values for those items that loaded in their PCA (n=45) with those items that did not (n=3: conventional, sensitive, quitting), that significant differences (both p<.01) exist for both ICC items. This suggests, at least to me, that individual level reliabilities *do* matter when selecting measures for PCA/FA.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

10 Dec 2019

PONE-D-19-26868R2

Infant rhesus macaque (Macaca mulatta) personality and subjective well-being

Dear Dr. Simpson:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Miquel Llorente

Academic Editor

PLOS ONE