Examining Early Childhood Dimensions of Emotional Awareness Using a Novel Coloring Task

Abstract

Emotional awareness supports emotion regulation. Psychologists have children “color in feelings” to assess emotional awareness, yet whether this relates to emotion regulation is unknown. The present study used a novel coloring task examining behaviors related to coloring in and dictating emotions to assess children’s (N = 95) emotional awareness. Furthermore, it was tested whether performance on this task predicted emotion regulation. Children’s coloring behaviors indicating physiological emotional awareness predicted better emotion regulation. Results may inform the use of emotional awareness tasks in community and clinical settings. Findings also suggest that physiological emotional awareness may be a more salient clinical target in children.

Keywords:

• Early childhood
• emotional awareness
• emotion regulation

Introduction

Emotional awareness, the ability to understand one’s own and others’ emotions (Lane & Schwartz, 1987), is a heterogeneous construct that comprises separate physiological and cognitive dimensions. Presently, many treatment protocols have children color in physiological sensations of emotions and generate emotional cognitions, based on an assumption that these facets of emotional awareness connect to children’s emotion regulation (Ehrenreich-May et al., 2018; Kendall & Hedtke, 2006). However, it is unknown whether a child’s performance on these tasks predicts their emotion regulation abilities, and whether some features are better predictors than others. The present study examined how 3.5 to 5-year-old children’s performance on a novel coloring task, modeled after emotional awareness exercises used in clinical practice, related to parent-rated emotion regulation. In what follows, we review emotional awareness as a construct, how child clinicians typically assess this ability, and gaps in our understanding of how to interpret children’s performance on these exercises.

Dimensions of emotional awareness

Emotional awareness is a heterogeneous construct that comprises separate physiological and cognitive dimensions. We define physiological emotional awareness as awareness of bodily sensations during emotions and cognitive emotional awareness as awareness of how thoughts are associated with emotions, specifically, how thoughts represent what is emotionally perceived and experienced. Lane and Schwartz (1987) were among the first to contend that physiological emotional awareness is a core dimension within the overarching emotional awareness construct. Since then, few studies have examined physiological emotional awareness across the lifespan, mostly in adults (Herbert et al., 2011) or older children (Rieffe et al., 2008) via questionnaire methods. In contrast, most of what is known about cognitive emotional awareness stems from the treatment literature. In Cognitive Therapy, for example, cognitions, emotions, motivations, and behaviors are postulated to work together, and treatments teach patients to become more aware of how cognitions (e.g., thoughts, self-narrations, mental images) can be influenced by emotional states and vice versa (Beck & Weishaar, 1989). Much of the research on cognitive emotional awareness has focused on its role in maintaining symptoms of psychopathology (Hankin et al., 2005), for example, the connection between rumination and depression (Whisman et al., 2020).

The relation between emotional awareness and emotion regulation

Researchers have hypothesized that emotional awareness supports emotion regulation capacity across the lifespan (e.g., Izard et al., 2008; Van Beveren et al., 2019). Van Beveren et al.’s (2019) suggested that emotional awareness is a specific skillset that is necessary before learning adaptive emotion regulation strategies. Indeed, commonly used interventions for childhood anxiety (e.g., problem solving training, exposure therapy; Kendall & Hedtke, 2006) have distinct modules early in the treatment sequence designed to probe and improve children’s physiological and cognitive emotional awareness. In these early modules, children may be instructed to illustrate physiological and cognitive responses to varying emotional states, such as coloring in emotions in cartoons or coming up with hypothetical emotion-based thoughts on worksheets. Children then transition from these emotional awareness modules to sessions focused on emotion regulation training. Although these early childhood treatments are efficacious in reducing symptoms of psychopathology (Lenz, 2015), a distinct link between emotional awareness and emotion regulation in young children remains, to our knowledge, untested. Currently, emotional awareness research in preschool-aged children is limited. Most recent work examining emotional awareness in youth utilize adolescent and middle-childhood samples. Of the limited work that has examined emotional awareness in early childhood, researchers predominantly focused on different, more basic features of emotional awareness, such as having children self-report valence and intensity of emotions using a series of paired comparisons between faces (Warren & Stifter, 2008). Indeed, it cannot be assumed that distinct physiological and cognitive emotional awareness dimensions are necessarily measurable in preschool children, given that these abilities are hypothesized to be nascent and under rapid development in early childhood (Cole et al., 2009). Basic empirical work on early emotional awareness that reflects how the construct is measured in therapy is needed to elucidate both the early development of emotional awareness and its clinical efficacy.

The present study

The goal of the present study was to test how preschool children’s demonstrations of physiological and cognitive emotional awareness related to individual differences in parent-rated emotion regulation. Children aged 3.5–5 years old were randomly assigned to a novel emotional awareness coloring task or control coloring condition. During the emotional awareness task, which was modeled closely after how a therapist might examine emotional awareness in a preschool child within a treatment setting, a research assistant prompted children to think of how a cartoon character was feeling, color in where the character might feel those emotions in their body, and then generate what emotion-related thoughts the character might have had. We hypothesized that characteristics of children’s coloring of physiological sensations (e.g., number of colors, amount of coloring) and generated cognitions (e.g., number of emotion-specific words) in the emotional awareness condition would be associated with parent-rated emotion regulation for both the negatively-valanced and positively-valanced cartoons.

Method

Participants

Ninety-five children aged 3.5 to 5 years old (M_age = 4.58 years, SD = .75; 47.4% female, 52.6% male; 74.7% White, 9.9% Black/African American, 2.2% Asian, 9.9% Multi-Racial, 3.3% did not report; 16.8% Hispanic or Latinx, 83.2% Not Hispanic or Latinx) were recruited from the community through advertisements on social media platforms and through flyers delivered to local preschools. Parents of participants reported their family’s annual income as follows: under $25,000 (11.6%), $25,001–$50,000 (28.4%), $50,001–$75,000 (14.7%), $75,001–$100,000 (22.1%), $100,001–$250,000 (15.8%), more than $250,001 (1.1%), and 6.3% chose not to respond. The present study was part of a larger study examining neural, physiological, and behavioral indicators of irritability and emotion regulation in a preschool sample (Grabell et al., 2019). Exclusionary criteria were history of head trauma and loss of consciousness (given overarching study had a neuroimaging component), preschool-level English, absence of developmental or intellectual impairments to participate, and absence of psychotic symptoms.

Procedure

Pilot study power analyses and prior work (Grabell et al., 2019) were used to determine the sample size of the present study. One child had missing expressive vocabulary data due to child’s refusal to complete the Expressive Vocabulary Test (EVT). Additionally, one child had missing coloring control data for the negative cartoon and two children had missing coloring control data for the calm cartoon because they did not color and thus could not be given a score. Caregiver informed consent was collected in person before data collection began. Families were given $60 for participation, and children were given a small toy. Ethical approval was received from the university’s institutional review board.

Emotion elicitation

Children first played a computer game called “Incredible Cake Kids” designed to elicit negative and positive emotions (Mateo Santana & Grabell, 2022), which were then explored in the emotion awareness task described below. Before playing, children watched a short video of a virtual baker named Gus, who told the child they would need to help take over the bakery while he was out of town. The video instructed the child that every time a customer came into the bakery, the child would see three cake options and it was the child’s job to choose the “most delicious cake.” The experimenter told the child that choosing the most delicious cake was an objective skill that some children are better at than others, and that they would be evaluated on while they played the game. Children first completed a practice version of the game to ensure they understood how to play. After practicing, children completed 30 trials of the Incredible Cake Kids game. During each trial, a cartoon customer would come onto the screen with three cartoon cakes (see Figure A1 in Appendix A). Children had 4 s to touch the most delicious cake, followed by 2 s of “anticipation,” then 2 s of feedback, and finally 2 s of rest. Feedback, unbeknownst to the child, was predetermined. Positive feedback customers smiled and indicated liking their cake (e.g., “Yummy!), and negative feedback customers frowned and responded with disgust (e.g., “Oh no!”). If the child did not choose a cake within the 2 s, an empty cake tray appeared indicating the customer did not receive a cake. The experimenter would then tell the child they needed to choose faster. The 30 trials were divided into three negative and three positive blocks, and each block was separated by 20 s rest periods; negative blocks contained four negative trials and one positive trial, whereas the positive blocks contained four positive trials and one negative trial. The entire task lasted approximately 10 min.

Figure 1. Example cartoons completed by participants. Left image is an example of the negative cartoon completed by a 3.5-year-old girl. Right image is the relaxation cartoon completed by a five-year-old girl.

Novel emotional awareness task

Immediately after the Incredible Cake Kids task, the researcher placed a child-sized table in front of the participant and told them they were going to “make some cartoons together.” Participants were shown two cartoons of a gender-matched, race indeterminate child character (see Figure 1). As described below, we designed the task after commonly used emotional awareness modules in child therapy that ask children to draw circles around body parts they feel different emotions in, and other modules that ask children to write and discuss what other children may be thinking in different situations (Kendall & Hedtke, 2006). Therefore, in the current study, child friendly techniques, such as displacement, and exploring physiological sensations of emotion through coloring, prompted children to demonstrate their understanding of physiological and cognitive emotional awareness. Given the very young age of the participants, as a validity check that the emotion-specific instruction changed the way children approached the task (as opposed to them not understanding the instructions), children were randomly assigned to the novel emotional awareness task (N = 62) and a non-emotion control version (N = 33).

Physiological emotional awareness measurement

The first cartoon depicted a child character playing the Incredible Cake Kids game in which the character picked the wrong cake and saw a “grumpy” customer on the screen. The rationale for this cartoon was for all participating children to recall an immediate, standardized, identical negative emotional challenge. Participants were first asked how they thought the cartoon character felt (e.g., “When the boy/girl saw that grumpy customer, how do you think s/he felt?”), followed by a series of closed-ended questions (i.e., “Do you think s/he felt angry?” “Do you think s/he felt sad?” “Do you think s/he felt worried?”) to give children multiple opportunities to articulate their thinking. The experimenter then instructed the child to color in where the cartoon character felt those emotions in his/her body with the colored markers using the same emotion words the child had articulated (i.e., “Pick out mad colors to color in where the child was feeling mad inside his/her body”). To prevent biasing children to adopt one particular coloring style, instructions were open-ended. Children were simply asked to “color in where the character is feeling those emotions,” and received no critical or corrective feedback so as not to influence their coloring style while they were coloring (see Appendix A for full script). As the child was coloring, the experimenter was trained to narrate and describe what the child was doing, reflect back anything the child said, and offer labeled praise, modeled after the standardized interactions used in Parent Child Interaction Therapy (PCIT; Galanter et al., 2012). For children who were not as engaged in the task, experimenters would regularly encourage the child to participate by asking a prompting question (e.g., “I wonder where the child was feeling those emotions”) throughout the interval. Children had 2 min to color in emotions.

Cognitive emotional awareness measurement

Next, children generated thoughts the cartoon character might have had during the scene. Based on the emotions the child identified previously for the particular cartoon, the experimenter asked the child what type of emotional thoughts the character might be having (e.g., “What type of mad thoughts do you think s/he had?”). The experimenter would reflect what the participant dictated out loud and then would write the words or phrase in the cartoon character’s thought bubble. The experimenter would summarize the phrase one more time after it was written in the thought bubble. The participants were prompted for additional thoughts for 2 min or until the child said they were done.

After the negative cartoon, the participant was shown a cartoon depicting the same gender-matched character wearing his/her bathing suit, floating in an inner tube. The rationale for this cartoon was to prompt the child to think about emotional awareness while feeling calm and relaxed, rather than high-arousal positive emotions such as excitement. The goal of the broader study was to examine how children transition from anger to relaxation during deliberate emotion regulation. Again, for the relaxation cartoon, the child colored emotions for 2 min and generated emotion thoughts for 2 min. Children received similar standardized interactions from the experimenter as during the negative cartoon. Children were instructed to rest for 30 s between cartoons.

Control condition

As a validity check, we randomly assigned a subset of participants to a control condition. The purpose of the control condition was to determine whether the emotion-specific instructions in the emotional awareness condition were understood and had an impact on children’s performance. That children would respond to the novel emotional awareness instructions could not be assumed given the sparse literature on early emotional awareness, and the fact that young children’s fine motor and vocabulary abilities are nascent (Conti-Ramsden & Durkin, 2012; Paris et al., 2019). Children in the control condition were given the same cartoons, in the same order, and with the same timing parameters. They also received narration, reflection, and labeled praise from the experimenter as a standardized form of interaction modeled by PCIT (Galanter et al., 2012). However, children in the control group were provided with no emotion-centered instructions. Control condition children were told to color in the cartoon any way s/he liked and come up with whatever words s/he wanted to be written in the thought bubble. Performance by children in the control condition could be compared to children in the emotional awareness condition to determine if there are differences based on the emotion content of the instructions provided to the latter group.

Emotional awareness coding

All cartoons were coded by undergraduate research assistants. Each undergraduate research assistant was trained by a laboratory coordinator. Coders needed to achieve and maintain a reliability coefficient of .7 or above with the other coders.

Physiological emotional awareness

We operationalized physiological emotional awareness based on the characteristics of how children colored emotions inside the cartoon character’s body. Specifically, we measured the number of different color types the child used, the number of body parts colored, and rated how controlled their coloring was (i.e., 0 = low, 1 = medium, 2 = high control). We also observed that some children colored highly specific areas within parts of the body (e.g., a smaller, clearly defined circle within the stomach), which we defined as “localized coloring.” The rationale for these coloring variables was that they were simple, easy to code, and therefore translatable to future research, clinical, and real-world settings (e.g., daycares, schools, homes) with minimal training. Thirty-six percent of participants’ physiological emotional awareness variables were double-coded; all coders were blind to hypotheses of the study, which group the child was assigned to, and was not in the laboratory when the child was tested and so did not have information on the child’s temperament. Intra-class correlations (ICCs) revealed excellent inter-rater reliability for number of colors used (ICC = .99), number of body parts colored (ICC = .99), and coloring control (ICC = .85). The localization variable, assessed using Cohen’s kappa due to the dichotomous nature of the variable, showed good inter-rater reliability (average kappa = .74).

Cognitive emotional awareness

We operationalized cognitive emotional awareness by counting the total number of words and number of emotion words that the child verbalized during the thought bubble prompt. Coders were trained to recognize emotion words (e.g., happy, mad, sad, calm) and the use of “like” or “love” when used as a synonym to an emotion word (e.g., “I really don’t like when I mess up”). Repeated instances of the same emotion word were counted as one emotion word. Ninety-nine percent of thought bubble transcripts were double-coded, and intra-class correlation was used to assess inter-rater agreement, which was excellent (ICC = .89).

Emotion regulation

All caregivers assessed their children’s temperamental emotion regulation using the Child Behavior Questionnaire’s (CBQ; Rothbart et al., 2001) Soothability and Anger/Frustration subscales. The CBQ examines kids’ emotion regulation by assessing their responses to frustration. Both subscales are scored on a seven-point Likert scale (e.g., 1 = “extremely untrue” to 7 = “extremely true”). The Soothability subscale consists of six items to assess a child’s ability to return to baseline following an emotional challenge (e.g., “Changes from being upset to feeling better within a few minutes”). The Anger/Frustration subscale consists of six items assessing how angry or frustrated children become following various stressors (e.g., “Gets quite frustrated when prevented from doing something s/he wants to do”). The CBQ has been shown to be a reliable and valid assessment of parent-reported child temperament (Putnam & Rothbart, 2006). In the present sample, reliability of the Soothability (α = .78) and Anger/Frustration (α = .83) subscales were acceptable.

Analysis strategy

We first examined descriptive statistics and inter-correlations for variables of interest in both groups, including testing whether these variables related to expressive vocabulary ability and age as potential covariates (Streubel et al., 2020). Next, we conducted a Multivariate Analysis of Covariance (MANCOVA) to test whether the control and emotional awareness groups differed in their task performance. To test our hypothesis that individual differences in emotional awareness coloring and thought generation would positively relate to parent-reported temperamental emotion regulation, we ran bivariate correlations separated by group. For any significant correlations in the emotional awareness condition, r to z transformations were used to assess whether correlation coefficients in the emotional awareness condition were significantly different from corresponding coefficients in the control group. In order to balance the risk of Type I and Type II errors we report uncorrected values. All analyses were run on IBM SPSS Statistics 27.

Results

Preliminary analyses

Descriptive statistics of all study variables are presented in Table 1. An initial one-way ANOVA showed no group differences for gender. Group differences were found for age, F(1, 93) = 3.91, p = .01, and expressive vocabulary, F(1, 92) = .05, p = .05, as assessed by the Expressive Vocabulary Test (EVT; Williams, 2007). Therefore, all subsequent group analyses controlled for age and expressive vocabulary. Preliminary bivariate correlations within each group revealed EVT scores were significantly associated with coloring control (r = .35, p = .05) in the negative cartoon for the control condition and with the sum of localized coloring (r = .27, p = .04) in the negative cartoon for the emotional awareness group. EVT scores were not significantly linked to any other variables of interest for the control or the emotional awareness conditions. Age was associated with number of colors in the negative cartoon and number of words in both the negative and relaxation cartoons (number of colors: r = .44, p = .01; number of words negative cartoon: r = .39, p = .03; number of words relaxation cartoon: r = .51, p = .002) in the control condition. Age was also associated with number of words and coloring control in the negative cartoon (number of words: r = .33, p = .01; coloring control: r = .31, p = .01), as well as with number of words, localized coloring, and coloring control in the relaxation cartoon (number of words: r = .45, p < .001; localized coloring: r = .27, p = .03; coloring control: r = .30, p = .02) in the emotional awareness group. We thus controlled for expressive vocabulary and age in all correlational tests.

Table 1. Correlations and descriptive statistics for all variables.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
1. Age (years)	1
2. Sex (1 = M)	−.02	1
3. Sooth	.19	.25*	1
4. Ang/Frus	−.002	−.11	−.53**	1
5. EVT	.10	.10	.17	−.13	1
6. Local color (N)	.14	.02	−.04	−.06	.16	1
7. Num colors (N)	.22*	.15	.22*	−.15	.04	.14	1
8. Color cont (N)	.32**	.32**	.01	−.17	.13	.33**	.18	1
9. Num body part (N)	.02	−.17	−.02	−.09	.06	.10	.24*	−.35**	1
10. Local color (P)	.26	.18	.34**	−.12	.07	.39**	.20*	.33**	.03	1
11. Num colors (P)	.18	.01	.11	−.21*	.12	−.02	.49**	.19	.15	.02	1
12. Color cont (P)	.26*	.27**	−.07	−.09	.15	.25*	.13	.66**	−.08	.30**	.27**	1
13. Num body parts (P)	.02	−.05	.24*	−.29**	.15	.05	.12	.01	.36**	.05	.06	−.34**	1
14. Num words (N)	.37**	.17	.13	.01	−.07	−.05	.11	.12	.03	.15	.20	.15	−.07	1
15. Num emo words (N)	.26*	−.01	−.15	.06	−.12	.02	−.003	.26*	−.07	.11	.06	.24*	−.14	.20	1
16. Num words (P)	.48**	.31**	.04	.06	.02	.03	.23*	.29**	−.06	.18	.23*	.35**	−.15	.60**	.21*	1
17. Num emo words (P)	.26*	.11	−.15	.02	−.09	.19	−.09	.36**	−.16	.04	.06	.30**	−.14	.04	.58**	.20*	1
Mean	4.58	1.47	4.65	4.75	109.03	0.58	1.94	1.15	4.43	.66	2.00	1.04	6.43	11.86	1.11	14.44	0.78
SD	0.75	0.50	1.06	1.25	18.04	1.08	0.95	0.75	4.17	1.42	1.42	0.81	5.15	9.07	1.32	12.74	1.12
N	95	95	95	95	94	95	95	94	95	95	95	93	95	95	95	95	95

Notes: ** indicates significance at the .01 level and * at the .05 level. “Sooth”: Soothability subscale of CBQ; “Ang/Frus”: Anger/Frustration subscale of CBQ; “EVT”: Scores on Expressive Vocabulary Test; “Local Color”: Average number of occurrences of localized coloring; “Num Colors”: Number of colors used; “Color Cont”: Level of coloring control; “Num Body Part”: Number of body parts colored; “Num Words”: Number of words in thought bubble; “Num Emo Words”: Number of emotion words in thought bubble; “(N)”: Negative-valence cartoon, “(P)”: Positive-valence cartoon.

Emotional awareness validity check

In order to confirm that children in the emotional awareness condition responded differently to the instructions than their counterparts in the control condition, a one-way MANCOVA test was performed on all physiological and cognitive emotional awareness variables for the relaxation and negative cartoons, while controlling for age and expressive vocabulary. The independent variable was group (e.g., control or emotional awareness). A Box’s M-test revealed that the assumption of homogeneity of covariance was violated (M = 123.33, F(36, 15228.97) = 3.06, p < .001) for the physiological emotional awareness variables. Another Box’s M-test also showed that the assumption of homogeneity of covariance was violated for the cognitive emotional awareness variables, (M = 65.63, F(10, 20377.71) = 6.21, p < .001). Thus, we report Pillai’s Trace criterion for subsequent analyses, which is a more robust test and preferred when the number of subjects is small and there is a significant difference in sample size (Tabachnick & Fidell, 2006). See Table 2 for multivariate and between-subjects tests results.

Table 2. Multivariate analysis of covariance (MANCOVA) for emotional awareness variables and covariates.

Source dependent variable	V	F	df	p	Sums of square	Mean square	ηp^2	95% CI [LL, UL]
^aGroup	0.28	3.85	8, 81	<.001**
Local color (N)		11.44	1, 88	.001**	11.71	11.71	0.12	0.33, 1.28
Num colors (N)		6.22	1, 88	.02*	4.89	4.89	0.07	0.11, 0.94
Color cont (N)		5.21	1, 88	.03*	2.50	2.50	0.06	0.05, 0.70
Num body part (N)		0.27	1, 88	.60	4.88	4.88	0.003	−1.46, 2.50
Local color (P)		4.57	1, 88	.04*	8.66	8.66	0.05	0.05, 1.34
Num colors (P)		0.001	1, 88	.98	0.001	0.001	<0.001	−0.64, 0.66
Color cont (P)		6.63	1, 88	.01*	3.81	3.81	0.07	0.12, 0.81
Num body part (P)		0.34	1, 88	.56	8.87	8.87	0.004	−1.70, 3.09
^cEVT	0.15	1.77	8, 81	.10
Local color (N)		5.31	1, 88	.02*	5.44	5.44	0.06	0.002, 0.03
Num colors (N)		0.11	1, 88	.74	0.09	0.09	0.001	−0.01, 0.01
Color cont (N)		2.49	1, 88	.12	1.19	1.19	0.03	−0.002, 0.02
Num body part (N)		0.19	1, 88	.67	3.35	3.35	0.002	−0.04, 0.06
Local color (P)		1.01	1, 88	.32	1.92	1.92	0.01	−0.01, 0.03
Num colors (P)		1.01	1, 88	.32	1.96	1.96	0.01	−0.01, 0.03
Color cont (P)		3.62	1, 88	.06	2.08	2.08	0.04	<0.001, 0.02
Num body part (P)		1.67	1, 88	.20	43.69	43.69	0.02	−0.02, 0.10
^cAge	0.11	1.26	8, 81	.28
Local color (N)		0.03	1, 88	.87	0.03	0.03	<0.001	−0.28, 0.32
Num colors (N)		6.55	1, 88	.01*	5.16	5.16	0.07	0.08, 0.60
Color cont (N)		3.54	1, 88	.06	1.70	1.70	0.04	−0.01, 0.40
Num body part (N)		0.01	1, 88	.94	0.10	0.10	<0.001	−1.20, 1.30
Local color (P)		2.58	1, 88	.11	4.89	4.89	0.03	−0.08, 0.74
Num colors (P)		1.61	1, 88	.21	3.11	3.11	0.02	−0.15, 0.68
Color cont (P)		1.74	1, 88	.19	1.00	1.00	0.02	−0.08, 0.37
Num body part (P)		0.004	1, 88	.95	0.10	0.10	<0.001	−1.56, 1.47
^bGroup	0.21	5.73	4, 87	<.001**
Num words (N)		0.04	1, 90	.84	2.91	2.91	<0.001	−3.55, 4.34
Num emo words (N)		18.90	1, 90	<.001**	25.87	25.87	0.17	0.64, 1.73
Num words (P)		0.05	1, 90	.83	6.07	6.07	0.001	−4.69, 5.83
Num emo words (P)		15.06	1, 90	<.001**	15.56	15.56	0.14	0.45, 1.39
^cEVT	0.01	0.30	4, 87	.88
Num words (N)		1.03	1, 90	.31	75.16	75.16	0.01	−0.15, 0.05
Num emo words (N)		0.11	1, 90	.74	0.15	0.15	0.001	−0.02, 0.01
Num words (P)		0.08	1, 90	.78	10.62	10.62	0.001	−0.15, 0.12
Num emo words (P)		0.05	1, 90	.82	0.05	0.05	0.001	−0.01, 0.01
^cAge	0.22	6.24	4, 87	<.001**
Num words (N)		11.86	1, 90	<.001**	863.79	863.79	0.12	1.83, 6.81
Num emo words (N)		1.64	1, 90	.20	2.25	2.25	0.02	−0.12, 0.56
Num words (P)		22.90	1, 90	<.001**	2962.78	2962.78	0.20	4.67, 11.31
Num emo words (P)		2.11	1, 90	.15	2.18	2.18	0.02	−0.08, 0.51

Notes: ** indicates significance at the .01 level and * at the .05 level. “V”: Pillai’s Trace; “LL”: lower limit; “UL”: upper limit; “Local color”: average number of occurrences of localized coloring; “Num colors”: number of colors used; “Color cont”: level of coloring control; “Num body part”: number of body parts colored; “Num words”: number of words in thought bubble; “Num emo words”: number of emotion words in thought bubble; “(N)”: negative-valence cartoon; “(P)”: positive-valence cartoon.

^a Physiological Emotional Awareness variables.

^b Cognitive Emotional Awareness group.

^c Covariates.

A Multivariate test revealed that there was a significant main effect of group for the physiological emotional awareness variables, Pillai’s Trace = .28, F(8, 81) = 3.85, p = .001, when controlling for age and expressive vocabulary. A between-subjects effects test revealed that there was a main effect of group on localized coloring for the relaxation, F(1, 88) = 4.57, p = .04, partial η² = .05, 95% CIs [.05, 1.34] and the negative cartoons, F(1, 88) = 11.44, p = .001, partial η² = .12, 95% CIs [.33, 1.28], such that children in the emotional awareness condition performed more localized coloring than children in the control condition. There was also a main effect of group on number of colors for the negative cartoon, F(1, 88) = 6.22, p = .02, partial η² = .07, 95% CIs [.11, .94], such that children in the control group used more colors than children in the emotional awareness condition, but only for the negative cartoon. Lastly, there was a main effect of group on coloring control for the relaxation, F(1, 88) = 6.63, p = .01, partial η² = .07, 95% CIs [.11, .81] and for the negative cartoons, F(1, 88) = 5.21, p = .03, partial η² = .06, 95% CIs [.05, .70], such that children in the emotional awareness condition displayed more controlled coloring than their peers in the control group.

For the cognitive emotional awareness variables, a Multivariate test revealed that there was a significant main effect of group, Pillai’s Trace = .21, F(4, 87) = 5.73, p < .001, when controlling for age and expressive vocabulary. There was also a main effect of the age covariate on the cognitive emotional awareness variables, Pillai’s Trace = .22, F(4, 87) = 6.24, p < .001. Lastly, a between-subjects effects test revealed that there was a main effect of group on number of emotion words for the relaxation, F(1, 90) = 15.06, p < .001, partial η² = .14, 95% CIs [.45, 1.39] and for the negative cartoon, F(1, 90) = 18.90, p < .001, partial η² = .17, 95% CIs [.64, 1.73], such that children in the emotional awareness condition generated more emotion words than children in the control condition for both cartoon types. A between-subjects effects test also showed that there was a main effect of age on number of words for the negative cartoon, F(1, 90) = 11.86, p = .001, partial η² = .12, and for the relaxation cartoon, F(1, 90) = 22.90, p < .001, partial η² = .20.

Associations between emotional awareness and emotion regulation

Within the emotional awareness group, localized coloring during the relaxation cartoon was significantly positively related to parent-reported soothability, (r = .45, p < .001) such that children who performed more localized coloring on the relaxation cartoon were reported to have higher levels of soothability. Similarly, number of colors used in the negative cartoon was positively correlated with parent-reported soothability (r = .27, p = .03), such that children who used more colors for the negative cartoon were rated as having higher levels of soothability. Number of colors used in the negative (r = −.29, p = .02) and relaxation (r = −.32, p = .01), cartoons were also inversely correlated with anger/frustration such that children who used more colors were rated as displaying lower anger and frustration. Additionally, coloring control in the negative cartoon was significantly negatively correlated with parent-reported anger/frustration (r = −.31, p = .02), such that children with higher coloring control in the negative cartoon were reported as displaying less anger and frustration. Number of body parts colored in the relaxation cartoon significantly positively related to parent-reported soothability, (r = .35, p = .01) such that children who colored in more body parts on the relaxation cartoon were reported to be better able to return to baseline following distress. Number of body parts colored in the relaxation cartoon was also significantly, negatively related to parent-reported anger/frustration (r = −.30, p = .02), such that children who colored in more body parts on the relaxation cartoon were reported to display less anger and frustration. All other physiological emotional awareness variables in the emotional awareness condition were not associated with parent-reported temperamental emotion regulation (all p > .06); furthermore, all cognitive emotional awareness variables (i.e., number of words, number of emotion words) did not predict parent-reported temperamental emotion regulation in the emotional awareness condition for either cartoon type (all p > .08). In the control condition, no study variables were significantly related to parent-reported temperamental emotion regulation (all p > .12). See Table 3.

Table 3. Correlations, means, and standard deviations of coloring task and emotion regulation variables in control (N = 33) and emotional awareness (N = 62) conditions controlling for age (years).

	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. Ang/frus	1	−.56**	−.13	−.29*	−.31*	−.06	−.24	−.32*	−.22	−.30*	−.08	−.01	.001	−.05
2. Sooth	−.45**	1	−.01	.27*	.12	.05	.45**	.16	−.05	.35**	.23	−.13	.07	−.16
3. Local color (N)	−.06	.01	1	.29*	.32*	.20	.35**	−.03	.22	.06	−.11	−.14	−.03	.07
4. Num colors (N)	.22	.09	.01	1	.34**	.13	.35**	.46**	.28*	.12	.12	.15	.20	.03
5. Color cont (N)	−.03	.11	.18	.07	1	−.22	.36**	.26*	.64**	.11	.15	.19	.28*	.38**
6. Num body part (N)	−.11	−.17	.004	.35*	−.50**	1	.06	.06	−.06	.45**	−.01	−.05	−.17	−.14
7. Local color (P)	.28	−.18	.23	.04	.05	.12	1	.02	.28*	.12	.18	−.01	.22	−.08
8. Num colors (P)	.04	−.03	.14	.57**	.06	.28	.04	1	.39**	−.01	.27*	.09	.27*	.11
9. Color cont (P)	.09	−.08	−.004	.03	.63**	−.07	.12	.03	1	−.25	.15	.12	.30*	.29*
10. Num body part (P)	−.22	−.01	.43*	.08	−.09	.24	−.03	.17	−.46**	1	.02	−.12	−.09	−.13
11. Num words (N)	.14	−.10	−.09	.19	−.01	.11	−.17	.07	.06	−.19	1	.13	.68**	−.02
12. Num emo words (N)	−.11	−.15	−.18	.01	.14	−.02	.04	.08	.24	−.06	.28	1	.15	.48**
13. Num words (P)	.11	.03	−.02	.40*	.26	.11	−.16	.17	.36*	−.20	.44*	.27	1	.16
14. Num emo words (P)	−.12	.11	−.11	−.14	−.07	−.20	−.10	−.05	−.11	−.03	−.04	.38*	.20	1
Mean (C)	4.47	4.76	.12	2.21	.91	4.94	.18	2.00	.76	7.30	10.21	.27	11.79	.12
SD (C)	1.13	.86	.33	1.02	.72	5.11	.53	1.39	.71	5.68	8.53	.57	13.14	.42
Mean (EA)	4.89	4.59	.82	1.80	1.28	4.16	.92	1.95	1.20	6.00	12.74	1.55	15.85	1.13
SD (EA)	1.29	1.15	1.25	.91	.73	3.58	1.66	1.44	.82	4.83	9.30	1.39	12.40	1.22

Notes: ** indicates significance at the .01 level and * at the .05 level. Control condition variables are below the diagonal; emotional awareness condition variables are above the diagonal. “Sooth” = Soothability subscale of CBQ; “Ang/Frus” = Anger/Frustration subscale of CBQ; “Local color”: average number of occurrences of localized coloring; “Num Colors”: number of colors used; “Color cont”: level of coloring control; “Num body part”: number of body parts colored; “Num words”: number of words in thought bubble; “Num emo words”: number of emotion words in thought bubble; “(N)”: negative-valence cartoon; “(P)”: positive-valence cartoon.

Finally, we tested whether significant correlation coefficients in the emotional awareness condition differed from the same coefficient in the control condition using r to z transformations. Findings indicated correlations between localized coloring on the relaxation cartoon and parent-reported soothability (z = −2.97, p = .003) were significantly different between groups. Additionally, correlations between number of colors and parent-reported anger/frustration were significantly different between groups for the negative cartoon (z = −2.33, p = .02), but not for the relaxation cartoons (p = .10). All other significant correlations in the emotional awareness group were not significantly different from the respective correlation in the control condition.

Discussion

The present study demonstrated that preschool-aged children could respond to instructions in the emotional awareness task by producing more localized coloring, higher levels of coloring control, more emotion words for both negative and relaxation cartoons, and fewer colors for the negative cartoon, compared to children in the control condition. In turn, within the emotional awareness group, specific aspects of how children colored in emotion predicted their parent-rated emotion regulation skills. More localized coloring on the relaxation cartoon was related to higher levels of parent-reported soothability, and higher levels of localized coloring on the positive cartoon was significantly related to lower levels of parent-reported anger/frustration. The number of body parts colored was associated with better parent-reported soothability and lower anger/frustration. Coloring control was also associated with lower anger/frustration. Our findings have implications for the interpretation of children’s performance on similar emotional awareness tasks administered in clinical settings, as well as implications for measuring emotional awareness in laboratory settings.

The present study suggests that preschool-aged children have the capacity to articulate an understanding of physiological and cognitive emotional awareness. Research on dimensions of emotional awareness in early childhood has been sparse to date. Indeed, studies have suggested that children under six years may lack the skills necessary to monitor and report on their own emotions (Chambers & Johnston, 2002), which would indicate emotional awareness is perhaps absent, or immeasurable, in early childhood. Here, differences between the study groups suggests that young children can understand and respond to prompts to demonstrate emotional awareness.

Our findings also suggest that specific facets of early emotional awareness relate to emotion regulation. Only the degree to which children engaged in localized coloring, controlled coloring, and how much of the character’s body they colored in, related to parent ratings of emotion regulation, whereas none of the cognitive emotional awareness variables related to emotion regulation. This is consistent with Lane and Schwartz’s (1987) contention that physiological emotional awareness is the foundation of emotional awareness development, and that children garner an increased capacity to verbally describe emotional states as they mature (Lane & Schwartz, 1992). Thus, physiological emotional awareness may emerge earlier in development compared to cognitive emotional awareness, making it a better predictor of emotion regulation in the preschool years. The possibility that physiological and cognitive emotional awareness differentially relate to temperamental emotion regulation has potential implications for widely used childhood treatments that scaffold these two dimensions in equal measure (Ehrenreich-May et al., 2018; Kendall & Hedtke, 2006). Physiological emotional awareness may be a more viable clinical target for supporting emotion regulation in early childhood. Therefore, interventions for young children may benefit from a more thorough exploration and building up of physiological emotional awareness in young children before engaging in tasks focused on cognitive emotional awareness.

The current findings may advance how early emotional awareness is measured in ways that may translate to clinical practice. For example, developmentally sensitive coloring tasks could track progression of emotional awareness in young children over time. To our knowledge, there are currently no tools to measure, in preschool-aged children, individual differences in dimensions of emotional awareness in school or clinic settings. This emotional awareness task takes approximately 15 min to administer, requires minimal training, and is easy to code. Thus, the present study sets the stage for research to investigate the validity and reliability of future emotional awareness tasks for young children based off typical clinical practices.

Strengths, limitations, and future directions

Strengths of the study included the use of a novel coloring task designed to assess physiological and cognitive emotional awareness in an early childhood sample and the use of a control group to test early understanding of these dimensions more robustly. Descriptive statistics suggest the open-ended instructions successfully prompted a wide range of individual differences in coloring (e.g., amount of coloring, use of localization) rather than a single uniform style. The variables coded for in the coloring task were operationalized on the basis that they would be easy to translate to future research, clinical, and real-world settings. However, it should be noted that these coloring variables (e.g., coloring control, number of colors used) may also be related to other child characteristics unrelated to emotion regulation (e.g., impulsivity, creativity). Future research that utilizes a coloring task like the present study’s task may benefit from accounting for such child characteristics.

While the current study utilized a standardized style of interaction modeled after an empirically supported treatment (i.e., PCIT), there was no data collected on fidelity of task administration to ensure consistency across trained experimenters. Future research utilizing a similar open-ended, novel task would benefit from incorporating such data collection to ensure integrity of the task. Additionally, considering the early developmental period of the study’s sample, future research examining similar constructs should consider incorporating other facets of vocabulary (e.g., receptive language, pragmatic use of language) to further identify if these other aspects of language ability play a role in emotional awareness understanding and expression.

The present study’s findings also suggest exciting next steps for work in the early development of emotional awareness. Here, temperamental emotion regulation was measured via parent ratings of their child’s anger onset and recovery from emotional challenges, setting the stage for future studies to explore links between early dimensions of emotional awareness and overt behavioral or physiological indicators of emotion regulation (e.g., heart rate, neural activation). It is important to note that temperamental emotion regulation provides insight into how children typically respond to frustration; it would be exciting for future research to also examine children’s emotion regulation in vivo and understand how they respond to an emotionally challenging situation in a lab setting. By modeling the emotional awareness task on commonly used clinical techniques, the present study has potential implications for the emergence of mental illness, although directly investigating this was outside the scope of the study. Future research using the novel emotional awareness task may be able to elucidate how dimensions of emotional awareness present in children with and without psychopathology, how these dimensions change with clinical improvement, and may be able to investigate whether different clinical disorders have distinct emotional awareness deficits.

Authors’ note

Findings were presented at the 31st Annual Association for Psychological Science Convention in Washington, DC.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data and materials can be made available if requested.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This work was supported by National Institutes of Health (K23 MH111708 PI: Grabell).

Appendix A

Coloring task script – emotional awareness condition

Here’s a cartoon of a child playing the Incredible Cakes Kids game, just like you were! This kid just picked the wrong cake and saw a grumpy customer. How do you think the child felt when s/he saw that grumpy customer? Do you think s/he felt mad? Sad? Worried? So, what you’re going to do is, after the beep, pick the mad, sad, and worried colors and color in the child where s/he is feeling those emotions, because sometimes we feel these things in our bodies.

Next, we’re going to fill in this thought bubble with what s/he was thinking when s/he saw this grumpy customer. You can let me know what mad, sad, and worried words this character was thinking when s/he saw this grumpy customer. Whatever you saw, I’ll write it here in this thought bubble. So, what mad, sad, and worried thoughts did s/he have when s/he saw this grumpy customer?

Instructions repeated for positive cartoon, but with a focus on positive emotions (e.g., happy, relaxed/calm).

Figure A1. Representation of the Incredible Cake Kids Task.