Introduction

Attempting to influence the experience and expression of emotion is known in scientific literature as emotion regulation [1]. Successful emotion regulation can aid adaptive social, occupational, psychological health, and physical health [e.g., 2–4]. One prominent conceptualisation of emotion regulation is Gross’ [1] process model of emotion regulation, which offers five major emotion regulation strategies, namely situation selection, situation modification, attentional deployment, cognitive change, and response modulation. Of particular importance, is the emotion regulation strategy of cognitive reappraisal (or cognitive change), which is the modification of one’s appraisal of a situation in order to alter its emotional impact [5]. Cognitive reappraisal is demonstrably one of the most effective [6], and well-studied [7] emotion regulation strategies, associated with adaptive psychophysiological [8], and neurological [9] outcomes.

The notion that cognitive reappraisal is a highly effective emotion regulation strategy is in line with dominant psychotherapies in which cognitive mediation is a key axiomatic principle [10]. Specifically, second wave cognitive behavioural therapies (CBTs), such as rational emotive behaviour therapy (REBT [11]), hold that in the face of an adverse situation, our thoughts about the event largely determines our emotional reactions, and thus, to regulate our emotional reactions, one can modify one’s thoughts about the adverse situation [12]. The hypothesis that our thoughts mediate between the environment and our emotional responses is supported by theories of emotion [13], and scientific evidence (see [14], for a review). In second wave CBTs it is typical to help patients and clients to understand the important role of cognition in their emotions, and to encourage them to take charge of their cognitions in order to enable greater emotion regulation [15]. For example, within REBT [11], cognitive mediation is represented by the ABC framework whereby core beliefs (B) mediate between adverse events (A) and emotional consequences (C). In REBT, clients and patients are introduced to the notion that their cognitions about events have a considerable impact on their psychological wellbeing. Recognising that thoughts, and not events alone, lead to emotions, is known as B-C thinking, and in contrast, believing that event directly causes emotions despite deliberative cognition, is known as Adversity-Consequences thinking. Beliefs-Consequences thinking enables the individual to exercise some volition over their emotions by regulating their thoughts [16].

Given the importance of cognitive reappraisal in regulating one’s emotions, predetermining or predicting factors of cognitive reappraisal are worthy of investigation. In other words, we can ask the question, what factors make cognitive appraisals attempts more likely? One potential answer is “emotion beliefs” ([17], p. 74), which are considered to be beliefs about emotion and emotion regulation. Individuals differ in how they think about their emotions and how these emotions can be regulated, and these differences in beliefs are consequential for emotion regulation [18, 19]. Amidst a call for more empirical research in the area of emotion beliefs [20], in a recent study [21] conceptualised four emotion beliefs. The four emotion beliefs capture the aforementioned idea of cognitive mediation that is the centre of second wave CBTs, and cognitive reappraisal specifically. The four emotion beliefs are:

1. Stimulus-Response (S-R) generation beliefs (emotions are caused by events)
2. Stimulus-Response (S-R) change beliefs (changes in the situation lead to emotion change)
3. Cognitive Mediation (C-M) generation beliefs (emotions are cognitively mediated)
4. Cognitive Mediation (C-M) change beliefs (changes in cognition lead to emotion change).

Cleary, S-R beliefs reflect Adversity-Consequences thinking, whilst C-M beliefs reflect Beliefs-Consequences thinking, in CBT (i.e, REBT) terms. That is, S-R beliefs reflect the idea that emotion is dependent upon situational events, and C-M beliefs reflect the idea that emotion is dependent upon the one’s cognitions about situational events. S-R and C-M emotion beliefs are considered to be superordinate and dispositional beliefs, not entirely unlike the superordinate emotion beliefs of goodness and controllability proposed by Ford and Gross [17]. Goodness beliefs concern the extent to which emotions are good versus bad (including desirability and usefulness) and controllability beliefs concern the extent to which emotions are controllable versus uncontrollable. Goodness beliefs are thought to guide the trajectory of emotion regulation (i.e., what people want to feel), whilst controllability beliefs are thought to guide the occurrence of emotion regulation (i.e., whether regulation is initiated). Goodness beliefs are associated with emotional and wellbeing outcomes, such that those who believe emotions are bad have heightened negative emotional responses to stressors [22] and worse psychological health (e.g., [22, 23]). Controllability beliefs are linked to important emotional and social outcomes (e.g., [24–26]). For example, Tamir et al. [27] found that participants who believe that people can control their emotions were more effective in controlling their emotions and report enjoying long-term emotional and social benefits.

Of course, goodness and controllability are not the only beliefs individuals hold about emotions [17], and S-R and C-M beliefs are not antithetical to goodness and controllability beliefs. Rather, the S-R and C-M beliefs offered by Turner et al. [21] are supplementary to those of goodness and controllability. Further, the notion that changes in cognition lead to emotion change (C-M beliefs) might be somewhat dependent upon, or co-occurring with, beliefs about the controllability of emotion per se. Therefore, whilst S-R and C-M beliefs differ to goodness and controllability beliefs, different beliefs may interact to determine emotion regulation and reactivity outcomes.

One of the notable features of S-R and C-M beliefs is that two-factor view of emotion is captured (e.g., [28, 29]), whereby one factor concerns the generation of emotion, and another factor concerns emotional change. This two-factor approach is useful for analysing processes, individual differences, and fashioning clinical interventions [30]. The importance of separating out S-R and C-M beliefs across generation and change was borne by the findings of Turner et al. [21]. Specifically, Turner and colleagues developed a psychometric, the cognitive-mediation beliefs questionnaire (CMBQa), for the proposed four emotion beliefs, and found that S-R generation beliefs and C-M change beliefs formed reliable factors in the factor analyses, whilst C-M generation beliefs and S-R change beliefs did not. Follow-up data indicated that greater C-M change beliefs and lower S-R generation beliefs were related to higher cognitive reappraisal tendencies (adaptive emotion regulation), a greater ability to control thoughts, more positive mental health outcomes, and lower emotion reactivity.

Initial findings for the four proposed emotion beliefs are promising. However, because C-M generation and S-R change beliefs did not form reliable factors in the factor analyses by Turner et al. [21], research needs to ascertain the factorial validity and utility of C-M generation and S-R change beliefs. If indeed the four proposed emotion beliefs are valid ideas relevant to the study of emotions, researchers and practitioners should have valid and reliable questionnaires that assesses all four beliefs. This will enable a more comprehensive assessment of how these beliefs interact with each other, and also the extent to which they collectively and or discriminately influence emotion regulation and reactivity.

Therefore, the chief aim of the current paper is to validity test a new psychometric questionnaire pertaining to C-M generation and S-R change beliefs, the CMBQb, using items produced by Turner et al. [21] in the development of the CMBQa. To be clear, in the current study we aim to validity test a questionnaire (CMBQb) that pertains only to the two factors that did not emerge in Turner et al. [21]. To achieve this, three studies in the current paper report empirical efforts that align with psychometric questionnaire development guidelines for a latent variable approach [31, 32]. Study 1 reports exploratory factor analysis (EFA), study 2 reports confirmatory factor analysis (CFA), and study 3 reports additional CFA and criterion validity. Fourth and finally, in a further study (Study 4), we report the results of a four-factor CFA inclusive of both CMBQa (S-R generation, C-M change) and CMBQb (S-R change, C-M generation) factors. In Study 4 criterion validity tests are also undertaken to examine the relationships between the four CMBQ factors and variables concerning emotion regulation and emotion reactivity.

We pose the question: to what extent do individuals’ beliefs about the generation and regulation of their emotions influence their emotion reactivity? Initially, we focus on the proposed C-M generation and S-R change beliefs of the CMBQb.

Study 1: Exploratory Factor Analysis (EFA)

For a detailed portrayal of the item generation process employed in the current study, see Turner et al. [21]. For the current study, we used items reflecting C-M generation and S-R change beliefs only (and no S-R generation and C-M change items), yielding a total of 24 suitable items for analyses. Following the psychometric questionnaire development guidelines (e.g., [32]), exploratory factor analysis (EFA) was used to assess the underlying two-factor structure of the 24-items [33]. We conducted EFA first, and then followed up with confirmatory factor analyses (CFA) with different samples [34]. We also recruited separate samples for each study in order to limit the attribution of results to cohort specific factors. For EFA the 24-items followed the stem, “When I experience an unpleasant or unwanted emotion, I believe that…”. Each item was scored against a five-point Likert-scale from 1 (strongly disagree) to 5 (strongly agree).

Results

In the EFA (see S2 File for the initial EFA) 9-items were deleted iteratively for failing to meet item retention criteria, and 15-items across the two factors were extracted accounting for 51.84% of the variance (Table 1). C-M generation (n = 7) and S-R change (n = 8) were negatively correlated (-.21). Although the factor loadings for items 7 (.595) and 21 (.599) were below .60, they were retained. We did not want to remove items that had loadings so close to the retention criteria, and these items can be further examined in CFA with a separate sample (study 2). EFA indicated the proposed two-factor approach to emotion regulation (e.g., [28]); C-M generation and S-R change.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. EFA outcomes for the two-factor model, with factor loadings, cross, loadings, and communalities.

https://doi.org/10.1371/journal.pone.0269928.t001

Study 2: Confirmatory Factor Analysis (CFA)

A separate CFA sample was recruited to the EFA sample (e.g., [32]) to test the two-factor structure emergent from the EFA in study 1. To do this, we collected data using the 15-items derived from the EFA of study 1, which has two factors (C-M generation, and S-R change). Again, no S-R generation and C-M change (CMBQa) items were used. Therefore, the aim of the CFA in the current study was to confirm the proposed two-factor structure of the CMBQb.

Results

The 15-item correlated two-factor model was not an acceptable fit, χ² = 285.909, df = 89, p < .001, RMSEA = .08 (90% CI = .072–.093), CFI = .89, NFI = .85, TLI = .87, SRMR = .06. After the removal of two items with low factor loadings (e.g., [43]), the correlated two-factor model was an acceptable fit, χ² = 114.006, df = 62, p < .001, RMSEA = .05 (90% CI = .036–.065), CFI = .97, NFI = .93, TLI = .96, SRMR = .05. The acceptable two-factor model was taken forward to bifactor analyses.

The two-factor bifactor model was an acceptable fit, χ² = 133.723, df = 52, p < .001, RMSEA = .07 (90% CI = .055–.084), CFI = .95, NFI = .92, TLI = .92, SRMR = .04. Although a good model fit was revealed, bifactor measures indicated that the CMBQb is not unidimensional and should not be assessed using a general factor (ΔCFI = .02). Specifically, for the bifactor model, PUC was .54, general ECV was .26, OmegaH was .22, and Relative Omega indicated that the general factor explained 24% of the variance. Additionally, all but one item had IECV coefficients below .50, indicating that items had stronger ties with their specific factors than the general factor [44]. Finally, the ARMPB for the factor loadings was 4.23% which is not acceptable [45], especially taking into account the other indices. In all, the ancillary bifactor measures for the two-factor models support the conceptualization of the CMBQb as multidimensional. Also, CFA indicated that a two-factor correlated factors model comprising 13 items (C-M generation = 7 items, S-R change = 6 items) produced the best fit to the data. For fit indices comparisons see S4 File, and for correlated and bifactor two-factor item loadings, Relative Parameter Bias, and IECV see S5 File.

Study 3: Second CFA, and criterion validity

The purpose of study 2 was to test and confirm the two-factor structure of the CMBQb using CFA with a separate sample of adults. The CFA supported a correlated two-factor structure, comprising 13-items (C-M generation = 7 items, S-R change = 6 items). In study 3, a separate adult sample was recruited to firstly reconfirm (replicate) the two-factor structure within a new sample, and secondly to conduct criterion validity tests on the CMBQb. We examined criterion validity in relation to emotion regulation and emotion reactivity (e.g., [21]). If the CMBQb assesses C-M generation and S-R change beliefs in the hypothesised manner, scores reflecting greater C-M generation and lower S-R change beliefs should be related to more adaptive emotion regulation (higher cognitive reappraisal). Further, scores reflecting greater C-M generation beliefs and lower S-R change beliefs were hypothesised to be related to lower emotion reactivity. Finally, we determined floor (lowest possible score) and ceiling (highest possible score) effects as an additional marker of quality. As in the previous studies of the current paper, no S-R generation and C-M change (CMBQa) items were used.

Data analysis

Data were screened for outliers (standardized z values > 3.29), and outliers were Winsorized (n = 32 from 10,260 cases = .31%). Data were Winsorized for CMBQ cases (n = 20), and ARI cases (n = 12). First, the 13-items of the CMBQb (C-M generation = 7 items, S-R change = 6 items) were subjected to CFA using SEM in AMOS version 25, in which a correlated two-factor model was tested (Table 2). The same goodness of fit indices posited by Schermelleh-Engel et al. [49] were used to determine acceptable fit, and the same MI guidelines by Rossier et al. [50] were used. Then, following similar research (e.g., [39]), criterion validity was tested by examining Pearson’s bivariate correlation co-efficients (using SPSS version 25) between the CMBQb subscale scores, and ERQ (cognitive reappraisal) scores, and ARI, and ERS scores (Table 3). It was hypothesised that greater C-M generation and lesser S-R change scores would be significantly related (p < .05) to greater ERQ (cognitive reappraisal), and lower ARI and ERS scores.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Item properties, internal consistency, inter-item correlations, and descriptives, of the 13-item CMBQb in study 3.

https://doi.org/10.1371/journal.pone.0269928.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Study 3 Pearson’s correlation coefficients.

https://doi.org/10.1371/journal.pone.0269928.t003

Results

Discussion

The results of study 3 confirmed the correlated two-factor structure of the 13-item CMBQb. Also, greater C-M generation scores and lesser S-R change scores were positively related to cognitive reappraisal, and greater S-R change scores were negatively related to affective and emotion reactivity, demonstrating criterion validity. However, C-M generation scores were unrelated to affective and emotion reactivity. Finally, descriptive statistics revealed acceptable floor and ceiling effects, which indicates some construct validity.

Study 4

The focus of the current paper thus far has been on the validation of the CMBQb as a complimentary assessment of emotion beliefs to the already validated CMBQa [21]. Whilst the CMBQa assesses S-R generation and C-M change, the proposed CMBQb assesses SR-change and C-M generation. Given the overlapping theoretical ideas between the CMBQa and the CMBQb, there are some important questions remaining as to a) whether and to what extent the two factors with each of the CMBQ questionnaires can form a four-factor solution, b) whether and what extent these four factors can tell us anything about emotion regulation tendencies and emotion reactivity outcomes, and c) whether and to what extent the four factors can collapse into two composite subscales of S-R beliefs ([S-R generation + S-R change] / 2) and C-M beliefs ([C-M generation + C-M change] / 2). To phrase point c) above differently, does the distinction between generation and change matter to emotion regulation and reactivity outcomes, or is it more useful to consider a single S-R subscale and a single C-M subscale? As an extension to the idea of synthesising the two CMBQs down to two subscales, and as a fourth consideration, d) it is also possible to consider a single discrepancy value (C-M minus S-R, e.g., hedonic balance; [52]). It could be the extent to which C-M beliefs predominate over S-R beliefs, or vice versa, that is most important for emotion regulation and reactivity.

To address the four points above, in Study 4 CFA is conducted with the four-factors offered within the CMBQa and CMBQb, namely S-R generation and C-M change (CMBQa), and C-M generation and S-R change (CMBQb). Following CFA, criterion validity tests are conducted to examine the associations between the CMBQ factors (four separate factors, and two combined factors) and prominent self-regulation tendencies, and emotion reactivity outcomes. Scores reflecting greater C-M beliefs (generation and change separately, and combined) and lower S-R beliefs (generation and change separately, and combined) were hypothesised to be related to more adaptive emotion regulation tendencies and lower emotion reactivity. Furthermore, a CM-SR discrepancy score (or a CM-SR index), is calculated such that higher scores indicate greater relative C-M beliefs and lower relative S-R beliefs. We then correlate this CM-SR index with prominent self-regulation tendencies, and emotion reactivity outcomes. We hypothesised that a greater index score would be related to more adaptive emotion regulation tendencies and lower emotion reactivity. Finally, to reflect the potential conceptual similarity between change beliefs (S-R change, C-M change), and controllability beliefs offered in past research (e.g., [27]), we hypothesised that S-R and C-M change beliefs would be related to more adaptive emotion regulation tendencies and lower emotion reactivity.

Results

CFA

The 28-item two-factor model was a somewhat acceptable fit, χ² = 590.954, df = 339, p < .001, RMSEA = .054 (90% CI = .047–.061), CFI = .91, NFI = .81, TLI = .90, SRMR = .064. See Table 4 for factor loadings. Similar to study 2 findings, item 16 of S-R change demonstrated a particularly low factor loading (.32), but all other loadings were between .48 and .79.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Item properties, internal consistency, inter-item correlations, and descriptives, of the 28-item CMBQc (CMBQa and CMBQb combined) in study 4.

https://doi.org/10.1371/journal.pone.0269928.t004

Criterion validity

Emotion regulation.

In the Pearson’s correlation analysis (Table 5), S-R generation and S-R change scores were negatively associated with reappraisal (respectively r = -.15 and r = -.20, p < .01), and positively associated with suppression (respectively r = .24 and r = .28, p < .001). S-R change was also positively associated with distraction (r = .13, p < .05). C-M generation scores were positively associated with engagement (r = 16, p < .02), acceptance (r = .22, p = .001), and putting into perspective (r = .27, p < .05) and C-M change scores were positively associated with reappraisal (r = .20, p < .01), acceptance (r = .25, p < .001), and putting into perspective (r = .27, p < .001), and negatively associated with suppression (r = -.16, p < .01).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Study 4 Pearson’s correlation coefficients (C-M comp = composite C-M score, S-R comp = composite S-R score, perspective = putting into perspective).

https://doi.org/10.1371/journal.pone.0269928.t005

For S-R composite and C-M composite scores, Pearson’s correlation analysis revealed that S-R composite scores were negatively associated with reappraisal (r = -.19, p < .001) and positively associated with suppression (r = .29, p < .001). C-M composite scores were positively associated with reappraisal (r = .16, p < .01), engagement (r = .13, p < .05), putting into perspective (r = .25, p < .001), and acceptance (r = .27, p < .001), and negatively associated with suppression (r = -.13, p < .05). In addition, Pearson’s correlation analysis revealed that CM-SR index scores were positively associated with reappraisal (r = .23, p < .001), putting into perspective (r = .14, p < .005), and acceptance (r = .23, p < .001), and negatively associated with suppression (r = -.30, p < .001).

Emotion reactivity.

In the Pearson’s correlation analysis (Table 5), S-R generation and S-R change scores were positively associated with affective reactivity (respectively r = .31, and r = .25, p < .001) and emotion persistence and sensitivity (respectively r = .33 and r = .31, p < .001). S-R generation was positively associated with emotion intensity (r = .18, p < 01). C-M generation scores were positively associated with emotion intensity (r = .15, p < 02), and C-M change scores were negatively associated with affective reactivity (r = -.13, p < .05).

For S-R composite and C-M composite scores, Pearson’s correlation analysis revealed that S-R composite scores were positively associated with affective reactivity (r = .30, p < .001) and emotion persistence (r = .35, p < .0010 and sensitivity (r = .32, p < .001), and emotion intensity (r = .17, p = .008). C-M composite scores were not related to ARI or ERS scores. In addition, Pearson’s correlation analysis revealed that CM-SR index scores were negatively related to affective reactivity (r = -.27, p < .001) and emotion persistence (r = -.25, p < .001) and sensitivity (r = -.24, p < .001).

Discussion

In study 4 the four-factor structure of the combined CMBQ (CMBQc) was tested for factorial and criterion validity. CFA results indicated a somewhat acceptable fit to the data, with one item (item 16) poorly loading onto the S-R change factor. It was also found that scores indicating greater C-M generation and change beliefs, and lower S-R generation and change beliefs, were related to more adaptive, and less maladaptive, emotion regulation tendencies. In addition, there was some evidence that greater C-M change beliefs, and lower S-R generation and change belief, were related to better affective and emotion reactivity outcomes. However, contrary to expectations, greater C-M generation beliefs were related to greater emotion intensity. Whilst this association was weak (r = .13), it is still worthy of note, and might reflect that C-M generation beliefs are not sufficient alone and may require C-M change beliefs to have an impact upon emotional outcomes. It is also possible that believing that emotions are generated by one’s own thoughts may lead to rumination that exacerbates emotion intensity, or encourages self-blame for one’s emotions and thus could be generative of further negative emotion. Of course, these explanations are speculative at this point, and more date are required to be able to test these conjectures.

Study 4 also moved the research into the CMBQc forward in terms of emotion regulation tendency assessment. Whereas Turner et al. [21] and studies 1–3 focussed on reappraisal due to its clear theoretical convergence with S-R and C-M beliefs, in study we also measured acceptance, engagement, distraction, suppression, and putting into perspective.

In study 4 we also formulate an S-R composite score and a C-M composite score, and a CM-SR index score (C-M minus S-R) to examine the utility of synthesising down the four factors of the CMBQc. Broadly, greater S-R and lower C-M composite scores, and greater CM-SR index scores (indicting higher C-M beliefs relative to S-R beliefs) were associated with more adaptive, and less maladaptive, emotion regulation tendencies, and better affective and emotion reactivity outcomes. Although, there are some nuances in these results that need to be elaborated. First, the associations between the S-R and C-M composite scales and affective and emotion reactivity indicate that it is S-R beliefs that are related to greater reactivity, rather than C-M beliefs being related to lower reactivity. This is consistent with the C-M generation and C-M change scores, which also do not relate, or are weakly related, to reactivity scores, whilst S-R generation and S-R change scores reflect stronger associations with reactivity scores.

In contrast, for the results concerning emotion regulation tendencies, C-M composite scores were associated with a variety of strategies including greater reappraisal, engagement, acceptance, and putting things into perspective, and lower suppression, whilst S-R composite scores were only related to lower reappraisal and higher suppression. It could be that S-R beliefs are not conducive to, or are prohibitive of, the particular emotion regulation strategies we measured. The CM-SR index scores support the findings for the separate S-R and C-M composite scores, in that a greater CM-SR index score was not related to distraction or engagement, but was related to the other emotion regulation strategies. As such, it appears that whilst there are some differences in the findings across the four separate S-R/C-M generation and change scores, S-R and C-M composite, and CM-SR index scores, there is a largely consistent narrative that when the results are taken together. It appears that the emotion beliefs combination that is most associated with adaptive emotion regulation tendencies and affective and emotion reactivity is one of high C-M beliefs and low S-R beliefs. Whilst nuanced relationships between CMBQ subscales and emotion regulation and reactivity are apparent, there appears to be some justification for utilising a CM-SR index. As such, it could be argued that it is the extent to which C-M beliefs predominate over and above S-R beliefs that are indicative of greater adaptive emotion regulation tendencies and lower affective and emotion reactivity.

Results of this more expansive emotion regulation assessment reveal some useful findings that are in line with what we would expect given the proposed implications of S-R and C-M beliefs. For example, it makes sense that S-R is related to less reappraisal, and more related to suppression and distraction. If I believe that emotions are only situationally determined, then a viable recourse for emotion regulation might be to distract myself from the situation and or suppress my emotion, since I might not see the viability of cognitive reappraisal. In contrast, it makes sense that C-M is related to more reappraisal, engagement, acceptance and putting into perspective, and less related to suppression. If I believe that emotions are cognitively mediated (C-M), then my recourse for emotion regulation might viably be cognitive reappraisal which allows me to engage with my emotions, rather than suppress them. I am also more likely to attempt to put the situation into perspective, since this strategy relies on reappraisal. Also, if I have high C-M beliefs and low S-R beliefs (which is not always necessarily the case) my focus might be less on changing the situation and more on accepting it, because I believe in that cognitive change leads to emotion change, rather than situation change, so focussing on the situation presents less emotion regulation potential. Of course, it is possible to endorse both C-M and S-R beliefs, and it is not necessarily the case that just because I have high C-M beliefs that I will have low S-R beliefs, and indeed, data in the present study reveals only a small correlation between C-M and S-R scores. In sum, the results of study 4 offers some support for the CMBQc, and extend the research by introducing C-M and S-R composite scores, and the CM-SR index.

General discussion

The chief purpose of the present paper was to develop and undertake initial validity testing of two emotion beliefs, namely C-M generation beliefs and S-R change beliefs. In studies 1–3 of the current paper, the factor structure of the CMBQb emerged and was validated through EFA (study 1) and CFA (study 2), confirmed in an additional CFA (study 3), and criterion validity testing (study 3). Greater C-M generation beliefs and lower S-R change beliefs were related to more adaptive emotion regulation tendencies. Greater S-R change beliefs, but not C-M generation beliefs, were related to greater emotional and affective reactivity. Results for studies 1–3, particularly for S-R change beliefs, are noteworthy when considering the proposed shortcomings of an S-R viewpoint of emotion (e.g., [13]). In other words, the finding that S-R change beliefs are related to less reappraisal and poorer emotional outcomes lends support to the idea that a reliance on situational change (or modification), and or the belief that emotions are the result of external events only, might be deleterious to emotion regulation and emotion reactivity.

A secondary purpose of the current paper was to test the validity of the CMBQc, which is a combination of CMBQa [21], and CMBQb (emergent in the studies 1–3 of the present paper). Turner et al. [21] found support for the two emotion beliefs of S-R generation and C-M change (CMBQa), but not for C-M generation and S-R change. That is, EFA and CFA in Turner et al. (2021) revealed two factors comprising S-R generation and C-M change items, which excluded all C-M generation and S-R change items. Therefore, in studies 1–3 of the present paper we took the excluded C-M generation and S-R change items from Turner et al. (2021) and applied separate validity tests to examine the factor structure and criterion validity of the CMBQb. Then in study 4 of the present paper we combined CMBQa and CMBQb to test the proposed four-factor structure of a CMBQc. The 28-item CMBQc offers an assessment of all four superordinate emotion beliefs proposed by Turner et al. [21]. That is, CMBQc assesses C-M generation beliefs (emotions are cognitively mediated), C-M change beliefs (changes in cognition lead to emotion change), S-R generation beliefs (emotions are caused by events), and S-R change beliefs (changes in the situation lead to emotion change).

The results of the present paper when considered alongside Turner et al. [21] provide a more comprehensive representation of how S-R and C-M beliefs might influence emotion regulation and emotion reactivity. Specifically, S-R beliefs (generation and change) were related to less adaptive emotion regulation tendencies, and greater emotion reactivity. In contrast, C-M generation and change beliefs were related to more adaptive emotion regulation tendencies, and C-M change beliefs were related to lower emotion reactivity. However, it remains unclear as to whether C-M generation beliefs are beneficial for emotion reactivity. In study 3, C-M generation beliefs were unrelated to affective and emotion reactivity, and in study 4, greater C-M generation beliefs were related to greater emotion intensity, which was counter to expectations. Therefore, whilst the current paper offers some initial promising findings for the validity of the CMBQc, more research is needed in order to establish exactly how the four superordinate emotion beliefs relate to and interact with emotion regulation to predict affective and emotion reactivity.

S-R generation and change beliefs reflect the notion that emotion is dependent upon situational events, whilst C-M generation and change beliefs reflect the notion that emotion is dependent upon the one’s cognitions about situational events. There is some clear convergence between the ideas of S-R and C-M beliefs, and the Adversity-Consequences and Beliefs-Consequences thinking present in REBT, and other CBTs. In REBT, clients are encouraged to understand that their emotions are cognitively mediated (Beliefs-Consequences), rather than the sole result of external events (or stimuli; Adversity-Consequences thinking). But the S-R and C-M emotion beliefs captured in the CMBQc also contain two additional and related ideas; (a) emotion is generated by the situation (S-R generation), or by one’s cognitions about the situation (C-M generation), and (b) emotion is regulated by changing the situation (S-R change), or by changing one’s cognitions (C-M change). Thus, one aspect involves a set of processes related to the generation of emotion, and a second aspect involves a different set of processes related to the management of emotion (e.g., [28, 29]). However, what remains to be examined concerning the CMBQc, is the extent to which the four factors interact with one another to influence emotion regulation and reactivity.

Some have argued that emotion generation and regulation may be inseparable [18], but distinguishing between emotion reactivity from emotion regulation is useful for analysing processes, individual differences, and developing interventions [30], but whilst the separation is practically appealing, it must be considered that emotion and emotion regulation cannot be separated because they are tightly intertwined [28]. As such, in the current study, and in Turner et al. [21], generation and change beliefs are separate emotion beliefs. If one holds C-M generation beliefs then it is perhaps more likely that one would hold C-M change beliefs, and consequently would be more likely to apply cognitive reappraisal. In contrast, if one holds S-R generation beliefs then it is perhaps more likely that one would hold S-R change beliefs, and consequently would be less likely to apply cognitive reappraisal. Indeed, cognitive change might be more effective in decreasing negative affect when emotions are generated in a top-down (reflecting C-M generation), rather than a bottom-up (reflecting S-R generation), fashion [55]. However, these hypotheses are yet to be tested, but now with the instantiation of questionnaires that assess the four proposed emotion beliefs, in future research it is possible to undertake such an investigation.

Indeed, it is of course possible that one’s generation beliefs are not indicative of change beliefs, but correlations between generation and change factors with S-R and C-M suggests orthogonality because S-R generation is positively related to S-R change, and C-M generation is positively related to C-M change. Also, C-M composite scores are unrelated to S-R composite scores, indicating some independence. However, C-M change is negatively related to both S-R generation and S-R change, indicating that rather than being independent, S-R beliefs are at odds but related with C-M change beliefs. The orthogonality of S-R and C-M beliefs needs to be studied further, but preliminary evidence is indicative of orthogonality. If this is the case, then it is not enough in practice to merely weaken a client’s S-R beliefs, one must also work to strengthen and reinforce C-M beliefs. One cannot assume that low S-R beliefs is illustrative of high C-M beliefs.

The composite C-M and S-R subscales appear to offer some value, in that they are related to emotion regulation and reactivity outcomes, but what is lost in the composite scores are the specific subscales for generation and change, which might differentially predict emotion regulation. Indeed, when comparing generation and change beliefs, it is perhaps the change beliefs that we would expect to be related to actual emotion regulation attempts. If I have strong C-M change beliefs then it stands to reason that I would be more likely to attempt cognitive change, whereas if I have strong S-R change beliefs then I might be more likely to attempt situation selection and or situation modification (e.g., [1]). Also, strong C-M change believers may be more likely to also believe that they can change their own emotions, compared to S-R change believers, who might be more likely to endorse environmental mechanisms in shaping their emotions (i.e., being around the right people). Indeed, in study 4 of the present paper, it is C-M change, not C-M generation, that related to emotion regulation tendencies, perhaps indicating for the C-M subscales, that it is C-M change that is the more likely to predict cognitive reappraisal attempts.

In the present study, greater S-R change beliefs related to greater suppression and lower cognitive reappraisal, but also worse emotion reactivity outcomes. Thus, an individual with high S-R change believes might be less likely to apply cognitive reappraisal. However, this does not mean that S-R change believers would be unable to regulate their emotions in situ via situation change. The limitations of a situation change approach are that one cannot always select or modify a situation, and avoidance of emotionally provocative situations, whilst potentially effective in the short-term, is no long-term strategy for emotional health.

In studies 3 and 4 of the current paper, it was quite clear that S-R change was related to less adaptive emotion regulation tendencies and poorer emotion reactivity tendencies, whilst C-M change in the current paper was related to more adaptive emotion regulation tendencies and greater affective reactivity, and in previous research has been related to poorer emotion reactivity. Thus, S-R change and C-M change do not appear to tell us the same thing, and as such, should be considered as distinct subscales. Similarly, S-R generation and C-M generation scores do not seem to relate to one another and do not relate to adaptive emotion regulation tendencies in the same way. Thus, there seems little rationale for conflating the two generation subscales. So whilst S-R and C-M composite scores, and the CM-SR index might be informative concerning emotion regulation and reactivity, and offer a concise way to perform analyses with the CMBQc, it is recommended that researchers utilise the four separate subscales for the purposes of predicting emotion regulation attempts and emotion reactivity.

The findings of the current paper, and the proposed of S-R and C-M emotion superordinate beliefs per se, should be considered alongside the foundational work that has already been conducted for other emotion beliefs. Specifically, an important line of future enquiry is to understand how S-R and C-M beliefs interacts with the goodness and controllability beliefs that have been the subject of past research (e.g., [22, 27]) and conceptualisation (Ford & Gross, 2018). Theoretically one could venture several hypotheses worth testing regarding S-R, C-M, goodness, and controllability beliefs. For example, we may expect those reporting greater S-R beliefs, alongside lower C-M, lower goodness (emotions are not good), and lower controllability (emotions are not controllable) beliefs, to report poorer psychosocial outcomes and worse emotionality (more emotional intensity, sensitivity, and persistence). Second, we may expect those reporting greater C-M change and S-R change beliefs to also report greater controllability beliefs, given that beliefs concerning change may be related to, dependent upon, or antecedent to, beliefs concerning controllability–I cannot change something that I cannot control.

However, from an emotion regulation perspective the picture may look a little more complex. It could be argued that greater C-M and controllability beliefs, alongside lower S-R beliefs, would lead to cognitive reappraisal attempts in order to regulate emotion. But goodness beliefs may determine whether emotion regulation is even attempted, because believing that an emotion is good may decrease the likelihood that the emotion is identified as needing regulation [17]. So, I may believe that the emotion is generated and regulated via C-M rather than S-R, and that emotions are controllable, but because I believe that the emotion is good, I may not attempt cognitive reappraisal because I do not desire to regulate the emotion. Thus, I may fail to regulate a controllable and dysfunctional emotion, because I view that emotion as good. Therefore, future research could focus on collected data across C-M, S-R, goodness, and controllability emotion beliefs and apply latent profile analysis (e.g., [56, 57]) to explore emotion beliefs profiles, and how these profiles relate to emotion regulation and reactivity outcomes.

In addition to goodness and controllability emotion beliefs, there may be other individual differences that are related to C-M and S-R beliefs. One potential avenue for further investigation would be to assess locus of control alongside C-M and S-R beliefs. Locus of control was posited by John Rotter [58, 59], and individuals can have an internal (we have the ability to control what happens to us) or external locus on control (what happens to us is determined by external forces). In relation to C-M and S-R beliefs it could be that those who endorse high internal locus of control will also endorse high C-M beliefs given that both ideas indicate that we ourselves can determine our destiny, emotional and otherwise. Indeed, one study [60] found that greater external locus of control was related to greater difficulties in regulating emotions and greater adaptive cognitive emotion regulation strategies.

The practical implications of the present study, alongside the findings of Turner et al. (2021), are manifold. In line with second-wave CBTs ([15, 61]), it is of course helpful to encourage clients to recognise the role their cognitions play in the aetiology and regulation of their emotions. C-M beliefs reflect greater volition of emotions, because cognitive change is often more possible when compared to stimulus or situation change (reflected by S-R beliefs). Indeed, given the weight of evidence in support of cognitive reappraisal, it is difficult to see the downsides of encouraging beliefs (like C-M beliefs) that make cognitive reappraisal more likely. It is not that clients are to blame for the external stimuli that often trigger emotion. Rather, we are suggesting that encouraging clients to understand that they can take some control of how their emotions occur and are regulated, will be beneficial for emotion reactivity and regulation. In line with second-wave CBT theory and practice, specifically rational emotive behaviour theory (REBT), we argue that by believing that emotion generation and change is cognitively mediated, one is more able to regulate one’s emotions. Thus, interventions informed by second wave CBTs should include a meaningful discussion with the client concerning their emotion beliefs, with a view to dissuading S-R beliefs and encouraging C-M beliefs. Alongside information psychoeducation concerning emotion beliefs, the CMBQ could be used as a marker of client A-C and B-C thinking pre-, post-, and during CBT to assess emotion belief change, and to assess whether these changes are related to therapeutic outcomes.

The above practical implications notwithstanding, just because an individual endorses C-M beliefs does not mean that they will execute cognitive reappraisal in everyday life, or in specific emotionally provocative situations. There are many factors to consider when predicting the selection and execution of emotion regulation strategies (e.g., social context; [62], regulation goals; [63]). For example, English et al. [64] found that reappraisal is used more when regulating for hedonic reasons (e.g., to feel better; pro-hedonic regulation), and were associate with particular instrumental goals such as getting work done. So, even if I have a strong emotion belief in C-M, my regulation goals will in part inform my selection of emotion regulation strategies. In other words, the influence of emotion beliefs upon emotion regulation has to be considered alongside the reasons why an individual is trying to change an emotion and also what they are trying to change (e.g., down or up regulation), which can be socially informed [19]. As such, C-M and S-R are just one potential piece of a complex puzzle that determined actual emotion regulation.

Conclusion

Across studies 1–3 the initial validity and reliability tests concerning the CMBQb reveal a correlated two-factor structure indicating C-M generation and S-R change beliefs. Results suggest that higher C-M generation lower S-R change beliefs relate to more adaptive emotion regulation, and that greater S-R change beliefs are related to poorer emotion reactivity. Study 4 brought the CMBQa and CMBQb together, revealing a correlated four-factor structure, and some criterion validity. The CMBQc is in need of further examination to confirm and or refine the factor-structure, and ascertain its predictive validity.

Supporting information