Trait self-control and beliefs about the utility of emotions and emotion regulation in self-control performance

ABSTRACT Three studies (N = 555) examined whether emotion regulation and emotions help people higher in trait self-control (TSC) to achieve their goals. Because emotion utility beliefs predict emotion regulation and performance, Study 1a examined whether TSC predicts emotion utility beliefs in two performance contexts, and Study 1b examined whether these beliefs predict preferences to regulate emotions. Study 2 examined whether TSC predicts choice to regulate emotions, and how choice and emotions influence self-control performance. While TSC did not predict emotion regulation, people higher in TSC considered positive (negative) emotions more (less) useful and experienced more (less) positive (negative) emotion after an emotion regulation task, which enhanced their self-control performance. This research underscores the role of emotion regulation and emotions in self-control.

evidence that people with high TSC are more effective at regulating emotions (Paschke et al., 2016), more likely to regulate emotions , and hold different beliefs about which emotions that are most useful to help them achieve their goals (Tornquist & Miles, 2019). Specifically, in three studies we test the proposition that TSC influences how people regulate emotions in performance contexts that require selfcontrol, and that this regulation leads to emotions that facilitate self-control performance.

Emotions and self-control
Several lines of research suggest that positive emotions are more adaptive for self-control than negative emotions, both in laboratory settings and in real-life (as reviewed in Tornquist & Miles, 2019; see also, Aspinwall, 1998;DeSteno, 2018). For example, Raghunathan and Trope (2002) found that inducing a positive mood in high (vs. low) caffeine consumers reduced their intentions toward caffeine consumption, and Winterich and Haws (2011) found that inducing a positive (vs. negative) mood in participants lowered their preferences for unhealthy snacks. Other research has similarly found that negative vs. neutral stimuli impaired performance on a self-control task (stop-signal and Stroop tasks: Frings et al., 2010;Kalanthroff et al., 2013), whereas positive vs. neutral stimuli facilitated performance on a self-control task (Stroop task: Liu et al., 2018; but see, Crossfield & Damian, 2021). In a more naturalistic setting, Niermann et al. (2016) found that higher positive affect predicted more time spent exercising that day, whereas the reverse was true for negative affect, and Vinci et al. (2017) similarly found that positive emotions were associated with a lower risk of relapse in smokers. Thus, this literature provides consistent evidence that positive emotions enhance self-control whereas negative emotions have the opposite effect.
The idea that positive emotions facilitate self-control is consistent with Fredrickson's (2001) broaden-and-build theory. This theory holds that, in contrast to negative emotions, positive emotions enhance peoples' personal resources (e.g., physical, intellectual, social) and motivate them toward future achievements, which ultimately make current challenges such as self-control dilemmas easier to manage (Wilcox et al., 2011;Winterich & Haws, 2011). The mood-maintenance theory similarly suggests that people who experience positive emotions are motivated to maintain these emotional states and therefore refrain from engaging in actions that may reduce them (such as immediate indulgence), which consequently promotes more virtuous choices (Andrade, 2005;Clark & Isen, 1982).
Moreover, as research has linked high TSC with increased and decreased daily experiences of positive and negative emotions, respectively (Hofmann et al., 2014), this might further suggest that these emotional experiences contribute to the success of people with high TSC. Tornquist and Miles (2019) argued that people with high TSC may not only benefit from their increased experience of positive emotions and decreased experience of negative emotions, but may actively pursue the experience of these emotions through emotion regulation in order to maximize these benefits. Emotion regulation involves altering which emotions we have, when we have them, and how we experience and express them (Gross, 1998(Gross, , 2015. While people often regulate emotions for hedonic reasons (i.e., to feel good; Gross et al., 2006;Larsen, 2000), the theory of instrumental emotion regulation holds that people also regulate emotions for instrumental reasons (i.e., to attain goals; Tamir, 2009a). In other words, people want to feel the emotions they believe will help them to succeed.
Indeed, Tamir and colleagues have found substantial evidence that people engage in this type of regulation and that it enhances performance, particularly in highly demanding contexts. For example, Tamir and Ford (2012) found that people instructed to confront (vs. collaborate) a partner in a negotiation context rated anger as more useful, attempted to upregulate (i.e., increase) their anger by choosing to engage in anger-inducing activities, and that anger enhanced their negotiation performance. Moreover, while this research examined the emotions people upregulated to achieve their goals, research has further suggested that people may also downregulate (i.e., decrease) their emotions if they believe it will facilitate goal pursuit, which is consistent with the wider emotion regulation literature showing that downregulating unwanted emotions is a hallmark of successful functioning across domains (Kalokerinos et al., 2017;Zaehringer et al., 2020). For example, Kalokerinos et al. (2017) asked participants to recall the most negative event of their day for seven days and to report their emotion regulation goals (e.g., whether they wanted to reduce or increase negative emotions) and instrumental motives (e.g., whether they regulated emotions to help them with their relationships or to help them to do something) during the event. They found that participants typically wanted to downregulate their negative emotions and that instrumental motives were most prevalent. Thus, while emotion regulation is sometimes considered a form of self-control in itself (Muraven et al., 1998;Paschke et al., 2016), the fact that emotional experiences influence self-control performance means emotion regulation could also be strategically used to enhance selfcontrol.
The hypothesis proposed in Tornquist and Miles (2019), which we expand upon here, is that strategic emotion regulation is a tool used by people with high TSC to achieve their goals. There is evidence that emotion regulation facilitates self-control performance, both in the lab and in real-life (Evers et al., 2010;Hennecke et al., 2019;Juergensen & Demaree, 2015). For example, one study found that instructing participants to adaptively regulate their emotions in the lab helped them to resist unhealthy snacks (Juergensen & Demaree, 2015), and another study found that participants who received instructions to regulate emotions performed better on a laboratory self-control task (the stop-signal task) than control participants (Alberts et al., 2012). An ambulatory assessment study similarly showed that emotion regulation in everyday life predicted success in daily aversive activities . These data indicate that emotion regulation can be used as a tool to achieve self-control success. More evidence that links emotion regulation with enhanced self-control comes from studies linking high TSC with the ability to regulate emotions, showing that people higher, relative to lower, in TSC regulate emotions more effectively in the lab (Paschke et al., 2016), that they are more likely to regulate emotions when faced with aversive activities in real-life , and that emotion regulation mediates the link between high TSC and greater subjective well-being (Nielsen et al., 2019). Bringing these lines of research together, we suggest that emotion regulation is a possible answer to the puzzle of how people with high TSC achieve their goals (Gillebaart & de Ridder, 2015).
Initial support for this hypothesis was found in Tornquist and Miles (2019), who found that people with high TSC were more likely to believe that positive emotions would help them to succeed in everyday self-control situations, while they believed that negative emotions would have the opposite effect on their self-control. Previous evidence on instrumental emotion regulation has found that when people believe that emotions are useful in a particular context, they are more likely to upregulate these emotions, and that these emotions subsequently facilitate performance (Tamir & Ford, 2012;Ford & Gross, 2018; see, Figure 1, for an overview of this theory), although people may also downregulate emotions they consider less useful to their performance in a given situation (Kalokerinos et al., 2017). However, we have not yet demonstrated that these differences in beliefs about emotion utility translate into differences in how people with high TSC regulate emotions in the context of self-control, or that this regulation and subsequent emotions facilitate their self-control performance. Thus, in the current research we test whether people higher and lower in TSC differ in their beliefs about which emotions they consider useful in situations that are high vs. low in self-control demand, whether they regulate their emotions differently in these contexts, and whether this regulation leads to emotions that improve their self-control performance.

Present studies
To test these central hypotheses, we conducted three studies, with each study drawing on a different method from research on instrumental emotion regulation to provide converging evidence for these hypotheses. Study 1a and Study 1b were conducted online via Amazon Mechanical Turk (MTurk); participants accessed the two studies through the same hyperlink but were then assigned to either take part in Study 1a or Study 1b; participants in Study 1a completed a measure about their beliefs about the utility of emotions, whereas participants in Study 1b completed a measure about their preferences to regulate emotions. All other aspects of the two studies were identical. Study 2 was conducted in a laboratory setting. Thus, Study 1a examined the emotions that people with high TSC considered useful in performance contexts that were high and low in selfcontrol demand ( Figure 1a); Study 1b investigated the emotions that people with high TSC preferred to regulate in these contexts ( Figure 1b); Study 2 assessed how people with high TSC chose to regulate emotions in these contexts (Figure 1b), and whether this choice lead to emotional experiences (Figure 1c), which then facilitated their self-control performance (Figure 1d). Materials and data can be found via the Open Science Framework (OSF; https://osf.io/2hte5/?view_only=66482c8b3c2a4a0b9099f5499fabde9f).  Figure 1. Key variables of the instrumental theory of emotion regulation. This theory holds that the more people believe that an emotion will be useful to their performance in a particular context (a), the more likely they are to attempt to upregulate that emotion (as measured by preferences and choices to regulate emotions) in this context (b), and that these regulatory attempts lead to increased emotional experiences (c), which then lead to improvements in performance in this context (d).

Study 1a
Following Tornquist and Miles (2019), we predicted that people higher, relative to lower, in TSC would consider positive emotions more useful and negative emotions less useful to their performance on both the high and low self-control demand tasks, and that these effects would be more pronounced in the case of the high self-control task.

Participants
In Study 1a, 212 participants (N = 101 high self-control demand condition; N = 111 low self-control demand condition) were recruited through MTurk and received $0.75 (M age = 36.65 years, SD age = 12.59; 137 females; 72% White, 13% Asian American, 7% African American, 1% Native Hawaiian or Pacific Islander, and 7% Other). To motivate performance, participants were told that they could win a $25 Amazon Gift Card if they performed well on the tasks. Adults living in the USA participated; all participants reported being fluent in English. The study was approved by the local Ethics Committee. An a priori power analysis in G*power (Faul et al., 2009) indicated that the required sample size for testing our key effects (emotion x self-control demand condition + TSC as a covariate) was N = 128 (power = .80, alpha = .05, effect size f = .25), although we set a target sample size of 200 based on the amount of money we were able to spend on the study. Our final sample exceeded this target sample because 12 additional people voluntarily completed the study via MTurk. After data collection was complete, we conducted a sensitivity power analysis in G*Power, which is thought to be the most informative power analysis as it allows researchers to report the minimum effect size their experiment had 80% power to detect (Perugini et al., 2018). Thus, a sensitivity power analysis (N = 212; power criterion = .80; alpha = .05) for our key test (emotion x self-control demand condition + TSC as a covariate) showed that our study was powerful enough to detect small to medium effects (f = .19), which are commonly obtained in this research area (e.g., Tornquist & Miles, 2019).

Current emotions
Participants rated "How are you feeling right now?" on a 7-point scale (1 = very bad, 7 = very good).

Trait self-control
The Brief Trait Self-Control Scale (Tangney et al., 2004) was used to measure TSC. Participants rated 13 items (e.g., "Sometimes I can't stop myself from doing something, even if I know it is wrong") on a scale of 1 (not at all like me) to 5 (very much like me). Final scores are the mean of the 13 items (9 items are reverse scored; Cronbach's α = .89), and higher scores indicate greater TSC.

Manipulation of expected self-control demand
Participants were assigned to expect to perform a task that was either high or low in selfcontrol demand. Based on self-control research that has suggested that retyping text while breaking the habit of using certain keys require self-control (Muraven et al., 2006), we used a retyping task as our high self-control demand task. Participants assigned to the high self-control demand condition were told that they would be presented with two paragraphs (each consisting of 150 words) taken from a chemistry textbook, and that they would retype as much text as possible in five minutes while following several rules, i.e., without typing a, e or t or hitting the space bar. To ensure that participants understood that the upcoming task would require self-control, they were presented with two sentences taken from a chemistry textbook and practiced the task by retyping the sentences into a textbox while applying the above rules. Participants assigned to the low self-control demand condition were told that they would be presented with a short paragraph (consisting of five sentences) taken from a children's science book, and that they would retype as much text as possible in five minutes. To ensure that participants understood that the upcoming task would require little or no self-control, they were presented with two sentences taken from a children's book and practiced the task by retyping the sentences into a textbox. Across conditions, participants were told that the sentences were taken from the longer paragraphs and that they would retype the remaining text later. Thus, participants received almost identical instructions, with only the self-control demands of the tasks being manipulated. A supplemental study showed that people (N = 26) judged the high self-control typing task to involve more self-control demand (M = 4.45, SD = .1.27) than the low self-control typing task (M = 2.78, SD = 1.14), t(25) = 5.39, p < .001, d = 1.06 (details of this study are available on the OSF website).

Expected emotion utility
Following prior work (Tamir, 2005(Tamir, , 2009bTornquist & Miles, 2019), we assessed how useful participants believed different emotions would be to their success on their assigned typing task (i.e., beliefs about the utility of emotions), given that these beliefs predict emotion regulation and subsequent performance and are therefore often used as an index of emotion regulation (Netzer et al., 2015;Tamir & Ford, 2012). Emotions were chosen based on the Modified Differential Emotions Scale (mDES), which defines each emotion by three adjectives (joy: joyful, glad, or happy; Fredrickson et al., 2003). Participants rated how much they thought positive (α = .91; hope, inspired, joy, gratitude, love, interest, pride, awe, amusement, and serenity) and negative (α = .94; fear, hate, sadness, embarrassment, guilt, disgust, shame, anxiety, anger, and contempt) emotions could help them to succeed on their assigned typing task (e.g., "To what extent do you think feeling hopeful, optimistic, or encouraged will help you to succeed on the typing task?") on a scale of 1 (not at all) to 7 (extremely). Scores were averaged to form one composite score for positive emotions, and one for negative emotions. Higher scores indicate greater utility beliefs for that emotion. 1

Attention checks
To ensure data quality on MTurk (Peer et al., 2014), we included three attention check questions in Study 1a and Study 1b that were embedded within the other measures. For example, to check if participants had read the instructions about the high self-control task, they were presented with the question "What is your goal with this task?", and were then asked to choose either "Retyping as much text as I can from two paragraphs while avoiding typing the letters a, e or t or hitting the space bar" or "Retyping a paragraph as quickly as possible and then take a quiz". Participants (N = 35) who failed to provide the correct answer were immediately thanked and dismissed by the software. That is, they did not complete the remaining tasks and their provided data was disregarded.

Procedure
Participants answered questions about demographics (age, sex, ethnicity), current affect, and TSC. Participants were then instructed that they would complete a typing task and were randomly assigned to the high or low self-control control demand condition. Across conditions, participants were told that they would practice the task by retyping two sentences taken from the longer paragraphs before proceeding to the next part of the study. Participants practiced retyping their respective sentences for five minutes. Following this manipulation, participants were instructed that they would complete a questionnaire before retyping the remaining text, and thus rated how much they thought various emotions could help them to succeed on the typing task. After completing this questionnaire, participants were told that they did not need to retype the remaining text, and thus rated their current affect and were thanked and dismissed. This was done because Study 1a was designed to test the emotions people with high TSC consider useful in performance situations that are high and low in self-control demand, rather than measuring their performance on these tasks. The study took about 10 minutes to complete.

Results and discussion
We conducted a 2(emotion: positive, negative) x 2(self-control demand condition: high, low) mixed ANOVA. Following Tamir (2005Tamir ( , 2009b, emotion was the within-subjects factor, self-control demand condition the between-subjects factor, and TSC (centered) a covariate, allowing us to examine interactions between our within-subjects variables and TSC without dichotomizing TSC. Expected utility ratings was the dependent variable.
The analysis revealed a significant main effect of emotion, F(1, 208) = 926.74, p < .001, η 2 p = .82; participants considered positive emotions (M = 5.05, SE = .08) more useful to their performance than negative emotions (M = 1.75, SE = .07), consistent with prior work (Tornquist & Miles, 2019). The main effect of self-control demand condition was significant, F(1, 208) = 6.63, p = .01, η 2 p = .03; participants considered emotions as more useful for the task low in self-control demand (M = 3.54, SE = .077) than the task high in self-control demand (M = 3.26, SE = .08). But, the interaction between emotion and self-control demand condition was not significant, F(1, 208) = 1.40, p = .24, η 2 p = .007, suggesting that people believe that positive emotions are more useful to their performance on tasks that are both high and low in self-control demand, as compared to negative emotions (see, Figure 2).
The main effect of TSC was not significant, F(1, 208) = 0.07, p = .79, η 2 p = .00, and the interaction between self-control demand condition and TSC was also not significant, F(1, 208) = 0.00, p = .98, η 2 p = .00. Importantly, the predicted interaction between emotion and TSC was significant, F(1, 208) = 28.26, p < .001, η 2 p = .12; people higher (+1 SD) in TSC considered positive emotions more useful for their task performance (estimated M = 5.02) than people lower (−1 SD) in TSC (estimated M = 4.66), β = .23, p = .001, consistent with our predictions. Moreover, people higher (+1 SD) in TSC considered negative emotions less useful for their task performance (estimated M = 2.06) than people lower (−1 SD) in TSC (estimated M = 2.59), β = −.28, p < .001, also consistent with our predictions. These data replicate Tornquist and Miles (2019), showing that people higher (vs. lower) in TSC consider positive emotions more useful and negative emotions less useful for their task performance. These data also extend this work by showing that these beliefs generalize to a lab-based self-control situation and to a performance situation that does not require self-control.
However, the interaction between emotion, TSC, and self-control demand condition was not significant, F(1, 208) = 0.09, p = .76, η 2 p = .00, suggesting that people higher (vs. lower) in TSC believed that positive emotions would be more useful and negative emotions would be less useful for their performance on both the high and low selfcontrol demand tasks. This is inconsistent with our prediction that people higher in TSC would consider positive emotion particularly useful and negative emotions particularly unhelpful for their performance on the high self-control demand task.

Study 1b
Consistent with the theory of instrumental emotion regulation, we expected that people's beliefs about the utility of emotions would translate into preferences to regulate these emotions. Hence, we predicted that people higher, relative to lower, in TSC would demonstrate greater regulatory preferences for positive emotions and lower regulatory preferences for negative emotions when expecting to perform both the high and low selfcontrol demand tasks, and that these effects would be more pronounced in the case of the high self-control task.

Participants
In Study 1b, 203 participants (N = 102 high self-control demand condition; N = 101 low self-control demand condition) were recruited through MTurk and received $0.75 (M age = 35.73 years, SD age = 11.67; 128 females; 74% White, 14% Asian American, 7% African American, 1% Native Hawaiian or Pacific Islander, and 3% other). As in Study 1a, participants were told that they could win a $25 Amazon Gift Card if they performed well. Adults living in the USA participated; all participants reported being fluent in English. The study was approved by the local Ethics Committee.
We conducted the same a priori power analysis and used the same target sample size as described in Study 1a (N = 200). Our final sample exceeded this target sample because three additional people voluntarily completed our study via MTurk. A sensitivity power analysis (N = 203; power criterion = .80; alpha = .05) for our key test (emotion x self-control demand condition + TSC as a covariate) showed that the study was powerful enough to detect small to medium effects (f = .20).

Current emotions
Participants rated the same statement as in Study 1a.

Trait self-control
We used the same TSC measure as in Study 1a (Cronbach's α = .89).

Manipulation of expected self-control demand
This manipulation was identical to the one described in Study 1a.

Regulatory preferences
People's regulatory attempts in performance contexts are often assessed by measuring their preferences or choices to engage in emotion-eliciting activities (e.g., recalling a personal event, watching film clips) that vary by emotion (Tamir, 2005(Tamir, , 2009b. This is because, when people wish to regulate their emotions, they often engage in activities that will increase these desired emotions (Tamir, 2009b). Thus, to assess participant's regulatory preferences, participants rated the extent to which they wanted to recall various personal events that differed in emotional tone, and were told that they would spend 10 minutes writing about one of their highest-rated events in detail right before completing the typing task. Autobiographical recall was chosen because this task is commonly used in research on instrumental emotion regulation and because it is a common and effective way to induce emotions (Lench et al., 2011). In addition, participants wrote a short description of each personal experience. To illustrate, participants were first instructed to "Write a short description of a recent personal event that made you feel hopeful, optimistic, or encouraged". After writing their description into a textbox, they rated the statement "Before you take part in the typing task, to what extent would you like to spend 10 minutes writing about the personal event you just described?" on a scale of 1 (not at all) to 7 (extremely). Participants completed this procedure for the same positive (α = .93; hope, inspired, joy, gratitude, love, interest, pride, awe, amusement, and serenity) and negative (α = .93; fear, hate, sadness, embarrassment, guilt, disgust, shame, anxiety, anger, and contempt) emotions as for the emotion utility survey in Study 1a. Scores were averaged to form one composite score for positive emotions, and one for negative emotions. Higher scores indicate greater preferences to increase (upregulate) that emotion. 1

Attention checks
Participants completed the attention checks described previously.

Procedure
The procedure was identical to Study 1a, with one exception. That is, instead of completing the expected emotion utility questionnaire after practicing the typing task, participants in Study 1b completed the regulatory preferences questionnaire. They rated their current affect and were dismissed. The study took about 10 minutes to complete.

Results and discussion
We conducted the equivalent analyses as in Study 1a, although regulatory preference ratings served as the dependent variable in Study 1b. The ANOVA revealed a significant main effect of emotion, F(1, 199) = 205.64, p < .001, η 2 p = .51; participants preferred to recall positive events (M = 4.28, SE = .10) over negative events (M = 3.03, SE = .10), suggesting that they preferred to regulate their emotions to increase their positive emotions more than their negative emotions. The main effect of self-control demand condition, F(1, 199) = 0.09, p = .77, η 2 p = .00, and the interaction between emotion and selfcontrol demand condition, F(1, 199) = 1.69, p = .20, η 2 p = .008, were not significant, suggesting that people preferred to regulate their emotions to increase their positive emotions more than their negative emotions when expecting to perform both the high or low self-control demand tasks (see, Figure 3).

Study 2
The findings that people higher in TSC considered positive emotions more useful in Study 1a, but did not report greater regulatory preferences for positive emotions in Study 1b, are puzzling given previous demonstrations that people regulate emotions to increase the emotions they consider useful (Tamir & Ford, 2012). Likewise, given that Study 1a showed that people higher in TSC considered negative emotions less useful, they should have reported lower regulatory preferences for negative emotions in Study 1b, although we did not find support for this. One explanation for the lack of association between TSC and regulatory preferences for positive and negative emotions might be that selfreported regulatory preferences is not a valid index of people's real attempts to regulate emotions. Indeed, Tamir and Ford (2012) argued that preferences may not always correspond with real choices and thus conducted tests of regulatory preferences and choices. Hence, in Study 2 we used a behavioral measure of people's regulatory choice, which should better approximate people's true regulatory attempts. Thus, in extending Study 1a and 1b, Study 2 examines whether people higher and lower in TSC differ in their choice to regulate emotions in situations that are high and low in self-control demand, how this choice influence subsequent emotions, and whether these emotions facilitate self-control performance.
Thus, due to the methodological improvements that were made in Study 2 and the large amount of research demonstrating links between beliefs about the utility of emotions, choice to regulate emotions, and subsequent emotions and performance (Tamir & Ford, 2012), we expected that people higher in TSC would choose to regulate their emotions in ways that would lead to emotional experiences that would improve their selfcontrol performance. That is, we predicted that people higher, relative to lower, in TSC would be more likely to choose to upregulate positive (vs. negative) emotions; that higher TSC would predict increased levels of positive emotions and decreased levels of negative emotions after the emotion regulation task; and that increased positive emotions and decreased negative emotions would improve the self-control performance of people high in TSC. We expected these findings to be evident for the tasks high and low in self-control demand, but that they would be more pronounced in the case of the former.

Participants
In Study 2, 140 psychology undergraduate students (N = 70 high self-control demand condition; N = 70 low self-control demand condition) participated in this laboratory study in exchange for course credit (M age = 21.61 years, SD age = 7.16, Range = 18-60 years; 121 females; 86% White, 7% Asian/Asian British, 4% Mixed/Multiple ethnic groups, 3% Black/ Black British, and 1% Other). Initially, 142 participants took part in the study, but two participants were removed due to failing the attention checks (described below), resulting in a final sample of 140. The study was approved by the local Ethics Committee. A sensitivity power analysis (N = 140, power = .80, alpha = .05) for our key test (the multiple regression with self-control demand condition, TSC, emotion, and their crossproducts as predictors, and baseline emotions as covariates) revealed a minimum effect size f 2 of 0.12, suggesting that our study was powerful enough to detect small to medium effects.

Current emotions
As in Study 1a and b, emotions were chosen from the mDES. Participants rated how much they currently experienced four positive emotions (proud, serene, hopeful, joyful), three negative emotions (angry, guilty, anxious), and two filler items (tired, concentrated; e.g., "Right now, to what extent are you feeling proud, confident, or self-assured?"), on a scale of 1 (not at all) to 7 (extremely). Scores were averaged to form one composite score for positive emotions and one for negative emotions before (positive emotions: α = .81; negative emotions: α = .59) and after (positive emotions: α = .89; negative emotions: α = .66) completing the emotion regulation task, with higher scores indicating greater experiences of that emotion.

Trait self-control
Participants completed the same TSC scale (α = .83) as in in Study 1a and b.

Manipulation of expected self-control demand
Participants were told that they would be assigned to a music (i.e., low self-control demand) condition or an anagram (i.e., high self-control demand) condition, but that they would practice both tasks first. We included this practice task to ensure that participants understood the level of self-control demand each upcoming task required, allowing participants to make more informed judgments about the self-control demand of their assigned task when choosing how to regulate emotions, as opposed to if we simply told them that they would perform a difficult or easy task. All participants listened to a neutral music clip (i.e., Baby Sweetcorn (Come Here) by Howie B; adapted from Tamir & Ford, 2012) for 30 seconds, and were told that they would listen to a longer music clip later, if they were assigned to the music condition.
Participants also practiced the anagram task, which asked them to generate words from two anagrams (e.g., OPOER; Calef et al., 1992) that were, unbeknownst to the participants, unsolvable, while following a number of rules, i.e., the words they generated needed to be real English words and consist of all letters, and no letters could be used twice. Participants were instructed to click the "next page" button when they decided they could no longer generate new words or if they wanted to continue the study. Participants were instructed that they would solve more anagrams later, if they were assigned to the anagram condition. As part of this manipulation, after practicing both tasks, half of the participants were told that they had been assigned to listen to more music (low self-control demand condition), whereas the other half were told that they had been assigned to solve more anagrams (high self-control demand condition). The supplemental study described in Study 1a showed that people (N = 26) judged the anagram task to involve more self-control demand (M = 4.86, SD = 1.33) than the music task (M = 2.07, SD = 1.08), t(25) = 7.99, p < .001, d = 1.58.

Emotion regulation paradigm
This paradigm consisted of three tasks in which participants indicated their preferences for completing different emotion inductions, chose one of these inductions to complete, and completed their chosen emotion induction (described in detail below).

Regulatory preferences
As in Study 1b, we measured people's preferences for regulating their emotions. Although our interest in Study 2 was people's regulatory choice, we included this measure to protect against the possibility that all participants would choose to regulate the same emotion, allowing us to, at the very minimum, replicate Study 1b's findings using a different self-control task. Participants were presented with seven short descriptions of everyday events, each event associated with an emotion (i.e., pride, serenity, joy, hope, guilt, anxiety, and anger), and were asked to write a short description of a personal event that matched each of these descriptions. For example, rather than explicitly asking participants to write a short description about an event that made them feel proud as in Study 1b, participants in Study 2 were asked to write a short description of a personal event in which they succeeded at something that they had worked hard for (descriptions adapted from Gilead et al., 2016;Winterich & Haws, 2011). Participants were presented with descriptions rather than actual emotions given that we only included seven emotions in Study 2, and thus to avoid drawing participants attention to the emotions of interest. Participants were asked to write short personal descriptions that matched the descriptions of everyday events to ensure that they considered events from their personal lives before they proceeded to the next task, which involved choosing one event to describe in detail. Similar to Study 1b, participants also rated how much they wanted to spend 10 minutes writing about each of the seven events in detail before completing their assigned task (e.g., "Before solving the anagrams, to what extent would you like to spend 10 minutes writing about the personal event you just described?") on a scale of 1 (not at all) to 7 (extremely). Scores were averaged to form one composite score for positive emotions (α = .63: pride, serenity, joy, and hope) and one for negative emotions (α = .40; guilt, anxiety, and anger).

Regulatory choice
After providing their regulatory preference ratings, to assess participant's regulatory choice Tamir & Ford, 2012), participants were presented with a list of the personal events that they wrote about in the previous part (and that were associated with either pride, serenity, joy, hope, guilt, anxiety, or anger), and were asked to select the one event that they wanted to write about in detail before solving more anagrams or listening to more music. Choosing pride, serenity, joy, or hope was considered a positive regulatory choice, and choosing guilt, anxiety, or anger was considered a negative regulatory choice.

Emotion induction
Another key feature of research on instrumental emotion regulation is that participants typically engage with their chosen emotion-eliciting activity, which leads to increases in the chosen emotion (e.g., Tamir & Ford, 2012). Participants in the current study regulated their emotions by recalling a past personal event (also referred to as "emotion regulation task" herein). Hence, after making their regulatory choice, participants were asked to vividly imagine themselves in that situation and to write about the event in as much detail as possible for about 10 minutes. Average time spent on the task was 8.65 minutes (SD = 1.91). Time spent on the writing task did not differ for participants who wrote about a positive (M = 8.54, SD = 1.97) and a negative (M = 8.99, SD = 1.74) event, t(138) = 1.23, p = .22, d = .24, or for participants assigned to the high (M = 8.70, SD = 1.89) and low (M = 8.61, SD = 1.95) self-control demand conditions, t(138) = .25, p = .80, d = .05. Thus, participants actively engaged with the task.

Self-control performance
Self-control research has suggested that attempting to solve unsolvable anagrams requires self-control (Muraven et al., 1998), and anagram tasks are therefore commonly used as dependent measures of behavioral self-control in laboratory settings (Hagger et al., 2010). Thus, we used an anagram task to measure self-control performance in Study 2; participants were asked to generate as many words as possible from seven anagrams, while following the rules described above. Unbeknownst to the participants, five of the anagrams were unsolvable (e.g., RATKN), whereas two were solvable (e.g., AHTRE; adapted from Calef et al., 1992), so as not to make the task seem impossible. Time spent on the task (i.e., persistence in the face of failure; Muraven et al., 1998), was our measure of selfcontrol performance. Average time spent on this task was 6.06 minutes (SD = 3.32).

Attention checks
We included two attention check questions in Study 2 that were embedded within the other measures (e.g., "Indicate a 3 on the scale if you are paying attention"). Participants (N = 2) who failed to provide the correct answers were excluded from the data.

Procedure
After reading and signing the consent form, participants answered questions about demographics (age, sex, ethnicity, native language), current emotions, and TSC. Participants were then instructed that they would be assigned to either solve anagrams or listen to music later, but that they would practice both tasks first. Thus, after practicing solving the anagrams and after listening to the short instrumental music clip (in that order), half of the participants were randomly assigned to the high selfcontrol demand condition and were told that they were going to solve more anagrams, whereas the other half were randomly assigned to the low self-control demand condition and were told that they were going to listen to more music. Participants in both conditions were instructed that they would complete a writing task before solving more anagrams or listening to more music. Hence, participants completed the three-part emotion regulation paradigm; they first provided descriptions of personal events and rated how much they would prefer to write about each (regulatory preferences), they then selected the event that they wanted to describe in detail (regulatory choice), and then spent about 10 minutes writing about their selected event (emotion induction). A timer was displayed on the screen to remind the participants of the time. Participants next rated their current emotions again, allowing us to examine how regulatory choice influenced subsequent emotions. Participants assigned to the low self-control demand (i.e., music) condition were then told that a change had occurred and that they had been assigned to solve more anagrams instead of listening to more music. This procedure allowed us to obtain comparable measures of self-control performance for participants who had regulated their emotions expecting self-control demand, compared to participants who had regulated their emotions not expecting self-control demand. Participants were told to click the "next page" button if they accepted the change, or to inform the experimenter if they wished to stop the study or ask any questions. All participants chose to continue, and thus performed the self-control task next. Participants were instructed to click the "next page" button when they decided they could no longer generate new words or if they simply wanted to stop the task. Once they clicked "next page", participants advanced to a new screen which informed them that they had completed the study. Participants were debriefed, thanked and dismissed. All tasks were completed on a desktop computer and took up to one hour to complete. Figure 4 demonstrates all measures and tasks administered at the various stages during the experiment.
These findings suggest that participants were more likely to choose to recall positive events than negative events overall, both when they expected to perform a task that was high and low in self-control demand, which might suggest that they were more likely to attempt to regulate their emotions to increase their positive, compared to their negative, emotions in these situations. Our findings further suggest that people higher and lower in TSC do not differ in their choice to regulate emotions in these situations, inconsistent with our prediction that higher TSC would predict choice to regulate positive (vs. negative) emotions.

Regulatory choice and subsequent emotions
To test whether regulatory choice influenced emotions experienced after the emotion regulation task, and whether this varied by TSC, we conducted two multiple regressions; TSC (centered), self-control demand condition (high, low), regulatory choice (positive, negative), and their cross-products were the predictors; baseline positive and negative emotions (centered) were covariates; and positive and negative emotions experienced after the emotion regulation task were outcomes. We expected that higher TSC would predict greater experiences of positive emotions and lower experiences of negative emotions after the emotion regulation task, particularly in the case of the high selfcontrol demand task.

Positive emotions
The multiple regression was significant, R 2 = .55, F(9, 130) = 17.84, p < .001. Participants who chose positive recall reported experiencing more positive emotions than participants who chose negative recall, β = −.48, p = <.001 (see , Table 1), while controlling for baseline emotions. The effect of self-control demand condition, β = −10, p = .16, and the regulatory choice x self-control demand condition interaction, β = .07, p = .41, were not significant. These findings suggest that the positive emotion regulation task successfully increased positive emotions in participants, and that it influenced participant's positive emotions similarly in the two conditions. Importantly, greater TSC predicted increased positive emotions, β = .18, p = .048, while controlling for baseline emotions. TSC did not interact with self-control demand condition, β = −.14, p = .12, or regulatory choice, β = .05, p = .58, to predict positive emotions, and the interaction between TSC, self-control demand condition, and regulatory choice was also not significant, β = −.008, p = .93. These data provide tentative evidence that, when controlling for baseline emotions, people higher (vs. lower) in TSC experience more positive emotions after completing an emotion regulation task, consistent with our predictions, although this was independent of the self-control demands of the expected task and regulatory choice, inconsistent with our predictions.

Negative emotions
The multiple regression was significant, R 2 = .48, F(9, 130) = 13.14, p < .001. Participants who chose negative recall reported experiencing more negative emotions than those who chose positive recall, β = .34, p = < .001 (see , Table 1), while controlling for baseline emotions. The effect of self-control demand condition, β = .05, p = .55, and the regulatory choice x self-control demand condition interaction, β = .10, p = .31, were not significant. These data suggest that the negative emotion regulation task successfully increased negative emotions in participants, and that it influenced participant's negative emotions similarly in the two conditions. TSC did not predict negative emotions, β = −.08, p = .42, or interact with self-control demand condition to predict negative emotions, β = −.03, p = .78, but TSC interacted with regulatory choice to predict negative emotions, β = −.28, p = .002, while controlling for baseline emotions, although the above effects were qualified by a significant interaction between TSC, self-control demand condition, and regulatory choice, β = .19, p = .035. Specifically, in the high self-control demand condition, TSC interacted with regulatory choice, β = −.36, p < .001, and follow-up analyses showed that greater TSC predicted lower levels of negative emotions after recalling negative events, β = −.61, p = .002, but not after recalling positive events, β = −.11, p = .35. However, in the low self-control demand condition, TSC did not interact with regulatory choice to predict negative emotions, β = .03, p = .76, showing that TSC did not predict negative emotions after recalling negative or positive events. These data provide preliminary evidence that people higher (vs. lower) in TSC experience less negative emotions after completing an emotion regulation task intended to increase these emotions, but only when the expected task is high (but not low) in self-control demand. 2

Emotions and self-control performance
A simple regression first showed that greater TSC generally predicted enhanced selfcontrol performance, β = .17, p = .04. To test whether emotions experienced after the emotion regulation task (i.e., higher levels of positive emotions and lower levels of negative emotions) enhanced the self-control performance of people with high TSC, we conducted multiple regressions; self-control demand condition (high, low), experiences of either positive or negative emotions (centered), TSC (centered), and their cross products were entered into multiple regressions to predict self-control performance, while controlling for baseline positive and negative emotions (centered).

Positive emotions
The multiple regression was significant, R 2 = .16, F(9, 130) = 2.67, p = .007. The simple effects of self-control demand condition, β = .10, p = .23, positive emotions, β = .09, p = .50, and TSC, β = .11, p = .35, and the interaction between self-control demand condition and TSC, β = .05, p = .64, did not predict self-control performance. But, self-control demand condition interacted with positive emotions, β = −.29, p = .01; greater positive emotions predicted enhanced self-control performance in the high self-control demand condition, β = .27, p = .025, but positive emotions did not predict self-control performance in the low self-control demand condition, β = −.15, p = .23. These findings suggest that positive emotions facilitate self-control performance, but only when people initially expected the task to be high in self-control demand. While the interaction between positive emotions and TSC was not significant, β = .16, p = .08, we conducted follow-up tests in order to test our hypothesis that positive emotions would improve the self-control performance of people high in TSC. Simple slope tests showed that people higher in TSC demonstrated better self-control performance (estimated M = 6.98) than people lower in TSC (estimated M = 5.67) when they experienced high levels of positive emotions (+1 SD), β = .24, p = .038. But, people higher in TSC did not demonstrate better self-control performance (estimated M = 6.86) than people lower in TSC (estimated M = 6.36) when they experienced low levels of positive emotions (−1 SD), β = .09, p = .42. The interaction between self-control demand condition, positive emotions, and TSC was not significant, β = −.16, p = .10.
These preliminary findings indicate that people higher (vs. lower) in TSC experience more positive emotions after completing an emotion regulation task, while controlling for baseline emotions, with one potential explanation being that they are more likely to upregulate these emotions, independent of whether they chose to recall a positive or negative event. These findings may further suggest that increased experiences of positive emotions then help people with high TSC to succeed at self-control, with people higher (vs. lower) in TSC performing better on a self-control task when they experience high (but not low) levels of positive emotions, although these findings should be regarded as tentative.

Negative emotions
The multiple regression was significant, R 2 = .13, F(9, 130) = 2.17, p = .028. The simple effects of self-control demand condition, β = .15, p = .09, negative emotions, β = -.25, p = .12, and TSC, β = .01, p = .93, and the interactions between self-control demand condition and TSC, β = .08, p = .51, and between self-control demand condition and negative emotions, β = .13, p = .37, were not significant. Importantly, the interaction between negative emotions and TSC was significant, β = −.31, p = .016, although these effects were qualified by a significant interaction between self-control demand condition, negative emotions, and TSC β = .28, p = .02. That is, in the high self-control demand condition, there was a significant interaction between negative emotions and TSC, β = −.36, p = .01. Simple slope tests showed that people higher in TSC demonstrated better self-control performance (estimated M = 8.29) than people lower in TSC (estimated M = 6.51) when they experienced low levels of negative emotions (−1 SD), β = .32, p = .025. But, people higher in TSC did not demonstrate better self-control performance (estimated M = 2.42) than people lower in TSC (estimated M = 4.31) when they experienced high levels of negative emotions (+1 SD), β = -.34, p = .12. In the low self-control demand condition, the interaction between negative emotions and TSC was not significant, β = .13, p = .33, suggesting that lower levels of negative emotions did not enhance the self-control performance of people higher in TSC when they initially expected this task to be low in self-control demand.
Thus, tentative evidence was found that people higher (vs. lower) in TSC experience less negative emotions after completing an emotion regulation task intended to increase these emotions, but only when expecting to perform a task high (but not low) in selfcontrol demand, with one speculation being that people with high TSC are more likely to downregulate these emotions, and that they do so in the context of self-control specifically. Preliminary evidence further showed that, when expecting to perform a task high in self-control demand, these decreased experiences of negative emotions then help people high in TSC to succeed at self-control, with people higher (vs. lower) in TSC performing better on a self-control task when they experience low (but not high) levels of negative emotions.

Replication of study 1b: TSC and regulatory preference ratings
We conducted the equivalent ANOVA as in Study 1b to examine the association between TSC and regulatory preference ratings for positive and negative emotions. Study 2ʹs findings were identical to Study 1b's findings, again suggesting that TSC does not predict differences in preferences to regulate emotions (results reported on the OSF website).

General discussion
Building on insights from the instrumental theory of emotion regulation (Tamir, 2009a), three studies examined how differences in TSC influenced beliefs about the utility of emotions and emotion regulation in performance contexts that were high and low in self-control demand. We found evidence that people believed that positive vs. negative emotions would be more useful to their performance in these contexts and that they demonstrated greater attempts to regulate their emotions to increase positive vs. negative emotions, although the self-control demand of the situation did not moderate these effects. We further found that increased positive emotions facilitated self-control performance when people initially expected to perform a self-control task. There was mixed support for our prediction that TSC would influence these processes. People higher vs. lower in TSC believed that positive emotions would be more useful and that negative emotions would be less useful to their performance across contexts, replicating Tornquist and Miles (2019). Moreover, while TSC did not predict attempts to regulate emotions, people higher vs. lower TSC experienced higher levels of positive emotions and lower levels of negative emotions after the emotion regulation task, which enhanced their subsequent self-control performance, although these findings should be regarded as preliminary. Tornquist and Miles (2019) showed that people consider positive emotions more useful for their everyday self-control success than negative emotions, and that people higher, relative to lower, in TSC consider positive emotions more useful and negative emotions less useful in these situations. Study 1a replicates and extends this work by showing that these beliefs may generalize to a lab-based self-control task, and a performance task that does not require self-control. Specifically, consistent with our predictions, the results of Study 1a provide preliminary evidence that people consider positive (vs. negative) emotions more useful to their performance on tasks high and low in self-control demand, and that people higher, relative to lower, in TSC consider positive emotions more useful and negative emotions less useful to their performance on these tasks, as measured by the extent to which people believed that different emotions could help them to succeed on their assigned typing task. But, the self-control demand of the situation did not moderate these effects, inconsistent with our predictions (potential reasons for this are discussed below). Thus, Tornquist and Miles's (2019) study and Study 1a suggest that people consider positive emotions more useful to their performance than negative emotions in various self-control situations and in one situation that does not require self-control, and that people higher, relative to lower, in TSC consider positive emotions more useful and negative emotions less useful to their performance in these situations.

Emotion utility beliefs
Hence, given the consistent evidence showing that positive (vs. negative) emotions are more adaptive for self-control and for performance in general (Aspinwall, 1998;Fredrickson, 2001), our findings might suggest that people's beliefs about how emotions influence performance in situations that are high and low in self-control demand are accurate, and that people with high TSC might possess a particularly good understanding of this, given that they were more likely to endorse these beliefs. Furthermore, because research on instrumental emotion regulation propose that people upregulate the emotions they consider useful in performance contexts, and that these emotions then are useful to their performance (Ford & Gross, 2018;Tamir et al., 2015), our findings may further imply that people might attempt to upregulate positive emotions in situations that are high and low in self-control demand, and that people with high TSC might be particularly likely to do so. Furthermore, while previous research has mainly emphasized the emotions people consider helpful and therefore upregulate in performance contexts, some research has suggested that people also downregulate their emotions if they believe it will facilitate goal pursuit (Kalokerinos et al., 2017). Our findings add to this work by showing that people higher (vs. lower) in TSC consider negative emotions less useful in situations that are high and low in self-control demand, suggesting that they might also attempt to downregulate negative emotions in these situations, although this speculation remains to be tested.

Attempts to regulate emotions
In extending Tornquist and Miles's (2019) study, Study 1b and Study 2 allowed us to examine people's attempts to regulate emotions, as measured by their preferences and choice to engage in emotion-eliciting activities (i.e., personal recall) that varied by emotion (this paradigm is discussed below). We found that participants demonstrated greater preferences, and were more likely to choose, to engage in activities aimed at increasing positive vs. negative emotions, when expecting to perform tasks that were high and low in self-control demand. One interpretation of these findings might be that people's beliefs that positive emotions help them to succeed in situations that are high and low in selfcontrol demand might translate into greater preferences and choice to regulate positive emotions in these situations, providing indirect support for the view that people upregulate the emotions they consider useful in performance contexts, and that this generalizes to performance contexts that require self-control, although these pathways need to be tested more directly in future studies. But, our findings do not support the view that these beliefs and regulatory attempts should be greater in highly demanding contexts , as self-control demand failed to moderate emotion utility beliefs and regulatory attempts in our study. Our findings therefore provide tentative evidence that people believe that positive emotions can help them to succeed in situations that are high and low in self-control demand, and that they might attempt to upregulate positive emotions in both situations.
One potential explanation for these findings could be that positive emotions are equally adaptive in situations that are high and low in self-control demand, and that participants' beliefs about the utility of positive emotions and attempts to regulate positive emotions were therefore similar in these situations. This is perhaps unsurprising given that positive (vs. negative) emotions are more adaptive for performance in general (Fredrickson, 2001). Another explanation might be due to the nature of the self-control task. Research has argued that laboratory self-control tasks require little or no self-control (Shenhav, 2017), which could explain why participants' emotion utility ratings and regulatory preferences were similar across the high and low self-control demand tasks. Likewise, it is possible that the typing tasks high and low in self-control demand in Study 1a and 1b were both relatively effortful, resulting in similar emotion utility ratings and regulatory preferences for these tasks. But we believe that this is unlikely, given that our task high in self-control demand was pre-rated as being high in self-control demand (M = 4.45), and as requiring significantly more self-control than our task low in self-control demand (M = 2.78; p < .001). Perhaps one challenge for future research involves identifying tasks that are even higher in self-control demand, and test whether the current findings replicate.
Moreover, while we used an emotion regulation paradigm frequently used in research on instrumental emotion regulation (e.g., Tamir & Ford, 2012), future research could consider using a more explicit emotion regulation task that instruct participants how to regulate emotions in various ways, and then assess their preferences and choice to regulate emotions in situations high and low in self-control demand. In addition, future studies could also assess why people might regulate their emotions in these contexts, given that people regulate emotions for different reasons (e.g., performance, social reasons: Kalokerinos et al., 2017).
Importantly, we further found that preferences and choice to regulate emotions did not vary by TSC, suggesting that TSC predicted differences in beliefs about the utility of emotions but failed to predict differences in choice to regulate emotions, inconsistent with the instrumental emotion regulation framework and our predictions. One reason for why TSC failed to predict attempts to regulate emotions might be that people with high TSC regulate emotions implicitly rather than making deliberate choices about how to regulate emotions. This is consistent with research suggesting that emotion regulation can operate on automatic or implicit levels, and that people with high TSC rely on habitual automatic behaviors to achieve their goals (Converse et al., 2019;Gillebaart & de Ridder, 2015;Koole et al., 2015;De Ridder et al., 2012). Thus, perhaps TSC failed to predict preferences and choice to regulate emotions because making such choices may not be a part of a high TSC person's daily routines, who may regulate emotions without awareness and conscious control to achieve their goals. Researchers taking an instrumental approach to emotion regulation should incorporate implicit measures of emotion regulation, and test whether people with high TSC regulate emotions implicitly in self-control situations, and whether this facilitates their self-control performance.
Another reason for why TSC failed to predict attempts to regulate emotions might be that people with high TSC do not need to regulate emotions in the context of self-control at all, given that they tend to experience increased and decreased positive and negative emotions, respectively (Hofmann et al., 2014), which might contribute to their self-control success. But this interpretation seems unlikely given that research has found that TSC predicts emotion regulation in real-life and in the lab Nielsen et al., 2019;Paschke et al., 2016), and because Study 2 showed that TSC predicted differences in emotions immediately after the emotion regulation task and right before performing the self-control task while controlling for baseline emotions. Specifically, greater TSC predicted increased positive emotions after recalling the events overall, and greater TSC further predicted decreased negative emotions after recalling negative events in the situation that was high (but not low) in self-control demand, after controlling for baseline emotions. Thus, it seems that people with high TSC believe that positive emotions are more useful and that negative emotions are less useful to their performance, and that they experience increased positive and decreased negative emotions after completing an emotion regulation task and right before performing a self-control task. These data provide preliminary evidence that the emotions people with high TSC consider useful in performance contexts may translate into how they feel in these contexts, although future studies need to test these links more directly.
One potential explanation for these differences in emotions may be that people higher and lower in TSC regulated their emotions differently, independent of their preferences and choice to regulate emotions. As suggested previously, people with high TSC may have regulated their emotions implicitly and habitually, resulting in increased positive emotions overall and decreased negative emotions when recalling a negative event and when expecting to perform a task high in self-control demand. Alternatively, they may have used other emotion regulation strategies (see, English et al., 2017 for strategies), resulting in these differences in emotions. This speculation is, for example, consistent with research showing that high TSC predicts greater subjective well-being through the use of cognitive reappraisal (Nielsen et al., 2019), which might suggest that people with high TSC used cognitive reappraisal to increase their positive emotions and decrease their negative emotions in our study. Thus, future research could examine which specific emotion regulation strategies people with high TSC use in the context of self-control, and how this influence self-control performance. For example, given that people with high TSC believed that negative emotions would be less useful to their performance, and that they experienced less negative emotions after recalling negative events and when expecting to perform a self-control task, an interesting direction would be to examine the emotion regulation strategies they use to decrease negative emotions (e.g., reappraisal, distraction: Gross, 1998Gross, , 2015, and how this influence self-control performance, perhaps by using the experience sampling method (Baumeister et al., 2019). Relatedly, Hennecke et al. (2019) found that people with high TSC were not only more likely to use emotion regulation strategies in aversive self-regulatory activities, but they were also more likely to focus on the positive consequences of performing these activities and to set goals. Future research should therefore examine which alternative strategies people with high TSC might use in self-control situations and how this influence their success.
Another interpretation could be that, rather than regulating emotions differently, people higher and lower in TSC responded differently to the emotion regulation task intended to increase emotions, resulting in differences in emotions. For instance, research has shown that adolescents higher and lower in TSC respond differently (demonstrate less reactivity) to daily stressors (Galla & Wood, 2015), indicating that they better cope with negative experiences, which could explain why participants with high TSC reported less negative emotions after the negative recall. Others have similarly examined individual differences in response to emotion inductions and have found that traits such as extraversion and neuroticism predict differences in emotions after an emotion induction task (Larsen & Ketelaar, 1991; see, also Blackburn et al., 1990;Scherrer & Dobson, 2009). These data underline the importance of considering individual differences when using emotion inductions to increase emotions, and warrant continued research into whether TSC predicts emotions in response to emotion inductions. Such research could disentangle whether the observed differences in emotions herein are due to emotion regulation or reactivity to emotional events (though it is debated whether such constructs can be distinguished; Gross & Barrett, 2011).

Self-control performance
Consistent with research showing that positive emotions facilitate self-control performance (Liu et al., 2018;Winterich & Haws, 2011), Study 2 found that higher levels of positive emotions predicted enhanced self-control performance, but only when people initially expected the task to be high (but not low) in self-control demand. However, inconsistent with previous research (Frings et al., 2010;Niermann et al., 2016), negative emotions did not predict worse self-control performance. Moreover, consistent with research suggesting that people higher, relative to lower, in TSC perform better on behavioral tasks that require self-control (Friese & Hofmann, 2009;Schmeichel & Zell, 2007), Study 2 showed that people higher in TSC persisted longer on a self-control task than people lower in TSC, although this effect was relatively small (β = .17). Importantly, consistent with our predictions, people higher in TSC demonstrated better self-control performance than people lower in TSC when they experienced high levels of positive emotions across situations, and when they experienced low levels of negative emotions in the situation high in self-control demand. While these data should be regarded as preliminary, they are consistent with research linking high TSC with increased and decreased daily experiences of positive and negative emotions, respectively (Hofmann et al., 2014), and therefore provide tentative evidence that these emotions might help people with high TSC to succeed on behavioral tasks that require self-control (Tornquist & Miles, 2019), although more research is needed to confirm these initial findings. These data might also explain why people with high TSC (presumably) regulated their emotions in ways that resulted in increased positive and decreased negative emotions after the personal recall; because these emotions facilitate their self-control performance. However, these interpretations are speculative and warrant further investigation.
Interestingly, throughout this research TSC seemed to be more strongly linked with negative emotions than positive emotions. For example, TSC had a stronger effect on negative emotions after recalling negative events in the situation that was high in selfcontrol demand (β = −.61) than on positive emotions after recalling the events overall (β = .18). Moreover, the finding that low levels of negative emotions enhanced the self-control performance of people with high TSC in the situation high in self-control demand (β = .32) was somewhat stronger than the finding that high positive emotions enhanced the performance of people with high TSC across situations (β = .24). These data are consistent with Tornquist and Miles (2019) who found more convincing evidence to link TSC with utility ratings of negative emotions than utility ratings of positive emotions, with one possible explanation for this being that downregulating negative emotions might play a particularly important role in improving the self-control performance of people with high TSC, providing an interesting direction for future studies.
Thus, consistent with the instrumental emotion regulation framework and Tornquist and Miles (2019), our findings provide tentative evidence that people with high TSC believe that positive emotions are more useful and that negative emotions are less useful to their self-control performance, that they experience more of these emotions after an emotion regulation task and right before performing a self-control task, and that these emotional experiences, particularly low levels of negative emotions, then are useful to their self-control performance, providing a basis for further investigations to confirm these initial findings. Our findings further suggest that people with high TSC may not achieve these increased experiences of positive emotions and decreased experiences of negative emotions by making choices about how to regulate emotions as predicted by the instrumental emotion regulation framework, perhaps because they may regulate emotions implicitly and are thus unaware of the emotion regulation strategies that lead to increased positive and decreased negative emotions, although other explanations are, of course, plausible.
Despite these promising findings, they are of correlational nature and should be regarded cautiously; future studies need to confirm whether increased positive and decreased negative emotions help people with high TSC to succeed at self-control, especially since the effect of emotions on self-control might depend on the context or self-control task. For instance, positive emotions may influence self-control differently depending on whether the task requires the initiation or inhibition of behaviors (initiatory vs. inhibitory self-control: Tornquist & Miles, 2019). We found that positive emotions boosted self-control performance for people with high TSC on an anagram task, which is assumed to involve initiatory self-control (Imhoff et al., 2014), indicating that our results may be specific to tasks that require initiation. This is consistent with research showing that people believe that positive emotions are more useful to their performance in situations that require them to initiate vs. inhibit behaviors (Tornquist & Miles, 2019), and research that has linked positive emotions with behaviors that require initiatory selfcontrol studying (de Boer et al., 2011) or with initiation in general (Albarracin & Hart, 2011). Moreover, because emotions may influence self-control differently depending on the task or behavior, different emotion regulation strategies might also be helpful in different self-control situations. For example, Alberts et al. (2012) found that participants who used acceptance as an emotion regulation strategy performed better on a selfcontrol task than a control group and participants who used suppression. It therefore seems plausible that people's emotion utility beliefs and preferences and choices to regulate emotions may differ depending on the self-control task or behavior. These speculations warrant further investigation; research could, for example, examine how positive emotions influence self-control performance when the task involves initiatory vs. inhibitory self-control, how people regulate emotions in these contexts, and whether this vary by TSC.
Finally, we note that, given that emotion regulation and self-control conceptually overlap, and that the association between TSC and emotion regulation is only correlational (Paschke et al., 2016), future research should investigate the exact associations between TSC, emotion regulation, and self-control performance.

Conclusions
This research tested the instrumental theory of emotion regulation to shed new light on whether people, especially people with high TSC, use emotion regulation and emotions to achieve self-control success. Our findings support this theory by providing tentative evidence that beliefs about the utility of emotions in performance contexts might translate into attempts to regulate emotions in these contexts, and extend this theory by showing that this generalize to performance contexts that require self-control and that are low in demand. But these processes may differ in people higher vs. lower in TSC. That is, while TSC predicted beliefs about the utility of emotions, emotions experienced after an emotion regulation task, and improved self-control performance, TSC failed to predict attempts to regulate emotions, providing preliminary evidence that their beliefs about the utility of emotions might translate into emotional experiences and subsequent performance. This highlights the importance of considering differences in TSC when testing this theory, and suggest that useful emotions can be achieved by strategies other than regulatory choices. Hence, researchers testing this theory should consider direct pathways from emotion utility beliefs to emotional experiences and performance, and include alternative measures of emotion regulation. Thus, we found preliminary evidence that people with high TSC might use emotions to achieve self-control success, but whether they achieve these emotions by making choices about emotion regulation remains unclear.

Notes
1. Participants also completed Goldberg's (1999) Big Five International Personality Item Pool scales in Study 1a and 1b for exploratory purposes. While we have not analyzed this data, the materials and raw data are available via the OSF website. 2. Given the interesting finding that greater TSC predicted more positive and less negative emotion after the recall when controlling for baseline emotions, we conducted exploratory content analyses (LIWC: Pennebaker et al., 2007) to examine whether these differences in emotions could be explained by the content people higher and lower in TSC wrote about (for example, perhaps people with high TSC had more positive experiences to write about, resulting in increased positive emotions). Thus, we analyzed whether TSC predicted the percentage of positive and negative words used in the essays. These data demonstrated no differences in content between people higher and lower in TSC, suggesting that differences in emotions after the recall cannot be explained by the content. Results are available at the OSF website.

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

Data availability statement
Data and materials are available in the Open Science Framework repository (https://osf.io/2hte5/? view_only=66482c8b3c2a4a0b9099f5499fabde9f).