Entertainment as a Creature Comfort: Self-Control and Selection of Challenging Media

ABSTRACT

A between-subjects experiment examined selective exposure to films in an imagined self-control scenario, and if exposure would be systematically related to perceptions of the film content as challenging, enjoyable, and a should versus a want choice. Across 3 measures of selective exposure—using open-ended choice, closed-ended choice, and prospective ratings—participants in the depletion condition were less likely to select films that were cognitively challenging, affectively challenging, or a should choice. In contrast to nondepleted participants, depleted participants were more likely to select films they expected being fun, suspenseful, and less appreciated. These results provide support for the proposition that users’ momentary self-control capacity and their perception of challenge provided by content predict media choice.

Media use has long been understood as a self-regulatory activity in that users may select media to maintain or intervene in existing mood states (Zillmann & Bryant, 1985). A specific subset of self-regulatory behavior is self-control, “the capacity for altering one’s own responses” to meet goals or obligations (Baumeister, Vohs, & Tice, 2007, p. 351). Self-control is theorized to be a limited resource; that is, after engaging in self-control for a prolonged period, people tend to experience fatigue and change their behavior and goals in predictable ways. Under conditions of depleted self-control, individuals have less willpower and persistence, but more impulsiveness (Baumeister et al., 2007).

Multiple studies suggest that low self-control may relate to media use, particularly with regard to selection of entertaining media content (Brosius, Rossmann, & Elnain, 1999; Hofmann, Baumeister, Förster, & Vohs, 2012; Panek, 2014; Reinecke, 2009; Reinecke, Hartmann, & Eden, 2014; Wagner, Barnes, Lim, & Ferris, 2012). Although this may be functional in the short term for mood management or recovery, it also suggests that depleted individuals may be systematically shying away from specific types of media content, particularly those that may promote long-term self-expansion or personal growth. Self-control allows people to override their impulsive inner animal and engage in the planning, cooperation, and culture that characterize humanity (Baumeister et al., 2007). When ability to control the self is diminished, will people avoid challenging media choices and turn to the creature comforts of easy, nonchallenging choices? Initial findings suggest that this may be the case (Eden, Hartmann, & Reinecke, 2015), however, these findings’ generalizability is limited by selective exposure methodological concerns (Clay, Barber, & Shook, 2013) and concerns about ego depletion laboratory paradigms (Hagger & Chatzisarantis, 2016).

Our study addresses these limitations by examining the role of depletion in media selection by using a three-part analysis of selective exposure in terms of the challenge afforded by media content, under conditions of an imagined loss of self-control. To do so, we employ recent conceptualizations of media content as affectively or cognitively challenging (Bartsch & Hartmann, 2017), as well as the notion of media options as reflective of should versus want choices analyzed in consumer decision-making literature (Bitterly, Mislavsky, Dai, & Milkman, 2015). The findings have implications not just for media psychology, but also for societal debates regarding the perceived virtue of selecting specific types of media content over others, as well as the role of media entertainment in our daily lives.

Self-control and media use

As noted, the link between self-control and media use has become an important concern for both psychologists and media psychologists. For example, Brosius et al. (1999) found that individuals who felt more stressed after work spent more time watching entertainment than others. Similarly, Wagner et al. (2012) showed that reduced self-control capacity is a strong predictor of personal media use during work hours, and Reinecke (2009) demonstrated that employees with high levels of work-related fatigue used video games more frequently during working hours. A recent time sampling study by Hofmann et al. (2012) showed that the more self-control a person exerts during the day, the less able they are to control subsequent behaviors such as media use, and media use was identified as the desire least likely to be successfully controlled. Panek (2014) found that trait self-control is negatively related to the amount of time students spend on social network services and online video use. Finally, Reinecke et al. (2014) found that low self-control capacity affected the use of entertainment media by viewers, and increased negative appraisals of that media use as guilt-inducing procrastination. However, this research has not yet accounted for content differences and how the presence of self-control may produce different patterns of media use.

Although scholars have started to illuminate the link between self-control and entertainment use, it is less clear what underlies that link. In a general sense, researchers have suggested that self-control depletion heightens emotional reactivity to stimuli (Wagner & Heatherton, 2013) or heightens a need for entertainment (Johnson, Ewoldsen, & Slater, 2015). However, whereas previous studies have examined media use as a failure of self-control processes, more recent studies suggest that specific media content may also be related to the amount of self-control required to engage with the media. This builds off prior work examining the perception of media content as challenging (Bartsch & Hartmann, 2017; Hartmann, 2011, 2013), media as a coping mechanism (Knobloch-Westerwick, Hastall, & Rossmann, 2009), as well as theoretical work suggesting that selective exposure to media is a function of various working components of the self-concept, including self-regulatory processes (Knobloch-Westerwick, 2015). Recently, Bartsch and Hartmann (2017) defined the types of challenges afforded by media content, suggesting that media content varies in how much it requires effortful self-regulation, in line with Hartmann (2011, 2013), who proposed a recreation/challenge model of entertainment. According to this model, more challenging content may offer personal growth (via meaningful or complex challenges), whereas less challenging content may offer recreation and restoration of depleted resources (via hedonic regulation).

We suggest that an individual suffering from diminished self-regulatory resources will be motivated to choose particular kinds of media content that allow for the restoration of self-control, or at the least not further deplete it. For these reasons, depleted people may avoid challenging media content, and exhibit a pattern of selective exposure in which less taxing fare is selected at a higher level than is typical when self-control is present. This process may fit under a broader conceptualization of mood management theory (MMT; Zillmann, 2000) in terms of the desire for homeostasis as a motivating factor in media selection. However, our present perspective differs from MMT in three ways.

First, the central question of MMT deals with how individuals use media for homeostatic affect regulation. As suggested by Knobloch-Westerwick (2015), MMT can be framed as a general expectancy-value model in which expectations of how content will influence mood have an impact on which content is then selected. However, self-regulation depletion and restoration is not driven by specific affective states, although it may be weakly associated with either negative states or low arousal (Hagger, Wood, Stiff, & Chatzisarantis, 2010). Indeed, MMT has been criticized for focusing on short-term hedonic regulation, particularly in terms of failing to explain the desire to watch tearjerkers or horror films (Knobloch-Westerwick, 2015), or its inability to explain counterhedonistic media choices (Oliver, 2003; Nabi, Finnerty, Domschke, & Hull, 2006). Although this has been addressed in terms of long-term perspectives of MMT (see Zillmann, 2000, discussion of telic hedonism), we believe our perspective may be more applicable in understanding why and when people will select affectively aversive content, based on the combination of self-regulatory resources available to the user, and self-regulation required by the content. Therefore, we propose that ego restoration is unique from affect management in terms of the end goal.

Second, MMT assumes that users often have “poor comprehension of both the causes of their choices and the ends that these choices serve” (Zillmann & Bryant, 1985, p. 161). An individuals’ media choices are assumed to start from an arbitrary selection point, which is then reinforced via repeated instances of remembered successful or unsuccessful hedonic regulation. While not disregarding the role of unconscious, long-term affective conditioning proposed in MMT, our perspective suggests people may also be well be aware of the extent to which media can meet long- and short-term goal states, including mood regulation, in that people may actively seek media to pursue different goals and motivations.

Third, we suggest that the media characteristics we identify as relevant to self-control (challenge and want/should, as articulated in the next two sections) are independent of the variables noted by Zillmann and Bryant (e.g., excitatory potential, absorption potential, affective valence, and semantic or behavioral affinity; 1985), and generate different predictions than traditional MMT variables. We describe these variables and their role in media selection below.

Cognitively and affectively challenging media

We define challenge as the extent to which users perceive they need to apply self-regulatory resources to control both cognitive and affective processes while processing media content. Hartmann (2011, 2013) proposed that media challenges are required to promote viewer psychological growth (see also Oliver, Bartsch, & Hartmann, 2014; Zillmann, 2000). These growth goals may be perceived as a long-term investment by the user, and, although effortful media may not generate short-term hedonic pleasure (and indeed may even be counter-hedonic, or provoke negative affect and sad mood in the short term), they may be associated with long-term benefits or growth of the self (Khoo, 2016; Oliver et al., 2014; Prestin, 2013; Zillmann, 2000). Therefore, rather than conceptualizing challenge as a feature inherent in media, we examine challenge as an interaction between viewer and content.

In a recent study, Bartsch and Hartmann (2017) defined two types of challenge present in media: cognitive and affective challenge. Cognitive challenges were defined as reflecting media content that is difficult for a user to process, due to the complexity or perceived novelty of the content (Lee & Lang, 2015; Silvia, 2005). Complexity in movies mainly results from the structure of the narrative or plot, i.e., how the story develops and how it is told, whereas novelty is determined by lack of familiarity with the plot structure, but also by aesthetics, casting, and unfamiliar topics or languages. A cognitively nonchallenging film is one that tells a very simple and familiar story, and does not draw heavily on self-regulatory resources needed to control cognitive processes such as maintaining attentional focus or engaging working memory.

Affective challenges in Bartsch and Hartmann (2017) were associated with media content inducing distressing emotions and aversive levels of arousal (Muraven, Tice, & Baumeister, 1998; Schramm & Wirth, 2008) that require regulation on the spot during viewing. We consider emotional challenge to largely result from aversive affect or mixed emotions present in media content, and thus, to draw on self-regulatory resources needed to control affective processes (e.g., emotion regulation; Gross, 2002; Wagner & Heatherton, 2013). We also note that previous research suggests that positively valenced media may serve to restore self-control (Tice, Baumeister, Shmueli, & Muraven, 2007), and thus would be less likely to be perceived as challenging when in a depleted state than content with predominantly negative valence. Although this is superficially similar to expected patterns of hedonic regulation proposed by mood management, it may also produce counter-hedonic patterns when self-control is high and challenge is able to be successfully met.

This conceptualization of challenge was supported in an experimental study examining the effects of cognitive and affective challenging content on the experience of fun, suspense, and appreciation in film viewers (Bartsch & Hartmann, 2017). Nonchallenging films led to greater experiences of fun in viewers, whereas both cognitive and affective challenges were predictive of stronger experiences of appreciation—a response associated with psychological growth. Although these findings are primarily concerned with the entertainment experience, rather than the self-control resources required by the user, they support the notion that challenge may reflect a required self-control capacity of users. In a state of lessened self-control, the individual has less ability to regulate distressing emotions or direct their cognitive resources toward effortful tasks (cf. Hagger et al., 2010). Therefore, we suggest that depleted individuals will exhibit selective exposure to less challenging media content, as well as avoiding challenge in content, to better match with their self-control capacity.

Regarding these predictions about preference and choice, Reinecke et al. (2014) found that low self-control capacity was associated with a preference for viewing less challenging television programs after a day of work or school. Building on this finding, Eden et al. (2015) examined the hypothesis that users with exhausted self-control resources would be less inclined to choose high-challenge films. Participants wrote an essay requiring self-control (i.e., requiring participants to refrain from using common alphabet letters e and n; Bertrams, Englert, & Dickhäuser, 2010) or not (i.e., allowing use of all letters). Subsequently, participants were offered a choice of a film from six options: three high-challenge and three low-challenge choices. Participants rank ordered films in terms of immediate preference. For both cognitive and affective challenge factors, ego depletion was involved in selective preference for low challenge films and avoidance of high challenge films (Eden et al., 2015).

However, the ranking procedure used limits the generalizability of those findings, as it was a comparative procedure versus free choice. Comparative rank ordering does not allow for examination of free-choice or noncomparative processes. Additionally, closed-ended choices with comparative options can bias choice toward more challenging fare (Bazerman, Tenbrunsel, & Wade-Benzoni, 1998). Alternate conceptualizations of media content besides the perception of cognitive and affective challenge (i.e., rating films along specific dimensions) may offer further insight into the regulatory processes required by media content. Specifically, perceptions of should versus want options researched extensively in the psychology of consumer decision-making (Bitterly et al., 2015) may be closely related to the conceptualization of challenge and selective exposure to media, as illustrated below.

Should versus want media

In many ways, the decision to watch a carefree, but maybe trivial, movie instead of a challenging highbrow movie seems to resemble other decisions about what one wants versus what one should do (e.g., in food consumption). Researchers have characterized this should versus want conflict as one between want options—desirable options with immediate gratifications (e.g., overindulging in sugary treats, procrastinating on a project) versus should options —those options that are less desirable immediately, but enable long-term benefits (e.g., exercising, eating a healthy diet; Bazerman et al., 1998; Bitterly et al., 2015). Milkman, Rogers, and Bazerman (2008) described a tension between should versus want choices, as the former provide greater future utility, whereas the latter provide greater immediate utility. One selection is not necessarily more optimal than the other. Sometimes, if the short-term payoffs of the want option outweigh long-term benefits of the should option, it may be rational to ignore the should for the want choice. For example, although healthy foods promote greater well-being and physical fitness, the occasional treat may keep dieters motivated and happy.

This should/want conflict has been linked to self-control failure and media in earlier studies. For example, when people subscribe to a Netflix-like service, they are likely to put many should movies in their rental queue. However, people are more likely to quickly watch and return want movies than should movies (Milkman, Rogers, & Bazerman, 2009). A study of shelf price versus subscription price for magazines showed significantly higher prices for leisure magazines (wants) than investment magazines (shoulds). The pricing scheme reflects the immediate temptation of want magazines at the newsstand (Oster & Scott Morton, 2005). Read, Loewenstein, and Kalyanaraman (1999) conceptualized virtue films (shoulds) as those which provide long-term cultural enrichment, and vice films (wants) as those providing a momentary pleasure. Lowbrow vice movies were primarily big box-office hits with attractive, popular stars and had an emphasis on action, comedy, or romance. Virtue movies usually had less familiar stars, often had subtitles, were intellectually challenging, and were typically less popular (but some were quite successful). Read et al. (1999) found that users were more likely to choose virtuous films for delayed consumption, and vice films for immediate consumption. However, their study was designed to test simultaneous/sequential choice options, not media selection per se.

More recently, Wang, Novemsky, Dhar, and Baumeister (2010) examined media choice as an outcome relevant to ego depletion and trade-offs in comparative decision-making. They found that depleted users would be more likely to choose a vice (i.e., a lowbrow film) than a virtue (i.e., a highbrow film). They found that 68% of participants chose lowbrow films after making a depletion-inducing decision condition, compared to 33% of participants in the nondepletion-inducing decision condition. This held when controlling for positive and negative affect as well as fatigue (Wang et al., 2010, Study 1). In regards to media choice, one may also interpret these results in terms of how highbrow versus lowbrow films are preferred under conditions of depletion. Cognitive challenge, affective challenge, and should perceptions are all partial indicators of what is often referred to as highbrow film (Bartsch & Hartmann, 2017; Milkman et al., 2009). Taken together, this line of reasoning suggests that should/want conflicts in content selection may also be affected by available self-control resources, in much the same way as cognitive and affective challenges. Yet, it is less clear whether the long-term rewards (such as cultural enrichment) represented by a should choice are conceptually distinct from those represented by cognitive and affective challenge.

This study

In the following, we present the results of an experiment designed to test the effects of reduced self-control capacity on selective exposure to entertainment media. Specifically, we consider films that vary in perceptions of being cognitively challenging, affectively challenging, and should versus want choices. Importantly, we use a three-step approach that measures selective exposure in free-choice versus comparative contexts (i.e., open- vs. closed-ended choices) while participants imagine themselves to be either in a state of reduced self-control resources (i.e., depleted) or with full resources available to them (i.e., non-depleted).

First, we predict that the level of self-control resources perceived to be available to users will generate distinct patterns of media preference in both free-choice and comparative selection of films. Specifically, film preferences will differ relative to cognitive and affective challenge, as well as the perception of the film as a should versus a want choice. These considerations lead to H1: Depleted viewers will show lower selection of (a) cognitively challenging films, (b) affectively challenging films, and (c) should films, but will show greater selection of (d) want films, compared to nondepleted viewers.

Next, given the association of affective challenges with suspense, cognitive and affective challenges with appreciation, and the lack of challenges with fun (Bartsch & Hartmann, 2017), we would expect similar audience response patterns to occur as the result of selective exposure. Although self-selected films should show similar levels of interest and enjoyment, those chosen under reduced self-regulation are less likely to be challenging films. So, we pose H2: Depleted viewers will select films that they believe will be (a) more fun, but (b) less suspenseful, (c) less moving/thought-provoking, and (d) leaving less of a lasting impression, compared to non-depleted viewers.

Method

An experiment tested how perceived self-control affected the selection of films that are cognitive challenges, affective challenges, should choices, or want choices. Participants were randomly assigned to a treatment condition, where they imagined a scenario involving low self-control capacity, or to a control condition, where they imagined a scenario involving the presence of self-regulatory resources. Selective exposure was measured three different ways, in three distinct steps: open-ended choice, closed-ended choice, and preference ratings for closed-ended options. This provided a robust test of the hypotheses, and made use of both sequential and simultaneous choice scenarios, which can influence decision-making (Bitterly et al., 2015). It also adds robustness by testing selective exposure through measures of both discrete choices and preference strength (Clay et al., 2013). In the open-ended measure (step 1), participants named a film they would watch in the scenario, and also reported their perceptions of the film. In the closed-ended measures (step 2), participants selected a film from a list of titles pretested to vary on challenges and should/want. In addition to choosing their favored film and reporting their perceptions of this title, they also indicated the film they would be least likely to view and their perceptions of this title. Finally, they prospectively rated (step 3) how much they would prefer to select each of 24 films pretested for level of cognitive and affective challenge.

Participants

Students (N = 179) were recruited from an entry-level communication course at a Dutch university, where they participated for course credit. The participants were M_age = 21.65 (SD = 2.16); 70.9% were women, 22.3% were men, and 6.7% did not indicate sex.

Stimuli pretest

Feature film titles were pretested to construct a set of films that systematically varied on perceived cognitive challenge, affective challenge, and should/want. First, film titles were screened by the authors for popularity and recency (as of September, 2014) from the Internet Movie Database (IMDB.com) list of most popular feature films. Examining the top 400 entries on the list produced a list of 84 films. To ensure a wide variety of genre and challenge, additional recent films were selected from underrepresented genres, drawing from separate IMDB lists of the most popular documentary, horror, musical, romance, and music films. This produced a full list of 110 films.

To evaluate perceptions of these films, an online survey following the procedure of Milkman et al. (2009) was administered to 201 Amazon Mechanical Turk users (55% men; M_age = 34.80, SD = 10.42). First of all, the concepts of intellectual challenge, emotional challenge, should film, and want film were introduced to participants to operationalize cognitive and affective challenges and should/want distinctions (see digital Appendix A, available online at https://osf.io/he4r9/?view_only=893eb93a87fa4a32887745362ae81968). Then, each film title was presented, along with its poster, director, year of release, and short IMDB synopsis. If participants indicated they had seen the film, they were asked to rate the film “based on your impression of the film.” If not, they were asked to rate the film “based on the description above” (i.e., the IMDB synopsis). Each film was rated with four different 7-point semantic differential scales, with the anchors no intellectual challenge–high intellectual challenge; no emotional challenge–high emotional challenge; not a strong “should” film–a strong “should” film; and not a strong “want” film–a strong “want” film. To prevent fatigue, each pretest participant only rated a random selection of 20 films; no fewer than 32 people rated each film.

These ratings were assessed by examining the mean scores for each film title on cognitive challenge, M = 3.69 (SD = 0.83), affective challenge, M = 4.43 (SD = 0.80), should, M = 3.51 (SD = 0.79), and want, M = 4.78 (SD = 0.57). Want and should perceptions were negatively correlated, r(108) = –.523, p < .001. To simplify the four variables of interest, should and want were collapsed into a single continuum by computing their difference, M = −1.26, SD = 1.19 (cf. Milkman et al., 2009). Using mean splits of the three scores of cognitive challenge, affective challenge, and should/want difference, films were then allocated to a 2 × 2 × 2 grid based on whether their scores were low or high for each variable. Three films were selected from each category in the grid, prioritizing (a) films with more extreme scores across the three variables, (b) genre diversity, and (c) face validity of categorization. This produced a set of 24 films in a 2 × 2 × 2 (cognitive challenge × affective challenge × should/want) crossed design (Appendix B: https://osf.io/he4r9/?view_only=893eb93a87fa4a32887745362ae81968).

Experimental procedure

In the main study, participants accessed an online experiment. First, they were introduced to the concepts of cognitive challenge, affective challenge, should films, and want films, using the definitions in the pretest study, plus a comprehension check (following Milkman et al., 2009), which no participant failed. This was done to aid comprehension of subsequent perception items, given that lay theories may use familiar terms with a different connotation (e.g., “challenge”).

Next, participants were randomly assigned to either an experimental treatment condition (n = 76, depletion group) or a control condition (n = 103, nondepletion group). In both conditions, participants were instructed to imagine a common scenario and place their self into that situation. For the depletion group, participants were asked to imagine coming home from “a long, exhausting day” that involved making “many choices and difficult decisions all day, and now [you] are completely wiped out.” In the nondepleted condition participants were told to imagine coming home after “a brief, exciting day” where they “successfully completed all your assignments today, and now you are completely energetic and full of vigor” (see Wang et al., 2010, studies 1 and 4; as well as similar procedures in Neubaum & Krämer, 2016; Oliver, 2008, study 2). Both scenarios also asked the participant to imagine that they would now sit down on their couch at home to watch a film.

After the induction, participants completed an open-ended measure of selective exposure, identifying a movie they wished to watch in the situation. The open-ended choice was also rated for strength of preference, film genre, cognitive challenge, affective challenge, being a should film, being a want film, and expected audience response (step 1).

Next, the same experimental induction was reemphasized to remind participants of the scenario, and participants were presented with a list of 24 films they could watch in the imagined scenario. Each film title was presented, along with its poster, director, year of release, and short IMDB synopsis. In this closed-ended selective exposure measure, participants indicated their most preferred choice (as well as their least preferred choice) and rated selections on cognitive challenge, affective challenge, being a should film, being a want film, and expected enjoyment (step 2). Choosing from an array of varied films, in contrast to the open-ended choice in step 1, provides an important test of the hypotheses, as previous research shows that simultaneous and sequential choices can produce different results (Bazerman et al., 1998; Bitterly et al., 2015). Specifically, simultaneous choices with contrasting should and want options tend to produce more should choices than a sequential choice scenario such as an open-ended measure of selectivity. After this, the imagined scenario was reemphasized one more time (the scenario of depletion vs. control), and participants then rated each of the 24 films on how likely they would be to select each film (step 3). Finally, participants were thanked and debriefed.

Measures

Open-ended selective exposure

After imagining the experimental situation, participants were asked to name a movie they would “like to watch in this scenario, if you could choose any movie available to you.” Each participant listed the title of a single film.

Perceptions of open-ended choice

Participants were then asked to provide more information about their open-ended choice. First, to ensure that any effects were not attributable to differences between groups in the ability to make a selection or the desire for media use in general, strength of preference was measured with a single-item measure, “How much would you like to watch [TITLE]?” 1 = none, 5 = very much, M = 4.08, SD = 0.76.

Next, participants indicated the film genre from a list of options (adapted from Oliver & Bartsch, 2010). Common genres were comedy (29.1%), action adventure (19.6%), romance (8.9%), drama (8.4%), thriller (8.4%), and science fiction (5.6%). Fantasy, crime, animation, documentary, sad/tear-jerker, history, and horror were each chosen by < 5% of participants; 7.3% specified other genres. Participants could only select one genre.

Perceptions of cognitive and affective challenge were each rated with single item measures indicating 1 = low to 7 = high how much the film was an “intellectual challenge,” M = 3.45, SD = 1.95, and an “emotional challenge,” M = 4.04, SD = 1.69. Similarly, single item measures indicated 1 = not strong “should” film, 7 = strong “should” film, M = 3.60, SD = 1.89, as well as 1 = not strong “want” film, 7 = strong “want” film, M = 5.51, SD = 1.36.

Finally, the 12-item audience response scale (Oliver & Bartsch, 2010) measured expected enjoyment and appreciation of the film. Participants indicated, 1 = strongly disagree to 7 = strongly agree, their expected fun, M = 6.16, SD = 0.68, being moved, M = 4.09, SD = 1.40, lasting impression, M = 4.96, SD = 1.34, and suspense, M = 4.06, SD = 1.62.

Closed-ended selective exposure

In this measure of selective exposure, 24 films that had been pretested based on a 2 × 2 × 2 (cognitive challenge × affective challenge × should/want; three films in each category) within-subjects design (as described previoulsy and presented in Appendix B) were presented for participants to choose from. The films were displayed as poster images on a single page, within a randomized presentation order. Participants were asked to indicate the film that they would select for viewing in the imagined situation. After reporting their perceptions of this selected film, they were asked to indicate the film they would be least likely to select, and to report perceptions of this title, too.

Perceptions of closed-ended choices

As in the open-ended measure, participants were provided their perceptions of the films they most preferred to watch from the list. Perceived cognitive challenge and affective challenge were rated from 1 = low to 7 = high regarding “intellectual challenge” (M = 3.42, SD = 1.89), as well as “emotional challenge” (M = 4.11, SD = 1.78). Participants rated their selection from 1 = not strong “should” film to 7 = strong “should” film (M = 3.75, SD = 1.93), as well as from 1 = not strong “want” film to 7 = strong “want” film (M = 5.30, SD = 1.54). In lieu of the audience response scale, expected enjoyment of the film (“How much would you enjoy watching this film if you had to?”) was measured with a single item, 1 = not at all to 7 = extremely (M = 5.68, SD = 0.96).

Additionally, the film participants were least likely to select was also rated on cognitive challenge (M = 3.28, SD = 1.82), affective challenge (M = 4.32, SD = 2.01), should (M = 2.96, SD = 1.68), want (M = 2.75, SD = 1.66), and expected enjoyment (M = 2.40, SD = 1.36).

Preferences

Participants indicated the likelihood that, given the scenario, they would choose each of the 24 films, 1 = very unlikely to 4 = very likely. Mean preferences for the 24 films ranged from 1.80 (SD = 0.87) for Ender’s Game to 2.99 (SD = 1.06) for 22 Jump Street.

Results

Step 1: Open-ended film choice

In a first step, we conducted between-subjects t-tests to compare the movie choice reported in the open-ended format by participants in the depletion versus control condition. As the results in Table 1 show, in support of H1a/b/c, participants in the depletion condition characterized their open-ended movie choice as less cognitively and affectively challenging, and less as a should film. The strength of their preference to watch the picked film did not significantly differ between conditions, nor did the characterization of the picked movie as a want film. Participants in the depletion condition selected a film that they expected to be more fun, less moving, less suspenseful, and leading to less of a lasting impression than the movie chosen by participants in the control condition. In sum, these findings provide support for our H1 and H2, except for (and in contrast to H1d) the fact that participants in both conditions did not differ in the characterization of their choice as a want film.

Table 1. Comparison of subjective ratings of open-ended film choice made by participants in depletion vs. control condition

	Depletion		Control
Selected film is….	M	SD	M	SD	|t|(df)	dCohen
Cognitively challenging	2.21	1.14	4.37	1.92	9.42(169.89)**	1.33
Affectively challenging	3.29	1.63	4.59	1.52	5.50(177)**	0.83
Should	2.92	1.83	4.11	1.78	4.35(175)**	0.66
Want	5.68	1.27	5.39	1.42	1.44(177)	0.22
Strength of preference	4.04	0.82	4.11	0.71	0.58(177)	0.09
Fun	6.26	0.69	6.08	0.67	1.75(177)^+	0.27
Moving	3.30	1.27	4.68	1.18	7.49(177)**	1.14
Suspense	4.36	1.48	5.41	1.03	8.73(177)**	1.33
Lasting impression	3.03	1.44	4.83	1.30	5.33(126.44)**	0.85

Note: **p < .01, *p < .05, ⁺p < .10. All applied measures were based on 7-point scales, except for the “strength of preference” measure that was based on a 5-point scale. A MANOVA testing H1a/b/c/d simultaneously was conducted with cognitive challenge, affective challenge, should, and want as dependent variables and condition as independent variable, Hostelling’s Trace Λ_LH = .463, F(4, 172) = 19.89, p < .001, η_p² = .32. A MANOVA testing H2a/b/c/d simultaneously was conducted with fun, moving, suspense, and lasting impression as dependent variables and condition as independent variable, Hostelling’s Trace Λ_LH = .512, F(4, 174) = 22.28, p < .001, η_p² = .34.

Relationships among the four perceptions of chosen films were also examined, regarding our assumptions linking challenge and the should/want dimensions. Cognitive and affective challenge were moderately correlated, r(179) = .512, p < .001. Should perceptions were also correlated with cognitive, r(177) = .488, p < .001, and affective challenge, r(177) = .344, p < .001. Perceptions of want had a small negative link with cognitive challenge, r(179) = –.184, p = .014, but none with affective challenge or should perceptions.

An additional analysis of the genres of the preferred films yielded further confirmatory evidence of these assumptions. A 2 (depletion vs. control) × 13 (film genre) cross tabulation yielded significant differences of the genres that were preferred in both conditions, χ²(13) = 57.36, p < .01. The most preferred genre among participants in the depletion scenario were comedies (53.9%), followed by romances (11.8%) and action adventures (11.8%). In contrast, only 10.7% of the participants in the control condition picked a comedy, and only 6.8% a romance (whereas 25.2% preferred to watch an action adventure). This result provides further evidence that participants imagining the depletion scenario anticipated a greater preference for a light-hearted, nonchallenging film than participants in the control condition.

Step 2: Closed-ended film choice

In a second step, we examined which film (of the 24 pretested options that we presented in the survey) participants in the depletion versus control condition would most prefer to choose, and which one they preferred least. When comparing the film choices made by participants in both conditions, we examined (a) how participants subjectively perceived the films they picked, and (b) how the chosen films differed based on our initial pre-tested categorization into three dimensions of cognitive and affective challenge and should/want.

Most preferred choice

Similar to the results obtained in step 1, between-subjects -tests (Table 2, left side) revealed that participants in the depletion condition rated their most preferred film as significantly less cognitively and affectively challenging, and less of a should film than control participants. Participants in both conditions did not differ in anticipated enjoyment of their film choice. Again, surprisingly, ratings of how much the most preferred film resembled a want film did not significantly differ. And, as with open-ended choices, there were moderate correlations among cognitive challenge, affective challenge, and should perceptions. Want perceptions were not correlated with the other three perceptions.

Table 2. Comparison of subjective ratings of most preferred and least preferred films selected from 24 presented film choices by participants in depletion vs. control condition

	Most preferred film						Least preferred film
	Depletion		Control				Depletion		Control
Selected film is….	M	SD	M	SD	|t|(df)	dCohen	M	SD	M	SD	|t|(df)	dCohen
Cognitively challenging	2.17	1.24	4.35	1.75	9.75(177)**	1.41	3.91	1.83	2.83	1.68	4.10(177)**	0.62
Affectively challenging	3.17	1.69	4.81	1.52	6.79(177)**	1.04	4.96	1.86	3.85	1.99	3.78(177)**	0.58
Should	2.78	1.80	4.47	1.70	6.42(177)**	0.98	3.32	1.75	2.70	1.58	2.47(176)*	0.38
Want	5.33	1.66	5.28	1.46	0.20(177)	0.03	2.84	1.68	2.69	1.66	0.61(177)	0.09
Enjoyment	5.68	0.93	5.67	0.98	0.10(177)	0.01	2.37	1.30	2.43	1.40	0.29(177)	0.04

Note: **p < .01, *p < .05, ⁺p < .10. All applied measures were based on 7-point scales. A MANOVA testing H1a/b/c/d simultaneously for most preferred film was conducted with cognitive challenge, affective challenge, should, and want as dependent variables and condition as independent variable, Hostelling’s Trace Λ_LH = .581, F(4, 174) = 25.27, p < .001, η_p² = .37. A MANOVA testing H1a/b/c/d simultaneously for least preferred film was conducted with cognitive challenge, affective challenge, should, and want as dependent variables and condition as independent variable, Hostelling’s Trace Λ_LH = .111, F(4, 173) = 4.80, p < .01, η_p² = .10.

In addition, we conducted three 2 × 2 cross-tabulations analyzing the most-preferred choice of participants in the depletion versus control condition based on our initial categorization into cognitively challenging versus nonchallenging, affectively challenging versus nonchallenging, and should versus want films. This analysis revealed that the percentage of participants picking a cognitively nonchallenging film as their most preferred choice was significantly higher in the depletion condition, 78.9%, than in the control condition, 42.7%, χ²(1) = 23.60, p < .01. The odds ratio (that compares the ratio of choosers to nonchoosers observed in the depletion vs. control condition) was 5.03, which shows that the odds that a cognitively nonchallenging film was selected (or the ratio of choosers to nonchoosers) were about 5 times higher in the depletion as compared to control condition. The share of participants in the depletion condition selecting an affectively nonchallenging movie as their most preferred choice, 59.2%, did not significantly differ from the share of participants making this choice in the control condition, 56.3%, OR = 1.13, χ²(1) = 1.51, p = .70. However, a significantly lower percentage (17.1%) of participants in the depletion condition selected a should film as their most preferred choice (vs. 31.1% in control condition, OR = 0.46), and thus a significantly higher percentage (82.9%) of participants in the depletion condition selected a want film as their most preferred choice (vs. 68.9% in control condition, OR = 2.18, χ²(1) = 4.53, p < .05).

Least preferred choice

Results of the comparison of participants’ least preferred film choice are also displayed in Table 2 (right side). Between-subjects t-tests revealed that the least preferred film by participants in the depletion condition was perceived as significantly more cognitively and affectively challenging, as well as more of a should film, compared to the film preferred least by participants in the control condition. Perceptions of the least preferred choice as a want film did not differ between conditions.

We also analyzed the least preferred choice of participants in the depletion versus control condition in three additional 2 × 2 cross-tabulations. This analysis showed that the percentage of participants choosing a cognitively challenging film as their least preferred choice was significantly higher in the depletion condition, 52.6%, as compared to the control condition, 29.1%, OR = 2.70, χ²(1) = 10.15, p < .01. The share of participants selecting an affectively challenging film as their least preferred choice was also significantly higher in the depletion condition, 78.9%, as compared to the control condition, 63.1%, OR = 2.19, χ²(1) = 5.21, p < .05. However, although the observed differences matched our expectations, participants in the depletion condition were not significantly more likely to pick a want film (65.8%) or a should film (34.2%) as their least preferred choice than participants in the control condition (want_control: 69.9%, OR = 0.83; should_control: 30.1%, OR = 1.21; χ²(1) = 0.34, p = .56). In summary, with only few exceptions provided by nonsignificant findings that were particularly related to missing differences in terms of how much the film choice resembled a want film, we interpret these results as providing further evidence for Hypotheses 1. Participants in the depletion condition displayed a relatively higher preference for non-challenging films and avoidance of challenging films and “should” films.

Step 3: preference across closed-ended film choices

In a third step, we examined participants’ average preference they expressed for those 24 presented movies that we categorized as cognitively challenging or nonchallenging, affectively challenging or nonchallenging, and should or want films. For this examination, we conducted three separate 2 × 2 mixed between-within subjects ANOVAs with participants’ preference for challenging versus nonchallenging movies (or should vs. want films) as a within-subjects factor and the experimental treatment (depletion vs. control) as a between-subjects factor. Results are displayed in Figures 1, 2, and 3.

Figure 1. Average preference (i.e., anticipated likelihood) of watching cognitively non-challenging vs. challenging films among participants in depletion vs. control scenario.

Note. Brackets indicate 95% CI around the means.

Figure 2. Average preference (i.e., anticipated likelihood) of watching affectively non-challenging vs. challenging films among participants in depletion vs. control scenario.

Note. Brackets indicate 95% CI around the means.

Figure 3. Average preference (i.e., anticipated likelihood) of watching “want films” vs. “should films” among participants in depletion vs. control scenario.

Note. Brackets indicate 95% CI around the means.

The mixed ANOVA analyzing the preference for cognitively challenging films yielded a significant challenge × condition interaction, Wilk’s λ = .90, F(1, 177) = 20.45, p < .01, η_p² = .10, which is displayed in Figure 1. As the figure suggests (and in contrast to H1a) participants in the depletion condition did not differ from control participants in their preference for cognitively non-challenging movies, t(177) = 1.61, p = .11, d = 0.25, but, consistent with H1a, they expressed a significantly lower preference than control participants for cognitively challenging films, t(177) = 3.46, p < .01, d = 0.53.

A second mixed ANOVA examined preference for affectively challenging films. This analysis also yielded a significant interaction effect for challenge × condition, Wilk’s λ = .88, F(1, 177) = 24.96, p < .01, η_p² = .12, as displayed in Figure 2. Participants in the depletion condition showed a marginally significant higher preference for affectively non-challenging, t(177) = 1.79, p = .08, d = 0.27, and a significantly lower preference for affectively challenging films than participants in the control scenario, t(177) = 3.76, p < .01, d = 0.57.

A third mixed ANOVA examined the preference for want and should films. This analysis yielded a weak albeit significant should/want × condition effect, Wilk’s λ = .97, F(1, 177) = 4.77, p < .05, η_p² = .03, which is displayed in Figure 3. Similar to the observations made in the previous analyses, participants in the ego depletion condition did not differ from participants in the control condition in their preference for want films, t(177) = 0.02, p = .99, d = 0.00, but expressed a significantly lower preference for should films than participants in the control condition, t(177) = 2.32, p < .05, d = 0.35. Taken together, we take the results of our third analysis step as further evidence for H1 and the insight that depletion is associated with greater avoidance of challenging and “should” content, respectively greater preference of non-challenging content.

Discussion

These results provide support for the central proposition that self-control and media choice are related in systematic ways, as predicted by the existing self-control of the media user and the perception of challenges provided by the content. Participants imagining themselves as depleted were, across three different measures of preference, by and large less likely to select affectively and cognitively challenging films, less likely to select should films, less likely to select films they perceived as suspenseful or meaningful, and more likely to prefer films they perceived as fun, or films in the comedy genre, than were participants imagining themselves with full self-regulatory capacity. Additionally, in terms of least preferred choices, depleted participants were more likely to avoid challenging films than the non-depleted.

The self-control related findings are in line with and support past empirical research indicating that self-control processes are related to entertainment media use (Hofmann et al., 2012; Panek, 2014; Reinecke, 2009; Reinecke et al., 2014; Wagner et al., 2012). In particular, this work extends and refines the initial relationships explored by Eden et al. (2015), to illustrate that self-control predicts a selective pattern of media entertainment use, in which greater self-control resources enable the selection of more affectively and cognitively challenging media content. The findings on challenge extend the research begun by Bartsch and Hartmann (2017) and are closely related to decision processes involved in the should/want conflict (Bitterly et al., 2015), as well hedonic processes (Zillmann, 2000).

Taken as a whole, these results also support the theoretical relationship between entertainment use and growth potential explored in Hartmann (2011, 2013)), Knobloch-Westerwick (2015), as well as Oliver et al. (2014) and Zillmann (2000), stating that media can act as both diversion and growth opportunity. The recreation/challenge model of media entertainment (Hartmann, 2011, 2013) describes how both fun, hedonic media and challenging, eudaimonia media each contribute to individual well-being. The former type of media allow for recreation, i.e., pleasurable distractions and amusements that allow for the replenishment of resources and provide rest and relaxation. The second type of media entertainment, which involves various types of challenges, are more difficult for media users to process but are highly rewarding in the long-term, as they contribute to personal growth and help the individual meet adaptive challenges in their environment (Hartmann, 2011, 2013). These findings illustrate that both kinds of media are selected as entertainment, but that the willingness of the user to engage with media at the recreational or challenging level quite clearly depends on the self-regulatory capacity available to the user.

Despite these strong findings, this study is not without limitations. First, we asked participants to simply imagine themselves in a particular state, and then imagine how they would feel about particular film choices. This may or may not map onto actual selection and preference patterns observed in the field where individuals have a multitude of options available to them. However, the use of three different measures of selective exposure, including both open- and closed-ended choices, helped to mitigate this limitation. Results were fairly consistent across all three measures, providing a robust demonstration of the effects of depletion on selective exposure. One inconsistency was in the lack of an effect between conditions in choosing a film categorized as affectively challenging or non-challenging in the closed-ended choice (step 2, most preferred film). However, this may stem from the methodological limitations of presenting comparative choices (Eden et al., 2015). Indeed, when asked to provide their perceptions of these same films, the depletion group reported their closed-ended choices as less affectively challenging, as expected.

Additionally, preference for challenge and for should versus want options can be shifted based on available choices, the way in which options are presented, and additional variables such as mood state (Bitterly et al., 2015; Zillmann, 2000). Therefore, the next step for this research is to move out of the lab to observe preference patterns for challenging media in individuals exhibiting natural variation in depletion states. For example, previous investigations have used diaries and experience sampling to illustrate how self-control and related variables increase general media exposure (Hofmann et al., 2012; Kubey & Csikszentmihalyi, 1990).

Next, the findings for the want dimensions were less conclusive than those for the challenge or should dimensions. We argue that this is an artifact of our methodology, by asking individuals what film they would want to see, it may have been difficult for them to disentangle their wanting to see the film from the conceptualization of the film as a want film. On the other hand, participants read a conceptual definition of want, and completed a comprehension check. So, findings may instead reflect different response patterns for challenge and should/want conflict, where depletion affects preferences for high and low challenges and for should choices, but not for want choices. It could also be that, rather than preference, we are seeing an avoidance mechanism in effect. Depletion may drive movement away from selecting challenging or should films. Previous research on should/want conflict (e.g., Milkman et al., 2009) has used should/want difference scores, which could have obscured differential effects for should and want choices. Future research can detangle should from want to understand how these drive short and long term selection. In addition, future research may consider existing preferences, habits, and traits as moderators of the link between self-control and selective exposure to challenging and should/want media.

Moreover, the mechanism underlying the concept of ego depletion has recently been under scrutiny. Lab studies have failed to replicate commonly cited effects (Hagger & Chatzisarantis, 2016; Xu et al., 2014), leading to concern that self-control may not function as conceptualized or only in specific domains such as cognitive performance (Carter, Kofler, Forster, & McCullough, 2015). However, others argued that reviewing the idea of self-control in terms of motivational and attentional focus may reconcile past findings (Inzlicht, Schmeichel, & Macrae, 2014). Additionally, studying self-control failure as a function of physiological and mental fatigue offers promising future perspectives for this line of inquiry (Evans, Boggero, & Segerstrom, 2016). Thus, although the concept of ego depletion (or its existing laboratory paradigm) may need revision, our findings are still valuable for media researchers in that they suggest that available goals and motivations shift as a function of perceived coping capacity and influence media selection in the process. To the extent that artificial lab inductions may be limited in their ability to produce real change in self-regulatory resources, an imagined scenario (cf. Wang et al., 2010) allows participants to draw upon experience with real-word depletion and fatigue (e.g., at the end of an exhausting day) and thus are reflective of prior media selections in those situations.

With regard to the dependent variable of selective exposure, this study made use of behavioral intention measures. In their review of selective exposure measures, Clay et al. (2013) suggested that behavioral intentions, in addition to being generally predictive of actual behavior, are less vulnerable to social desirability and recall error than retrospective reports of media use. However, intentions are limited (Clay et al., 2013; Knobloch-Westerwick, 2015) in that they may not fully account for availability of media content, strength intention, or time spent with media. To that end, the design used both open-ended and closed-ended measures of behavioral intentions, alleviating issues regarding availability of titles or unrealistic choices. We also accounted for strength of intention through step 3, employing a continuous rather than discrete measure of selection. In terms of time spent, the stimuli all represented feature-length films, but future research could capture time spent with challenging and want/should media through observational measures (cf. Knobloch-Westerwick, 2015).

This study adds to an ongoing societal debate on the placement of leisure in our daily lives. Today’s individuals face stress and burnout from overwork (Jansen, Kant, van Amelsvoort, Nijhuis, & van Den Brandt, 2003). Our research suggests that this increased exhaustion may result in individuals selecting media for diversion rather than personal expansion, or media offering short-term hedonic benefits versus media offering long-term eudaimonic benefits. This is in line with high/low culture debates suggesting that some media are better at promoting intellectual growth and flourishing (Latorre & Soto-Sanfiel, 2011; Read et al., 1999). Although we would not suggest that people engaging in nonchallenging media selections are, as the saying goes entertaining themselves to death, we suggest that over time, these selections may lead to long-term patterns of using nonchallenging and diversionary media, which limit the personal growth and self-expansion of the individual. Over time and magnified across societal strata, this may result in less opportunity for users to engage in challenge and personal growth via media. On the other hand, the findings show that individuals in possession of self-control resources were able to choose cognitively challenging, affectively challenging, and should media that allow for personal growth.

Initial evidence from forced-exposure experiments shows that challenging media can produce personal growth (Khoo, 2016; Oliver, Hartmann, & Woolley, 2012; Prestin, 2013). Future research should test not only selective exposure and immediate responses to recreational and challenging media, but also the long-term effects of selective media repertoires on personal growth and well-being. For example, media have been suggested to aid in coping strategies (Knobloch-Westerwick et al., 2009; Nabi et al., 2006). The selection of affective and cognitive challenge might well map onto the distinction between emotion-focused and problem-focused coping strategies (Lazarus & Folkman, 1984). Nonchallenging media may aid emotion-focused coping but also avoidant coping strategies (Knobloch-Westerwick et al., 2009). Conversely, affectively challenging media may offer the opportunity to alter existing dysfunctional emotions (such as regret) via observing successful coping strategies in others (Nabi et al., 2006). Although beyond the scope of this work, media-as-coping under high and low self-regulatory capacity conditions offers an interesting perspective for media as personal growth research.

Another intervention opportunity comes from examining media in terms of should/want dimensions. Making long-term plans involving media diets over time may help individuals avoid short-term hedonic choices in favor of long-term growth potential (Gollwitzer, 1999). Choosing to watch critically acclaimed lists of films or television, versus whatever the local network is offering, for example, is a way for individuals to manage media exposure in a way to promote long-term goals. Limiting access to want options (e.g., installing antiprocrastination software) may also help keep long-term goals salient via managing the personal environment of the media user (Ariely & Wertenbroch, 2002). Finally, people might also benefit from pairing short-term hedonic pleasures with long-term goals, for example by listening to music while studying or by only allowing oneself to indulge in want choices while at the gym (Milkman, Minson, & Volpp, 2014; Wang & Tchernev, 2012).

On the other hand, literature on the should/want conflict suggests, there are times when the short-term benefits of want choices outweigh the long-term benefits of should (Milkman et al., 2008). We see that reflected in the work of Muraven and Baumeister (2000), Reinecke (2009), and Reinecke et al. (2014), who suggested that use of entertainment media can restore depleted self-control and act as a recovery process for fatigued individuals. Therefore, we argue that our findings do not condemn the choices of the creature seeking comfort in media. Instead, we endorse a more nuanced understanding of how, why, and under what circumstances the individual—whether feeling fully human or depleted to feeling like a base creature—makes their choices is required. This study begins to achieve these goals in terms of defining challenge in self-regulatory terms, and illustrating how this interacts with the existing resources of the media user to exert pressure on preference and selection of media content.