The Character Engagement and Moral Adjustment Model (CEMAM): A Synthesis of More than Six Decades of Research

ABSTRACT

Character liking, identification, and parasocial interaction/relationships are terms used in various literatures to describe character engagement. The current paper synthesizes more than six decades of research in media psychology and communication science to organize and delineate four processes related to character engagement with fictional characters: Attention, Appraisal, Affiliation, and Assessment. In addition to defining and distinguishing these four processes, we describe how they are influenced by narrative, character, and viewer features, leading to moral adjustment – that is, a viewer’s own morality being shaped and molded through exposure to fictional personae. We endeavor here to diminish conceptual confusion and to clarify causal, temporal, and reciprocal relationships between the four factors regarding moral adjustment in viewers. By uniting these processes under a single conceptual model, we provide a framework for understanding moral adjustment through character engagement that can serve as a launch point for more focused research projects.

What is a character? Characters are narrative constructions in a fictional text, but also mental constructs in the audiences’ minds (Bálint & Tan, 2019). Characters in fiction serve as action agents of the narrative who drive narratives forward via decision making and conflict generation/resolution (Weijers, 2014). Characters may also come to embody certain attitudes, beliefs, and behaviors for viewers – a process known as exemplification (see, Zillmann, 2006). Thus, characters should be the central element in research examining the role of media in shaping and changing viewer morality. Yet, theoretical models that attempt to predict how media exposure can alter a viewer’s morality (see for example, Tamborini, 2013; but see also extensions, Tamborini, Grizzard et al., 2021) have largely overlooked how viewers engage with characters, and consequently, how character engagement might explain the specifics of moral adjustment. Such an oversight requires some consideration from the social scientific community, as a substantial body of research on character engagement exists in media psychology and communication sciences.

Research on characters has become heavily siloed, with various fields including social psychology, film studies, and media psychology, all bringing unique perspectives. Thus, research examining the role of characters and their relationship with viewers (particularly when it involves the formation or altering of attitudes and beliefs), which we term character engagement, has become a free-for-all where terms may be invented and discarded without great concern for their similarity with past concepts or their predictive utility. This conceptual mélange is partially the result of diverse subfields examining similar concepts and processes. But even within subfields, researchers will use different terms to refer to similar processes, resulting in definitional obfuscation.

In this paper, we synthesize past work on character engagement and moral adjustment in media psychology and communication science and attempt to bring clarity to understanding how engagement with characters can lead to moral change in viewers. Our focus here is on characters in fiction, but some processes may extend to non-fictional media personae. By presenting a model that results from this synthesis, we explicate the necessary and sufficient conditions for moral change via character engagement, and do so at both proximal and distal levels of analysis. Toward these goals, we identify and define four overarching character engagement processes discussed by various disciplines. These processes can be grouped in a somewhat chronological manner (see, Figure 1) and are Attention (i.e., how much of a viewer’s attention is directed toward characters), Appraisal (i.e., how viewers perceive a character’s traits/attributes), Affiliation (i.e., how viewers take sides for and against characters), and Assessment (i.e., how viewers morally evaluate the actions of characters and the events that befall them). Attention always begins the model, as we describe in subsequent sections. However, the remaining three processes – Appraisal, Affiliation, and Assessment can occur in varying orders – and can all feed into Moral Adjustment (i.e., the extent to which the viewer’s own morality is shaped and molded through exposure to fictional personae). Finally, we close by suggesting the utility of our Character Engagement and Moral Adjustment Model (CEMAM) for organizing character processes across multiple disciplinary domains relevant to media studies.

Figure 1. The character engagement to moral adjustment model (CEMAM) depicts a semi-sequential process that encapsulates the subprocesses that lead to moral adjustment. The model begins with attention, through which certain characters receive greater and lesser attention of the viewer. The next three stages are not in any particularly serial ordering, but they involve appraisal and assessment of characters, as well as affiliational dynamics of the viewer with characters. The final stage is adjustment during which individuals alter or maintain their moral values.

Attention

The first process we address in our synthesis is Attention. Consumers of narrative content are unlikely to devote their limited attention indiscriminately toward all characters. Certain characters will stand out more than others, and absent a viewer’s attention, no other processes can occur (e.g., one cannot affiliate themselves with a character that they do not pay any attention to). Thus, the entry point for the CEMAM relates to how characters garner more and less attention from viewers. That said, once a character has received some level of attention, and therefore activated the other processes of the CEMAM, these processes may feedback into attentional processes. Yet, even when this occurs, for change to occur in affiliation, appraisal, or the outcomes of the other processes described in the CEMAM, one must reattend to the character. We draw on Bandura’s (2001) social cognitive theory (SCT) in order to explain how (a) associational patterns, (b) features of the character, and (c) features of the viewer separately and in combination influence attentional processes.

Associational Patterns: Commonality and Prominence

SCT’s associational patterns relate to the commonality and prominence of characters. Commonality relates to how often one encounters a specific character (or perhaps a specific type of character; e.g., TV Tropes). Characters and character types that are repeatedly encountered are likely to draw the viewer’s attention more so than characters that are rarely encountered; this is especially true with regard to characters for whom the viewer has already paid close attention and developed some level of affiliation. For example, perennially popular characters, such as Batman and Joker, consistently draw the attention of large audiences. In addition to repeated exposure, plot structures prioritize some characters over others, and major characters are more likely than minor ones to play a role across the entire plot. Prominence, a distinct but similar attribute as commonality, describes the importance of the character either to plot or the viewer. Whereas major characters are likely to be more prominent than minor characters, even extremely minor characters can rise to prominence. Consider for example, Boba Fett from the Star Wars series, who is credited as a supporting character in the first film he appears in, yet has become a prominent character in the broader Star Wars universe. This increased prominence may be due to the centrality of the character to the multi-film narrative’s overall conflict, which has resulted in a streaming series devoted to his character in 2022, forty years after his first appearance.

Commonality and prominence are two important determinants of attentional processes, and recent forays into data mining approaches may help quantify their contribution to attention (Hopp et al., 2020). Commonality can be quantified by narrative attributes, such as screen time, lines of dialog, or number of scenes. Characters with greater screen time, more lines of dialog, and who are present within a larger number of scenes should elicit greater attention. Prominence can be quantified by examining the network centrality of a character to the conflict present within a narrative (see, Hopp et al., 2020). Characters who are closer to the conflict are likely to garner greater attention than those who are farther from the conflict.

Features of the Character

Features of characters are also likely to direct a viewer’s attention (Hoffner & Cantor, 1991). Bandura (2001) lists several aspects of characters and events that are likely to determine attention to the character. These include salience, the importance the character to the viewer (Appiah, 2002); affective valence, how the viewer feels about the character (Zillmann, 2000); complexity, how difficult or complex the behavior of the character is to understand (Raney & Janicke, 2012); prevalence, a concept similar to commonality and prominence; accessibility, how easily the behavior or character comes to mind (Busselle & Shrum, 2003); and functional value, perceptions related to how useful the character or their behavior is to the viewer (Aubrey, 2004). Another important feature is distinctiveness, which should be considered in relation to other characters within a narrative. Characters who seem somewhat distinct from the other related characters (e.g., Boba Fett being one of several bounty hunters in Empire Strikes Back, but the only one with dialog) should garner greater attention (Hoffner & Cantor, 1991).

Some of these factors should be related in a linear fashion to attention. For example, more salient and more prevalent characters are likely to garner more attention (Busselle & Shrum, 2003). However, other factors, such as affective valence, are likely to be related to attention in a quadratic manner. Viewers are likely to pay the greatest attention to characters that are either highly liked or highly disliked, whereas characters that garner ambivalent or middling affect are unlikely to drive viewer’s attention (Tamborini et al., 2010). Similarly, novelty follows an inverted-u shaped curve with attention, with characters appearing moderately novel attracting the most attention (McCall & McGhee, 1977).

Many of the features of characters that determine a viewer’s evaluation of their salience, affective valence, and the like are situated within the narrative itself. These features can include attributes of the character and plot (e.g., the character’s behaviors) or of their presentation (e.g., formal features). We discuss the features of character and plot in our affiliation section. However, here we wanted to highlight how formal features, such as costume design choices, lighting, or mise-en-scene, might influence a viewer’s perception of a character’s importance and thus drive their attention. Just as costumes and outfits can communicate importance in the real world (e.g., the Pontiff’s hat, a general’s uniform), the visual depiction of characters can communicate narrative importance to viewers (Hoffner & Buchanan, 2005). Indeed, recent research has found that visual depictions can elicit perceptions of characters as heroes or villains absent any other cue (see, Grizzard et al., 2018).

Features of the Viewer

A final determinant of a viewer’s attention relates to attributes of the viewer and how they interact with attributes of the character. It is uncontroversial to say that the same character can be highly appealing to one viewer and highly unappealing to another. As viewers consume narratives, certain characters will garner more of their attention than others, and this variance in attention is likely to be related to some aspect of the character being relevant to the viewer. For some viewers, similarity of the character to themselves will drive attention (Hoffner, 1996; Hoffner & Cantor, 1991). The point here is that one must consider both attributes of the character and attributes and motivations of the viewer to fully understand which characters will garner the attentional resources of the viewers, leading to subsequent processes.

Appraisal

When we discuss character appraisal, we mean basic person perception and perceived similarity. Appraisal (unlike affiliation which we turn to next) is a psychological process that does not involve high levels of deliberation. For example, social categorization can include thinking of a character in stereotypical terms wherein they are perceived as a member of a social category (e.g., nurse, soldier, man, woman) rather than an individual (see, Frazer et al., 2022). Such social categorization often occurs spontaneously and – according to character evaluation models such as Sanders (2010) character impression formation model – lacks attribute-by-attribute assessment. Thus, appraisal processes are argued in most character perception models to occur very quickly once a character has garnered attention. Raney’s schema activation perspective (Raney, 2004) tested by Grizzard et al. (2018; see, also, Francemone et al., 2022; Grizzard, Francemone, et al., 2020) provides another example of how appraisal can occur without strong awareness or careful consideration. Appraisal is often the launching point for understanding moral change in viewers.

Person Perception

Perhaps the most basic form of learning about others falls into what psychology has termed person perception, which is a general tendency to form impressions of other people and to categorize others based on learned information (Moskowitz & Gill, 2013). Person perception incorporates automatic processes that occur without conscious attention, such as categorization, stereotyping, schema-activation (see, Raney, 2004), and trait inference. For example, clearly observable physical traits can lead to quick categorization of characters as good or bad, heroes or villains (Grizzard, Fitzgerald, et al., 2020; Grizzard et al., 2018; Hoffner & Cantor, 1991; Sanders, 2010). These categorizations are facilitated by past experience with characters in films and stories. To some extent, viewers build “prototypical” characters (e.g., heroes and villains, see, Kinsella et al., 2015; Eden et al., 2015; morally ambiguous characters, see, Eden & Tamborini, 2017) in their mind and compare new characters to these prototypes.

Person perception and categorization are thought to be cognitively efficient. Interestingly, the prototypes that are activated when observing a single character can extend to other characters. Grizzard, Francemone, et al. (2020) found that evaluating a high typicality character (i.e., a clear hero or clear villain) prior to evaluating a low typicality character (i.e., a more neutral appearing character) was capable of shifting perceptions of the low typicality character. When evaluated after a clear hero, a neutral character was evaluated as villainous, whereas when evaluated after a clear villain, a neutral character was evaluated as heroic. These findings suggest that schema activation processes might relate not only to specific characters, but also the narrative structures in which different types of characters are more or less likely to be present (see, Raney, 2004).

Similarity

One of the fundamental ways we use our perceptions of others is by deciding if they are similar or dissimilar to us. Byrne and Rhamey (1965) stated the correlation was so strong between similarity and liking that it could be considered a law of attraction. Similarity is best conceptualized as homophily between viewer and character, and it can be measured either as actual similarity or perceived similarity. Actual and perceived similarity show comparably large effect sizes when associated with interpersonal attraction (Montoya et al., 2008). Similarity as a predictor of liking for media characters has been examined in terms of demographic characteristics such as race and age, (e.g., Appiah, 2001), as well as cultural proximity (Trepte, 2008) and gender (e.g., Eyal & Rubin, 2003; Hoffner & Buchanan, 2005). However, whereas similarity may “prime the pump” toward character engagement, it is not sufficient to determine which characters will be perceived as possible objects of affiliation.

Affiliation

By affiliation with characters, we primarily mean “side-taking” (see, DeScioli, 2016), rather than identification, wishful identification, or parasocial relationships. We feel that in terms of moral adjustment, side-taking – the alignment of viewer attitudes with a media character’s goals – is most relevant in terms of affiliative processes. In much the same way that fans root for their team and against others, narrative consumers come to root for some characters and against others through side-taking. We note here that affiliation is a higher-level cognitive process as compared to appraisal. One can recognize a character-type and categorize or compare a character (i.e., appraisal) without any strong understanding of the character’s motivations and desires. Affiliation on the other hand requires some understanding of the character’s motivation and desires. For an individual to side-take requires some understanding of which side the character is on. Once the character’s side is understood viewers can then determine whether they align their own desires with the characters (as is often the case for protagonists) or against the characters (as is often the case for antagonists).

However, what explains side-taking and how side-taking is experienced has led to considerable theoretical arguments. Some theoretical perspectives argue that side-taking is the result of character evaluations (specifically moral judgment of the character’s motives and behaviors; see, Zillmann, 2006). These perspectives argue that viewer’s form character affiliations much the same way individuals do in real life; we form liking and disliking toward characters, who we perceive as separate entities from us, capable of, for example, pseudo-relational feelings (e.g., Dibble et al., 2016). Other perspectives assume that side-taking is the result of a form of ego-confusion wherein the narrative consumer views themselves as the character (see the concept of identification as described in Cohen, 2001; see also, Klimmt et al., 2009), which we address first in the following subsections.

Identification

Identification has a central role in many scholars’ understandings of fictional character engagement. For example, Oatley and Duncan (1994) states that through identification, the reader adopts the character’s goals and plans, and then simulates events that happen to the character and experiences emotions in consonance with the success or failure of these imagined plans (Oatley & Duncan, 1994, p. 69). Slater et al. (2014) have centered identification and transportation as key elements in narrative attraction via temporary self-relief. de Graaf et al. (2012) found that identification with protagonists could override preexisting attitude similarity, shifting attitudes in the direction of the character with whom readers identified (see, also, Klimmt et al., 2009). According to Cohen (2001), identification is “a mechanism through which audience members experience reception and interpretation of the text from the inside, as if the events were happening to them” (p. 245). This definition explicitly assumes that the viewer feels as though they are part of the narrative rather than simply being an observer of the plot. Identification involves four dimensions: empathy (sharing feelings with the character); cognition (sharing the character’s perspective); motivation (internalizing the character’s goals); and absorption (losing one’s self-awareness during exposure; Cohen, 2001; Cohen & Klimmt, 2021). Similarly, Brown (2015) proposes that liking, homophily, familiarity, and realism lead to transportation into a narrative and identification with a character, which leads then to worship of a character.

However, we – like others before us (most notably Zillmann, 1995) – question whether identification as a concept is necessary to explain the role of character engagement in moral adjustment. First, it is often unclear to what extent identification is distinct from liking, empathy, similarity, worship, or wishful identification (see, Cohen & Klimmt, 2021; see also, Brown, 2015, for review). It is also unclear what kinds of outcomes identification would predict that these other, more parsimonious accounts of character affiliation would be unable to predict. For example, it is not necessary to feel that you are Jamie Lee Curtis in Halloween to feel afraid for her, or to feel you are Oskar Schindler in Schindler’s List to feel horror and shame at the fate of your employees. Nor is worshiping a character required when forming, for example, exemplars of particular moral actions. One can form an exemplar of what not to do by examining the behaviors of detested and despised others. Liking of characters, perspective taking, and side-taking seem quite capable of predicting when a viewer will feel concordant emotions with a character (i.e., empathy for liked characters) or discordant emotions (i.e., counter-empathy for disliked characters; see, Zillmann, 2006). We thus agree with Zillmann (1995) in suggesting identification may obfuscate rather than illuminate relationships.

Empathy

An alternate way of thinking about affiliation from identification, with more relevance for moral adjustment, is by focusing on the role of empathy. Zillmann (1991) suggested that empathy, including motor mimicry, excitation, and dispositional considerations, facilitates the type of process meant when scholars use the term identification. Empathy in this conceptualization consists of three factors, which lead to a viewer feeling with or feeling for a character. Feeling with is mostly an involuntary motor mimicry response, such as moving a leg when a character is bitten by a snake, whereas feeling for is determined by dispositional considerations toward characters.

Liking

Zillmann (1991) states that positive affective dispositions, or liking, toward characters allow empathic reactions (i.e., feeling similarly valenced emotions as the character), whereas negative affective dispositions impair or reverse these reactions into counter-empathy (i.e., feeling oppositely valenced emotions as the character). The dominant role of character dispositions in determining empathy versus counter-empathy makes it impossible to understand viewer’s responses to characters without also discussing dispositions (or liking) toward them. Zillmann talks about liking and disliking as positive and negative affective dispositions, which in turn lead to “the approval or disapproval of good fortunes or misfortunes experienced by observed persons or personas” (Zillmann, 1991, p. 44). The correspondence between affective disposition and approval/disapproval of outcomes for the character is based on appraisals that entail moral judgment. Therefore, liking (or dispositions) and morality are so intertwined in narrative that disentangling the effects of morality and liking is nearly impossible (see, Grizzard, Huang, et al., 2020).

Therefore, when considering affiliation with characters, we propose that the empathic concept of “side-taking” may be better poised to explain viewer reactions than identification. For example, side-taking has been explicitly linked to moral judgment (DeScioli, 2016). Such a perspective has numerous benefits, but foremost among them is (a) it is more parsimonious than identification and (b) it maintains separation between concepts. We also note that such a perspective is not only consistent with moral judgments of mediated events, but also with non-mediated moral judgment processes.

Parasocial Interaction and Parasocial Relationships

The semantically related concepts of parasocial interactions (PSIs) and parasocial relationships (PSRs; see, Horton & Richard Wohl, 1956; see, also, Giles, 2002; Hartmann & Goldhoorn, 2011; Dibble et al., 2016) capture the idea that narrative consumers sometimes feel as if they can or are having interpersonal interactions with a character (PSI) or developing a friendship with a character (PSR). Although these interactions and relationships are more likely to result from repetitive contact with the same character, they are consistent with a process by which the character and the receiver remain separate entities. Although PSI and PSR are important broadly for understanding character affiliation, as they explicitly acknowledge a distance between self and character, we address them only briefly here. We will revisit PSI/PSR as they pertain to moral adjustment when we consider third party moral judgment in the Adjustment section.

Assessment

By assessment, we mean the process by which viewers assess moral judgment of characters (e.g., computing the rightness/wrongness of actions, announcing moral judgments, condemning character’s moral violations). Zillmann proposed that viewers’ basal morality, or fundamental notions of justice and appropriate retribution for wrongdoing, is the underlying cause of empathic reactions to media characters (Zillmann, 1991, 2006). That is, viewers will empathize with characters acting in line with a viewer’s own morality, and conversely will not allow themselves to feel empathy for characters who act outside the bounds of the viewer’s moral sanction. Past research has been generally consistent with this explanation. For example, both Eden and Tamborini (2017) and Kjeldgaard-Christiansen et al. (2020) found interactions between a viewer’s morality and character moral/immoral behavior on moral judgments. Viewers more strongly approved/disapproved of moral/immoral behavior, when it was more relevant to the viewers’ own sense of morality. Of course, the psychological development of the viewer will also determine these effects. Research indicates that children’s media have simpler depictions of moral conflict than media targeting older adults (see, Hahn, 2021), and past research from Zillmann and Bryant (1975) implicates psychological development as central to the evaluation of moral content. These ideas form the core of affective disposition theory (Zillmann, 2006), which has been supported in research in several media genres (see, Raney, 2004).

Recent work on character morality has focused on moral intuitions based on moral foundations theory (Graham et al., 2011). MFT posits that human morality is defined by sensitivities related to five distinct areas of morality (care, fairness, loyalty, authority, and purity). In terms of characters, Eden and Tamborini (2017) found that individuals high in fairness responded more strongly to immoral actions related to fairness. In other words, individuals for whom fairness was highly salient condemned a fairness violation more so than individuals for whom fairness was not highly salient. However, this moral judgment may also be a function of attributions made about the character. A recent study by Tamborini et al. (2018) found that manipulation of character behaviors – to be either more consistent, more distinctive, or to show greater audience consensus – altered the attributions viewers made about characters. Thus, the evidence seems to suggest that not only does a viewer’s basal morality influence moral evaluations of characters, the inferred basal morality of a character will also play a role.

Grizzard, Francemone, et al. (2020) in a recent study found that comparisons between characters could also facilitate moral judgments. In their data, moral evaluations of the protagonist and antagonist covaried in such a way that evaluations of one character helped to explain evaluations of the other. Grizzard, Francemone, et al. (2020) termed this kind of interaction between characters as character interdependence, which they defined as “mutual covariation between two or more characters wherein the perceived attributes of one character influence and interact with the perceived attributes of (an)other character(s). In other words, viewers’ judgments of one character may depend on the attributes of (an)other character(s)” (p. 275).

Assessment of Narrative Outcomes

According to Zillmann’s moral sanction theory of delight and repugnance (MSTDR; Zillmann, 2006), viewers evaluate outcomes that befall characters in conjunction with their hopes for positive outcomes for liked, moral characters, and their hopes for negative outcomes for disliked, immoral characters. Central to this process is the moral sanction of the resolution, which is chiefly concerned with whether good things happened to characters deserving of such outcomes and whether bad things happened to characters deserving of such outcomes. Punishment of characters is perhaps more central than reward in these situations, as witnessing others being punished leads to neural synchrony between individuals (see, Weber et al., 2011).

Punishments befalling immoral characters must neither exceed nor fall short of an individual’s threshold regarding what is appropriate or fitting for their misdeeds, else enjoyment suffers. Punishments that exceed such a threshold (e.g., execution for petty theft) or those that fail to meet the threshold (e.g., a bloodied nose for murder) likely disturb the viewer’s enjoyment of a narrative (see, Zillmann & Bryant, 1975; see, also, Grizzard et al., 2021a, 2021b). Zillmann is careful to note that his threshold argument is not related to a specific punishment or even a specific level of punishment. According to Zillmann, punishment is at first determined as an all or nothing process (either a bad thing happened to the character or not), and it is only after observing such a punishment that the viewer begins to wrestle with the appropriateness of the punishment. Synthesizing recent work on this topic (see, Grizzard et al., 2021a, 2021b), we can see that outcome assessment hinges upon (a) the severity of the transgression and (b) the severity of the punishment. So long as there is no major discrepancy between these two in the eyes of the viewer, the viewer is likely to be satisfied with the outcome. With regard to reward, the relationships are far less understood. For example, there is no research to our knowledge examining whether a threshold for positive outcomes exists. A recent reexamination of Zillmann’s outcome evaluation hypothesis from ADT points out a greater need for specificity about outcomes and their evaluations (see, Tamborini, Grizzard, et al., 2021).

Integrations of moral psychology with character research have helped to specify which types of actions are likely to lead to the development of positive or negative dispositions toward characters (see for example, Eden et al., 2015; Grizzard, Fitzgerald, et al., 2020). Tamborini, Grizzard, et al. (2021) argue that applying theories of human thriving (e.g., social determination theory, see, Deci & Ryan, 2000; Schwartz’s theory of basic values, see, Schwartz, 2012) can help to explain whether a viewer perceives that good or bad things happened to the character. According to Tamborini, Grizzard, et al. (2021), witnessing a character’s intrinsic needs being satisfied should lead to perception that a positive event has befallen the character, whereas witnessing a character’s intrinsic needs being thwarted should lead to perception that a negative event has befallen the character.

Although Tamborini, Grizzard, et al. (2021) focus on the importance of intrinsic needs to the viewer, we highlight that outcome assessment will also involve the perceived importance of the intrinsic needs to the character and the extent to which those needs are satisfied or thwarted. The same action (e.g., a character being thrown into a briar pit) can be perceived as either intensely punishing or a respite from danger. We argue that thwarting a character’s intrinsic needs will be perceived as more punishing in direct relationship to how important those needs are for the character. Similarly, satisfying a character’s intrinsic needs will be perceived as more rewarding in direct relation to how important those needs are for the character. Consider, for example, how rewarding a viewer would perceive a working-class person’s winning the lottery compared to Bill Gates’ winning the lottery.

Summary of the First Four Processes

In the prior sections, we synthesized and began to unite in a single conceptual model how character evaluation is influenced by attention, appraisal, affiliation, and assessment processes.

The model suggests that the entry point for these four processes is Attention. However, once a viewer has attended to a character, the three other processes (Affiliation, Appraisal, and Assessment) can operate serially, in tandem, or in a reciprocal manner (see, Figure 1). The combination of the four processes can begin to explain Moral Adjustment.

Adjustment

The main argument of this paper is, essentially, that character engagement, via attention, appraisal, affiliation, and assessment, form the determining factors underlying viewer moral adjustment. By moral adjustment, we mean the extent to which viewers adjust their own sets of moral standards and codes after encountering and engaging with characters. We do not specify if the viewer will adjust toward or away from the morals presented on screen. In many cases, we suspect adjustment will be toward the morals presented. But in others, characters may serve as exemplars of what not to do, and in these cases we would suspect adjustment would be away from the morals presented. We outline below several ways in which moral adjustment has been examined, or could be examined, based on the processes we outlined above.

Character as Moral Exemplars

First, characters can serve as exemplars and character types as prototypes of various moral values. Exemplars are specific instantiations of a class or category. For example, Abraham Lincoln might serve as an exemplar of an honest dealer, C-3PO as an exemplar of a cowardly sidekick, and Atticus Finch as an exemplar of a justice warrior. It is important to note that for exemplars, there is considerable variance between characters who might serve as an exemplar of specific moral traits. For example, Frodo Baggins and Indiana Jones might both be considered exemplars of bravery, even though the characters are extremely different from each other. Similarly, variance in exemplification can exist within the same characters. Indiana Jones might be considered an exemplar of bravery in some instances, an exemplar of anti-authoritarianism in others, or even an exemplar of overconfidence. Single characters thus can be exemplars of various traits, and various characters can be exemplars of singular traits.

In contrast to exemplars, prototypes represent an average of various instantiations (for overview of exemplars and prototypes, see, Mervis & Rosch, 1981). Prototypes do not exist as specific characters, but rather are the result of the averaging of features across various characters. Individuals might select specific characters (i.e., exemplars) at times as a representation of moral behaviors or they might select character types (i.e., prototypes).

We might expect the salience of moral values within an individual, what Zillmann describes in terms of morality subcultures (Zillmann, 2006), would predict which moral values a character is interpreted as representing and which characters are selected as representing moral values. Given the considerable variance within and between characters, we might expect these characters to serve as alternative viewpoints to the viewer. As viewers develop attachments toward some characters, they may begin to use these characters as points of comparison for themselves or others. Thus, both exemplars and prototypes help shape viewer moral judgments by altering perceptions of what is important based on what the narrative emphasizes.

As an example, consider how characters might have helped shape attitudes toward LGBTQ+ individuals (see, Schiappa et al., 2006). First-time viewers of Will & Grace may have entered the program with certain expectations of how a gay man might behave. If they had little contact with gay men in their everyday lives, then Will may have come to serve as an exemplar. As the viewer consumed more of the content, they might have come to have different conclusions and perceptions of what the concept of “being a gay man” included or did not include. When exemplar characters and their behaviors diverge and become discrepant from the viewer’s concept of the exemplar, the viewer is likely to update their understanding of the category the exemplar represents. Overtime, this type of discrepancy resolution can lead to greater acceptance of the actions of the characters (see, Schiappa et al., 2006).

Characters can also alter morality in viewers by making some morals or moral concerns more salient to viewers. This process can be understood in similar terms to how the Model of Intuitive Morality and Exemplars (MIME, Tamborini, 2013) describes the influence of media content more broadly. We would argue that this process takes place essentially through the characters central to dramatic narratives. For example, in one study Eden et al. (2014) found that over 8 weeks of soap opera exposure the villain’s behavior was related to subsequent shifts in moral domain salience in viewers. When the villain violated care, fairness, and purity, it made these domains more salient for viewers (when compared with a control group who did not watch the show). However, no such relationships were found for the hero character. Another example of character centrality in MIME processes is seen in Hahn et al. (2021). There, the manipulation of moral salience was deployed through character dialog. These instances indicate that it is not simply the presence or absence of moral concerns in a narrative that can alter view moral salience, but rather, how they are expressed and realized by the characters in the narrative. Characters thus seem to act as a sine qua non for the altering of moral salience in viewers.

Wishful Identification and Social Comparison

Characters may also display characteristics or behaviors that viewers find admirable and wish to emulate. Such a process represents wishful identification (Brown, 2015; Hoffner, 1996) and is likely to promote moral adjustment. The early study of potential determinants of wishful identification focused on character attributes (see, Reeves & Greenberg, 1977), and more recent work (see, Hoffner & Buchanan, 2005) has identified several character attributes that have been linked to wishful identification, including success (see also, Bandura, 2001), intelligence, physical attractiveness, aggressiveness, and humor (see, Hoffner, 1996, for complete list). The desire to be like, or behave in ways similar to, the character may immediately arise in the exposure situation, but may also extend far beyond (Rosengren et al., 1976). Admiration and other expressions of affection demonstrated by characters in a narrative can also have an influence on wishful identification with media characters (Reeves & Greenberg, 1977). When such feelings of admiration are expressed, they can serve as a reinforcement of a viewer’s wishful identification (Onu et al., 2016; Thomson & Siegel, 2017). Admired characters may function as an example or role model for future actions and may, thus, trigger imitation of attitudes and behavior (Konijn et al., 2007).

This type of admiration to emulation pathway echoes processes outlined in social comparison theory (Festinger, 1954), which suggests that social comparison can be categorized into (a) upward comparisons, in which we compare ourselves with others who we perceive to be or perform better than ourselves in a given situation; (b) lateral comparison, in which we compare ourselves with others who are performing similarly to ourselves; and (c) downward comparisons, in which we compare ourselves with others who we feel are performing worse. Social comparisons with characters have been examined by Mares and Cantor (1992) who found that downward comparisons can improve self-perceptions and the enjoyment of narratives. Lewis and Weaver (2016) similarly found downward comparison to lead to hedonic outcomes for viewers. Further supporting the role of social comparison in moral engagement, Krakowiak and Tsay-Vogel (2015, 2019) found that the salience of viewer morality led to differential media selection, such that those for whom vices were made salient preferred morally ambiguous characters (MACs) to villains, and those for whom virtues were made salient preferred heroes over MACs. However, these findings must be taken with caution, as in a follow-up study Tsay-Vogel and Krakowiak (2019) found that viewers did not differentially select characters for downward comparison when comparing heroes, MACs, and villains.

Character Moral Growth

Another way characters can make morals salient to viewers is by demonstrating moral growth and change. By examining audience responses to points in the narrative at which the protagonists made morally-relevant choices, Kleemans et al. (2017) found that the upward moral trajectory (from immoral to moral) of one protagonist (Léon from The Professional) caused audiences to engage more with the character, whereas the flat morality of another (Patrick Bateman from American Psycho) did not. Thus, the extent to which a narrative allows a character to change, followed by the character either changing or refusing to change, seems particularly important for how a narrative will lead to moral adjustment in viewers. The points of possible moral change in a narrative can be amplified by creative story telling. For example, at each point in Léon when the character faced a moral quandary, the camera paused on Léon’s face. Such pauses can allow viewers to also pause and reflect on the potential for moral change.

Character change can also be examined through the lens of SCT’s description of human agency (see, Bandura, 1989). SCT describes goal setting and goal achievement as a reciprocal process of discrepancy production and discrepancy reduction. Discrepancy production consists of comparing one’s current state to an aspirational state. The discrepancy provided puts an individual into a state of disequilibrium. As one advances one’s capabilities and skills toward the aspirational state (i.e., discrepancy reduction), disequilibrium is reduced. Once an individual has moved close to this aspirational state, they are likely to choose a new aspirational state and repeat the process. A character’s shifting morality can be considered a form of discrepancy production/reduction. Witnessing this kind of character growth can encourage a viewer to engage in a similar process. Moreover, characters can serve as the aspirational state that viewers seek to achieve. By selecting a character to emulate that is different from oneself, one is engaging in discrepancy production. When one moves one’s own standards toward that character, one is engaging in discrepancy reduction.

Normative Processes

A final pathway to moral adjustment may also be that repeated exposure to characters with similar moral judgment processes can alter viewers’ perceptions of social norms. Norms are the perceived ubiquity of behaviors (or approval of behaviors) of a referent group (Lapinski & Rimal, 2005). Descriptive norms are perceptions of the prevalence of certain behaviors within a group (Lapinski & Rimal, 2005), whereas injunctive norms are perceptions about what the group thinks its members ought (and ought not) to do, regardless of their actual prevalence (Cialdini et al., 1990; Lapinski & Rimal, 2005). Both types of normative influence may act as cues for appropriate behavior in a given situation (Cialdini et al., 1990).

Individuals learn about group norms via observation and communication from a referent group of important others. Characters may increase the social exposure to specific moral messages, and thereby change viewers’ perceptions about the prevalence of these behaviors in real life (Mead et al., 2014). Relatedly, the perception of how common specific moral beliefs are can be shaped by observing the behaviors of others (Lindström et al., 2018). This process has been demonstrated for health behaviors, including smoking, drinking, and sexual behavior (see, Bleakley et al., 2019). However, the theory of normative social behavior also suggests that group identity aspiration (i.e., the desire to emulate members of a group) and perceived similarity with the group, can increase modeling of observed behaviors directly (Crawford & Tobacco Control Network Writing Group, 2001).

Norms relate to outcomes that befall characters as well. The outcomes that befall characters and the extent to which they meet or fail to meet an individual’s threshold of acceptance is likely to have important influences on shaping an individual’s perceptions of what are acceptable or unacceptable levels of punishment. For example, Zillmann (2000, p. 60) argues that moral sanctions regarding different levels of punishment “are considered greatly variable within the reward-punishment dichotomy, their range being narrowed though by social validation. The ability to employ moral reasoning to ‘fit a punishment to a crime’ is thus seen as initially arbitrary matching that is eventually communally endorsed.” Repeatedly witnessing death as a fitting retribution could potentially increase acceptance of the death penalty (see, Schultz & Huet, 2001).

Although normative influences likely occur at the individual level, the popularity of any particular piece of media will drive this process forward and make it observable (see, Tamborini, 2013). Recent research has suggested that moral emotions in particular, are likely to be shared among ingroups (Brady et al., 2017), which leads to the formation and preservation of moral norms (Harms & Skyrms, 2008). If characters, who are invited nightly into viewers’ homes and lives, communicate particular attitudes or feelings night after night, new moral norms may be formed unobtrusively in viewers’ minds.

The Value of the CEMAM

In this paper, we have attempted to illustrate the mechanisms through which engagement with fictional characters may adjust or alter viewers’ own morality. We argue that four overarching character engagement processes lead to moral adjustment in viewers. In the remaining section of this paper, we address the utility of this model of character engagement. We begin by acknowledging that our model builds upon extant models of character engagement and morality from media psychology and communication science. Most notably, we drew on Bandura’s (1989, 2001) social learning theory and Zillmann’s exemplification theory (Zillmann, 1999), moral sanction theory of delight and repugnance (Zillmann, 2006), and the three-factor theory of empathy (Zillmann, 1991), as well as Tambornini’s MIME (Tamborini, 2013) to discuss appraisal and assessment. In terms of appraisal, we also drew heavily on Sanders’ (2010) Character Impression Formation model (CIF) and Konijn and Hoorn’s (2005) Perceiving and Experiencing Fictional Character (PeFiC) models. However, we would argue that the CEMAM’s integration offers improvements over these individual models.

First, the CEMAM is more comprehensive than these models, as each model noted above contributes to only one or two segments of our model. Additionally, our model includes aspects of moral adjustment via character engagement that have not been addressed or included in previous models, such as models of moral formation via normative processes, social comparison, and moral side-taking, retribution equity, among others. We have attempted to synthesize and connect many disparate perspectives on character engagement to the extent that these perspectives contribute to our understanding of characters as forces for moral change.

Second, the CEMAM is a guiding framework built predominantly on empirical observation, which can form the foundation for future inquiries. Although the discerning reader may suggest that the processes we delineate above are covered by existing models, we suggest that our contribution is to connect many disparate data points representing hundreds of empirical studies. This contribution is an advantage as previous theorists did not have access to such a trove of data from which to base their observations. In addition, it provides a comparative advantage over models from disciplines such as aesthetic criticism and literary understanding, which commonly lack empirical tests. Although there is great value in arguing from logic, when that logic does not lead to testable hypotheses that can be falsified, we are left unclear about what we have learned. Therefore, we have attempted to select ideas from past research in order to build a testable, yet parsimonious and falsifiable model of character engagement.

Why is this important? We argue that to move our understanding of character engagement forward, we must start by synthesizing the extant literature, clearly defining and describing our terms, and explicating how they fit together in specific fashion. A more careful, precise, and consistent description of the various mechanisms (rather than variables) will help improve predictions related to moral change from media exposure. If such an approach is carefully and consistently followed, it should be possible to weight these processes and compare to what extent they separately or together lead to moral adjustment via character engagement. We hope to work toward a computational model, such as the Disposition Theory Vector (DTV; Weber et al., 2008; Tamborini et al., 2010; see, also, Grizzard et al., in press, for an alternative account), which could be applied to character engagement and viewer moral adjustment. Thus, this is the first step toward that larger goal. However, even en route toward that destination, we argue that a model which insufficiently or incorrectly defines its components is not terribly useful. Therefore, we have taken the step here to define our component parts in order to form a testable, working model, from which we can move forward with greater shared understanding of our terms and expectations.

Disclosure Statement

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

Additional information

Notes on contributors

Matthew Grizzard

Matthew Grizzard (PhD, Michigan State University) is an Associate Professor in the School of Communication at The Ohio State University. His research examines moral emotions and moral judgment processes related to the consumption and reception of narrative and interactive media entertainment.

Allison Eden

Allison Eden (PhD, Michigan State University) is an Associate Professor in the Department of Communication at Michigan State University. Her work centers on understanding media enjoyment, particularly the role enjoyment plays in attention to and selection of media content, and the effects of entertainment on user behavior and well-being.