Putting Perspective Taking in Perspective 1 Putting Perspective Taking in Perspective

We present a new framework for the discussion of perspective taking, particularly with reference to the processing of literary narrative. In this framework, adopting a perspective entails matching evaluations with those of the narrative character. This approach predicts that perspectives should be piecemeal rather than holistic, dynamic rather than consistent, effortful rather than automatic, and reactive, in the sense that they are a function of the reader’s on-line processing as it interacts with narrative technique. We describe evidence from an interpolated evaluation method in which readers are periodically interrupted and asked to rate evaluations from a character’s perspective. The results indicate that interpolated evaluations interact with narratorial stance to determine a character’s transparency – that is, the extent to which she is rational and understandable. These results demonstrate that perspective taking depends on the details of a reader’s processing over the course of the story. Putting Perspective Taking in Perspective 3 Putting Perspective Taking in Perspective The ability to take into consideration another person’s point of view is a fundamental requirement of many aspects of social interaction. Thus, it is hardly surprising that perspective taking has become a central concern in a wide variety of different disciplines. Discussions and studies of this ability abound in fields such as business and management, sociology, education, philosophy, literary studies, and several subfields of psychology, including developmental (where it is discussed in the context of theory of mind), social, clinical, and personality. In spite of this wealth of research, our understanding of how perspective taking in literature is related to this ability is still rudimentary. The term “perspective taking” is commonly understood to refer to the adoption of another person’s physical vantage point, that is, the ability to imagine what something looks like from where someone else is standing. However, it can also allude to the ability to understand or consider someone’s psychological viewpoint – their thoughts, feelings, or emotions (Epley, Keysar, Van Boven, & Gilovich, 2004; Johnson, 2012). In literary studies, it is typically this understanding of the term that is borrowed. The problem is that this simple interpretation is imprecise and leaves unexplained the textual and cognitive requirements and complexities that underly the process of psychological perspective taking. In this paper, we begin by discussing what seems to us to be a common, implicit account of the nature of perspective taking. Although intuitive, this account has a number of obvious deficiencies. As an alternative, we provide a novel framework for the study of literary perspective taking and how narrative technique affects readers. In particular, we argue that perspective taking involves a process of matching evaluations generated during the course of reading. This view entails that perspectives Putting Perspective Taking in Perspective 4 are reactive in the sense that they are determined by the moment-to-moment dynamics of readers’ processing as it is elicited by the details of narrative technique. We then provide some empirical findings that demonstrate both the reactive nature of perspective taking and the role of narratorial guidance. An Implicit Account of Perspective Taking We argue that empirical investigations of perspective taking are often based on assumptions of homogeneity, consistency, and automaticity. First, the nature of the expression, “to take a perspective,” suggests that there is “a perspective” that is homogenous and complete: At some point during reading, readers’ evaluations align with those of a character, and there would seem to be no possibility of not taking that perspective to any degree. Second, “a perspective” seems to presume that it is consistently aligned with a character, so that it can be identified independent of the dynamics of the reader’s processing over the course of the text. Third, there is typically little discussion of mechanisms, so that perspective taking in fiction would seem to occur naturally without motivated effort or risk of failure. In this sense, it is often associated with identification (Green, Brock, & Kaufman, 2004; Kuiken et al., 2004; Tal-Or & Cohen, 2010). This automaticity assumption seems to exclude the possibility that a perspective will not be taken and minimizes the processing complexities that might be involved. Indications of these assumptions can be found in many examples. In the field of media studies, Tal-Or and Cohen (2010) claim viewers “enter the fictional world and adopt the perspectives of characters” (403-4). Busselle (2009) argues that “identification is purported to be the adopting of the perspective and emotions of a character” (338). Green (2004) indicates that Putting Perspective Taking in Perspective 5 “to identify with a character means seeing the character’s perspective as one’s own, to share his or her existence” (319). Although these (and other) researchers were not explicitly arguing for an account of perspective taking that is homogenous, automatic, and consistent, we argue that such properties are typically unexamined, implicit assumptions. To our knowledge, there is no empirical study that challenges them or presents alternatives. Related issues arise in the discussion of perspective taking in real life. Research in social psychology has often been concerned with the effects of taking someone’s perspective, where it is associated with behaviors related to moral development (Kohlberg, 1969) and altruism (Cialdini, Brown, Lewis, Luce, & Neuberg, 1997). Galinsky (2004), for example, found that asking subjects to take someone’s perspective in an essay-writing task had an effect on out-group prejudice. More generally, Galinsky and Moskowitz (2000) assume that “perspective-taking has been shown to lead to a merging of the self and other” (709; cf. Davis, Conklin, Smith, & Luce, 1996). (Whether such a merger is possible, or whether perspective taking rather entails an analogical inference from one’s self to others has been the center of a long-standing debate, e.g., Goldman, 2006). This research is focused on the results of perspective taking. Nevertheless, it would seem to presuppose that perspective taking is straightforward and can be engaged volitionally (by, for example, complying with experimental instruction). Consequently, it neglects the question of what that perspective entails and the nature of the perspective taking process itself. Moreover, the assumptions of homogeneity, consistency, and automaticity are inconsistent both with the properties of good literature and with what we know about perspective Putting Perspective Taking in Perspective 6 taking in real life. To begin with, it is easy to find literary examples in which the possibility of taking a character’s perspective seems to be relatively unlikely: Artistic texts can present their readers with complex characters whose own evaluations and opinions are inconsistent, indeterminate, uncertain, or ambivalent, as, for example, with unreliable or naïve narrators. Presumably, such complexity would inhibit the facile taking of a perspective, and readers might share some aspects of a character’s perspective but not others. Thus, it would be unreasonable to think of perspective taking as homogenous. Instead, a perspective is piecemeal in that a reader’s perspective may match a character’s to some extent but is unlikely to match perfectly in any given situation. Similarly, it seems clear that a readers’ understanding of a complex character must change over the course of a story. Thus, perspectives must be dynamic and developing rather than consistently aligned with a character. Finally, the assumption that literature naturally fosters perspective taking implies that there is something about literature, such as specific textual features or techniques, that safeguards readers from the numerous roadblocks that can thwart individuals’ perspective taking in everyday life. Although some researchers have alluded to the complexity of perspective taking in literature (Harding, 1962; Oatley, 1999; van Peer & Chatman, 2001), empirical work on literary response has generally failed to address the reader processes involved in perspective taking. The idea that perspective taking is simple or automatic is surprising given the wealth of psychological research on perspective taking in real life. This work demonstrates that perspective taking entails “a complex and critical set of cognitive abilities” (Barnes-Holmes, McHugh, & Barnes-Holmes, 2004, p. 18) which are prone to failure (Hoffmann, 2000) due to the many Putting Perspective Taking in Perspective 7 requirements that must be met in order for one to succeed in adopting someone else’s perspective. Some of these include: motivation to understand a target’s mindset (Epley & Caruso, 2012); attention (Wondra & Ellsworth, 2015); the ability to overcome the “egocentric default” (Epley & Caruso, 2012, p. 298) and to bracket out the influence of past experience, cultural schemas, biases, misinformation, and so on; access to adequate information; and the ability to draw sound inferences and to “create a complex and fleshed-out representation of the other person” (2012, p. 281). Even under optimal conditions, factors such as morally ambiguous situations can trump even the most skilled perspective takers (Cervone & Tripathi, 2009). There is no reason to think that these types of complexities would not apply in understanding literature. Perspective Taking as Matching Evaluations To move to a deeper analysis of perspective taking as a process, it is necessary to first define the noun “perspective.” In the popular, intuitive sense of the term, “perspective taking” is equated with a matching, or sharing, of perspectives. Such a match can occur as a result of preexisting similarity (individual A recognizes that individual B shares similar values or beliefs). However, a more precise understanding of perspective taking is obtained by noting that it is generally related to evaluation (cf. Harding, 1962; Dixon & Moore, 1990; Lamm, Batson, & Decety, 2007; Igl, 2016). In keeping with this research, we understand a psychological perspective in a literary context as a set of evaluations, that is, some body of assessments of the characters and events of the story world. For example, in Moby Dick, the narrator, Ismael, regards Captain Ahab as mad; if the reader comes to the same assessment, we can say that the reader and the narrator share his or her perspective. We use the term “perspectival resonance” to Putting Perspective Taking in Perspective 8 refer to this cognitive matching of evaluations (which is not to be confused with the concept of “memory resonance” (e.g., O'Brien, Rizzella, Albrecht, & Halleran, 1998)). In contrast, one is perhaps unlikely to take the perspective of Captain Ahab regarding his preoccupation with finding the great white whale. We refer to this mismatching of views as “perspectival dissonance.” Matching evaluations are related to the analysis of Davis, Conklin, Smith, and Luce (1996) in which perspective taking increases the overlap of the self and the target; presumably, an increase in matching evaluations would constitute such an increase in overlap. In addition, it also seems clear that it is possible to take another’s perspective conditionally, even if one’s own beliefs might lead to a different evaluation. In this case, one forms evaluations that are conditional on another’s knowledge and situation. This situation often arises in reading fiction because the story may involve situations or circumstances that the reader has not experienced. For example, many readers would not share, a priori, Ishmael’s desire to go to sea as a sailor. However, given the development in the text, they might come to appreciate that evaluation by Ishmael under his circumstances. Matching evaluations conditionally in this sense would be of the general form, “Yes, that would be my evaluation if I were in that situation with that knowledge and background.” Building on this definition, we propose the term “conditional perspectival resonance” for evaluations that match given hypothetical scenarios based on another’s circumstances. A closely related conceptualization of this reaction is the term “simulation,” which is described as the running of another’s situation in one’s mind (Oatley, 1999; Mar & Oatley, 2008), resulting in the actual experiencing of the other person’s plight (Kaufman & Libby, 2012). Another possibility, though, is that one imagines another’s Putting Perspective Taking in Perspective 9 evaluations – thoughts, feelings, reactions in a foreign situation – from a detached, spectator perspective (Harding, 1962; Carroll, 2001; Busselle & Bilandzic, 2008). Our concept of conditional perspectival resonance suggests this possibility. Thinking of perspective taking as a process of matching evaluations leads to several predicted properties. First, the evaluations need to be based on information provided in the story world: No evaluations can be made if the relevant information for forming an evaluation is not available. We refer to this information as the evaluation basis. Second, perspective taking must be piecemeal rather than all-or-none because only some of the possible evaluations will match; other possible evaluations may not. Thus, taking a perspective must be regarded as an incremental process that unfolds as each evaluation is identified and matched. Third, perspective taking may not occur at all due to indecision; there are generally a large number of possible evaluations that might be relevant in a given context, and a potential perspective taker will not have an opinion on many of these. Some evaluations may be unresolved and other may not have even been considered. We use the term “perspectival flux” to refer to this state of not having formed or matched the evaluations that might contribute to perspectival resonance or dissonance. Our notion of “flux” corresponds to what in quantum cognition is referred to as a superposition state (Wang, Busemeyer, Atmanspacher, & Pothos, 2013). Informally, in this context, a superposition state can be thought of as two mutually inconsistent evaluations that are maintained simultaneously without selecting either one. In our analysis, one is in a state of perspectival flux (superposition state) while uncommitted with respect to different possible evaluations (and their respective matches). Flux is not quite the same as simply having a bias or Putting Perspective Taking in Perspective 10 predisposition that is not verbalized: Such a predisposition could presumably only lead to one possible evaluation, while we think of a superposition state as being consistent with several possible evaluations depending on the subsequent nature of the processing. Fourth, perspective taking is dynamic; evaluations can be revised, and the revised evaluations may match or mismatch. Flux can decrease as some relevant evaluations are matched or increased as new evaluations become relevant. A final prediction, central for the current investigation, is that perspective taking is reactive. Reactivity is a product of flux (the matching evaluations may be undecided at any given point) and dynamics (evaluation matches change over time). Because of these properties, perspective taking will vary with the information provided in the narrative and how the reader responds to that information. These responses will be a function of both the details of the reader’s knowledge, biases, motivation, attention, and inferences as well as the manner in which the textual information is presented (or withheld) by the narrator. For example, a reader may have a fear of water, and that may affect the evaluations that match or do not match in reading Moby Dick. Or, more critically, simply reminding a reader of a previous boating experience could affect those evaluations. Thus, the nature of the information provided by the narrator will often interact with the specific knowledge that is currently active for each reader, producing or not producing these kinds of remindings. For example, drawing an inference early in a story may have very different implications for memory and processing than if the same inference is drawn later. Other possible processing variations include what Oostendorp (2001) refers to as the “skipping hypothesis” the tendency to skip information when one thinks one has enough – and Putting Perspective Taking in Perspective 11 the “rejection hypothesis” – the difficulty in relinquishing information when a perspective has been established. Potentially, these relatively minor variations in processing can have large effects on perspectival resonance and the ultimate interpretation of the narrative. Narratorial Guidance As some scholars have pointed out, readers interact not only with characters, but also with narrating agents (Oatley, 1999; Niergarden, 2002; Skar, 2013; Igl, 2016), and such agents may guide processing in a variety of ways. For example, narratorial techniques may guide the reader towards identifying a central character and devoting mental processing to understanding that character and his or her evaluations (but they need not unequivocally; Keen, 2007). These include access to a character’s mind, the use of a focal character through whose eyes the events of the story world are seen, and free indirect speech that intertwines the voice of the character and the narrator. More generally, narratorial guidance may indicate that some aspects of the story should be processed more extensively than others. For example, Mullins and Dixon (2007) found that when the narrator indicates that a particular piece of information in mystery is notable, readers are more likely to solve the mystery. This may lead to less processing of other aspects of the story world. For example, in The Murder of Roger Ackroyd (Christie, 1926/1997), the narrator focuses on some aspects of the story world to the exclusion of others, and it is precisely these other, neglected aspects of the story world that provide clues that the narrator is the murderer. Certainly, taking a character’s perspective can make a reader more attentive to the information about that character, but can also cause the reader to overlook other information (Morrow, 2001). Putting Perspective Taking in Perspective 12 Another aspect of narratorial guidance is narratorial stance, that is, the attitude of the narrator towards and evaluation of characters, events, and milieu (Lanser, 1981). Narratorial stance can be suggested through a variety of techniques. In particular, the narrator may offer explicit or implicit evaluations, attributed either to the narrator or a central character. When these match those of the reader, this would lead to an increase in resonance. Presumably, this would occur if those evaluations were well justified in the context of the story world. On the other hand, if the narrator offers evaluations that seem ill justified, as with, for example, unreliable narration, the reader is less likely to share these evaluations and perspectival dissonance may result. Furthermore, narratorial stance can be comprised of competing techniques; a story may be presented from the perspective of a main, focal character, but that character may be slowly revealed to be naïve, suggesting an ironic distance between narrator and focal character. Narratorial stance indicated by explicit, well-justified evaluations is typical of certain kinds of “light reads,” or what Barthes (1974) calls “readerly texts.” The process of matching evaluations would then be simple: We can easily identify the “bad guy,” and we take the perspective of the “good guy.” For example, in the Philosopher’s Stone, it is clear that Voldemort is the antagonist and that we should adopt Harry Potter’s perspective in general. What makes it possible to easily form matching evaluations is the stance of the narrator towards the characters: The narrator makes it clear that he or she sides with the good guy and denounces the bad guy. Therefore, perspectival resonance increases when readers can generate evaluations that match those of the character and the narrator, and some kinds of fiction promote this effect. Putting Perspective Taking in Perspective 13 However, this form of simple narratorial guidance seems implausible for many complex literary texts. Zeman (2016) noted that “perspectives are primarily implicit and are made explicit only under certain circumstances” (19). One could argue that more complex texts are still intentional and purposeful, and that certain techniques guide us towards the adoption of a particular perspective. However, good literature often employs narratorial techniques to create levels of complexity and ambiguity that thwart facile perspective taking and promote deeper reflexion. To make the story world more like real life, with all its unresolved dilemmas and complexities, narrators can provide information about characters that is incomplete, inconsistent, ambiguous, and dynamic. Iser (1978) described the “multiple perspectives” and “constantly shifting constellation of views” (96) that are common in literary narratives, leading him to conclude that “it is not possible for the reader to embrace all perspectives at once” (97). Decades later, Oately (1999) explained that “most of our great writers encourage a moving back and forth along the spectrum of aesthetic distance, identification with different characters in turn ... or identification with a character and then a view from the exterior perspective of the narrator” (446). More generally, in complex texts the stance of the narrator with respect to a character can be ambivalent or ambiguous (Igl, 2016; Lanser, 1981}. Ambiguous narratorial stance may be created in a variety of different ways. A narrator may express evaluations that are sometimes consistent and sometimes inconsistent with those of a character; or a narrator’s attitudes may change over the course of the story (Oatley, 1999); or, in cases of external narration, the implied author’s perspective can be “refracted” among different textual voices and characters (Bakhtine, Putting Perspective Taking in Perspective 14 1934/1981; Zeman, 2016). Ambiguous narratorial stance requires more active inferencing on the part of the reader. Barthes (1974) coined the term “writerly” to refer to texts that require such construction on the part of the reader. The more writerly the text, and the less consistent or absolute the narrator’s stance, the greater the fluctuations in the reader’s perspective taking endeavors, and the more likely it is that readers remain in a state of perspectival flux so that they do not simply embrace any one character’s perspective (Miall & Kuiken, 2001; Skar, 2013). Thus, in such literature, narratorial guidance promotes greater flux or uncertainty with respect to the relevant evaluations rather than increasing resonance in a simple fashion. Whether such flux is ultimately resolved in terms of matching or mismatching evaluations will depend on the details of the reader’s processing. Of course, conceptualizing and measuring this fluctuating, dynamic reader response would not be possible on the naïve view of perspective taking as homogeneous, consistent, and automatic. Incorporating such ideas into a theory of perspective taking is thus an important challenge for empirical studies of literary response. Testing for Reactivity and Guidance The main goal of this research is to provide evidence that, contrary to the assumption that perspective taking is a homogenous, consistent, and automatic process, readers often remain in a state of perspectival non-commitment, or flux, and will only form matching or mismatching evaluations when they are motivated to do so. Such motivation may be self induced: For example, readers may read for the pleasure of engaging with minds very different from their own, or they may want to analyze characterization techniques to write a good essay in a literature course. Similarly, taking a perspective may be needed in order to understand the events that Putting Perspective Taking in Perspective 15 transpire in the story world and the information being conveyed about them. As outlined above, the extent of the perspectival flux and how it may resolved can depend critically on narratorial guidance. However, it can also be provided externally by researchers in experimental settings. For example, in previous research, instructing their participants to read a story or passage from a given character’s perspective increased what we call perspectival resonance (Cupchik, Oatley, & Vorderer, 1998; Hakemulder, 2000; Hakemulder, 2001; Hakemulder & Zyngier, 2008). Generally, though, in the absence of these kinds of motivations or external prompting, it is quite possible that readers would simply have remained in a state of perspectival non-commitment. Here, we describe a study in which we used a more focused approach to promoting perspectival processing. In order to demonstrate that perspectival resonance is dynamic and reactive, we used a method of “interpolated evaluations”: Readers were periodically interrupted while reading a story and were asked to rate evaluations made by one or another character. Specifically, they were given an evaluation that one of the characters might have (e.g., “Rosie knew that she would have to work hard to understand haiku in Japanese”) and asked to rate the extent to which it was true. Our supposition was that generating such ratings would often require additional processing of information from the story world. Readers would have to retrieve and elaborate relevant textual information, generate an evaluation based on that information, and then assess the extent to which that evaluation matched the item to be rated. We argue that this is precisely the kind of processing that would be involved in generating conditional resonance with that character. In other words, to perform the interpolated evaluation task, readers would need to generate a representation of the character’s understanding of his or her world, including Putting Perspective Taking in Perspective 16 matching conditional evaluations. Thus, processing the interpolated evaluations should affect the processing of evaluations of characters and events in the story world and potentially change the extent to which readers take a character’s perspective. In our interpolated evaluations, we attempted to select aspects of the story world that were orthogonal to the important elements of the plot. In particular, the evaluation generated by a reader could not be simply matched to an evaluation explicitly mentioned in the text. Rather, the intent was to stimulate further processing of the story world and the character in order to form the basis for additional matching at some later point. We did not have strong presuppositions concerning what the effect of such interpolated evaluations should be on perspective taking. Presumably, when the interruptions lead readers to generate evaluations that match those of a character, perspective taking should be enhanced. On the other hand, it is possible that the evaluations that are generated mismatch those of a character, leading to perspectival dissonance. Finally, the additional processing may be irrelevant, in that it does not lead to either matching or mismatching evaluations for a character. We conjecture that these possible effects may be mediated by variations in narratorial guidance. Our measure of perspective taking was character transparency, the extent to which the behavior of a character was reasonable and understood by the reader (cf. Kotovych, Dixon, Bortolussi, & Holden, 2011). In our current perspective-taking framework, transparency can be considered as an index of resonance. In particular, when readers have high perspectival resonance, there are many evaluations that they share with a character. We suppose that such shared evaluations make it straightforward to interpret and explain the thoughts and actions of Putting Perspective Taking in Perspective 17 that character. In turn, this would be reflected in higher ratings of transparency. Using transparency as an index of perspective taking is related to the work of Regan and Totten (1975) on the effect of perspective taking on situational attributions for explaining behavior. We assessed character transparency at the conclusion of the story. These predictions can be contrasted with what might be expected if perspective taking were homogeneous, consistent, and automatic. If such assumptions were correct, readers would normally have adopted a perspective (because perspective taking is automatic) and that perspective would include the relevant evaluations (because it is homogenous). Readers could then simply retrieve the information they have about the character’s perspective in order to decide whether the supplied evaluation was correct or not, and make the rating response. Further, because the perspective is assumed to be consistent, the act of producing such ratings would have no impact on later ratings of character transparency. Instead, we argue that requiring readers to rate evaluations prompts additional story world processing that could lead to new or different evaluations. Thus, reactivity of perspectival resonance is a central prediction of the idea that perspective taking requires the construction (and matching) of evaluations. Method Subjects read the short story “Seventeen Syllables” (Yamamoto, 1994) prepared in a loose-leaf binders. The story primarily concerns the relationship between Rosie, the adolescent daughter of Japanese immigrants, and her mother. The story is related in third person, but the focal character is Rosie, with a significant amount of perceptual and mental access to that character; informally, we see the mother, father, and events through Rosie’s eyes. The use of a Putting Perspective Taking in Perspective 18 focal character is one technique that would presumably encourage readers to take that character’s perspective. The story is 5,119 words long. Two experiments were run with different versions of the story: The first experiment used the original story; the second experiment used a version in which explicit attitudes of the narrator (or Rosie) and descriptions of Rosie’s thoughts were eliminated as much as possible. We were especially concerned about eliminating those textual segments that implied an evaluation of characters and events. However, we retained information that seemed necessary for the logical flow of events in the story world. We refer to the modified version as “reduced guidance” because it had less information to guide the the reader with respect to evaluations of characters and events. Table 1 provides an excerpt from the two versions. Although the second experiment was run at a later time, the two are reported and analyzed together for comparison and ease of exposition. Putting Perspective Taking in Perspective 19