1
Empathy
Mark H. Davis
EckerdCollege
What is empathy? Although seemingly simple, this questionhas proven surprisingly difficult to answer. For over 200 years, thoughtful people have tried to understand the general phenomenon in which one individual, through observation of another, comes to experience some change in his or her thoughts or feelings. These efforts have typically fallen into one of two broad categories. One approach has been to consider empathy an essentially emotional phenomenon, with the defining feature of the empathic experience consisting of observers either coming to share the target’s emotional state (e.g., Eisenberg and Strayer 1987), or to experience some emotional state in response to the target’s (e.g., Batson 1991). The other approach has been to consider empathy an essentially cognitive phenomenon, with the defining feature of the experience consisting of observers coming to discern accurately the target’s internal state, but without necessarily experiencing any emotional change themselves (e.g., Wispe 1986).
Recent years have seen growing acceptance of a third approach, which is to explicitly treat empathy as a multidimensional phenomenon that inevitably includes both cognitive and emotional components (e.g., Davis 1983; Hoffman 1984). In this vein, I have previously proposed a model designed to organize all of these approaches into a comprehensive treatment of the empathy phenomenon (Davis 1994); Figure 1 contains a somewhat revised and updated version of this model. In contrast to much previous work, the spirit of this model is deliberately inclusive, designed to emphasize the connections between these constructs. Thus, empathy is broadly defined here as a set of constructs having to do with the responses of one individual to the experiences of another. These constructs specifically include both the processes taking place within the observer and the affective and non-affective outcomes that result from those processes. Based on this definition, the model conceives of the typical empathy "episode" as consisting of an observer being exposed in some fashion to a target, after which some response on the part of the observer--cognitive, affective, motivational, and/or behavioral--occurs. Four related constructs can be identified within this prototypical episode: antecedents, which refer to characteristics of the observer, target, or situation; processes, which refer to the particular mechanisms by which empathic outcomes are produced; intrapersonal outcomes, which refer to cognitive, affective, and motivational responses produced in the observer that are not necessarily manifested in overt behavior toward the target; and interpersonal outcomes, which refer to behavioral responses directed toward the target. One critical feature of this model is that it considers both cognitive and affective outcomes to be part of empathy.
Because of its breadth and versatility, in this chapter I will use the model as a framework for selectively presenting and discussing past and current work on the general topic of empathy. At the conclusion of the chapter I will also offer some suggestions regarding promising areas for future research. To begin, however, let us consider in more detail the various elements contained within this model.
Empathy: An Organizational Framework
Antecedents
The first component of this framework encompasses antecedent factors—features of the observer, target, or the situation that may influence in some way the subsequent empathy episode.
The observer. All observers possess certain characteristics that have the potential to influence the empathy episode. One of these is the biological capacity for empathy-related processes and outcomes. Almost all members of our species possess these capabilities, although sometimes serious deficiencies do occur in the ability to imagine other perspectives (e.g, autism) or to experience compassion for distressed others (e.g., sociopathy). Of more importance for the purposes of this chapter are the individual differences that exist in non-clinical populations in the tendency to engage in empathy-related processes or to experience empathic outcomes. A variety of individual difference measures have been developed over the years for the purpose of assessing the dispositional tendency to engage in empathy-related processes such as perspective taking (e.g., Hogan 1969) or to experience empathy-related affective responses (e.g., Mehrabian & Epstein 1972). Of special note here are individual differences in the tendency to experience two particular affective reactions to the distress of others. Specifically, the tendency to experience feelings of sympathy for a person in distress, and the tendency to experience personal unease in such cases, are especially important antecedent characteristics that have considerable relevance for social interactions.
The situation. All responses to another person, whether cognitive or affective, emerge from some specific situational context, and these contexts vary along certain dimensions. One such dimension is the strength of the situation, defined as its power to evoke an emotional response from observers. For example, a situation that includes a clear display of negative emotion by a weak or helpless target is particularly able to engender powerful observer emotions, and would be classified as a “strong” situation. In contrast, situations lacking such evocative emotional cues would be characterized as relatively weak. A second situational feature is the degree of similarity between the observer and target. Although similarity is of course affected by characteristics of the observer, it is really a joint function of the target and observer, and is thus considered a feature of the situation.
Processes
The second major construct within the framework consists of the specific processes that generate empathic outcomes in the observer. Based on the work of Hoffman (1984), and Eisenberg (Eisenberg et al. 1991), the model identifies three broad classes of empathy-related processes, chiefly distinguished from one another by the degree of cognitive effort and sophistication required for their operation. In a sense, it is potentially misleading to characterize these processes in terms of dimensions like “cognitive” and “affective”. It is really the outcomes of these processes that can be more clearly identified in this way, and each process is capable of producing both cognitive and affective outcomes. However, given the clear differences in the level of cognitive sophistication required for their operation, it seems reasonable to use this dimension to describe these three broad classes.
Noncognitive processes. Some processes that lead to empathic outcomes require very little cognitive activity. The apparently innate tendency for newborns to cry in response to hearing others cry, which Hoffman (1984) refers to as the primary circular reaction, is one example. Another noncognitive process is motor mimicry, the tendency for observers automatically and unconsciously to imitate the target. Although early conceptions of mimicry viewed it as a somewhat deliberate strategy for "feeling into" the other (e.g., Lipps 1903), more recent approaches (e.g., Hoffman 1984; Hatfield, Cacioppo, and Rapson, 1994) have treated it as a relatively automatic, largely noncognitive process.
Simple cognitive processes. In contrast, other processes require at least a rudimentary cognitive ability on the part of the observer. Classical conditioning is an example; if an observer has previously perceived affective cues in others while experiencing that same affect (perhaps because both observer and target are simultaneously exposed to the same unpleasant stimulus), then the affective cues of targets may come to evoke that emotional state. Similar processes of comparably modest sophistication—direct association (Hoffman 1984) and labeling (Eisenberg et al. 1991)—have also been proposed.
Advanced cognitive processes. Finally, some processes require rather advanced kinds of cognitive activity. One example is what Hoffman refers to as language-mediated association, in which the observer's reaction to the target's plight is produced by activating language-based cognitive networks that trigger associations with the observer's own feelings or experiences. For example, a target who says "My manuscript has been rejected" may exhibit no obvious facial or vocal cues indicating distress, but an observer may respond empathically because her relevant memories (perhaps of an especially undiplomatic review) are activated by the target's words. Eisenberg et al.’s (1991) elaborated cognitive networks refer to a very similar process. The most advanced process, however, and the one that has received the most empirical attention, is role taking or perspective taking: the attempt by one individual to understand another by explicitly imagining the other's perspective. It is typically considered an effortful process, involving both the suppression of one's own egocentric perspective on events and the active entertaining of someone else's.
Intrapersonal Outcomes
The model’s third major construct deals withintrapersonal outcomes—the cognitive, affective and motivational responses of the observer that result from exposure to the target. These outcomes are thought to result primarily from the various processes identified at the previous stage in the model.
Cognitive outcomes. One cognitive outcome is interpersonal accuracy, the successful estimation of other people's thoughts, feelings, and characteristics; typically, such interpersonal judgments have been viewed as resulting to a considerable degree from role-taking processes (e.g., Dymond, 1950). Empathy-related processes have also been implicated in affecting the attributional judgements offered by observers for targets' behavior (e.g. Regan and Totten, 1975). More recently, perspective taking has been linked to changes in the cognitive representations that perceivers form of targets—in particular, the degree to which these representations resemble the cognitive representations of the self (Davis et al.1996).
Affective outcomes. This category consists of the emotional reactions experienced by an observer in response to the observed experiences the target, and is further subdivided into two forms: parallel and reactive outcomes. A parallel emotion may in a sense be considered the prototypical affective response: an actual reproduction in an observer of the target's feelings. This sort of emotional matching has clearly been the focus of several historical approaches (Spencer 1870; McDougall 1908). Reactive emotions, on the other hand, are defined as affective reactions to the experiences of others that differ from the observed affect. They are so named because they are empathic reactions to another's state rather than a simple reproduction of that state in the observer. One response clearly falling into this category is the feeling of compassion for others referred to variously as sympathy (Wispe 1986), empathy (Batson 1991), and empathic concern (Davis 1983); another example would be personal distress, the tendency to feel discomfort and anxiety in response to needy targets.
Motivational outcomes. A third category of intrapersonal outcomes, somewhat related to the second, are motivational states produced in the observer by empathy-related processes. For example, forgiveness is often conceptualized as a transformation of motivation toward a transgressing partner in which desires for revenge are reduced and desires for reconciliation are increased (McCullough, Worthington, and Rachal 1997). More generally, empathic processes have also been linked to increased motivation to value the other’s outcomes (Batson et al. 1995).
Interpersonal Outcomes
The final construct in the model encompassesinterpersonal outcomes, defined as behaviors directed toward a target that result from prior exposure to that target. The outcome that has attracted the most attention from empathy theorists and researchers is helping behavior; both cognitive and affective facets of empathy have long been thought to contribute to the likelihood of observers offering help to needy targets. Aggressive behavior has also been linked theoretically to empathy-related processes and dispositions, with the expectation that empathy will be negatively associated with aggressive actions. The effect of empathy on behaviors that occur within social relationships—a topic that has only recently begun to attract consistent research interest—also falls into this category.
Research Evidence Relevant to the Model
It should be apparent from this brief overview that this model aspires to incorporate most, if not all, of the social psychological research carried out in the name of empathy. Thus it is well-suited for use as an organizing device, and in this section I will selectively discuss past and contemporary research on empathy, using the organizational model as a framework. The goal will be to provide some sense of the history of research efforts in each area, but to also highlight some of the most interesting and provocative lines of research currently underway.
Antecedents
As Figure 1 suggests, there are several ways to think about the antecedents of empathic processes and outcomes. At one level, explanations may focus on the inherent human capacity for empathic responding, and therefore grapple with the issue of why such capacities would evolve in humans at all. At another level, explanations may focus on our dispositional tendencies to utilize the capacities we possess. Finally, explanations focusing on empathy as it occurs within specific situations must examine features of both the particular setting and of the individuals involved. At each level of analysis, of course, it will be important to carefully distinguish among the various kinds of cognitive and affective processes and outcomes.
Evolutionary origins of empathy. The idea that an empathic capacity may have its roots in humans’ evolutionary history has been around for some time. The earliest impetus for such thinking came from theoretical attempts to reconcile altruistic behavior with evolutionary theory. Self-sacrificing behavior, which on the surface seems incompatible with the notion that we are all engaged in a struggle for survival, has been found in many species, and has been explained by such additions to evolutionary theory as inclusive fitness (Hamilton 1964), genetic similarity theory (Rushton, Russell, and Wells 1984), and reciprocal altruism (Trivers 1971). Each of these approaches accounts for altruistic behavior by arguing that the genes contained within an individual “benefit” from behavior that increases their survival chances. This holds true even if we are talking about identical genes residing in others—especially close relatives. Thus, a genetic tendency to offer help—even costly help—to those who share our genetic makeup can be evolutionarily advantageous.
Empathy comes into this discussion because of the need for some proximate mechanism to produce altruistic behavior. It is one thing to say that genes "for" altruism produce altruistic behavior, which in turn leads to greater survivability for those genes in the population. However, the altruistic behavior is still undertaken by the individual organism and not the gene; some mechanism must exist within that individual—between the gene level and the behavioral act—to prompt the individual to act against its own short-term interest. Hoffman (1978) has made the case for empathy’s role by arguing that any mechanism responsible for producing altruism in humans must be reliable, but also flexible. That is, it should not be so automatic in operation that behavior could not be modified as a result of environmental conditions. In particular, the mechanism should allow the behavior to be affected by judgments regarding costs to the individual and benefits to the recipient(s). Thus, Hoffman argues that what must have been selected for during evolution was a biologically-based predisposition to act altruistically, but one that was still subject to control by cognitive processes. In his view, empathy, defined as a vicarious affective response to the experiences of others, meets these criteria. This analysis therefore suggests that the empathic response selected for by eons of evolutionary pressure is the sharing of negative affect.
Individual differences in empathy. Another way to think about the antecedents of empathy is to focus on individual differences in empathy-related constructs—that is, the degree to which individuals possess the ability and/or the motivation to think, feel, or act in an empathic fashion. Several approaches have been taken in attempts to measure these differences. One early and influential technique was that of Dymond (1950), who defined empathy in terms of the accurate transposition of the self into the thinking, feeling, and acting of others. Thus, her method consisted of assessing the accuracy with which observers could estimate how targets describe themselves on trait rating scales. Unfortunately, high levels of accuracy in this technique can result from several different factors, most of which have nothing to do with empathy (Cronbach 1955; Gage and Cronbach 1955). Once this was recognized, this method rapidly fell from favor.
The most widely used contemporary measure based on a cognitive definition of empathy is no doubt Hogan's (1969) empathy scale, which was developed based on a definition of empathy that emphasized the intellectual attempt to imagine another’s point of view without experiencing any affective response. The 64-item scale is made up of items drawn from other psychological instruments (e.g., the MMPI, and the CPI). Just as Hogan’s scale is the most widely used measure employing a purely cognitive definition of empathy, the Questionnaire Measure of Emotional Empathy (QMEE; Mehrabian and Epstein 1972) has been the most widely utilized instrument adopting an affective definition. The QMEE was designed explicitly to assess the chronic tendency to react emotionally to the observed experiences of others. Its 33 items assess the likelihood of experiencing such affective responses in a variety of contexts.