Why We Should Reject the Restrictive Isomorphic Matching Definition of Empathy

From the time of its major emergence in psychology around 1950 until relatively recently, the empathy construct has commonly been conceptualized by psychologists as a multi-faceted process of understanding another person by imaginatively “feeling oneself” into that person's subjective inner perspective (Allport, 1961; Cottrell, 1950; Dymond, 1949, 1950; Main et al., 2017; Rogers, 1959, 1975; Wispé, 1987). This “feeling into” the other person's consciousness is flexible and evolving, with the empathizer sometimes feeling the same emotion as the other person and sometimes feeling a different but complementary emotion, such as warm caring (Rogers, 1975) or curiosity (Halpern, 2003; Main et al., 2017). It is an unfolding dynamic process, not reducible to a fleeting static moment (Main & Kho, 2019; Rogers, 1975). Typically, this empathic process has been embedded in a context of other-oriented caring rather than self-oriented distress (Aspy, 1975; Batson et al., 1987, 2005; Rogers, 1975), with approach motivations outweighing avoidance motivations. The functional outcome of empathy, in this traditional view, is not merely to understand and sympathetically appreciate what another is feeling, but also the often complex reasons why that person feels that way (Aspy, 1975; Rogers, 1959, 1975). This traditional scholarly conceptualization is similar to the meaning of “empathy” as it often used in broader lay discourse, such as in President Obama’s (e.g., 2006) frequent proselytizing of it as: “the ability to put ourselves in someone else's shoes, to see the world through the eyes of those who are different from us.”

In recent decades, though, this traditional meaning of empathy has been increasingly challenged by a radically narrower redefinition. A growing cadre of scholars (e.g., Bird and Viding, 2014; Bloom, 2017b; Coll et al., 2017; Jolliffe and Farrington, 2006; Preckel et al., 2018) has adopted a circumscribed definition of empathy as restricted only to feeling the same emotion that one perceives another is feeling. In doing so, they (a) exclude approach-motivating feelings of compassion, caring, and curiosity from the empathy construct, as these emotions do not isomorphically mirror those of the person who is being perceived, and (b) reduce empathy to a momentary state rather than an unfolding process (cf. Main et al., 2017).

For example, if a father sees that his child is throwing a temper tantrum, these researchers would call his reaction “empathy” only in a moment when he responds by feeling similarly angry; if the father is calming and compassionately comforting towards his child, that would not be considered empathic responding. Similarly, a therapist would be “empathizing” with a severely panicked client if she herself becomes panicked, but not if she is responding in a reassuring, validating manner. Perhaps unsurprisingly, highly publicized recent criticisms of empathy, characterizing it as a harmful phenomenon plagued by emotional bias, innumeracy, potential burnout, and even the capacity to be exploited for hateful purposes, hinge heavily on it being defined in this narrow manner (Bloom, 2017a, 2017b).

As a representative example of this restrictive definition, de Vignemont and Singer (2006, p. 435, emphasis added) defined empathy as follows: “(i) one is in an affective state; (ii) this state is isomorphic to another person's affective state; (iii) this state is elicited by the observation or imagination of another person's affective state; (iv) one knows that the other person is the source of one's own affective state.” Similarly, Bird and Viding (2014, p. 521) stated, “for empathy to have occurred, the empathiser must be in the same affective state as the object of empathy.” This alternative conceptualization of empathy appears to be gaining in popularity (e.g., Bloom, 2017b; Jordan et al., 2016; Lamm et al., 2019; Lockwood, 2016; Preckel et al., 2018), and is now considered by some as the “standard definition” (e.g., Coll et al., 2017). It is the “unusually narrow” (Zaki, 2017, p. 59) conceptualization, for instance, promoted by Bloom (2017a) in his bestselling book Against Empathy. Hereafter we will refer to it as the RIM (“restrictive isomorphic matching”) definition.

We take issue with the RIM definition. For reasons to become apparent, we strongly side with Main et al. (2017, p. 358; also see Main and Kho, 2019), who argued that “empathy is best characterized not by a finite point in time of mutual affective experience, but rather as a dynamic process that involves cognitive and emotional discoveries about others’ experiences…. empathy involves the act of imagining what is significant from another person's perspective…and an affective match may or may not occur.” The RIM content domain, which Murphy et al. (2018) termed “empathic contagion,” is a narrow aspect of a broader, multi-faceted empathy construct. It should not, however, be considered the totality of, focus of, or absolute requirement of “empathy.” Although fleeting moments of empathic contagion may frequently occur as empathy unfolds, empathy cannot be reduced to only such moments.

Empathy, the imaginative understanding of another's subjective inner perspective, is based in an intrapersonal process, but frequently manifests as an interpersonal process as well. Intrapersonally, it is a multi-faceted unfolding process, somewhat similar to creativity, in which observation, imagination, inferential reasoning, and emotional sensing/experiencing dynamically interact to sympathetically understand the inner perspective of another. As with creativity, the intrapersonal process of empathy cannot be reduced to a narrow static moment; moreover, there is probably no uniform temporal sequence in which the various components emerge sequentially or simultaneously. Sometimes the intrapersonal aspects of empathy will be reflected in moments of emotion matching, whereas in other cases the empathy process will be moved forward effectively via moments of non-matching, such as when the empathizer feels caring for the other person or curiosity about the person's perceptions.

If the target of empathy is present and the empathizer can interact with him/her, the intrapersonal process of empathy may flow into and alongside an interpersonal process as well. Interpersonally, empathy can be seen as “an iterative process involving feedback from the other, and subsequent adjustment of one's behavior in response to such feedback” (Main et al., 2017, p. 362; also see Rogers, 1975). Interpersonally, empathy is an engaged, imaginative, curiosity-driven way of being with another person (Main et al., 2017). Perhaps prototypically, the interpersonal process of empathy flowers in conversation, with the empathizer asking questions and offering guesses as to the target's subjective perspective (Rogers, 1975; cf. Kupetz, 2020). ¹

In this article, we aim to (a) buttress the support for conceptualizing empathy as a broad unfolding process, a la Main et al. (2017), rather than a narrow state, and (b) argue against the RIM definition. Although a number of scholars have offered direct and indirect arguments against conceptualizing empathy in a narrow isomorphic manner, such as pointing out how its narrowness is inconsistent with broader scientific usage of the term (e.g., Zaki, 2017), we marshal a more extensive and focused case for rejecting the RIM definition.

In the first section, we summarize the major emergence of the empathy construct in psychology circa 1950–1980. In most traditional conceptualizations from this period, empathy is (a) an unfolding process that is (b) strongly tied to functionality. Though Main et al. (2017) made no significant reference to the intellectual history of the empathy construct prior to the 1980s, their “unfolding process” view is strikingly close to conceptualizations of empathy in academic psychology 1950–1970. This “back to the future” perspective buttresses the argument for the Main et al., perspective, illustrating the degree to which the RIM definition veers away from traditional conceptualizations.

In the second section, we describe various ways in which the RIM definition sacrifices empathy's functionality while aiming to enforce an isomorphic form. We briefly examine empathy as (a) a spontaneously elicited phenomenon, (b) a cognitively effortful phenomenon, and (c) governed by approach/avoidance in a sustained process (e.g., Cameron, 2018; Zaki, 2014). We argue that the RIM definition's focus on isomorphic form is inconsistent with functionality from all three perspectives of analysis.

In the third section, we argue that the RIM definition's two main pillars, isomorphic emotional contagion co-occurring alongside high self-other awareness, tend to be in conflict. As awareness increases that one's emotional state, such as anxiety, has been contagiously caught from another person, isomorphic matching of that other person's emotion will tend to decrease. Thus, the RIM definition appears to be a compound construct consisting of two (typically) negatively correlated domains. If both emotion-matching and self-other awareness are high, we argue that co-occurrence frequently implies an additional element motivating the empathizer to maintain this cognitively taxing and/or aversive state, such as compassion or curiosity about the other's perspective. In other words, RIM empathy implicitly requires non-isomorphic emotional content even though it explicitly seeks to exclude such content.

Back to the Future: The RIM Definition Conflicts with Empathy's Traditional Meanings

Precursors of the “empathy” construct extend across hundreds of years in intellectual history, such as discussions of “sympathy” in Smith (1759) and Hume (1740); Schopenhauer’s (1840) “mitleid”; Nietzsche’s (1881) neglected discussions of “einleid” and “mitempfindung”; Herder’s (1774) “einfühlen”; and treatments of “einfühlung” by Lipps (1903), Stein (1917), and others. Pragmatically speaking, though, “empathy” as a prominent psychological term only emerged much more recently, around 1950, due to heavy promotion by Carl Rogers, Rosalind Dymond Cartwright, and others. We speculate that one major cause of the current confusion in empathy research is that many contemporary researchers have lost sight of this critical era in the intellectual history of the construct (e.g., Carl Rogers is not discussed in Bloom [2017a]).

In translating the German term “einfühlung,” Titchener (1909) coined the term “empathy” and later preliminarily described it: “We have a natural tendency to feel ourselves into what we perceive or imagine…This tendency to feel oneself into a situation is called empathy” (Titchener, 1915, p. 198). Nonetheless, he did relatively little to raise the term after birthing it (Debes, 2015). Although empathy was sometimes mentioned in the psychological literature on aesthetics in the succeeding decades (e.g., Brown, 1915; Howes, 1913; O’Neill, 1914), namely, as a process in which individuals project their own feelings into objects and works of art, it was rarely mentioned in other domains of psychological inquiry.

On the infrequent occasions that “empathy” was referenced in psychology outside of aesthetics, it was typically referenced as a knowledge-generating method or a process by which human beings can gain superior holistic knowledge about others, not as a narrow momentary state (e.g., Allport, 1924). Among the few psychologists who mentioned the term in these early decades, “empathy” as an epistemological method came to be viewed by some as a form of unscientific intuitionism (e.g., Cattell, 1937; Sarbin, 1944). This kind of negative characterization may partly explain the minimal discussion of empathy in psychology in the first half of the 20^th century (cf. Cottrell, 1950).

As documented by Lanzoni (2018), it was not until the 1950s that the term “empathy” became common in public discourse. Empathy's popularity within academic psychology, or lack thereof, largely tracked its popularity in general usage, with only a handful of references in psychological literature prior to 1948 (Buchheimer, 1963).

Empathy's eventual emergence in psychology was strongly tied to its use in sociology. As described by Edwards (2013), ethnographic methods promoted by the “father of American ethnography,” Franz Boas, involved stepping outside one's own perspective and imaginatively participating in the perspectives of those one is studying. As Burgess and Cottrell (1939, p. 334, emphasis added), a past and a future president of the American Sociological Association, explained, “This method, if it may be called a method, is that of empathy, of sympathetically entering into the experiences and attitudes of another person through the medium of an interview or a personal document and thus, for the time, identifying oneself with the other and taking his role.” For this method of empathy to produce accurate understanding, though, the researcher must contend with the tendency to (a) project his/her/their own feelings or biases onto the other person and/or (b) become excessively emotionally submerged in the perspective of the other person (cf. Cottrell, 1941).

In the 1940s, the late Rosalind Cartwright (née Dymond, 1922–2021) was a psychology doctoral student at Cornell University. Stepping beyond the psychology department, she found a mentor in the sociologist Leonard Cottrell, and together they worked to reintroduce the empathy construct. Cottrell and Dymond (1949) lamented that essentially no research attention had been paid to empathy within academic psychology and argued that empathy was of critical importance in analyzing and treating personality disorders; interpersonal dynamics and communicative processes; and self-insight. Cottrell (1950, p. 708) suspected that the neglect of empathy stemmed from the fact that it was inherently difficult to study with the empirical tools then available, while observing that empathy needed to be adequately defined rather than left “in the realm of mysticism” or “circular reasoning.” Cottrell and Dymond (1949, p. 357) concluded that “whatever the reasons, it is our contention that research in this area can no longer be avoided.”

Cottrell and Dymond (1949) extended empathy as a dispositional trait domain, not just an academic case study method. Reporting on a series of studies that were admittedly “crude” and “none too rigorous” (p. 357), Cottrell and Dymond (1949, p. 359) suggested that people high in trait empathy “appeared to be emotionally expressive, outgoing, optimistic, warm people who had a strong interest in others,” whereas those low in empathy were “rather rigid, introverted people whose emotional life appears inhibited but who are subject at times to poorly controlled outbursts of emotional behaviour.”

In preparing her doctoral thesis on empathy, Dymond noted that few people had heard of the term and she had to define it in nearly every conversation (Lanzoni, 2018). Building on the definitions of Burgess, Cottrell, and others, and distinguishing empathy from projection, she defined it as “the imaginative transposing of oneself into the thinking, feeling and acting of another and so structuring the world as he does” (Dymond, 1949, p. 343). At least in the early years of empathy research, this definition appears to have demonstrated some degree of unifying convergence (Parker, 1955; also see Allport, 1961).

Dymond (1949) attempted to devise objective tests of some aspects of dispositional empathy. Although the validity of Dymond's empirical methods was strongly questioned (e.g., Cronbach, 1955; Lindgren and Robinson, 1953; Murstein, 1957), her work spurred further attention to the empathy construct (Buchheimer, 1963). In 1951, Dymond was hired by clinical psychologist Carl Rogers, and the two worked together on conceptualizing empathy and other processes in therapeutic practice (Lanzoni, 2018), producing, for instance, the classic Psychotherapy and Personality Change (Rogers & Dymond, 1954).

Among psychologists, at least until recent decades, Rogers has long been the figure most closely associated with the empathy construct. As Wispé (1987, p. 29) concluded, “Without doubt, the present popularity of empathy as a construct comes from Rogers's emphasis on it.” Surprisingly, although Rogerian empathy is still frequently discussed in scholarly works in psychotherapy and counseling psychology, Rogers is now only rarely referenced in contemporary empathy research articles outside of such contexts (but see Hall and Schwartz, 2019; Naor et al., 2018; Silani et al., 2013).

Much of Rogers's non-directive approach to therapy, which he began to term “client-centered therapy” in 1945, strongly resembled empathy as a process; it was not until 1948, however, that he began referring to “empathy” as a part of his approach (Lanzoni, 2018). His conceptualization of “empathy,” though, evolved substantially over the decades. At a middle stage in his explorations of the construct, for instance, Rogers (1959, pp. 210–211) defined empathy as follows:

The state of empathy, or being empathic, is to perceive the internal frame of reference of another with accuracy and with the emotional components and meanings which pertain thereto as if one were the person, but without ever losing the ‘as if’ condition. Thus it means to sense the hurt or the pleasure of another as he senses it and to perceive the causes thereof as he perceives them, but without ever losing the recognition that it is as if I were hurt or pleased or so forth.

Critically, Rogers later clarified that empathy was an unfolding “process” and that he had erred in previously referring to it as a “state” (e.g., Rogers, 1975). In the Rogerian view, empathy is dynamic and cannot be reduced to a fleeting occurrence in a moment of time.

The multi-faceted complexity of empathy as an unfolding process is exemplified in the late-stage definitional description presented by Rogers (1975, p.4, emphasis added):

The way of being with another person which is termed empathic has several facets. It means entering the private perceptual world of the other and becoming thoroughly at home in it. It involves being sensitive, moment to moment, to the changing felt meanings within this other person, to the fear or rage or tenderness or confusion or whatever, that he/she is experiencing. It means temporarily living in his/her life, moving about in it delicately without making judgments, sensing meanings of which he/she is scarcely aware, but not trying to uncover feelings of which the person is totally unaware, since this would be too threatening. It includes communicating your sensings of his/her world as you look with fresh and unfrightened eyes at elements of which the individual is fearful. It means frequently checking with him/her as to the accuracy of your sensings, and being guided by the responses you receive. You are a confident companion to the person in his/her inner world.

This Rogerian view of empathy is directed primarily at therapists, but Rogers argued that it can be re-articulated in “terms which are perfectly understandable by contemporary youth, or citizens of a beleaguered inner city” (p. 5). At the same time, Rogers noted that his conceptualization is “hardly an operational definition, suitable for use in research” (p. 4); it would need to be re-formulated for such purposes.

For Rogers, empathy was not merely accurate prediction of a person's thoughts and behaviors, but an emotional stance toward the other person, one that is sincerely interested in the other's inner life, for the other's sake rather than for one's own. Rogers argued that empathic understanding is not separate from actual caring or compassion: “It might seem that we have here stepped here into another area, and that we are no longer speaking of empathy. But this is not so. It is impossible to accurately to sense the perceptual world of another person unless you value that person and his world – unless you in some sense care” (p. 7). Moreover, Rogers (1975) indicated that the caring stance of the observer is what leads others to lower their guard and better allow the observer to enter into their private experiential worlds. In other words, in the Rogerian view of empathy, empathic caring is essentially a requirement for both the intrapersonal and interpersonal elements of an empathic process to unfold effectively.

The Rogerian view of empathy diverges from the RIM definition in three fundamental ways. First, it posits empathy as an unfolding process rather than a static phenomenon. Second, it explicitly rejects the idea that empathy encompasses only isomorphic emotion sharing. Third, it makes the empathic motivation of the empathizer critical to the process.

Largely as a result of the work of Rogers, Dymond, Cottrell, and their collaborators and interlocutors, empathy research emerged in the 1950s and 1960s, and has remained popular among psychological scientists. The proselytizing of empathy by Rogers, his students, and his colleagues was so successful that their empathic approach to psychotherapy eventually lost its distinctiveness, with empathy becoming a generally accepted and adopted aspect of many approaches to psychotherapy (Lanzoni, 2018).

At the same time that empathy was becoming a key concept in psychotherapy, it also came to be viewed as a broadly positive term among much of the general public (Lanzoni, 2018). Much as with its Rogerian meaning, its meaning in general usage has been far from circumscribed. Recent research indicates, for instance, that lay conceptions of empathy strongly encompass a heterogeneous mixture of perspective taking, prosocial emotional response (e.g., caring, compassion), and interpersonal perceptiveness (Hall et al., 2021a). Similarly, regarding lay and clinician conceptions of clinician empathy, characteristics such as “shows care and compassion” and “understands patient's feelings” are most strongly endorsed as representative of empathy, whereas emotion matching (e.g., “gets sad when the patient is sad” and “can't help feeling what the patient is feeling”) are regarded as much less representative (Hall et al., 2021b; also see Sanders et al., 2021).

How did the RIM Definition Become Ascendant?

The RIM definition is dramatically narrower than the conceptualizations of empathy presented by Rogers, Dymond, and other major early proponents of empathy research. It is also much narrower than prototypical lay understandings of the term empathy. How and why, then, has it become such a popular definition among empathy researchers in recent years?

The conceptual narrowing and popularity of the RIM empathy definition have increased in a gradual fashion over generations. A few scholars suggested this kind of definition even during Rogers's era (e.g., Feshbach, 1975), but it seems to have never gained heavy traction in those earlier decades. Somewhat similar definitions seem to have become more frequent in the 1980s and 1990s. Yet, they tended to be less restrictive than current RIM definitions, typically allowing emotional responses to be “congruent with” or “appropriate for” the target of empathy's state, not necessarily directly isomorphic to it. For example, see some of the definitions highlighted in Eisenberg & Strayer’s (1987, ps. 3–4) influential edited volume on empathy: “an affective response more appropriate to someone else's situation than to one's own” (Hoffman, 1987); “requires that an observer share the general emotional tone of another – whether or not there is direct emotional match” (Thompson, 1987); “an affective state stemming from apprehension of another's emotional state or condition and which is congruent with it” (Eisenberg & Miller, 1987); “the vicarious experiencing of an emotion that is congruent with, but not necessarily identical to the emotion of another individual” (Barnett, 1987).

In the 2000s, isomorphic matching definitions were frequently presented in more explicitly circumscribed forms (e.g., de Vignemont and Singer, 2006). At the same time as these particularly narrow RIM definitions were being put forth, a growing number of researchers began to shift towards measuring empathy in accordance with the RIM perspective. For example, since the 1970s, the overwhelming majority of studies of empathy have employed self-report questionnaires (Hall & Schwartz, 2019). Almost all the popular empathy questionnaires up until the mid-2000s presented non-isomorphic elements of emotional responding, such as compassion and caring, as key aspects of the empathy construct (e.g., Baron-Cohen and Wheelwright, 2004; Davis, 1983; Mehrabian and Epstein, 1972). Jolliffe and Farrington (2006) argued, though, that these existing empathy questionnaires did not actually measure empathy, given that they encompassed non-isomorphic elements such as caring and compassion. Their Basic Empathy Scale (BES: Joliffe and Farrington, 2006) aimed to measure the RIM definition of empathy, prioritizing isomorphic emotion matching and explicitly removing non-isomorphic elements from the empathy construct. Since then, the BES has become one of the most extensively used measures of empathy (e.g., see Jolliffe and Farrington, 2021) and additional measures have been developed to operationalize the RIM conceptualization (e.g., Jordan et al., 2016; Raine and Chen, 2018). Though non-RIM empathy questionnaires are still more widely used (Hall & Schwartz, 2019), it is possible that RIM empathy measures may eventually supplant them.

There are a number of possible contributing factors behind the rise of the RIM definition. Perhaps most obviously, the more traditional definitions adopted by Rogers, Dymond, Allport, Cottrell, and others in the 1950s-1970s are complex, abstract, multi-faceted, and difficult to operationalize, as some of these early figures themselves readily admitted. Furthermore, confusion regarding the empathy construct may also have been exacerbated by the fact that empathy was initially conceptualized and promoted as a professional method in sociology and psychology (Cottrell, 1941, Rogers, 1959), but ended up being translated into more basic psychological domains, both within academic research and popular discourse. Given the many “moving parts” at play in empathy as a method, it should come as little surprise that it emerges in observational or trait conceptualizations as confusingly heterogeneous and challenging to operationalize. ²

The drive to refine the empathy construct to operationalize it in empirical research is surely a contributing factor to the rise of the RIM definition. In comparison with the opacity of more classical definitions, the RIM definition of empathy is, at least facially, clear and easier to measure or manipulate: “identify the emotion → feel the same emotion.”

In addition to its facial clarity, though, we suspect that the RIM definition's growing popularity in recent decades may also have been fueled by the rise of functional magnetic resonance imaging (fMRI) methods and interest in the scientifically controversial (Hickok, 2014) phenomenon of mirror neurons (“In the last two decades, empathy research has come to be dominated by neuroscience-oriented researchers,” Eklund & Meranius, 2021, p. 301). As suggested by various scholars (e.g., Debes, 2015; Edwards, 2013; Lanzoni, 2018), the fate of the empathy construct has been pulled in different directions depending on which methodologies and movements were in vogue at the time, as well as depending upon the perceived limitations of different measurement methods.

Compared with its predecessors, the RIM definition, static and restricted rather than unfolding and highly complex, is easier to operationalize in a neuroimaging paradigm. Two seminal studies (Decety & Jackson, 2004; Singer et al., 2004) demonstrated that watching another experience pain activated much of the same “pain matrix” that is activated by experiencing pain oneself. Since then, hundreds of fMRI studies have been conducted on this kind of “empathy,” primarily “pain empathy” (see meta-analysis by Timmers et al., 2018). Under an “unfolding process” definition of empathy, these studies would, at best, only be capturing a modest or minor sliver of “empathy.” Under the RIM definition, though, these fMRI studies can claim to be capturing the core of empathy. As neuroscientific investigation of empathy has blossomed, the RIM definition's amenability to such methods may have subtly increased its popularity (cf. Zaki and Oschner, 2012, for other conceptual concerns about the growth of the neuroscience of empathy).

The RIM definition offers simplicity and clarity, but deviates dramatically from traditional scholarly, clinical, and lay meanings of the term “empathy.” In so doing, the RIM definition not only creates substantial risks of communicative confusion, but also, as we will explain further, sets “empathy” up to be lampooned or characterized as fundamentally harmful. However empathy should be defined, it should be understood as a broader unfolding process (Barrett-Lennard, 1981; Main et al., 2017). Momentary states of isomorphic matching may be some of the primitive building blocks of this process, but empathy is much more than only such moments. ³ For an analogy/homology, just as “Eureka” moments may be part of the creativity construct, creativity cannot be reduced to or even adequately encapsulated in such moments by themselves.

The RIM Definition Unnecessarily Sacrifices Function for Emphasis on Form

Main et al. (2017, p. 364) argued that recent empathy research has been hindered by “considerable emphasis on the form of empathy at the expense of its function….We believe that a definition of empathy based on its function is important to move the field forward…” In offering their “unfolding process” view of empathy, in which isomorphic matching is not required, Main et al., proposed that their definition has the flexibility to encompass empathy as a functionally valuable interpersonal process, across varying contexts. In supporting their perspective, we offer a more extensive, direct line of argumentation that the RIM definition conflicts with the functionality of empathy as a process.

As Wittgenstein (1953, §43) argued, terms can usually be understood as “tools” and the meaning of a term is typically substantially revealed in how it is used to serve a particular purpose or collection of purposes. Batson (2009), summarizing the field, contended that there are two main questions motivating empathy research: (1) how do human beings come to accurately perceive the subjective perspectives of others?; and (2) what motivates human beings to care about others, and subsequently behave in a caring or helpful manner toward them (especially in regard to altruistic motivation)? Many researchers have aimed to use the empathy construct to deal with both issues simultaneously (Baron-Cohen & Wheelwright, 2004; Hogan, 1969). In particular, in clinical practice, empathy has long been discussed as a way of both coming to understand a client's inner perspective and also as a way to build a validating, caring relationship with the client (Rogers, 1975; also see Halpern, 2014, in the broader medical context).

If empathy is restricted only to isomorphic contagion, then it loses much of its value related to all of these different functions. In many cases, the RIM empathy definition even veers into absurdity. As noted by Hall and Schwartz (2019), would we call it “empathy” if a person feels murderous rage after observing a person feeling murderous rage? Or, as Nietzsche (1881) asked, what would we call it if we encountered a self-hating person and, in isomorphic matching, hated him too? If a therapy client comes to an appointment in a delusional euphoric state, would a therapist be empathizing only if he/she experienced euphoria too? These reductio ad absurdum examples point to a fundamental problem with the restrictiveness of the RIM definition.

RIM is Inconsistent with Empathy's Functionality as a Spontaneously Elicited Phenomenon

In many cases, empathy may emerge spontaneously upon observing the emotional state of another, such as seeing a person crying. In such cases, RIM empathy theorists frequently conceptualize empathy as building on more automatic physical mimicry and emotional contagion processes, leading to isomorphic emotion matching (Bird & Viding, 2014; Hatfield et al., 2009; Lockwood, 2016). In this view, empathy emerges “mostly automatically and without conscious awareness,” and “in most cases, mimicry or emotional contagion precedes empathy…” (Singer & Lamm, 2009, p. 88, 82).

Similar to Main et al.’s (2017) functional perspective on empathy, though, social-functional accounts of emotions more broadly have emphasized how the rapid interpersonal influence of emotions between individuals can be better understood in terms of social functionality rather than form (Keltner & Haidt, 1999; Keltner & Kring, 1998). In this view, emotions in interpersonal contexts are “adaptations or solutions to specific problems related to the formation and maintenance of social relationships” (Keltner & Kring, 1998, p. 321).

Keltner and Kring (1998) noted that, although research attention had been heavily directed toward instances in which rapid emotional responses are isomorphic to those of another person, for many emotions the spontaneous functional response will tend to be complementarily different rather than isomorphic (e.g., feeling fear in rapid response to seeing another's anger). The functionality of an isomorphic versus non-isomorphic spontaneous response depends on (a) the specific emotion perceived in the other and (b) the functionally valuable response in a specific context. For example, if a loved one is angry one might spontaneously experience guilt, whereas if a stranger is angry one might spontaneously experience fear. Although seeing an angry look on the face of a stranger or a loved one might spontaneously generate a flicker of anger in the observer as well, such isomorphic contagion will be far less consciously experienced (and far less reliably observable or reportable) than more socially functional emotions (e.g., guilt, fear) in those contexts. In other words, even if there is a mechanism that generates isomorphic matching, the degree to which it dictates rapid, spontaneous emotional matching may often be far weaker than the rapid non-matching dictates of social functionality.

Similarly, to the extent that empathy emerges in rapid, spontaneous interpersonal emotion influence, its functionality may sometimes be served by non-matching emotions and sometimes by matching emotions between empathizer and target. This functionality depends on the particular emotion perceived in the target and the particular context.

Consider spontaneous parental responses to their upset children. Whereas securely attached mothers exhibit approach or reward-related brain activation in response to their infant's sadness, insecurely attached mothers exhibit more aversion-related brain activation (Strathearn et al., 2009; see also Riem et al., 2011). Warm, approach-related parental responding to one's infant crying is critical to the infant's survival, whereas negative parental emotion may lead to anger and even child abuse or neglect (Soltis, 2004). The RIM definition, however, would seem to argue that researchers (e.g., Riem et al., 2011) are wrong in designating the warm, former response as “empathic” compared with the latter, aversive response. In a functional conceptualization of empathy, though, a parent immediately feeling warm caring in response to a baby's tears, rather than bursting into tears himself or herself, is better situated to understand (a) why the infant is crying and (b) what can be done to help.

RIM is Inconsistent with Empathy's Functionality as a Cognitively Effortful Phenomenon

Many proponents of the RIM definition extend empathy beyond what is directly observed, such as to what one imagines another might feel, even if that person is not present (e.g., de Vignemont and Singer, 2006; Lockwood, 2016). Bloom (2017b, p. 25), for instance, in defining empathy as isomorphic matching, noted, “Unlike in emotion contagion, this person does not have to be present, or even exist – we can have this feeling toward fictional characters.”

If empathy can be experienced toward individuals who are not present to be observed, such as if you are at home thinking about a colleague who recently lost her husband, then it is difficult to see how mimicry, mirror neurons, and the like could play a role defining enough to justify the inflexibility of the isomorphic matching requirement. Moreover, even in many cases where an individual is present to be observed, the target may be suppressing emotion or camouflaging emotion, such as a grieving widow who is smiling to try to keep her children upbeat (cf. Main et al., 2017). In both such cases, empathy will emerge, not automatically from mimicry or contagion, but from a more cognitively effortful imaginative process.

Even when empathy may appear to emerge from a spontaneous, automatic response to the transparently observable emotions of another, non-automatic cognitive reflection will often quickly emerge to play a role, perhaps the dominant role (cf. the appraisal theory of empathy: Wondra and Ellsworth, 2015; also see Coplan, 2011). This idea has a long pedigree in the intellectual history of the empathy construct. A number of the RIM proponents, for instance, have highlighted Adam Smith’s (1759) description of “sympathy” ⁴ as similar to their meaning of empathy (e.g., Bloom, 2017a; Jordan et al., 2016; Singer and Lamm, 2009). In passages that are infrequently quoted or referenced, however, Smith (1759/2009, p. 16, emphasis added) argued that our “fellow feeling” is more strongly tied to how one appraises the situation of the other person than to immediate mimicry or contagion:

If the very appearances of grief and joy inspire us with some degree of the like emotions, it is because they suggest to us the general idea of some good or bad fortune that has befallen the person in whom we observe them…The first question which we ask is, What has befallen you? Til this be answered…yet our fellow-feeling is not very considerable. Sympathy, therefore, does not arise so much from the view of the passion, as from that of the situation which excites it.

In Smith's model, the mere observation of someone in a particular state, such as joy or fear, only weakly generates “fellow feeling” emotions in us; it is the subsequent cognitive appraisal of the person's situation that more powerfully determines our emotional state. As a result, Smith (1759) argued that, in many cases, our “fellow-feelings” will be non-isomorphic to the feelings of the other. Furthermore, Smith pointed out that, although some emotions witnessed in others will tend to spontaneously elicit similar emotions in us, such as happiness and sadness, others, such as anger, will only tend to elicit similar emotions in us, of any degree, if we subsequently appraise the situation to be worthy of anger.

We neither endorse nor reject appraisal-based theories of empathy (e.g., Smith, 1759; Wondra and Ellsworth, 2015). Rather, we reference them to make a more general point: effortful cognitive processes play a substantial role in empathy, even when one can directly observe the transparently displayed emotion of the other, and these processes will often generate non-isomorphic emotional states in the empathizer when such non-matching emotions are more functional. When empathy is conceived in terms of such “top-down” processes, rather than primarily emerging from mimicry and contagion, there appears to be little justification for sharply delineating isomorphic from non-isomorphic responses in the empathy construct (cf. Wondra and Ellsworth, 2015). As a result, elevating form over functionality in defining empathy appears even less defensible.

Although much of their focus is often on automatic bottom-up processes like mimicry and contagion, RIM proponents readily acknowledge that more effortful top-down processes play a role in empathy (Bird & Viding, 2014; de Vignemont & Singer, 2006; Singer & Lamm, 2009). These top-down processes (a) distinguish RIM empathy from mere emotion contagion, via self-other awareness, and (b) allow empathy to imaginatively extend beyond situations in which a person is present to be observed. Moreover, these top-down processes allow empathy to be flexibly regulated, such as by terminating RIM empathy and moving to a different emotional state such as compassion (e.g., Eisenberg and Eggum, 2009; Singer and Lamm, 2009). In other words, top-down processes decrease the vulnerability of RIM empathy to being lampooned through reductio ad absurdum.

It seems to us, though, that adhering to the RIM empathy definition while also pointing to top-down processes as a way to extend and regulate empathy borders on a “having one's cake and eating it too” situation. Mimicry and emotional contagion are the linchpins of thinking of empathy as restricted to isomorphic matching; as top-down processes assume a greater role, mimicry and emotional contagion become less central. Put differently, the more one endeavors to distinguish empathy from mere emotional contagion and to buttress it with cognitive regulation, the less justification there is for retaining the isomorphic definitional requirement. Instead, from this cognitively-regulated perspective, empathy should be defined more in terms of function than in terms of a restrictive isomorphic-matching form.

RIM Undermines Functionality in Approach/Avoidance of Empathy as an Unfolding Process

In the “unfolding process” view of empathy promoted by Main et al. (2017), which generally aligns with the earlier traditional conceptualizations of empathy promoted by Rogers, Dymond, and others, empathy cannot be reduced to a momentary state, whether of rapid, spontaneous emotion sharing or of more effortful cognitive appraisal. At an intrapersonal level, empathy involves a dynamic interplay of imagination, inference, emotion, and so on, evolving over time. At an interpersonal level, empathy prototypically involves back-and-forth communication or other interactions between the empathizer and the target. In both cases, approach (e.g., “I care about this person's wellbeing”) and avoidance motivations (e.g., “emotional conversations stress me out”) play a major role in whether the process quickly terminates or, instead, continues fruitfully (cf. Weisz and Zaki, 2018, p. 70; also see Cameron, 2018; Ferguson et al., 2020; Zaki, 2014).

Especially when empathizing with someone in emotional distress or another unpleasant emotional state (whether in an in-person dialogue or in one's imagination), the ability to functionally transition between isomorphic-matching and non-matching affective states enables the empathizer to reduce avoidance motivations and support approach motivations, permitting the empathic process to unfold in a functionally beneficial manner. Limiting empathy to only isomorphic matching, in contrast, would seem to imply that empathy for the distress of others must quickly drop away if the encounter is to produce good and understanding for anyone.

Paralleling the process model of emotional regulation described by Gross (1998), empathy can be approached or avoided both antecedent to a full emotional empathic response as well as subsequent to the emergence of that response (also cf. Gross and John, 2003). Zaki (2014; also see Cameron, 2018) provided an excellent account of the various stages at which empathy can be approached or avoided. Individuals can choose, in many cases, whether they will enter into empathic encounters (e.g., talk to the grieving co-worker or not). When individuals enter into an empathic encounter, whether voluntarily or involuntarily, they can choose to change it into a different kind of encounter (e.g., change the subject, crack jokes) or divert attention away from its salience (e.g., “zone out” or start thinking about a grocery list). Similarly, an individual can engage in cognitive reappraisal to diminish the empathic salience of the encounter, such as placing blame on people for their own troubles or focusing on the “bright side” (e.g., “your spouse might have run off, but at least you still have the kids!”). Alternatively, individuals can choose to suppress their emotional response to the empathic encounter. In sum, there are myriad ways in which individuals can approach or avoid empathy (Cameron, 2018; Zaki, 2014), much as with any other kind of emotional process (Gross & John, 2003).

Proponents of the RIM definition have tended to push issues of approach/avoidance out of the empathy construct almost entirely, treating them as merely “outcomes” (e.g., personal distress, sympathy, compassion) of empathic states rather than as fundamental aspects of empathy itself (e.g., Christov-Moore and Iacoboni, 2014; Eisenberg and Eggum, 2009; Marsh, 2018; Singer and Klimecki, 2014). In other words, in the RIM conceptualization, empathy is just the initial transient phenomenon of isomorphic emotion matching, which then transitions into some other non-empathic phenomenon, whether adaptive or maladaptive. In this regard, the RIM definition further reduces “empathy” to a momentary state rather than an unfolding process.

In the context of empathy for another's distress or pain, transitioning away from an isomorphic-matching state is necessary for other-oriented caring (approach motivation) to fruitfully move the empathic process forward, whereas remaining locked in isomorphic matching of the other person's emotional plight is associated with self-oriented distress (avoidance motivation) in the perceiver (cf. Batson, 2009). In promoting the RIM definition, Preckel et al. (2018), for instance, described compassion and empathic distress as potential outcomes only of empathy: compassion, characterized by helpful non-isomorphic emotional responses; and empathic distress, characterized by detrimental isomorphic emotion matching.

In other words, if the RIM definition is accepted, it will tend to cripple empathy's functional value in an unfolding process setting. For example, if emotion-matching is required for empathy, it may tend to remove empathy as a major functional component of therapist behavior. From its earliest prominent uses as a concept in therapeutic practice, empathy has often been associated with refraining from experiencing the same emotions as a client (e.g., “the counselor is perceiving the hates and hopes and fears of the client through immersion in an empathic process, but without himself, as counselor, experiencing those hates and hopes and fears,” Rogers, 1949, p. 86), or at least strongly regulating such matching so that the therapist is capable of remaining fully present and attentive to the client's inner life rather than being swept away in his/her own self-oriented distress. Similar problems would seem to prevent the RIM definition from being held in any other kind of “unfolding process” view of empathy, such as in parent-child, romantic partner, or ethnographic interview contexts: heightened approach and reduced avoidance motivations, as well as other aspects of empathy as a process (e.g., calm thinking, focused attention), require that the empathizer flexibly shift among different emotional states rather than only isomorphically mirroring the target person.

The RIM's Two “Frenemy” Pillars: Isomorphic Matching and Self-Other Awareness

In addition to the requirement that the empathizer's emotion must isomorphically match the emotion of the other person, RIM proponents generally add a second major stipulation in defining empathy: the empathizer must be aware that “the other person is the source of one's own affective state” (de Vignemont & Singer, 2006, p. 435). Emotions must be isomorphically shared “without losing sight of whose feelings belong to whom” (Decety & Meyer, 2008, p. 1053)

This stipulation is not unique to the RIM definition (see Cuff et al., 2016). From the earliest decades of empathy's discussion as a method in sociology and clinical psychology, this kind of self-other awareness has been important because it buttresses the adaptive functioning of the empathic process. It allows empathy to be a form of controlled emotional investigation. Self-other awareness is what prevents a clinician from “getting lost” in the world of the other, such as being overcome with discomfort when working with a distressed client (Rogers, 1959, 1975). It also clarifies that empathy is not projecting one’s own feelings onto the other person (Cottrell, 1941; Dymond, 1949). In other words, at least in the more traditional conceptualizations of empathy, self-other awareness assists the empathizer (a) in being a calm and effectively helpful companion and (b) facilitating the accurate interpretation of the inner experience of the other person. Whether as a clinician, a sociologist, or a parent, being mindful of the “otherness” of the inner life of another is key to unlocking empathy's value.

The self-other awareness requirement is critical for proponents of the RIM definition because it is the primary concept distinguishing RIM empathy from mere emotional contagion, a phenomenon present across many animal species. For example, Singer and Klimecki (2014, p. R875, also see Decety and Lamm, 2006; Eisenberg and Eggum, 2009) noted that “if this self–other distinction is not present, we speak of emotion contagion, a precursor of empathy that is already present in babies.”

Self-other awareness and emotional contagion are not merely different, but often opposing, phenomena. Although solid data on this issue are lacking, it is highly plausible that isomorphic emotion-matching and self-other awareness tend to be negatively correlated: as one's awareness of the self-other distinction increases, the isomorphism of his or her emotional response to another's emotion will tend to decrease. First, from a basic phenomenological perspective, experiencing an emotion as explicitly belonging to someone else will inherently tend to differ from experiencing an emotion as belonging to oneself (Zahavi & Rochat, 2015, p. 544).

Second, self-other awareness will tend to shift isomorphic matching from an automatic phenomenon to a more controlled, motivated one. Intentionally sharing the emotional experience of another, however, is generally perceived as cognitively taxing and difficult to do effectively, even when the target's emotion is positive in valence (Cameron et al., 2019; Ferguson et al., 2020). As a result, individuals tend to avoid intentionally feeling the emotions of others (Cameron et al., 2019), unless they are particularly motivated to do so (Ferguson et al., 2020).

The motivated tendency to avoid matching the emotion of another is especially likely when confronted with another's distress (Klein & Hodges, 2001; Zaki, 2014). When one encounters a person in distress, whether an anguished coronavirus patient in a hospital bed or a stressed-out spouse breaking into tears, sharing in that person's emotions will tend to be aversive and uncomfortable. High awareness that the other's suffering is not one's own, however, increases one's ability to transition to a less aversive non-matching state. Combining emotional contagion and self-other awareness into a single construct and calling it “empathy” does not make the two conceptual enemies automatically become friends.

For the two requirements to cohere, an additional component is required to pull them together. When self-other awareness and isomorphic emotion-matching are both high, that configuration is probably either innately pleasurable to the empathizer (e.g., sharing someone's joy), or the person is motivated to voluntarily maintain an innately aversive or taxing emotional state (e.g., sharing someone's pain because you care about the person), or the person is intrinsically curious about the other's subjective emotional experience. In a therapy context, if a therapist's self-other awareness is high, then, if the therapist does isomorphically match the aversive emotion of the client, it will tend to be a motivated choice, perhaps serving a goal of building connection with and/or better understanding the client. Much the same is true of any empathizer who encounters a person suffering and then, with full self-other awareness, chooses to approach the distress and feel it oneself, rather than avoid it.

Hints that the RIM definition may implicitly require a motivational component can be found in elaborations of empathy in RIM definitional contexts. Bird and Viding (2014), adhering to the RIM definition, postulated that the distinction between empathy and emotional contagion is that the former draws on a set of processes they refer to as the “Self-Other Switch” to tag the empathizer's affective state as appropriate for the target person rather than for himself or herself. They further stated:

Switching the SELF-OTHER SWITCH to process the other is likely to be caused either by a conscious volitional decision, or by a motivational goal/task set which requires other-focussed processing. In the case of empathy, the goal may be to gain insight to the experience of other…” (Bird & Viding, 2014, p. 526, emphasis added).

If self-other awareness and isomorphic emotion matching tend to be opposing elements, this would render the RIM definition operationally incoherent in many, but not all, contexts unless another implicit component is elaborated: motivation. The motivation to choose to share the emotion of another, even though one is fully aware that the emotion is the other person's rather than his or her own, could take any number of forms. It could be the emotion of curiosity in some cases (Halpern, 2014; Main & Kho, 2019). It could be a desire for relatedness with another person, hoping to bask in the glow of another's success, or sadistically wanting to bask in the agony of another's tragedy. In many cases, it will be an other-oriented caring for another person, the kind of other-oriented motivation that clinicians and laypersons typically imply when using the term “empathy” (see Hall et al., 2021a; 2021b) and which has been widely measured in empathy research (see Zaki, 2017). Whatever the nature of the motivation for an empathizer in any particular case, it will likely entail emotional content that is not isomorphic to the emotion of the other person. As a result, the RIM definition seems to implicitly require some kind of non-isomorphic emotional content, even though it explicitly seeks to exclude such content.

Discussion

The “hard core” of the empathy construct is the unfolding process of imaginatively experiencing the subjective consciousness of another person, sensing, understanding, and structuring the world as if one were that person (Main et al., 2017; Dymond, 1949). In this way of coming to “know” another person, the empathizer experientially senses the subjective experiences of another, the qualia (Nagel, 1974) of the other person, to some limited and highly fallible degree, in the “mind's eye” (Dekeyser et al., 2009). As Halpern (2003, p. 673) explained, “empathy is an experiential way of grasping another's emotional states, a ‘perceptual’ activity that operates alongside logical inquiry.” Although one can accurately recognize and predict another person's feelings and behaviors in a purely abstract, propositional manner, much like one can know a car engine or a statistics processor (cf. the “theory theory” of our understanding of others; e.g., Carey, 1985; Gopnik and Wellman, 1992, 2012), empathy involves “knowing” a person's subjective experience more “feelingly,” as a consciously experiencing personality rather than just an object in a world of things. As many scholars have argued, this process also requires maintaining some degree of awareness of the “otherness” of the other's consciousness and modulating the degree to which one unknowingly attributes one's own feelings/perspectives to the other's consciousness.

Empirical measurement of this core of empathy, which manifests within the subjective consciousness of a person and is extremely difficult to distinguish from other phenomena, such as projection and emotional contagion, has thus far presented a stubbornly challenging problem. As a result, most studies of empathy have focused on measuring the more observable or self-reportable theorized close correlates, responsiveness in both feelings and behaviors, of this core construct. To some extent, these aspects of empathy-related responsiveness can be conceptualized as indications of empathic approach. As such, they can also be viewed as important aspects of empathy as an unfolding process rather than as merely a flickering state phenomenon.

Of the varieties of potential non-isomorphic emotional responses to another person, warm caring or other-oriented concern (i.e., compassion) are most commonly associated with traditional scholarly as well as lay conceptualizations of empathy. The arguments for considering emotional caring for others as closely aligned with core empathy can be briefly summarized. First, imaginatively stepping into another's person consciousness may often, but not always, tend to “humanize” the other person, increasing one's valuing of the feelings of that person (Wang et al., 2019). In other words, empathic processing makes the feelings of others more experientially “real” and intelligible to the observer, rather than remaining distantly abstract and alien, allowing them to be more readily emotionally valued by the perceiver.

Second, in the opposite causal direction, valuing the feelings of others may be the most common motivator for empathy. A self-centered, solipsistic person may be less motivated to contemplate the feelings of others, much less “feel” himself/herself into them, except perhaps in situations when doing so has a strategic self-seeking purpose. A more generous, caring person, however, may tend to be more motivated to approach empathic opportunities, expend effort in imagining the perspectives of others, and de-prioritize his or her own feelings and judgments in considering situations (Rogers, 1975; also see Clark et al., 1987).

The RIM definition, which excludes caring and other non-isomorphic emotional responses from the empathy construct, can seem appealing. Whereas more traditional, widespread conceptualizations of empathy are multi-faceted, with fuzzy boundaries, the RIM definition tidies the confusion up into a simple statement of “identify the emotion → feel the same emotion.” Moreover, if empathy can be reduced to momentarily sharing the same emotional state of another person, rather than understood as a dynamically unfolding process, then brain imaging and other psychophysiological techniques can more easily be said to “capture” the core of empathic experiencing.

Yet, the RIM definition's simplicity comes at great cost. It reduces the unfolding, flexible process of empathy to a momentary, inflexible state. It unmoors “empathy” from its widespread meaning/use (Hall et al., 2021a; Zaki, 2017), divorcing it from thousands of studies extending back to the 1950s as well as from generations of clinical practice. Restricting “empathy” to only isomorphic emotion-matching lacks psychological justification and sets it up to be lampooned (cf. Nietzsche, 1881) or characterized as fundamentally harmful (Bloom, 2017a). Finally, adding the self-other awareness requirement to the isomorphic matching requirement creates an unwieldy compound of negatively associated dimensions, which can only be held together by non-isomorphic emotional content that motivates a person to isomorphically share another's emotional state.

The RIM definition is not tenable. The domain of empathic contagion, while of substantial value, should not be considered as the totality of, focus of, or requirement of empathy.

We propose that, moving forward, the broad construct of empathy be understood as a combination of: (1) core empathy, the complex unfolding process of imaginatively experiencing the subjective consciousness of another person, sensing, understanding, and structuring the world as if one were that person; and (2) the broader context of interpersonal responsiveness within which this process is typically embedded. This loose framework can be understood as proto-Rogerian in form. Core empathy represents the central, but extremely difficult to operationalize, meaning of the construct, across many domains (see Wispé, 1987). Responsiveness represents the broader emotional stance that is commonly associated with “empathy” both in clinical practice and in layperson usage. ⁵

That said, consistent with the concerns of many who have adopted the RIM definition, aspects of empathy-related responsiveness should not be understood as “empathy” by themselves. Rather, they are the responsive qualities that motivate empathic approach, allow empathy to unfold, and cement core empathy into positive interpersonal relationships. They should be referred to as “empathy-facilitating” constructs rather than “empathy” in isolation (cf. similar recommendations by others: Decety and Cowell, 2014; Hall and Schwartz, 2019). Nonetheless, sharply demarcating these manifestations of responsiveness as outside of the empathy construct goes much too far, essentially throwing the conceptual baby out with the terminological bathwater (cf. Zaki, 2017). The RIM definition should be abandoned, and empathy research should firmly maintain a substantial degree of definitional compatibility with its academic roots and the widespread usage of “empathy” by both psychologists and laypersons.