Towards a Clear and Fair Conceptualization of Empathy

ABSTRACT Empathy is operationalised and measured in various different ways in research. I have identified several trends in empathy research that have resulted in what I refer to as neurotypical gatekeeping of the concept of empathy. Narrow assumptions on the relationship between experiences and expressions have made the concept exclusive to those who are perceived as neurotypical. In several ways, this has biased our knowledge of empathy, especially regarding autism. This does not only invalidate autistic empathy, but also sustains a harmful and stigmatizing narrative of autism. In this paper, I expand on the neurotypical gatekeeping of empathy as a matter of epistemic injustice and argue why and how neurodiversity calls for a reconceptualization of empathy. I continue by building a proposal for a clear and fair notion of empathy. I argue that we need to settle the dispute on empathy and morality by accepting the value associated with empathy in society, and use an anti-discriminatory normative conceptualization accordingly. I propose to understand empathy as appropriately attending to experiential differences and similarities, balancing between – what I introduce as - distantism and proximism. I discuss conceptual and methodological implications of this approach to empathy, as well as its application to neurodiversity. Video Abstract Read the transcript Watch the video on Vimeo © 2023 Informa UK Limited, trading as Taylor & Francis Group


Introduction
The knowledge that we have and generate is shaped by the concepts we hold, how we understand them, and how we use them. Feminist scholarship has demonstrated how power inequalities can, through the exclusionary use and meaning of concepts, create and uphold said inequality, both in power and knowledge (Fricker 2007). This paper will explore how this is the case for the concept of empathy and the exclusion of neurodivergence -particularly autism. Crucially, empathy is not merely academic jargon, and its colloquial understanding often differs from its meaning in academic context. In society, the concept is usually associated with virtue (Morris 2019). To illustrate, when someone we know calls us unempathetic, this is typically an insult, while the contrary, calling someone empathetic, is considered a compliment. Because of this normative connotation, the consequences of exclusion surrounding the knowledge generated on the concept is even more troublesome. As such, it is of utmost importance to rethink how we conceptualize empathy in research.
A reconceptualization implies there is a conceptualization to begin with, one that needs revision. However, like with many important concepts, there is a wide variety of ways in which empathy is being used and understood right now (Bollen 2023). While there seems to be an intuition of what empathy as a phenomenon is, there is no consensus on how to exactly define it. As such, a number of different interpretations are currently at use in academia, in itself resulting in misunderstandings and confusion (Cuff et al. 2016). This alone can be a motivation to rethink empathy. A conceptual revision aimed at resolving the descriptive and normative ambiguity surrounding the concept would benefit our progress in researching the concept, especially in an interdisciplinary and collaborative context (as most recently attempted by Eklund and Meranius 2021). I will aim to do so as well, however, importantly, with the specific additional aim to correct for the injustice done to those who are currently unfairly excluded from the concept. Within the assortment of popular conceptualizations of empathy, there are some trends that, I argue, require reconsideration. Namely, some elements of these conceptualizations do not align with the moral normative connotation the concept holds in society, and additionally, some elements invite exclusive operationalisations (both will be expanded upon in more detail). These issues are by themselves troubling, but the combination of the two amplifies the concerns.
In the way empathy is being operationalised and measured, I have identified several issues to which I refer as neurotypical gatekeeping of the concept of empathy. Neurotypicality refers to the range of neurocognitive functioning that is considered 'normal', as opposed to neurodivergent, which refers to all functioning outside of this range. As of recently, neurodiversity scholars have been pointing out how such implicit norms of neurocognitive functioning shape the way we think of concepts (Chapman 2020). Exposing such assumptions and rethinking their meaning is of special importance considering concepts that have normative connotations, such as empathy. Implicit assumptions about the relationship between expression and experience are abundantly present in empathy scholarship. However, such generalizations are unfair when applied to the neurodivergent. In several ways, this has biased our knowledge of empathy, especially regarding autism. There is a dominant narrative that holds that autistic people lack empathy. However, in fact, many firstperson accounts and testimonies contradict the findings that autistic people lack empathy (see for example, Welch et al. 2020;Hens and Langenberg 2018;Smith 2009). Here are a few quotes to illustrate this: "Imagine being told you can't feel empathy, even though you feel people's emotions so much it bleeds into you" Vrana as quoted in Welch et al. (2020) He wrote that it is like there is "a sequence written on the eyes of the other person that tells my brain and my emotions exactly how they should be feeling at that point. These feelings are inevitably always what that other person happens to be feeling" (p. 54). McKean suggested that in autism, this capacity is "a rather cruel practical joke of nature" Smith (2009) on McKean (1994 While there is a vast number of such examples, both recent and from decades ago, autistic empathic experiences keep getting overlooked or even invalidated. Stenning (2020) explored the history of autistic testimonies on empathy, and how biased conceptualizations and assessment methods for empathy uphold a systemic underrepresentation of autistic empathy in scholarship on the concept to this day. I will go into what this means in practice and how this comes about in the section methodological exclusion.
This use of the concept of empathy not only invalidates autistic empathy, but also sustains a harmful and stigmatizing narrative of autism (Fletcher-Watson and Bird 2020). Because, in folk psychology, and also often in academia, being empathetic is seen as a desirable characteristic, important for being a moral agent, or even essential to being human (Decety and Cowell 2014). Noteworthy is that this normative connotation is being debated in philosophy, especially in relation to similarity bias and the role of empathy in dealing with otherness. A concept of empathy that accepts similarity bias as a property of this phenomenon, does not align with the use of empathy as an ethical concept. Whereas empathy understood as overcoming similarity bias and better attending to alterity, would. So, whether it is fair to consider empathy a virtue or not (which it is colloquially, and plays an important role in the stigmatization of autism), highly depends on how one exactly understands it.
The role of similarity bias in empathy is striking in the context of neurodiversity. It can be particularly difficult for empathy to bridge neurocognitive differences, imagining what it is like to differ on such a fundamental level as processing stimuli and other information -overcoming 'neurosimilarity bias'. This particular challenge is referred to as the double empathy problem (N. P. Chown 2014;Milton 2012). So, while autistic people may struggle to empathize with non-autistic people, the same goes vice versa. However, as I will further argue in this paper, current conceptualizations overemphasize the difficulty autistic people face, while excusing those who are not autistic for not extending empathy to the autistic (see N. Chown, Hughes, and Baker-Rogers 2020 for a striking example of this imbalance in empirical research). Paradoxically, concepts of empathy that unfairly favour neurotypicality hold back empathy from its potential as a moral concept, as it undermines the exact values that are typically associated with why empathy is related to morality in the first place (which I will expand upon in section 2).
This poses the question how to engage with the concept of empathy in research in a way that takes into consideration its normative connotation and impactful potential. In this paper I will lay the foundation for an account of empathy to do so. In section one, I will expand on why a revision of empathy is so important and urgent. The normative and conceptual ambiguities and inconsistencies surrounding empathy will be further discussed, and I will demonstrate how this, combined with exclusive methodological practices, is creating problems I recognize as epistemic injustice (which I will define and clarify there), in the form of neurotypical gatekeeping. Then, I will argue why a revised understanding of empathy that responds to these issues should be an anti-discriminatory and normative one. In section two, I develop such an account. I propose to understand empathy as appropriately attending to experiential differences and similarities. This can be understood as a balance between what I term proximism and distantism. Proximism refers to disregarding experiential differences, by projecting one's own experiences on the other, and/or appropriating the other's experiences. Contrastingly, distantism refers to the overestimation of experiential differences, overlooking what is shared between oneself and the other. I will clarify my proposal in relation to other notions of empathy by discussing its position on different dimensions of current debates on the concept of empathy, that I first introduced in the first section. Finally, I will discuss some conceptual and methodological implications of this proposal, and its application to autism.

The Need for a Revision of the Concept of Empathy
The need to revise the concept of empathy comprises two different but deeply related concerns: the ambiguity around the concept, and the exclusion of neurodivergence. I will start by sketching an overview of the conceptual differences that are present in empathy research, followed by a discussion of the ambiguous relation between empathy and morality. These discussions are necessary as the confusion is not only a problem in and of itself. Rather, an understanding of the landscape of what is meant by empathy, and why it should or shouldn't be related to morality, is needed before revising the concept. Then, I will reflect on methodologies used to measure empathy and how they not only further confuse the meaning of the concept, but are often exclusive to neurotypicality. Lastly, I will explain how I identify this exclusion as epistemic injustice.

Conceptual Ambiguity
The concept 'empathy' can refer to various different phenomena. This is in itself problematic as it makes it difficult to interpret and integrate works of different authors, specifically when working cross-disciplinarily. Fletcher-Watson and Bird (2020) called attention to how this confusion specifically impacts research on the link between autism and empathy. But more generally, the ambiguity of the concept leads to misunderstandings and holds back scientific and philosophical progress in this domain (Cuff et al. 2016). I will shortly touch upon the most salient areas of confusion to sketch out the variety of what can be meant by using the word empathy (Bollen 2023). 1 • There is disagreement on the cognitive and/or affective nature of various aspects of empathy (Aaltola 2014b;Cuff et al. 2016;Fernandez and Zahavi 2020;Smith 2009). There are multiple ways of demarcating cognitive and affective empathy, based on, for example, the experience of empathy itself or the experience one empathizes with. Empathy is sometimes understood as an emotion responding to another's emotion, or as a cognitive process of understanding the others mind, or another combination of cognitive and affective states of a 'self' and an 'other'. • Another debated aspect of empathy is the strategy we use to get insight in the other's lived experience. Specifically whether this ought to be spontaneous or deliberate (Cuff et al. 2016).
Often being conceptualized as spontaneous, automatic, or natural, this relates problematically to a similarity bias -as overcoming this bias, so empathizing across differences (race, gender, neurotypes) can require more effort. Conceptualizing empathy in this way recognizes only ingroup empathy as empathy, which does not align with the use of empathy as a something praiseworthy. This includes a neurosimilarity bias (Bollen 2023) and the asymmetry in the societal acceptance of empathic challenges from autistics to neurotypicals compared to the other way around (Milton 2012). • Related to this, the role and importance placed on the ability to read facial expressions varies.
The salience of this particular way of inferring the other's experience in both conceptualizations and operationalisations of empathy is noteworthy (for example Bons et al. 2013;Golan and Baron-Cohen 2006). As this can be more challenging for autistic people, the dominance of this strategy, and lack of openness to other mechanisms of attending to another's experience, is problematic, as will be expanded upon in the section 'methodological exclusion'. • There is disagreement on the function of empathy; should it manifest certain behaviours, motivations, or 'merely' create a connection (Cuff et al. 2016)? As touched upon in the introduction, depending on the definition of empathy one employs, the concept can fundamentally loose or gain moral relevance. For example, empathy is often understood as a psychological feature that has developed from its evolutionary benefit of enabling parentchild bonding and enjoying group protection. In this narrative, one often finds empathy to be inherently biased towards individuals belonging to the same social group, community, or family (Preston and De Waal 2002). However, if one understands empathy as actually a way to overcome such social in-group biases and differences, considering a similarity bias as an intrinsic characteristic of empathy does not make sense (and would actually be contradictive).
As argued in the introduction, descriptively specific but normatively vague conceptualizations of empathy cause friction with the normative use of the concept in society. • The role of self-other distinction is also a point in which conceptualizations of empathy tend to differ from each other (Cuff et al. 2016). And connected to this, the function of self-other distinction in the manifestation of empathy as concern for the other rather than distress in oneself (compassion fatigue) is debated (De Coster et al. 2018;Pouw et al. 2013;Senland and Higgins-D'Alessandro 2013;Smith 2009). • Lastly, there are different ways to conceptualize the importance of self-awareness and selfreflection for empathy (Robinson 2020;Robinson and Elliott 2019;Tordjman et al. 2019).
Later on in the paper in section 3, I will revisit these areas of disagreement on what empathy means and explicate how my own account of empathy relates to these debates in a manner that avoids the concerns I raise in this paper. Crucially, as I will now go on to discuss, the ambiguity of the conceptual meaning of empathy confuses the deliberation on the moral relevance and normative potential of the concept.

Empathy and Morality
While empathy is often held as being a good, beneficial, and morally important phenomenon, there is a complex debate surrounding the relationship between empathy and morality. First of all, on whether empathy is in fact of moral import, and secondly, if it is, what this relationship exactly entails (Aaltola 2014b). A well-known opponent of the moral value of empathy is Jesse Prinz. He argues that empathy is not needed, or indeed detrimental for morality (Prinz 2011). His arguments for this view include the idea that empathy supports in-group bias and narrow considerations of out-group persons. 2 His stance has attracted quite some resistance. For example, Passos-Ferreira argues empathy to be quintessential to morality in its capacity of extending care to others and escaping egocentrism (Passos-Ferreira 2015). Similarly, Masto accepts empathy to be of moral import as a sometimes necessary motivator to do the right thing (Masto 2015). Morris points out that Prinz' arguments mainly point to the pitfalls of empathy if not executed properly, which actually emphasizes the importance of extending empathy further than most of us do intuitively (Morris 2019). Prinz bases his arguments on psychological empirical findings, and understands and accepts empathy to be what is found to be the experience of the average person, while others use a more idealistic notion of the concept. As such, I find a naturalistic fallacy in Prinz' concerns against the normative power of empathy as understood as the observed average human capacity of (or lack thereof) empathy. This discussion is ongoing, but the societal impact of empathy research and its association with moral agency is notable. If one accepts the dominantly held narrative that autistic people lack empathy and if one understands empathy as essential to morality, one might come to the conclusion that autistic people are inferior moral agents. This is at odds with experiences of autistic empathy and moral agency, suggesting the possibility of autism sometimes associated with exemplary morality (Jaarsma 2013; Stenning 2020). 3 For example, a recurring theme in autistic testimonies regarding empathy and morality is a care for non-human animals and the environment, Stenning writes: The possibility of autistic concern for other species offers a chance to 'reverse' the assumption that cognitive empathy is essential to moral behaviour, and to turn the gaze towards what might be missing in 'neurotypical' morality. (2020) Further exploring autistic versus neurotypical morality is out of the scope of this paper, but the role theories of empathy play in the acknowledgement of the very possibility of autistic morality shows the significance of the need to revise the concept. Whether to include autistic people in our moral community or not has serious consequences, including exemption from both blameworthiness and praiseworthiness, as well as worthiness of being treated and respected as a moral agent (Richman and Bidshahri 2018). Aaltola (2014a) defends autistic morality on the notion that only affective empathy is of moral importance, which, according to the empathy imbalance hypothesis (Smith 2009) is well-developed or even superior in autistic people. However, the separation between affective and cognitive empathy is also conceptually unclear, with different, often implicit, approaches being used (Bollen 2023). Kennett advocates the possibility for autistic moral agency by proposing that empathy informs moral agents, not moral agency itself, in the sense that it is not a prerequisite (Kennett 2002).
To conclude, the role of empathy in morality is under debate. However, the dubious narrative that autistic people are inferior human beings because they lack empathy is, sadly, actively present in society -and it is used to support the dubious idea of prevention and 'cure' of this way of being (Bovell 2020). This showcases how while debated in academia, the concept of empathy holds normative power in society.

Methodological Exclusion
What contributes to the conceptual confusion surrounding empathy is that specifics in empathy assessment methods, the operationalisations of the concept, more clearly shape what is actually meant by empathy in empirical research, moving it further away from its moral dimension, while keeping this connotation (Bollen 2023). Crucially, in the most popular conceptualizations of empathy, having an 'appropriate response' to someone else's experience is included in the definition (Fletcher-Watson and Bird 2020). Consequently, most, if not all, methods to assess empathy are founded upon certain norms of behaviour, expression, and experiences. These norms dominate both what empathy should look like in the empathizer, how it should be expressed as well as what social stimuli should enable empathy, as will be explained in this section. This can be understood as a form of what has been introduced by the neurodiversity movement as 'neurotypical domination'. This refers to neurominorities (those whose neurocognitive functioning fall outside of what is considered normal) being marginalized and oppressed by the systematic favouring of behaviours and experiences that are considered neurotypical (the 'normal range' of neurocognitive fu(nctioning) (Chapman 2020). This can be seen in the often implicit, rarely contested, assumptions made in research operationalizations of empathy.
"Body Language 101 is the importance of mirroring your subject's posture or body language as a show of empathy and means of establishing connection. So if you have a clinical suckage at doing that very thing . . . maybe that's where some of the 'autistic people have no empathy' thing comes from." (Aspergia as quoted in Welch et al. 2020) In terms of how empathy should be expressed, social norms influence research methodology when empathy is operationalized as a specific set of responses to social and emotional stimuli. Responses that are considered to be 'appropriate' are then used as measures for empathy, such as which facial expressions ought to be made or which things ought to be said. However, appropriateness is subject to the personal, emotional, social and cultural context, which is rarely reflected upon in the context of empathy assessment in research settings (Harrison et al. 2022). As such, these methods do not allow for diversity in how to express empathy. They rather assess how one fits into a predefined behavioural norm -representing the majority or dominant group. Other ways in which norms on the expression of empathy influence research methodology concern quantitative measures such as physiological responses (heart rate, skin conductance) and neurological activity. Such measures might seem to be less objectionable in this context because of their quantitative and objective nature, but if used in the context of a value-laden concept as empathy, such physiological characteristics suddenly are awarded with normative power. For example, applying results of studies exploring the neural underpinnings of empathy and empathic differences (for example Klapwijk et al. 2016;Lassalle et al. 2018;Stroth et al. 2019) 4 to individuals and their empathic capacities, and capacities as a moral agent, oversimplifies and overlooks the complexity of contextualized social and moral behaviour in a reductionist manner.
Furthermore, norms on expression and behaviour shape stimuli used in empathy assessments. For example, one common qualitative way to measure empathy is through exercises on reading facial expressions, or interpreting scenarios, stories or movie clips. However, the diversity of expression and behaviour that exists in society is usually not represented in these scenarios. Assumptions are made on what expressive content underlies the verbal and non-verbal cues present in the stimuli -and as such, they enforce norms on the relationship between experience and expression. Giving the 'right' answer means recognizing and translating cues in a certain way, a way that represents a social majority, for example reading typical facial expressions.
These methodologies all concern what is typically referred to as 'state empathy'; assessing the response to a certain stimuli. Trait empathy assessments, by contrast, aim to evaluate empathic capacity independent of certain stimuli, for example through interviews or questionnaires. Strikingly, a recent systematic review on a variety of self-report questionnaires on empathy used in autism research found that both evidence for content validity (whether these assessments actually measure what they claim to do) and measurement invariance (whether they assess the same in autistic individuals as in a neurotypical sample) are lacking in all the most well-known trait-empathy questionnaires (Harrison et al.). Furthermore, the review concluded that the questionnaires were highly culturally specific, nonliteral, and vague. For this reason, it is unlikely that measurement invariance can be assumed when using these methods in a neurodiverse group. To top it off, the most popular empathy questionnaire -the empathy quotient -was 'validated' partially by showing that an autistic sample had a lower score on this questionnaire (Baron-Cohen and Wheelwright 2004); and now it is often used to show autistic people have diminished empathic capacities. This is circular reasoning. Potentially, these tests could be a method to test for autism, or one's skill and sensibility in adhering to social conventions in their specific cultural context. But as measures for empathy, this questionnaire fails to reliably assess what it aims to (Harrison et al.).

Neurotypical Gatekeeping of Empathy
Because of these conceptual and methodological confusions in academic research, the knowledge that is being generated on empathy is biased in favour of neurotypicality. Both who is considered to be empathetic and who deserves empathy is exclusive to specific behaviours and neurocognitive characteristics. As a result, this state of affairs can be understood as, what I call, neurotypical gatekeeping of the concept of empathy.
I recognize this phenomenon as epistemic injustice -a concept referring to the idea that we can be harmed as knowers, as introduced by Fricker (2007). This, because of the way biased understandings of empathy shape what we know of empathy, how we get to know things about empathy, and who gets to know something about empathy, as I will further argue and explain in this section. The two types epistemic injustice is typically divided into -hermeneutical and testimonial injusticeare both at stake with regards to how empathy as a concept is being understood and attributed to people, and crucially, these types reinforce each other. Hermeneutical injustice relates to the accessibility and/or inaccessibility of the concepts by which we understand the world. When these are founded upon certain privileges, the result is that some experiences/perspectives cannot be understood or reflected upon using these concepts. With regards to empathy, this applies as it is operationalized upon neurotypical norms of communication and self-expression, leaving little room for autistic people to use the concept of empathy to understand and assess their own autistic (or otherwise neurodivergent) empathic experiences. This not only harms this minority as knowers and users of the concept of empathy, but it limits everyone in our knowledge of empathy. Testimonial injustice refers to knowledge held by certain individuals/groups not being heard, respected or taken seriously because of their social identity. As argued by Stenning (2020), autistic empathic experiences are systematically being excluded from informing the academic knowledge of the concept. This is partly because of the exhaustion of a narrow and exclusive notion of empathy and corresponding methods (hermeneutic injustice), but also by invalidation and erasure of testimonial evidence of neurodivergent empathy (testimonial injustice). She offers examples where the dogmatic conviction that autism and empathy do not go together have made researchers question either the empathic experience or the autism diagnosis, because the existence of autistic empathy is omitted from the realm of possibilities. She refers to this phenomenon as 'the self-fulfilling prophecy of the neurotypical gaze on an autistic subject'. This clearly demonstrates how testimonial and hermeneutic injustice are deeply intertwined here, and reinforce each other. Stenning argues for the importance of starting to let autistic life-writing inform our understanding of empathy and its relation to morality and neurodivergence, as for example done by Welch et al. (2020). This move is also referred to as the empathic turn in relation to autism, in which, for example, technologies that allow non-speaking individuals to give their unique testimonial input as well can play a vital role (van Grunsven and Roeser 2021).

Revising Empathy
To summarize, while there is an academic debate on whether and how empathy is related to morality, in society the concept is generally valued (Morris 2019). Meanwhile, there is a complex conceptual confusion, leading to misunderstandings, incongruences and misinterpretations of research findings -stagnating academic progress in understanding of empathy. Some elements of popular conceptualizations of empathy are at odds with its normative societal use and power, and invite problematic exclusive operationalisations. These elements include spontaneity or lack of effort as a feature of empathy, which conflicts with praiseworthy efforts to overcome similarity bias; the inclusion of 'appropriate' responses as an essential part of empathy, inviting limiting behaviourism in operationalisations; and a commitment to specific mechanisms such as interpreting facial expressions, not leaving enough room for diversity in ways in which we can be empathetic. Furthermore, due to systematic methodological and epistemic exclusion, the concept is being withheld from neurominorities, a process I have referred to as neurotypical gatekeeping. As empathy is often suggested to be fundamental in our social world and at the essence of being human, it is all the more important to reflect on the injustice done to those who are being excluded a priori from the very concept of empathy.
I will argue that, to resolve these issues, a revision of empathy is needed by means of an antidiscriminatory approach. With anti-discriminatory I mean that while some people are more empathic than others, this should be evaluated only on directly relevant factors, and a concept of empathy should not invite or afford operationalisations that confuse this and are unfairly exclusive. What then should be relevant factors for what is empathy (and what is not), is the main question here. Furthermore, I argue that a concept of empathy should be explicitly normative as so align it with the normative significance of how it is used. The problems caused by inconsistency and misunderstandings surrounding empathy could be solved by simply deciding upon one account and sticking to it (conceptually, methodologically and normatively). However, I also want to respond to the undesired social impact of empathy research, in a society where this concept is valued. This comes down to a mismatch between research conceptualizations and operationalisations and how empathy is understood colloquially. To resolve this misalignment, one could adjust to the other. So, either A) academics could use a conceptualization of empathy that matches its connotation in society or B) the entire society could change the intuition that empathy is valuable, accepting an operationalization according to which empathy is morally irrelevant or even anti-moral (as, for example, argued by Prinz 2011). In principle, both options would work to resolve this mismatch (see Jorem and Löhr 2022 for a general argument of the following approach to conceptual engineering). However, the issues at hand do not only exist on an abstract and theoretical level, but concern real-life problems that require urgent attention. It is a priority to adequately respond to the consequences of this conceptual confusion and exclusion. Therefore, from a pragmatic perspective, I reason in favour of option A for this specific case. I argue that we need to settle the dispute on empathy and morality by accepting the value associated with empathy in society, and use a fair normative conceptualization accordingly, which would in turn resolve the dispute on conceptual specifics. Thus, my aim for the remainder or the paper is to sketch out a foundation for such an account.

Boundary Conditions
I will start my proposal by exploring some terms and conditions for an anti-discriminatory and normative account of empathy. As discussed in the previous section, I argue that we ought to use the concept of empathy in a normative way, that is, I argue that we need to align the concept of empathy as understood in academia with its colloquial use, to respond to the societal impact of how empathy is talked about in academia, by adopting a concept of empathy that is associated with virtue. As discussed in the previous section, conceptualizations of empathy that include in-group bias and exclusivity are not suitable in this regard. As many conceptualizations and methodological operationalisations of empathy tend to undermine the values of diversity and inclusivity, an account of empathy that justly connects it to morality, contrastingly, needs to embed these values.
While empathy can mean a variety of things, the proposal needs to be in line with the common core found in existing conceptualizations of empathy for it to be intuitive and practical. I propose that the array of definitions agree with each other on the following: empathy refers to a relational process that connects the inner lives of different individuals. This is the only statement I expect all studying empathy would agree upon (i.e. any attempt to further specify this sentence will likely point at an area of disagreement about the concept). Some understandings of empathy in the academic literature stay at this general level, others provide a detailed model clarifying which intra-and interpersonal processes are included and which are excluded in their definition of empathy. However, each mention of empathy faces the following issue: how can one access or yet catch a glimpse of someone else's inner life? Despite all conceptual ambiguities, empathy is considered to be an experience related to another's experience. On that note, it has the potential to support, or even be at the heart of, a diverse and inclusive society. Empathy could be seen as a bridge connecting our experiences, overcoming individualism. Some authors hold this to be the sole objective of empathy (rather than the objective being to support or facilitate certain moral/prosocial behaviours). Between conceptualizations that are founded upon this objective, I found an interesting dichotomy, seemingly paradoxical. On the one hand, this interpersonal bridge can be built upon identification with the other, appreciating the similarities you share (for example Komeda et al. 2015). On the other hand, it involves recognizing and appreciating the differences between you and the other (for example Jurecic 2006). These contrasting aspects are both important and valuable, if applied appropriately (Taipale 2014). If taken to an extreme, both processes are problematic. What I will now go on to argue is that it is useful to understand this as a virtue balancing between two vices. This virtue-vice structure captures the nuance and struggle that comes with balancing between two compelling but morally undesirable alternatives. 5

Understanding Empathy as the Balance Between Proximism and Distantism
I propose to understand empathy as appropriately attending to experiential similarities and differences. This definition combines the two intuitions about empathy that are both common -while seemingly opposing -as described above. Namely: 1. identifying yourself with the other, creating a bridge between two life worlds, sharing experiences, and 2. perspective taking, acknowledging that the other has a different experience from yours, and trying to make sense of the other's life world. If taken to an extreme, both of these phenomena can be harmful, which I will term distantism and proximism. Distantism and proximism refer to a person disregarding similarities (distantism) and differences (proximism) between themselves and the other, respectively. Empathy, then, can be understood as the balance between the two (Table 1).
Proximism is failing to have a proper self-other distinction, considering the experiences to be closer to each other than they are. This can happen either by placing another's experience too close to one's own, or by placing their own experience inappropriately close to the other's (or a combination of both). I refer to these inclinations as ego-projectionism and experiential appropriation respectively. Ego-projectionism refers to the tendency to believe that someone else has the same experience as you. Consider the following example. Frieda hears Jamie's favourite song, which she in fact doesn't like so much, on the radio, and turns up the volume. Jamie runs away and slams the door. Frieda feels irritated, because she thought she was doing a nice, considerate, and actually empathetic thing. She would have started dancing happily if Jamie did the same thing for her. The Empathy 'Appropriately attend to experiential differences and similarities. Balancing between proximism and distantism'. Proximism 'Mistakenly disregarding experiential differences'. Ego-projectionism 'A form of proximism. Extending your experiences to another'. Experiential appropriation 'A form of proximism. Extending another's experiences to oneself'.
volume, however, was painful for Jamie, which is why he quickly ran away to escape from the sound. Frieda is, in this case, failing to understand that Jamie experiences something different when hearing the loud music than she would. Experiential appropriation, on the other hand, can be understood as projecting the experience of the other onto oneself. This happens, for example, when internalizing the suffering of another, what is sometimes colloquially interpreted as being 'too empathetic'. Experiential appropriation has also been brought to the fore as a risk of 'empathy' in the context of racism and sexism -despite the good intentions that often underlie the attempt to empathize with a marginalized group from a position of privilege (Davis 2004). The opposite of proximism is distantism; failing to see one's similarities to the other. In an extreme form, this means dehumanizing another, disregarding the most basic level on which one could identify with the other: on being human. 6 Distantism can also, in a less extreme but still harmful form, entail reducing someone to a certain characteristic, diagnosis, or status, omitting the richness of someone's inner life world. While Frieda very much dislikes oranges, it would be inappropriate for her to feel bad for Jamie while they were eating an orange -which she would do if she assumed an orange tastes the same to them as it does to her. On the other hand, they both share what it is like to experience a nice taste, both being human tasters (of course assuming here, they both have typical taste perception) -so it would also be inappropriate to be apathic towards their pleasant sensation. If empathy is about both similarities and differences between each other's life worlds in interpersonal interaction, it would, applied to this case, be best to acknowledge their differences in the taste sensation they perceive from eating oranges, while connecting upon their similarities on what it is like to eat something that tastes good.
Empathy, I propose, then, is the careful balance between distantism and proximism. One needs to be an ego-projectionist to some extent. Having to ask everyone everything without making any assumptions that their experience might be similar to yours is highly unpractical (it is safe to assume that if you hit someone the other will feel pain, as you would). Moderated experiential appropriation is of value as well, as it helps to understand another's perspective to feel with them and place oneself in the other's shoes. While these aspects are often included in accounts of empathy (simulation, perspective taking, emotion contagion etc.), the next one is rarely acknowledged. Namely, that appropriate distantism puts some humility in the mix, knowing what you don't know about the other's experience, what you can't ever know, but what you nevertheless try to take into account.
To develop and refine this as a virtue, one needs to learn from experiences of diverse interactions; some with people who are more like you, and some with people with strongly different experiences. A virtue is understood here not as a static capacity, skill or characteristic, but as a dynamic equilibrium. To borrow a metaphor from chemistry; a virtue is like a buffer. A buffer solution manages to keep its pH stable even when a strong acid or base is added to it -a virtuous person manages to resist to fall for one of the vices the virtue lies in between, even when the situation poses a challenge by making one of the vices even more tempting. For example, when interacting with someone with whom there appear to be more substantial differences than you are used to, it might be difficult to connect on the similarities that there are (at the very least being human), while not inappropriately using projection.

Clarifying Conceptual Ambiguities
As mentioned in the beginning of the paper, conceptualizations of empathy diverge in many aspects. In the previous section I proposed to adopt a notion of empathy as appropriately attending to intersubjective experiential differences and similarities. In this section, I will expound my proposed conceptualization of empathy and its relation to other commonly highlighted dimensions of the concept of empathy, as discussed in section 1, by clarifying my position on areas on which concepts of empathy tend to deviate.

The Cognitive versus Affective Nature of Empathy
Scrutinizing the duality between affect and cognition is out of scope for this paper. I avoid this distinction with the use of 'experience', because of the following concern I have in this domain. In practice, what is considered to be an emotion or thought and how this distinction is made in research operationalisation is again, often based on a narrow neurotypical dominated idea of how to experience and express these states and their differences. For example, by reducing the experience of an emotion to it being expressed by a certain facial expression. With the use of 'experience' I aim to avoid such misunderstandings. This notion is used here to include the diverse ways to experience emotions, sensations, thoughts, beliefs and perspectives and acknowledge this variety in the manifestation of empathy.

Empathy as Spontaneous or an Effort
In my approach to empathy, the effort (or lack thereof) put in empathy is not a requirement for it to be called empathy. The effort required for empathy depends not only on the person (and to what extent this person developed the virtue of empathy), but also on the specific relationship with/to the other subject. If you don't know the other very well, or when encountering someone who differs from you more or in different ways than other interactions you've had had, more effort might be required to find this balance, and not fall into the 'traps' of ego-projectionism, experiential appropriation or distantism. These situations raise an empathic challenge, and, with this, an opportunity to strengthen the virtue. When connecting with individuals who are familiar or have a lot in common with you or with other individuals you've known (but not so much that it creates another empathic challenge: dealing with an unknown level of similarity), accommodating this balance might be effortless.

The Status of (Facial) Emotion Recognition in Empathy
While some understand facial emotion recognition as essential to empathy, or even actually what empathy is, this is one of the aspects that narrows down the concept and makes it unnecessarily exclusive. In most circumstances, the skill to recognize and identify emotions as typically expressed in facial movements or behaviour is very informative to get insight in the other's experience -as a source of information about the other's experiences. However, it is also important to recognize when to not rely on this skill -as it is attuned to behaviour and expressions of the majority in the sociocultural context one is situated in. When interacting with someone who expresses their experience in a different manner (so for example in the case of neurodivergence, but also in cross-cultural communication), this requires a different way of translating behavioural cues into the underlying expressive content -as only the latter is informative and relevant to empathy.

The Function of Empathy and the Place of Similarity Bias
In my account, the function of empathy is to appropriately deal with intersubjectivity. Not attending to both experiential similarities and differences, involves a disrespect of either aspects you share or aspects you do not, and in result disrespecting part of the others subjectivity. Consequently, my understanding of empathy does not include similarity bias as one of its characteristics. On the contrary, empathy should help one to overcome on their biases in approaching another. It might indeed be easier to appropriately attend to intersubjectivity when there is more similarity, especially if one is used to relying on projection. However, if one often fails to empathize with someone whose experiences are more different, this is a feature of the empathic ability of the person (having room for improvement) rather than a characteristic of empathy itself.

The Role of Self-Other Distinction
In various conceptualizations of empathy, having proper self-other distinction, namely being aware that the experience of the other is not the same as yours, is an essential part of empathy. In some this is used in an even more specific way, namely being aware of the causal relationship between the other's experience and yours. So, for example, recognizing that when you feel sad when the other expresses sadness, your sadness is a response to the other. The other's sadness is not yours. Both of these uses of self-other distinction play an essential role in my account, as they correspond to preventing the two aspects of proximism: ego-projectionism and experiential appropriation, respectively.

The Manifestation of Empathy as Concern
Both the manifestation of empathy as a certain emotion as well as resulting behaviour or expressions are not included into this definition of empathy. Its role in bridging between different life worlds has an intrinsic value in a diverse society, not only an instrumental one to promote certain behaviour. Furthermore, requiring empathy to manifest in a certain way would inevitably result in again an exclusive definition of the term, considering the immensity of the variation in individuals, situations and relationships. For example, while offering help might often be a desired result of empathy, one could successfully empathize with a serial killer without condoning their behaviour.

The Role of Self-Awareness in Empathy
The importance of being aware of one's own experience, and how this relates to the other, is frequently brought up in relation to one's ability to empathize. This aligns with the concept of empathy as proposed here, meaning that without having any sense of one's own experiences, one cannot relate it to the other's in terms of differences and similarities. However, self-awareness is meant here in a broad and holistic sense, not exclusively in a cognitive and/or reflective manner. The ability to label one's emotions or put experiences into words might be helpful, but not being able to do this does not imply that one does not have a sense of one's own experience. And consequently, that it prohibits one from relating that experience to another's -empathizing (similar to the role of being skilled in reading facial emotion expressions). To further improve and refine one's empathic ability, some self-reflection on one's limitations and tendencies towards proximism or distantism, is needed. But also here, it is important to be wary of limiting what is considered and recognized as reflection (for example by holding an exclusively cognitivist/intellectualist understanding of reflection).

Implications
I have argued that, to responsibly engage with the concept of empathy in research, an antidiscriminatory notion of empathy is needed. I have made a proposal to understand empathy as appropriately attending to experiential differences and similarities. This entails working on a balance between proximism (ego-projectionism or experiential appropriation) and distantism. Adopting this approach to empathy has several conceptual and methodological implications, which I will discuss here.

Conceptual Implications
It might seem like this account of empathy includes a big move away from most well-known definitions of empathy, especially ones used in cognitive and social science. 7 However, applied to practice, it will often lead to the same conclusions about 'this is empathy' and 'this is not'. This will mainly be the case when considering interactions between people who belong to the same group or share many characteristics as such. The latter is because most accounts of empathy tend to disregard the pitfalls of proximism, or even favour proximistic tendencies (for example when empathy is understood as emotion-contagion, simulation, or 'tuning in' to the other's experience). This is less of an issue in intra-community interactions. When experiential differences are relatively small, the extrapolation of experiences does not have to lead to any trouble. Though, applied to cross-neurotype interactions or other situations with relatively big experiential gaps to bridge (for example cross-cultural interactions), the same strategy appears to be, what I call, proximism. Consider, for example, someone with a perfect EQ score who fails to empathize with their autistic son. Their ego-projectionist tendency is suddenly exposed by the cross-neurotype interaction. According to many operationalisations of empathy, this person would not to be considered less empathetic -as their son falls out of the norm one is supposed to empathize with. Paradoxically, the same ego-projectionist tendency would be considered lack of empathy in the son. What would not be considered empathy by others because of atypical expression, effort required or strategies used, would be appreciated as empathy in my account.
As each interaction, context and situation comes with a unique set of experiential differences and similarities, empathy can take many forms in terms of its manifestation in behaviour, emotions, expressions etc. The proposed conceptualization of empathy acknowledges and accommodates this immense diversity. There are no requirements on output (what empathy should look like), input (what is needed to enable empathy in another) or whether it should or should not require effort. As a result, this notion of empathy is free of the exclusive, privileged and discriminatory (implicit) characteristics and effects of most other conceptualizations. This does not mean that 'anything goes'. On the contrary, the only requirement there -the balance between distantism and proximism -is a very stringent one. Considering the diversity in people, relationships, experiences and contexts, the absence of other conceptual restraints facilitates the variety of shapes and forms empathy has to take on to actually fulfil this one requirement. So, while this concept of empathy may seem wider because it is neutral about the mechanisms used to empathize, it is more precise in its normative dimension and moral demands. 8 Adopting this notion of empathy is not only of applied ethical import (considering the epistemic injustice done to those who are currently being excluded from it), but also of importance on a moral theoretical level. Unjustified narrow understandings of empathy rob the concept of its normative potential. By understanding empathy in the way I propose, and including only that as a conceptual requirement, empathy has a normative power that actually meets the positive connotation it intuitively enjoys. Appropriately attending experiential differences and similarities, empathy, is in itself of moral significance. It is intrinsically valuable, in a sense of respecting another's dignity, the uniqueness and richness of their (and one's own) experiences. Additionally, some examples of its instrumental value are improving communication in relationships, promoting prosocial behaviour that is actually appropriate to the unique person and situation, and facilitating a nonviolent diverse society/community. An important topic in the debate around the moral import of empathy concerns its relation to prosocial behavior. In my account of empathy, it supports the expression of care in the following way. Someone with proximistic tendencies may care very deeply about others, but in lacking the ability to recognize the difference in how others experience things compared to them, they may fail to properly act upon their feeling of care, in truly anticipating the other's needs. On the other hand, a distantist might also care deeply about, for example, their disabled child, while failing to respect the child as being fully human, with a deeply meaningful and interesting inner life.
Lastly, similarly to letting go of too narrow notions of what empathy looks like, we need to reconsider the strategies, mechanisms and processes that are supposed to be needed and used by someone in order to empathize. On the one hand, we need to address the limits of mechanisms often included in conceptualizations of empathy (Zahavi 2010). For example, simulation, a common way of cashing out empathy, is limited in its danger for ego-projectionism (lack of humility/awareness of what is the unknown), as well as the risk of experiential appropriation (confuse the simulated experience for one's own, and adopt it). Theorizing about the other's mind (typically considered a more 'cognitive' approach to empathy, opposed to the 'affective' nature of simulation) puts one in the risk of ego-projectionism as well, but also distantism (stereotyping, tokenization, objectification). The skill to read facial emotion expressions and other emotional cues has its limits as well. As this skill is most often attuned to the norm in the specific sociocultural context, one cannot rely on this entirely when it comes to interacting with an individual person (similar to applying a statistic to an individual case). On the other hand, we need to assess which strategies to make sense of another's experiences are currently not considered, underrepresented, neglected, or marginalized. Specifying which mechanisms and processes are considered to be useful to or even part of empathy, is, again, unnecessarily exclusive. It does not do justice to neither human diversity, nor human creativity. Moreover, it creates the false illusion that the included strategies are actually sufficient. On the contrary, we need to accept that the experience of another will never be fully accessible to us -and therefore, neither is perfectly assessing nor dealing with experiential differences and similarities. As a result, no one can ever be flawlessly empathetic. This is important to make explicit. Neurotypical empathy is currently used in a normative way in research, clinical and pedagogical practice, e.g. finding ways to make those who diverge from this norm reach this goal is a priority. Instead, we should accept this 'norm' as a mere average, while perfect (even though unreachable) empathy should be the thing to strive for. This would do more justice to those who are currently excluded from the concept, while humbling and challenging those who are currently being excused from further developing empathy, namely those who already fit into the 'norm'.

Methodological Implications
Changing how empathy is understood as a concept, should be accompanied with appropriate methodological shifts. After all, many of the problems that occur with other conceptualizations have to do with the corresponding methodologies, as discussed in the first section. Most methods that are currently being used to measure empathy are either measures for neurodivergence or social literacy (in that specific sociocultural environment). This does not undermine the value of these methods or research questions, but the conclusions drawn about empathy based on studies using these should be reconsidered. For example, those experiments demonstrating similarity bias (as discussed by Bloom 2017), point out the limited empathic capacity of the average person, rather than the limited moral capacity of empathy. Similarly, low scores on the EQ or IRI questionnaires, indicate challenges in (British) social literacy. Social literacy is attuned to the majority, as they simply make up most of the social environment. One can have challenges in this area, and benefit from therapy or other interventions to improve these skills. However, it does not have the same moral connotation as empathy, and should therefore not be confused with it. It often is, which exactly leads to the issues described before, including stigmatization or even dehumanization of neurodivergent individuals, as well as excusing those who excel in social literacy from developing actual empathy (for example, towards neurodivergent individuals). Oddly, many methods are linked to the amount of effort needed and challenges faced in social situations where empathy is required. As empathy will always be a challenge, as discussed before, absence of the experience of challenge could actually be a sign of shortcomings in self-reflection in this regard, and, actually, lack of empathy. It could also relate to living in a homogenous social environment. The concept of empathy proposed here challenges some of the research currently being done on what is called 'empathy'. Given the moral connotation and societal implication of the notion of empathy, one should be careful with using term empathy when actually studying specific skills or mechanisms.
Then, which measures are appropriate to assess empathy? As mentioned before, empathy is a relational concept, not an individual one. Another subject's experience to empathize with is essential. Only then there are experiential differences and similarities -unique to the subjects and context. As a result, empathy does not exist in isolation. That being said, some individuals have developed the virtue of empathy further than others. They do this by interacting with others, in a diverse environment, and respond to their mistakes. Due to the complexity of social interactions and the phenomenon of empathy, quantitative measures cannot suffice. They require oversimplifications that inevitably lead to, again, exclusion and bias. Engaging with a moral concept in a research setting comes with a lot of responsibility, and should be done extremely carefully, mindfully and critically. If done at all, more suitable qualitative measures designed with utmost critical reflection and creativity should be considered, including explicit discussion of all limitations of the methods and, even more important, of the results (in research dissemination). To assess empathy in a conversation, for example, one could perform a discourse analysis looking for signs of critical reflection on experiential differences and similarities, humility in this regard, and responsiveness to signs of proximism and distantism. But then, from such an experiment, one can only draw conclusions within all the limitations of the research setting, the analytic framework, and the biases, privileges, and blind spots of the research team itself.

Application to Autism Research
Empathy plays a central role in various theories of autism, diagnosis, research, interventions and societal narratives. Adopting the conceptualization of empathy defended here would have strong implications for autism research and its societal impact, as the concept of empathy is currently most often founded upon neurotypical norms and studied with exclusive measures. Most issues could be solved by using terminology more carefully: taking caution of when the term empathy is actually appropriate, and when it is not. A significant problem arises when neurotypicality is mistaken for virtue (not necessarily by the researchers themselves, but in the societal narrative) -which occurs when misusing a value-laden term like empathy. The body of research done on autism and empathy strongly indicates that there are differences between neurotypical and autistic 'empathy', namely, concerning input, output, and used strategies. However, as I argued, these things should not be included in our understanding of empathy itself. Understanding these differences can be very valuable, if used, interpreted, and framed appropriately. Next to changing the vocabulary, compensating the overrepresentation of neurotypical behaviour, experiences, and expressions, and the prescriptive use of those, would make the knowledge generated on empathy less exclusive and, in fact, more empathic itself. 9 The question arises whether there are differences between autistic people and neurotypical people in possessing and developing the virtue of empathy. 10 The use of the concept empathy in the notion of the double empathy problem is in line with the one proposed here -as a way of dealing with intersubjectivity. It states that empathy between different neurotypes is challenging -for autistic people to empathize with neurotypicals and vice versa (Milton 2012). Because autistic people are a minority, their challenges to empathize (with the majority) are more apparent. Additionally, developing and practicing 'easier' empathy (with people of the same neurotype) is made harder as well, since, as a minority, they can have less interactions with their peers (while neurotypicals do) (N. P. Chown 2014). In parallel, opportunities for development of empathy towards autistic experiences are scarce for neurotypicals -as well as encouragement to do so. Since neurotypicality is the norm, not being able to empathize with neurodivergent groups is excused -while the other way around is seen as a deficit. 11 Both should be considered equally normal or a deficit: just a person being a human who has to develop and refine a virtue. Tendencies towards proximism as well as distantism towards autism are present in society. For example, ego-projectionism underlies the expectation and demand put on neurodivergent individuals to mask -to pretend to be/to act neurotypical. Without the confusion of neurotypical egoprojectionism, masking wouldn't be needed. On the contrary, distantism occurs when individuals are infantilized, stigmatized, or even dehumanized. Rather than focusing on empathy on an individual level, research on empathy in inter-neurotype interaction, and factors that can facilitate or disturb this, is needed.

Conclusion
Adopting an anti-discriminatory notion of empathy would remedy the undesired impact of research that uses this term without attention to its societal power and its potential to exclude, discriminate, and stigmatize neurological minorities. Accepting its colloquial connotation, and building upon the intuition of connecting different subjects through identification or perspective taking, I have proposed to understand empathy as the balance between proximism and distantism -appropriately attending to experiential differences and similarities. Acknowledging the immense diversity in people, relationships, and contexts, we need a notion of how empathy that does not limit how it should be experienced, expressed, executed, and developed.