Autism as a Disorder of Affective Empathy Open Access

Empathy is a basis for social interaction as it enables an understanding of other persons [1]. It is considered a core feature for intersubjectivity, cooperation, and prosocial behavior. Through empathic understanding, the mind of another person and his current experience are given in direct perception, including emotions, intentions, and attitudes [2], namely because we always perceive the other’s body as a lived, expressive body [3]. This is based on the phenomenological assumption that we are psychophysical beings as an expressive unity (Ausdruckseinheit) [4] rather than private minds hidden in object bodies. As the experiences of other persons express themselves in many ways, empathy can also take many forms [5]. Nevertheless, there is still a long way to go before there is a consensus on the nature of this capacity, especially in terms of distinguishing it from similar concepts such as emotional contagion, empathic concern, or sympathy. Accordingly, two dimensions of empathy are mostly distinguished: cognitive and affective empathy.

Since autism spectrum disorder (ASD) presupposes impairments in reciprocal social interactions and patterns of communication, it is reasonable to suspect a disorder of empathy. According to Uta Frith, a lack of empathy was identified as “the most general description of social impairment in autism” [6]. While reduced empathy is widely acknowledged, it has not yet been clarified whether affective empathy, cognitive empathy, or even both dimensions are affected. This problem is due to the complexity of the phenomenon itself [7]. Possible reasons for this knowledge gap are, on the one hand, the multilayered and multidimensional structure of empathy, and on the other hand, its contextuality [8]. Crucial is how both dimensions are defined. In the present paper, cognitive empathy is defined as the ability to explicitly understand the thoughts, intentions, and feelings of other people using inferential or imaginative processes, which include processes such as taking the perspective or role of the other. In contrast, affective empathy is defined as experiencing an emotional state that consists in feeling another person’s emotions. Central to understanding the individual dimensions of empathy is that they influence each other. Empathy can never be reduced to just affective or cognitive processes. It always occurs as an interplay of both dimensions, whereby the focus can be set differently. Knowing the facts about another person can influence our compassion for them just as much as a pre-reflective, affective experience influences our conclusions about other minds.

Even if numerous phenomenological papers have already dealt with empathy and some also refer to autism [9], as far as we know, there is no separate phenomenological analysis of empathy in ASD. In this paper, we will first summarize the empirical findings on empathy in ASD, followed by an examination of which of the two dimensions is affected. We will draw on phenomenology as our central method, which can be regarded as a basic science of subjective experience [10] and focuses on its central structures such as corporeality, temporality, or intersubjectivity [11, 12].

Although empathy in ASD has been studied in numerous studies using a variety of methods, it could not be clarified conclusively whether there is a disorder of the affective, cognitive, or both dimensions. The following section will summarize the empirical findings on studies of empathy in ASD:

Conceptually, cognitive empathy overlaps with the theory of mind (ToM), which is why both terms are often used interchangeably [13]. ToM assumes that a form of mentalization or theory is necessary to recognize the thoughts, feelings, and sensations of other persons, since one cannot perceive them directly. It is considered to be one of the most influential neuropsychological models for explaining ASD [14], so it is obvious that, consistent with ToM, an impaired cognitive empathy is often assumed.

There are several performance tasks available to study ToM, which can be distinguished into first- or second-order tasks and advanced ToM tasks [15]. First-order tasks assess the ability to assign other people’s thoughts and feelings that differ from one’s own, like the false belief tasks [16] or the Smarties task [17]. Second-order tasks, such as the second-order belief attribution task [18] or the ice cream van story [19], involve the capacity to infer what one person thinks about another person’s states. Finally, advanced tasks comprise different capabilities, including the attribution of complex mental states to others, social understanding, and emotional- and mental-state recognition. Examples for these tasks are the Strange Stories Test [20], the Faux Pas Test [21], or the Reading the Mind in the Eyes Test [22]. All these experiments were carried out on people with ASD (for a review, see [15]), demonstrating that they have impairments of ToM. Even after the age of four, an age from which the corresponding ability for first-order tasks is normally developed [23, 24], children with ASD can have difficulties accomplishing such tasks. This has led to the thesis that individuals with ASD lack the ToM [14, 16].

It is important to note that impairment in ToM is neither specific nor universal for ASD: various groups of individuals with intellectual disability also show difficulties in these tasks, e.g., children with specific language impairment [25], Prader Willi syndrome [26], or Fragile X [27]. On the other hand, there are also a considerable number of autistic children who pass the ToM task [28, 29]. Thus, some seminal findings about a deficit of ToM in individuals with ASD could not be reproduced, such as the False Belief Task [30, 31] or the Strange Story Test [32‒34].

However, individuals with ASD can be limited in their abilities to understand beliefs, intentions, thoughts, and feelings that are different from their own [35, 36] or in abilities like perspective taking [37]. Accordingly, at least some of the individuals with ASD can be said to have impaired cognitive empathy, although there are conflicting views on how basic this impairment may be considered (see the next two sections). Regarding affective empathy in ASD, it is often assumed to be preserved or even enhanced. Adam Smith described the constellation of diminished cognitive and increased affective empathy as empathy imbalance hypothesis [38] and refers mainly to the following three studies in relation to affective empathy:

• 1. Capps et al. [39] examined the response of autistic and non-autistic children to videos designed to elicit empathy. Although the facial expressiveness in autistic individuals implies that they do react to other people’s emotions, the authors concluded that it may be challenging for them to appropriate this response. This is crucial because affective empathy is not just an emotional response to another’s emotion but an appropriate emotional response triggered by the other person’s emotion [23].

• 2. The second study measured facial electromyography (EMG), which is considered an index of sensorimotor contagion [41], following the presentation of emotional (happy and fearful) facial expressions and the presentation of audiovisual emotion pairs [42]. The results showed a heightened EMG activity in response to happy and fearful faces in adults with ASD in comparison to a control group, which may be taken as evidence for an increased affective empathy in ASD. However, these results are in contrast to recent EMG studies showing decreased EMG activity in individuals with ASD: in contrast to the control group, individuals with ASD did not automatically mimic facial expressions when looking at happy and angry facial expressions [43].

• 3. The third study found appropriate psychophysiological responses, measured via electrodermal responses, to images of distressed people in children with ASD [45]. But at the same time, this study also found a decreased responsiveness to the threatening stimuli, thus providing evidence not for increased but for decreased affective empathy.

While there are a few studies suggesting intact or enhanced affective empathy there is broad evidence against such an assumption [46]. Numerous studies show that the first abnormalities in autistic children can be observed at 12 months. Studies of home videotapes of infants later diagnosed with ASD found that many of these infants had abnormalities in or absence of social smiling, spontaneous imitation, visual attention, and reactivity [47‒51]. 2-year-old autistic children showed less empathic concern and self-distress than the control group [52, 53], were limited in affective responses [54] and did not express facial concern in response to feigned distress [55]. These abnormalities in the first 2 years of life, which are associated with rudimentary or already more advanced forms of empathy, are, on the one hand, related to the affective dimension of empathy, and on the other hand, they manifest too early in development, i.e., at a time when cognitive abilities such as ToM are not yet developed, which does not appear until the fourth year of life [29, 56].

“Disturbances of affective contact” [57] are already described by Leo Kanner regarding ASD and are confirmed by further studies from later stages of development. Autistic individuals often respond to external social stimuli in weaker ways, as evidenced by decreased imitation of another person’s action [58, 59], less emotion contagion [60], decreased eye contact [61, 62], or limited affect expression in social interactions [63]. The perspective of embodiment, or the corporeality of mental processes, indicates that facial expression or posture cannot be separated from emotional experience [64, 65], which is why diminished expression in social interactions in ASD is most likely to be associated with a reduced emotional experience. Numerous studies document the difficulties of autistic children in sharing the feelings of others: they showed decreased attention to negative emotional expressions of others [66, 67], but less positive affect combined with attention to social partners [68, 69]. In observational studies, autistic children showed less visible emotional arousal in response to the injury of an investigator [67, 70, 71]. When examining the embodied side of empathy, individuals with ASD did not show the neurophysiological response as if they were in pain themselves when their reaction to the pain of the examiner was assessed [72]. Restrictively, for less severe stimuli, such as losing a watch, there was no difference in visible emotional arousal between autistic and non-autistic individuals [73]. Furthermore, many individuals with ASD exhibit alexithymia [74], a condition characterized by difficulties in recognizing and describing one’s own emotional state. If they have difficulty experiencing their own feelings in general, their capacity to be emotionally resonant with others will be also restricted.

According to parental reports, autistic children showed fewer empathic responses [75], a diminished response to other people’s nonverbal and verbal communication [76, 77], and lesser degrees of concern or guilt toward other people [78]. Autistic children express less empathic guilt, instead describing situations factually rather than emotionally in terms of rule-breaking, disruptiveness, and damage to property [79].

Abnormalities were also found when physiological parameters were considered; studies used respiratory sinus arrhythmia (RSA) as an index of central nervous system regulation of heart rate. Higher RSA is considered as an indicator of emotion regulation and readiness to engage with different environmental stimuli [80]. 4-month-old children, who later received a diagnosis of ASD, showed a lower RSA in response to an encounter with a stranger [81]. This result could be confirmed for 8- to 12-year-old autistic children [82]. In contrast to children with developmental delays, 3–5-year-old children with ASD failed to show a heart rate response to the feigned distress of an examiner [66].

One of the most commonly used method for studying affective and cognitive empathy are self-report questionnaires, which are suitable for empirical studies due to their practicable feasibility and they are considered to be the most intuitively appealing mode of assessment, which is why they were also often used in ASD. Some studies, using such questionnaires, could confirm the empathy imbalance hypothesis [83‒85], while others also revealed a decreased affective empathy in addition to a reduced cognitive empathy [36, 72, 86‒88]. As a possible explanation for these different results the number of participants were mentioned since a closer examination of at least the study by Dziobek and colleagues [84] revealed a trend (p = 0.051) toward significant deficits in affective empathy, while a study with a larger number of participants found a significant difference [36].

In view of the limited ecological validity of questionnaires, the Multifaceted Empathy Test (MET) was developed to evaluate cognitive and affective elements of empathy separately [84]. The test consists of photographs showing people in emotionally charged situations. Participants were asked to infer their emotional state by picking one of four response options (cognitive empathy) and scored their own level of affective concern for the persons seen on a 9-point Likert scale to test affective empathy. The results revealed an impaired cognitive empathy in individuals with ASD, but no difference to non-autistic individuals for affective empathy. Although the MET is an improvement over self-report questionnaires, it still has several issues. The use of photographs is still not very ecological, as in the real world, emotional expressions are not static entities but embedded in an interpersonal situation and constantly moving [89], while facial, vocal, and bodily cues are presented at the same time. More crucially, however, a Likert scale statement about how much one is concerned about someone is a rather course parameter which does not reflect all facets of affective empathy. It certainly does not capture the differentiated perception of emotions [46]. Moreover, a response for reasons of social desirability is not ruled out. The results are therefore not suitable to prove a lack of cognitive empathy as the primary disturbance in autism.

According to the empathy imbalance hypothesis, individuals with ASD are restricted in their ability to recognize the emotions of others (cognitive empathy deficit), but not in their ability to respond adequately to these emotions (affective empathy). This raises the question of how to adequately resonate with or respond to an emotion if the emotion itself is not even properly recognized. One explanation for this could be that the affective reaction is automatic without being conscious; however, the proponents of this theory have not yet presented an explanation or evidence for this. Moreover, there is a broad body of evidence demonstrating impaired emotional connectedness and responsivity in ASD, so that, based on empirical research, a limited affective empathy in individuals with ASD must be assumed, while there is some evidence of a reduced cognitive empathy as well. In the following, we will discuss which of the two dimensions can be regarded as a basic disorder of ASD.

A deficit in ToM is considered to be one of the most influential neuropsychological models for explaining ASD [14]. From a phenomenological perspective, ToM has been criticized because it assumes a fundamental inaccessibility¹ of other people [90, 91]. ToM follows the Cartesian dualism, which distinguishes between mind and body [92]. According to Descartes, a subject recognizes the world primarily on a cognitive level [93]. Feelings are conceived as emotional or psychological states that are located in the subject’s inner world and can only be understood indirectly from external bodily behavior, such as facial expression, via mentalization. Social understanding thus becomes a projection of inner representations onto others. Such an approach conceives the psychic realm intersubjectively, in that, a perspective is taken from one mental state to another, but it creates the impression that such a perspective taking is basically disembodied.

Proponents of ToM like Baron Cohen usually suggest that it is hard for us to make sense of behavior in any other way than via the mentalistic (or “intentional”) framework. We mindread all the time, effortlessly, automatically, and mostly unconsciously. In the words of Sperber (1993), “attribution of mental states is to humans as echolocation is to the bat.” It is our natural way of understanding the social environment” (14, p. 3–4). In doing so, however, they overlook the implicit and immediate understanding of expressive characters of another person in an embodied social interaction, especially in the first few years of life. On the other hand, it is obvious that ToM is based on additional cognitive abilities that develop only gradually from the second year of life onwards [94]. There are situations in which we can use perspective taking and knowledge about the other person to enhance our empathic understanding (this is termed “Theory of Mind”, although there is no theory involved), but this does not change the fact that empathy starts with interbodily resonance and direct perception of others’ feeling states. Furthermore, ASD is an early developmental disorder that must have manifested in the early developmental period. In non-autistic children, ToM does not appear until the fourth year of life [29, 56]. Since ToM manifests after ASD, a deficit of ToM cannot explain ASD.

Ironically, an analysis of ASD teaches that in everyday life, we usually do not depend on some form of mind reading in order to understand other persons, but that such is used explicitly by high-functioning autistic individuals in form of a rule- and knowledge-based system [9, 33, 95], namely in order to compensate for their difficulties in embodied social interactions [96]. Nevertheless, such compensatory strategies are not sufficient for intuitive social understanding, since human expression and language are seldom unambiguous. In ASD, a second-person perspective, based on pre-reflective embodied attunement with others, is replaced by a third-person perspective, i.e., the position of an observer using algorithms and abstract rules [97]. Pointedly, one could say that people with ASD interact socially as one would expect all people to interact if ToM was correct and the others were not directly accessible.

Regarding this body obliviousness represented by ToM, in recent decades, based on developmental psychology, the individual has been conceptualized as being in implicit bodily relationships from birth [98‒100]. From a phenomenological point of view, feelings or intentions to act are not hidden within a person; rather, they manifest themselves in facial, gestural, and linguistic expressive behavior.

In an embodied encounter, there is a tacit connection between bodies, which is conveyed through eye contact, gestures, facial expressions, and touch, since the body represents the medium to the world and other people as well as the permanent primordial horizon of all our experiences (Merleau-Ponty [101]). Because one’s own and another’s body belong to one encompassing structure, both are experienced as related from the outset [102] and the body can extend through the body schema to the body of another person (Merleau-Ponty [101]). The body schema encompasses innate and acquired sensory motor skills and habits that enable the body to interact with the environment. The body transcends itself continuously and thus also incorporates other people, as it were. This dyadic relationship, referred to by Merleau-Ponty [103] as intercorporeality, is the basis of affective empathy, characterized by an unthematic interplay of expression and impression.

Each body affects the other, functioning both as an organ of emotional expression and as a resonance organ for the impressions created by the other’s expressions [104]. Resonance includes all forms of local or general body sensations, such as activation of the autonomic nervous system, muscle activation, physical postures, and their corresponding kinesthetic sensations [105]. Through a mutual expansion of the body schema, the bodies of interactive partners affect each other as if they were being drawn into a force field of interaction. The embodied counterpart is understood pre-reflexively, as every expression of one person creates an impression in the interactive partner [106]. The reciprocity of mutual expressions and impressions results in an interbodily resonance in which the body of the other can be experienced through one’s own [107]. Thus, from early childhood, an implicit bodily relational knowledge of how to behave in social situations and how to empathize with others is formed as a basis for affective empathy [108, 109].

Developmental psychology has been able to confirm the phenomenological approach of embodied, nonverbal communication by examining the interactions between a newborn/toddler and his or her caregiver, and to demonstrate the relevance of implicit, body-mediated learning processes since early childhood [102, 110]: Even a toddler differentiates between inanimate objects and people [111] and shows a stronger preference for faces than for other stimuli [112]; he or she can already recognize a human face in a set of environmental objects and imitate the facial expressions of the caregiver [113, 114]. Imitation already represents a social action in young children, implying knowledge of the sociality of the other. Because physical imitation also elicits corresponding feelings, a mutual affective resonance gradually develops within the dyad. Through mediation of his or her body, the toddler can incorporate and imitate the body of another person. From birth, emotions are not an intrapsychic phenomenon, but rather located in the area of the “in-between,” embedded in primarily bodily communication with the caregiver.

Trevarthen names a willingness to communicate and an affective ability to resonate between child and caregiver as a prerequisite for embodied interaction from birth [115]. If these prerequisites are inadequate, the child cannot develop social skills and early developmental disorders manifest themselves, as is the case in ASD [116].

Phenomenological studies assume a disorder of intercorporeality in individuals with ASD [90, 91, 117], since the pre-reflexive understanding of other persons fails. In ASD, the social interplay of expression and impression as the foundation of affective empathy is reduced. The implicit, body-mediated knowledge which otherwise enables prelinguistic understanding is limited [98]. Therefore, the emergence of an overarching system in which both communication partners are integrated “like organs of one single intercorporeality” [118] is thwarted. The body in autism does not extend to the body of the other and, thus, cannot generate a basis for an intuitive, empathic understanding. Consequently, individuals with ASD have difficulties understanding nonverbal forms of communication such as eye contact, facial expressions, and body language [119]. Their attention is on single details of social interactions rather than on the integration of information in a holistic and contextual manner which allows grasping the higher level meaning or gestalt [120].

Ms. K., a woman diagnosed with ASD, describes her difficulties dealing with other people as follows:

Ms. K. describes how she cannot affectively empathize with other people. The body’s function as an organ of resonance, which characterizes affective empathy, is limited, so that the body-mediated expression of other people does not lead to a pre-reflective understanding – their body cannot be experienced through one’s own [98]. Even if a situation is explained to Ms. K., the unspoken rules of social interaction remain incomprehensible, which addresses common sense [122]: the implicit embedding in a self-evident context of meaning is missing, so that the other person remains a mystery to her. Instead, people with ASD are looking for explicit, cognitive strategies to interact with other persons in order to “unravel” them.

The lack of implicit, body-mediated knowledge also makes it more difficult for individuals with ASD to empathize affectively with others. Since they often do not know what the other person is thinking or feeling, they cannot respond to them accordingly. Such an account is in line with the double empathy problem [123], according to which the social and communication difficulties, presented in an encounter of an autistic and a non-autistic person, are based on a mutual lack of understanding and bidirectional differences in communication style. The disorder of intercorporeality emphasizes the space of the “in-between”; i.e., the disorder is not located in an individual itself, otherwise manifesting in the social interaction between an autistic and a non-autistic individual.

Hoffmann assumed in his influential theory about the development of empathy that human beings have an innate capacity to experience at least primitive forms of empathy [124, 125]. He distinguishes between different modes of empathic arousal, of which mimicry, conditioning, and direct association are passive, involuntary, and preverbal, and therefore tend to be assigned to an affective dimension of empathy. They enable a primitive form of empathy in young infants, which further consolidates over the second and third years of life. Cognitively more complex mechanisms develop later (e.g., role-taking). At the age of 12 months, autistic children exhibit less social smiling and spontaneous mimicry, and at the age of 2 years, they exhibit less self-distress and empathy. A feeling for the other person’s body and the perception that it is similar to one’s own are necessary conditions for being able to imitate that person [126], yet these conditions do not appear to be fulfilled in the same way in children with ASD. Imitation of another person is often accompanied by a corresponding affective experience. Since individuals with ASD have difficulty with imitation, this must also diminish their ability to empathize with others.

A resulting lack of interest in social interactions, because they are not experienced as affectively valuable or meaningful, can even increase the children’s social withdrawal. Without adequate interaction, they are unable to understand or develop the rules of complex social interactions, be it implicit social rules or the development of friendships. Such affective interactions between toddlers and their caregivers not only deepen their relationships, but also linguistic and cognitive (e.g., perspective or intelligence) skills [127].

The notion of a primary limitation of affective empathy is consistent with Greenspan´s account of autism, according to which, people with ASD miss the critical developmental stage to enter into a reciprocal and affective interaction with other people and to connect the affect to motor planning [127, 128]. Greenspan developed his view into a therapeutic approach for autistic children. His “Developmental, Individual-Difference, Relationship-Based Model” is based on the assumption that cognition, language, social, and emotional skills are learned through relationships that involve an emotionally meaningful exchange [129]. Unlike other therapeutic approaches, it emphasizes the child’s emotional rather than cognitive development. The caregiver uses the child’s natural interests and tries to get into a mutual communication while playing. The aim is to find ways to reach the child emotionally through interest, motivation, and curiosity. This gives the child the opportunity to understand the full range of feelings, but he or she is also encouraged to develop abstract thinking skills. Such a therapeutic approach, when used in the first few years of life, may enable autistic children to implicitly understand other people, which can lead to an improvement in their disturbed affective empathy.

According to Hobson, autistic children are limited in experiencing and expressing “person-centered” feelings such as concern [78]. They exhibit a lower propensity to perceive or even assimilate and respond to another person’s bodily anchored experience. Limitations in interpersonal and interaffective experiences might be regarded as “basic” because they cannot be reduced to other terms. This is demonstrated by the early symptoms that allow ASD to be diagnosed. In terms of development, they come before many facets of later social-cognitive skills, such as a ToM or cognitive empathy [130, 131]. Although affective and cognitive empathy may be both impaired in people with ASD, the limits in affective empathy arise earlier in the developmental process due to a reduced intercorporeality. As a result, ASD can be referred to as a disorder of affective empathy.

A phenomenological analysis considering empirical evidence assumes a disorder of affective empathy in individuals with ASD. An impaired intercorporeality, which is thought of as the foundation for affective empathy, can be used to explain irregularities in social smiling, spontaneous mimicry, empathic concern, and affective reactions that are observed as early as the first 2 years of life. Since the bodies of individuals with ASD do not transcend themselves to resonate with and incorporate the bodies of others, individuals with ASD do not implicitly understand other people’s feelings. The mutual interplay of expression and impression is interrupted, so the body of an autistic person can only affect the body of the counterpart to a limited extent and vice versa, resulting in an impaired affective empathy.

Although individuals with ASD may also struggle with cognitive empathy, this cannot explain their fundamental peculiarities in social interactions. Cognitive empathy, overlapping with ToM, considers social interaction as a kind of cognition. Assuming a fundamental inaccessibility of other individuals, it requires a form of mentalization to deduce from an external physical behavior or expression the emotional experience behind it. Proponents of ToM usually assume that most of empathic understanding can be adequately explained by such indirect cognitive operations, while ignoring the implicit and immediate comprehension of another person’s expressive characters in an embodied interaction, which is a hallmark of social interaction, particularly in the early years of life. Since ASD is an early developmental disorder, the abnormalities must manifest themselves in the first years of life, when even typically developing children have not yet acquired ToM or cognitive empathy. The autistic abnormalities in social interaction can therefore not be explained by a deficit of ToM, but by an impaired affective empathy based on a disturbed intercorporeality.

We want to thank Valeria Bizzari for her very careful proofreading of this text and philosophical suggestions.

The authors have no conflicts of interest to declare.

This research was not funded by any external funding agency.

Tim Schnitzler was responsible for the initial conception, first draft, and subsequent rewriting. Thomas Fuchs supervised and structured the text throughout the various stages of development.