Background
Empathy has been defined as the drive to identify and respond appropriately to emotions and mental states in others [
1,
2]. It plays a vital role in human relationships and allows an individual to make sense of and predict the behaviour of another [
3]. Empathy involves both the ability to recognise and understand emotion in others [
3] as well as an affective response to another’s emotional state [
4,
5], respectively cognitive and emotional empathy [
4,
6].
Autism spectrum conditions (ASC) involve empathy deficits [
6‐
9] and are characterised by communication and social difficulties as well as repetitive behaviours or restricted interests [
10]. Empathy dysfunction in autism has been demonstrated via research noting a theory of mind (ToM) impairment in children with ASC [
11]; that is, that individuals with autism have difficulty reading the beliefs and intentions of others [
11,
12]. ToM is often used interchangeably with cognitive empathy, perspective taking and ‘mentalising’ [
13]. However, as noted above, empathy has long been defined as a multifactorial construct including not only the representation of another’s emotional state (ToM or cognitive empathy) but also an affective response (emotional empathy).
The Empathising-Systemising (E-S) theory [
14,
15] expands the concept of ToM to include this affective component of empathy. The E-S theory argues that the social and communication difficulties seen in ASC can be accounted for by an empathy impairment (including both cognitive and emotional components) and the repetitive behaviours and narrow interests by an inclination for systemising (the drive to understand and derive rules about a system) [
16]. A recent factor analytic study by the authors [
17] found support for the E-S model. This study, based on the same individuals as the research reported here, identified two factors representing empathy and systemising. These factors were found consistently across individuals with autism, first-degree relatives and general population controls [
17]. In concordance with the E-S theory, individuals with ASC showed elevated scores on the latent systemising factor and low scores on the empathy factor. This previous study included questionnaire measures of empathy and systemising only. However, other studies have indicated that individuals with autism also have difficulty with performance-based tasks involving the identification of emotions and perspective taking [
18‐
20]. As these tasks involve the identification of emotion, they are generally conceptualised as performance-based tasks of cognitive empathy.
Although there is much evidence to suggest that individuals with autism display difficulties with ToM or cognitive empathy, there is more debate about the role of emotional empathy in autism. While mirror neuron theory [
21] argues that individuals with ASC have weak emotional empathy, Dziobeck and others [
22] claim that emotional empathy is intact in autism. Other theorists have proposed that it is due to heightened emotional empathy that individuals with ASC find the social world more challenging, arguing that it is overwhelming rather than difficult to understand [
3,
23,
24].
First-degree relatives of individuals with an ASC diagnosis may also show some mild traits of the autism spectrum [
25], also referred to as the broader autism phenotype (BAP) [
26,
27]. The finding of the BAP fits with the notion that autism is under polygenic influence, and that at least part of these genetic influences are inherited (rather than
de novo genetic events) and can also be found in undiagnosed relatives displaying the broader phenotype [
28]. The BAP has also been shown to apply to empathy, with parents and siblings of affected individuals scoring lower on performance-based tasks involving emotion recognition [
20,
29‐
31] and questionnaire measures assessing empathy [
20,
27]. It is therefore important to examine cognitive and emotional empathy not only in clinical samples, but across the full range of genetic variability, including individuals on the autism spectrum, their relatives and general population controls.
A number of quantitative measures of empathy have been used in previous research, including the Interpersonal Reactivity Index [
32] and the Empathy Scale [
33]. However, one of the most widely used measures is the Empathy Quotient (EQ) [
6], a self-report measure of empathy assessing both cognitive and emotional components. The EQ has recently been studied in detail across three studies. Two studies highlight a three-dimensional structure including cognitive empathy, emotional empathy and social skills [
34,
35], with the third highlighting a single dimension [
36]. The first two studies were based on student and general population samples, with the third including individuals with autism and first-degree relatives. Although individuals with ASC and family members were included in the third study, the factor structure and utility of the EQ was examined for the whole sample and not for each of the three groups (individuals on the spectrum, first-degree relatives and general population controls) separately.
Although the EQ and performance based measures of cognitive empathy have been studied quite extensively by themselves in previous studies, the relationship between subscales of the EQ (a questionnaire-based measure), and performance-based measures of empathy have not been comprehensively assessed to date. The current study aims to evaluate the multifactorial nature of empathy utilising both behavioural and performance-based task data. It was assessed whether the latent structure of empathy differs across samples stratified by genetic vulnerability (individuals with ASC, first degree relatives and controls).
Discussion
Factor analyses in data from a large sample of individuals with ASC, parents and controls, using both questionnaire and performance-based measures of empathy, suggested a four-factor latent structure of empathy encompassing Cognitive empathy, Emotional empathy, Social skills and a Performance-based measurement factor. This structure was consistent across individuals deemed to have a high (individuals on the autism spectrum), medium (parents) or low (controls) genetic vulnerability for autism, indicating that the overall latent structure of empathy is consistent across both clinical and general population samples. However, there were some differences in the factor loadings and factor correlations across the three groups.
The latent structure identified in this study is consistent with previous research in that it identifies both a cognitive and emotional component of empathy [
3,
4,
13,
22]. In addition, the analyses also identified a separate Social skills factor. Items measured by the Social skills factor of the EQ assess specific empathising skills within a social situation. For example, ‘I find it hard to know what to do in a social situation’ and ‘I often find it difficult to judge if something is rude or polite’. Future research utilising other measures is needed to further assess the theoretical implications of this Social skills factor, which is shown to be separate from cognitive and emotional empathy.
It was expected that the performance-based emotion recognition tasks would be related to the Cognitive empathy factor. However, factor loadings of the Eyes and KDEF on the Cognitive empathy factor were low, with a model including a separate performance-based task component providing a better fit. Interestingly, the relationship between the Performance-based test factor and the other empathy factors was different across the three groups under study. In the control group, the performance tasks were not significantly correlated with any of the questionnaire-based empathy factors. Within parents and individuals with autism, the performance measures were related in almost equal magnitude to all three components, rather than solely to cognitive empathy. The finding that these performance tasks do not directly and exclusively assess cognitive empathy is new. Previous research has operated on the assumption that these tasks are performance-based measures of cognitive empathy. The findings of the current study indicate that rather than being a direct measure of cognitive empathy, scores on performance-based tasks like the Eyes and the KDEF have a bearing on empathy more widely. Our results suggest that completion of either of these tasks requires engagement of more than just cognitive empathy abilities. Rather, impairment on these performance-based tasks is indicative of a broader impairment across all facets of empathy. This has important implications for future research involving the implementation of such tasks.
Individuals with autism showed greater impairment (as indexed by lower mean latent factor scores) across the Cognitive and Emotional empathy, Social skills and Performance-based empathy factors compared with controls. Similarly, the ASC group displayed greater impairment than parents across all factor means, with the exception of the Performance-based factor. This fits with the notion of autism as a disorder of empathy [
7‐
9]. In contrast with some previous research [
3,
22‐
24], there was no evidence that individuals with autism exhibited intact or heightened emotional empathy.
Parents also showed mild impairment across all four factors compared to controls. However, with the exception of the Performance-based factor, impairment was not as strong as observed in the ASC group, placing their difficulties somewhere in between the clinical and the control group. This is consistent with previous accounts indicating that first-degree relatives show some difficulties on tasks of empathy [
20] compared with controls. Moreover, it fits with the notion that characteristics related to autism are distributed as quantitative traits rather than discrete entities [
37,
60] and are likely to be influenced at least in part by common genetic variation [
28].
Limitations
As mentioned previously, a number of fathers had missing data on the Eyes and KDEF tasks (n = 104). To assess whether these missing data had any effect on the results, all analyses were run both by imputing data for these individuals as well as excluding the missing cases. As there were no substantive changes within any of the models, it is highly unlikely that these missing data were a confounding factor.
The parent group also consisted of a larger proportion of mothers (n = 298) than fathers (n = 141). To ensure these differences would not bias the analyses, any sex effects on the means were regressed out prior to conducting the factor analyses. Moreover, the evaluation of sex differences in the latent factor structure indicated that it was similar across both sexes. Future studies including very large sample sizes would be of interest, as these could explore any possible sex difference in the latent factor structure in more detail than the current sample size permitted.
Lastly, the control group included in this study had completed a somewhat higher level of education than the parent and ASC groups. We can therefore not exclude the possibility that differences in educational level may explain some of the differences in factor structure of empathy observed between controls and the parent and ASC groups.
Conclusions
The current study assessed the latent structure of empathy across individuals with a low, medium and high genetic vulnerability to autism. Results highlighted that empathy shows evidence of multidimensionality, in which four factors can be distinguished irrespective of genetic vulnerability, including three components of empathy and a performance-based factor. Unexpectedly, performance-based measures of empathy were related in almost equal magnitude to Cognitive empathy, Emotional empathy and Social skills, rather than solely to Cognitive empathy. This has implications for the nature of impairment indicated by performance on such tasks, suggesting that these effects are much wider than impairments in cognitive empathy alone. Individuals with autism displayed impairment on all four components of empathy, confirming the notion that autism is characterised by difficulties with multiple facets of empathy. Parents showed intermediate impairments of empathy, providing evidence for the BAP and highlighting the importance to assess characteristics of autism on quantitative scales.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
RG, RH and AB designed the study. Data reported on in this study were collected by CA and SB-C. RG undertook all statistical analyses and drafted the manuscript. All other authors contributed to subsequent drafts and the final submitted manuscript. All authors read and approved the final version of the manuscript.