Journal of Intellectual & Developmental Disability, June 2007; 32(2): 8293

Empathy and theory of mind in offenders with intellectual disability

TRACEY PROCTOR1 & NIGEL BEAIL2,3

1

Derby Adult Learning Disability Service, UK, 2Barnsley Learning Disability Service, UK, and 3University of Sheffield, UK

Abstract Background Little research has been carried out on empathy and theory of mind in offenders with intellectual disability (ID) and these concepts are often poorly defined. Method Various models of empathy and theory of mind are discussed and scores on 2 empathy and 3 theory of mind tasks are compared for 25 offenders with ID and 25 non-offenders with ID (all male). Results Differences were found in empathy and theory of mind performance of offenders and non-offenders with ID. Offenders performed better than non-offenders on a second order theory of mind task and on emotion recognition. They required fewer prompts to mention emotions, and gave empathic/caring responses more often than non-offenders when observing happiness (but not sadness or anger). Conclusions Results suggest that offenders with ID may have better, rather than poorer, empathy and theory of mind abilities than non-offenders, and that empathy training is therefore not indicated for this group.

Keywords: Empathy, theory of mind, offenders, intellectual disability

Introduction Research on offenders with intellectual disability (ID) has undergone considerable development in recent years. In 20012002, three journals published special issues containing new research findings in this area (British Journal of Forensic Practice, edited by Frankish, 2001; Journal of Applied Research in Intellectual Disabilities, edited by Lindsay, 2002a; and Journal of Intellectual Disability Research, edited by Fraser & Taylor, 2002). While there is research interest in the empathy of offenders (particularly sexual offenders), this area has only recently begun to be investigated in offenders who also have ID. This delay in transfer from the mainstream to intellectual disability services is echoed in treatment programs, which often aim to increase empathy and evaluate victim empathy at outcome. According to one survey, over 90% of North American programs included empathy training as a principal component (Knopp, Freeman-Longo, & Stevenson, 1992), however this approach appears to have only recently been applied to offenders who also have ID (Rose, Jenkins, OConnor, Jones, & Felce, 2002). The lack of research focus on empathy in people with ID is surprising, given that it has become increasingly common to investigate the related concept of theory of mind with this group.

Although early research focused on the measurement of theory of mind abilities in the under-5s and in those with autistic spectrum disorders, more recent research has looked at this concept in adults and in individuals with ID (Sullivan & TagerFlusberg, 1999; Tager-Flusberg & Sullivan, 2000; Yirmiya, Solomonica-Levi, Shulman, & Pilowsky, 1996). Exactly how empathy and theory of mind are related, however, is still unclear. Empathy has been defined as sharing the emotional state or context of another (Eisenberg & Strayer, 1987). Historically, however, the term has not been used or measured in a consistent way, and has often been used without detailed consideration or explanation of exactly what it means. There is ongoing debate around whether empathy is affective, cognitive or both (Jolliffe & Farrington, 2004). The definition can be expanded to include a cognitive component by adding a phrase that indicates understanding as well as sharing the emotional state of others. However, this may represent an oversimplification. In an earlier review of the literature, Marshall, Hudson, Jones, and Fernandez (1995) summarised their findings by reconceptualising empathy as a 4-stage process, involving emotion recognition, perspective taking, emotion replication and response decision. This model makes it clear that deficits could occur and be measured at any

Correspondence: Professor Nigel Beail, Clinical Psychology Unit, University of Sheffield, Sheffield S10 2TP, UK. E-mail: n.beail@sheffield.ac.uk ISSN 1366-8250 print/ISSN 1469-9532 online # 2007 Australasian Society for the Study of Intellectual Disability Inc. DOI: 10.1080/13668250701373331

Empathy and theory of mind in offenders with ID stage in the empathy process. However, despite providing a useful framework for thinking about empathy, the components remain rather simplistic. Similar attempts to operationalise the concept of empathy have been published elsewhere (Geer, Estupinan, & Manguno-Mire, 2000; Goldstein & Higgins DAlessandro, 2001), but little empirical research has been carried out to formally evaluate the hypothesised components. The related concept of theory of mind is generally defined as the understanding that other people have a mind separate to ones own, along with the capacity to understand those mental states (Yirmiya, Erel, Shaked, & Solomonica-Levi, 1998). At first glance, this appears similar to the perspective taking component of empathy referred to above. In order to clarify current opinion about the definitions of empathy and theory of mind, a number of componential models of empathy, theory of mind, and social skills are summarised in Table 1, with common features across the models highlighted. Four empathy models are considered (Davis, 1983; Geer et al., 2000; Goldstein & Higgins DAlessandro, 2001; Marshall et al., 1995), together with a model of general social skills (McFall, 1990), which has been included to show how these fields overlap, and two theory of mind models (Keenan & Ward, 2000/ Ward, Keenan, & Hudson, 2000; Tager-Flusberg & Sullivan, 2000). The aim of this tabulation is to draw together the strengths of existing models, using the detail of some to elaborate on omissions in others, and to assemble a model with the highest level of detail presently available. In the construction of this table, it has been necessary to interpret the ideas expressed in two related papers (Keenan & Ward, 2000; and Ward et al., 2000) and to present these in a format consistent with the ideas of the other researchers. Marshall et al. (1995) proposed an emotion recognition component of empathy, which would appear to be the earliest of all the suggested components of the empathy process and to be equivalent to McFalls (1990) reception and perception aspects of decoding skills and the social-perceptual component of Tager-Flusberg and Sullivan (2000). These have been labelled the perceptual component in Table 1, which attempts to integrate all the suggestions into one overarching model of empathy (subsuming theory of mind). The next common component relates to perspective taking (Davis, 1983; Goldstein & Higgins DAlessandro, 2001; Marshall et al., 1995) or the ability to understand anothers point of view (Geer et al., 2000). This may be the equivalent stage to theory of mind, since it can be broken down into

83

the realisation that others have their own mental states, the comprehension of these states, and being able to predict the persons behaviour based on these (Keenan & Ward, 2000; Ward et al., 2000). Table 1 also shows that this understanding can then be used in a circular way to interpret further social information. The exact relationships between these components are not clear, so this part of the empathy model has been rather vaguely labelled cognitive comprehension, as it appears to be about the ability to understand the social information that comes in from the previous perceptual stage. There is generally believed to be an emotion replication component of empathy, which involves vicariously experiencing others emotions (Geer et al., 2000; Goldstein & Higgins DAlessandro, 2001; Marshall et al., 1995). This has been labelled the emotional component in Table 1, but it is clear that this concept requires breaking down based on further empirical research. Finally, an action component of the model is often included (Geer et al., 2000), sometimes separated into a response decision aspect (Marshall et al., 1995; McFall, 1990) and the enactment of the behaviour itself (McFall, 1990). Given that empathy must now be seen as a composite of component elements, it is clear that many existing questionnaire measures, which are based on an outdated understanding of empathy as a unitary model, are now unsatisfactory. Such measures include the Hogan Empathy Scale (Hogan, 1969) and the Questionnaire Measure of Emotional Empathy (Mehrabian & Epstein, 1972). An early factor analysis of the Hogan Empathy Scale (Johnson, Cheek, & Smither, 1983) suggested that it was measuring four factors that clearly do not correspond to the theories of empathy described above: social self-confidence, even-temperedness, sensitivity and non-conformity. In addition, this measures validity and reliability (as a measure of a trait construct) were not supported in a recent analysis (Froman & Peloquin, 2001). Given that most existing research is based on such scales, it is important to critically assess what can be inferred from each study, taking into account the measures used. Explanations of criminal or antisocial behaviour (both violent and/or sexual) have often hypothesised that lack of empathy reduces the inhibition to cause harm to others. Consequently, offender treatment programs (particularly, though not exclusively, those for sex offenders) often include empathy training components (see Jolliffe & Farrington, 2004). However, empirical evidence on this matter is not conclusive. There is some evidence that sex

84

T. Proctor & N. Beail

Table 1. Summary of componential models of empathy, theory of mind, and social skills

offenders (without ID) obtain decreased scores on unitary empathy scales compared to non-offenders, although their scores remain within normal limits (Marshall et al., 1995). There is also research suggesting that offenders show deficits in the areas of emotion recognition and replication of victims feelings, (posited as components of empathy by Marshall et al., 1995). However Jolliffe and Farrington (2004) point out that other researchers have not found lower empathy in offenders (e.g., Bush, Mullis, & Mullis, 2000), and their metaanalysis found that a relationship between low affective empathy and offending disappeared when intelligence and socioeconomic status were controlled for. By contrast, there is no evidence to draw upon comparing offenders and non-offenders on theory of mind.

In people with ID, the evidence on empathy is again conflicting. Even when empathy is broken down into components, studies contradict one another about whether or not those with and without ID differ on emotional and behavioural aspects (Dyck, Ferguson, & Shochet, 2001; Kasari, Freeman, & Bass, 2003). In the field of theory of mind in people with ID, it can be difficult to interpret results because those who do not manage to follow the tasks are often classified together with those who only fail on theory of mind aspects (e.g., Bowler & Strom, 1998). It is therefore not possible to distinguish whether there is a theory of mind deficit or a general difficulty because of lower IQ. Due to the scarcity of research in these two fields, even combining the evidence from both sheds little light on how we might expect offenders with ID to

Empathy and theory of mind in offenders with ID perform in areas of empathy and theory of mind. Lindsay (2002b) points out that a major flaw of research on offenders with ID is that studies are often based on clinical samples with no control group of people with ID who have not offended. It was the aim of the current study to remedy this shortcoming, and at the same time to avoid a number of the pitfalls that make the existing literature difficult to interpret. First, empathy was treated as a componential concept, and as far as possible, tests were selected on the basis that they could be interpreted in this way. Second, failures of general comprehension were separated from failures specific to theory of mind or empathy. Third, commonly used and validated measures were used where possible. It was hypothesised that the offender and nonoffender groups would differ on empathy and theory of mind scores. It might be predicted that offenders would show deficits in these scores, but existing evidence is contradictory, and some researchers have hypothesised that offenders might actually be more skilled in these areas (Covell & Scalora, 2002). Further, in the interests of clarifying the relative roles of empathy and theory of mind, the empathy scores for those who were found to have good theory of mind skills were compared with those found to have poor theory of mind skills. The aim of this comparison was to facilitate a discussion of whether theory of mind is a sub-component of empathy, as hypothesised in Table 1.

85

Participants in the control group were selected from community-based day services or clinical psychology services for adults with ID. There were 25 participants in the control group, with a mean IQ of 60.8 (SD56.1) and a mean age of 41 years (SD512). All participants were male due to a lack of females in the participating organisations. If IQ data were not available, the Wechsler Abbreviated Scale of Intelligence (WASI: Wechsler, 1999) was administered. IQ scores of the groups did not differ significantly. Individuals with serious mental health problems or autism were excluded from the study, as these conditions may account for atypical empathy/theory of mind skills. The vast majority of participants in both groups had ID of unspecified cause, meaning that there were no records of specific syndromes or conditions. The group as a whole is therefore representative of a group with general intellectual disabilities, but it should be remembered that such disabilities vary across individuals and may have a profound impact on a number of areas, including social skills. Measures Interpersonal Reactivity Index (IRI: Davis, 1983). The IRI was selected because of its common usage in the existing literature (Bovasso, Alterman, Cacciola, & Rutherford, 2002; Burke, 2001; Bush et al., 2000; Goldstein & Higgins DAlessandro, 2001; Smallbone, Wheaton, & Hourigan, 2003), and because it attempts to measure four separate components of empathy (perspective taking, fantasy, empathic concern and personal distress). It has been demonstrated to have sufficient internal reliability (alpha levels of 0.710.77) and test-retest reliability (0.620.71) (Davis, 1983). However, for the purposes of this study, the measure was adapted to make it more suitable for people with ID by removing unnecessary wording, simplifying concepts, and eliminating ambiguity, and therefore existing reliability and validity data did not apply. The IRI was completed by interview using a 5-point scale on a response sheet (no 7, a little bit 3, sometimes 3, quite a bit 3, a lot ). Informal checking was carried out using non-emotional questions such as Do you like chocolate? to ensure that each participant was able to use various points on the scale.

Method Participants Due to the lack of existing research, power calculations were based on a study that compared individuals with ID with and without challenging (rather than offending) behaviour (Moffatt, HanleyMaxwell, & Donnellan, 1995), and therefore could provide only a rough guide for sample size. It was determined that for a power of 80% in the Test of Emotional Perception (TEP), sample size should range from 10 to 30 per group, depending on subtest. Participants in the experimental group were recruited from organisations providing a secure environment for offenders with ID. Volunteers were screened to include only individuals with ID who had at some point offended against another person1. A total of 25 participants were recruited for this group, with a mean IQ of 64.2 (SD57.3) and a mean age of 31 years (SD511). Empathy training was not a part of the treatment program for these offenders.

Test of Emotional Perception (TEP: Moffatt et al., 1995). This is the only empathy measure which has been found to objectively measure concrete components of empathy in people with ID.

86

T. Proctor & N. Beail could not answer any of the questions, to have no theory of mind (if they followed the task but said the new person would think the box contained a pencil), or to have good theory of mind (if they answered correctly). Location Change (Sally-Anne) Task (Wimmer & Perner, 1983). This is another commonly used theory of mind task (e.g., Reed, 1994; Yirmiya & Shulman, 1996; Yirmiya et al., 1996), which has also been found to have moderate reliability (e.g., Charman & Campbell, 1997). However in order to maximise the understanding of participants with ID, a video version was developed. In the first video-clip, two men are sitting on a couch reading. One of them gets up and goes through a door marked with a Mens toilet sign. After a brief pause, the man comes back and the tape pauses as he closes the door. The participant is asked what happened in the tape and what might happen next. This is to check that they understand the concept of What happens next? and that they can demonstrate this, so that any mistakes on the next sequence can be attributed to theory of mind difficulties. The second video-clip corresponds to the SallyAnne task. The two men are sitting on the couch eating sandwiches and one of them (the other one this time) gets up to go through the toilet door, putting his sandwich in his bag before he goes. While he is gone, the remaining man takes the sandwich and puts it in his own bag. The first man then returns from the toilet and the tape pauses as he closes the door. Again, the participant is asked what happened in the tape and what might happen next, as well as where the sandwich really is. For both video-clips, the participant is helped to understand the question with prompts such as What will that man [point] do now?, allowing participants who find it difficult to communicate to point their answers on the screen. The experimenter was interested mainly in whether the participant knew that the returning man would have to look in his own bag before realising the sandwich had gone. If the participant said Hell look in the bag, they were prompted with Which bag?. As long as participants indicated that the man would look in his own bag, they were scored as having good theory of mind. If they were unable to indicate the general gist that he would come back and sit down and try to finish his sandwich, they were scored as lost. If they said there would be an argument or the man would be angry, they were prompted to talk the experimenter through the scene, to determine whether they understood that he would have to look in his own bag before realising that the other man

Components include emotion recognition, prediction of an appropriate emotional response to anothers emotion, and expression of ones own emotional response. A training video was shown (up to a maximum of three times) to ensure that participants understood the concept of What happens next? in terms of selecting from photographic options. For the test, after viewing each of six video vignettes (two depicting happy news, two sad news, and two angry news, shown in different order), participants were asked What did you see?. If their response did not mention an emotional response, a vague prompt (Anything else?) was given (up to a maximum of two times), followed by a more precise prompt (What happened after he/she answered the phone/ opened the letter?), and then if necessary, the participant was finally asked How did he/she feel? The number of prompts given was recorded. Next, three photographs depicting different emotional responses of the characters friend were shown in random arrangement. The participant was then asked to choose the one showing what might happen next. Their selection and the time taken to make that selection (measured using a stopwatch) were recorded. Finally, participants were asked how they would feel if one of their friends received similar news. If no feeling was mentioned, the participant was prompted: But how would you feel?. The response was coded as either the same as the emotional response they identified in the first part of the test (e.g., they would feel sad if their friend was sad), as a concerned response, an incongruous response or no emotional response given (if they failed to give a response even after prompting), and the number of prompts required was recorded. Deceptive Box (Smarties) Task (Gopnick & Astington, 1988). This test was selected due to its repeated use in theory of mind research (e.g., Charman & Baron Cohen, 1995; Perner, Frith, Leslie, & Leekam, 1989; Prior, Dahlstrom, & Squires, 1990; Yirmiya et al., 1996; Zelazo, Burack, Benedetto, & Frye, 1996). Reliability with people with ID has been found to be moderate, which is no lower than in typically developing children (e.g., Charman & Campbell, 1997). Participants were shown either a Smarties or a toothpaste box and asked: What do you think is in here?. They were then shown that the box actually contained a pencil and asked: What is in here really?. The next question asked: If somebody else comes in here, what will they think is in here? The participant was considered not to have followed the task (scored as lost) if they

Empathy and theory of mind in offenders with ID had taken his sandwich. Only if they omitted the critical element of the man looking in his own bag were they scored as having no theory of mind. Second Order Location Change (Ice Cream Van) Task (Perner & Wimmer, 1985). This task was selected because it has been commonly used in existing research (e.g., Baron Cohen, 1989; Dahlgren, Dahlgren Sandberg, & Hjelmquist, 2003; Doody, Gotz, Johnstone, Frith, & Owens, 1998). Although reliability and validity data could not be obtained for this specific test, good test-retest reliability and internal consistency have recently been demonstrated in a number of tests based on the same premise (Hughes, Adlam, Happe , Jackson, Taylor, & Caspi, 2000). The test involves a plan view of a town, including a park, church, two houses and roads between them (Figure 1). Two small figures and an ice cream van are initially placed in the park area. An interaction is acted out whereby the ice cream man says he will be in the park all day and the male character goes home to get some money, but the ice cream man changes his mind and goes to the church, after telling this to the female character. Next, the female character wants to see

87

her friend but when she knocks on the door, she is told that he has gone to buy ice cream. Participants are then asked where she will go to find her friend. The participant either answers The park (correct) or The church (incorrect). Again, failures were classified in terms of general failure or theory of mind failure. Ethics Ethical approval for this study was obtained from the local NHS Research Ethics Committees of Sheffield Teaching Hospitals, Southern Derbyshire, Hull and East Riding, Doncaster and Barnsley. Informed consent was obtained from participants prior to the implementation of all procedures. As agreed in NHS ethics approval, participants were read an information sheet and asked to sign a consent form if they wished to take part. If IQ data were not available, the Wechsler Abbreviated Scale of Intelligence (WASI) was used. The theory of mind and empathy tests were then given in random order and participants were given a chance to discuss the process. The authors have no financial interest in publication of this paper.

Figure 1. Diagram used in second order theory of mind (Ice Cream Van) task.

88

T. Proctor & N. Beail Test of Emotional Perception (TEP) The TEP results for each emotion (happy, sad or angry) were considered separately. Non-continuous data provided by the variables in the TEP did not meet the criteria for parametric analysis. However, a randomisation ANOVA (RANOVA) is a powerful, non-parametric equivalent of ANOVA in which significance is assessed in terms of random permutations rather than an F distribution. The only requirement for a RANOVA is that observations within each sample are independent of one another. All RANOVA tests were carried out using 5000 permutations of the data. Means, standard deviations and RANOVA F values are summarised in Table 2 (none of the interactions was significant). For emotion recognition, a significant effect of offender status (offenders performed better) and type of emotion was found. Post-hoc RANOVA tests (with Bonferroni correction) showed that happy was correctly identified significantly more often than sad or angry (F(1, 49)513.72, p5.001, and F(1, 49)529.63, p5.001 respectively). There was also a significant effect of offender status (offenders required fewer prompts) and type of emotion on number of prompts required. However as before, due to this effect only just reaching significance, post-hoc RANOVA testing was performed, and found that no significant differences. For the picture selection task (emotional response prediction), there was a significant effect of emotion but not offender status. Post-hoc RANOVA tests found the correct response was chosen for the happy condition significantly more often than for the sad or angry condition (F(1, 49)526.64, p5.001, F(1, 49)518.66, p5.001 respectively). For response latency, there was a significant effect of type of emotion and offender status (offenders took significantly longer). However, post-hoc RANOVA testing indicated that none of the individual emotion comparisons were significant, which is not surprising given that the overall result only just achieved significance. Finally, participants own emotional response expression if the vignette situations hypothetically happened to a friend were compared. As too many cells had an expected frequency of less than 5, the data were condensed in order to fit a 262 contingency table (empathic vs. non-empathic). Fishers Exact Test showed that only the happy condition showed a significant difference between offenders and non-offenders (p5.003), with significantly more offenders showing empathic responses

Results Interpersonal Reactivity Index Group comparisons on the four IRI subscales (perspective taking, fantasy, empathic concern and personal distress) were conducted separately using individual analyses of variance (log transformation was required for the Personal distress subscale). No significant differences were found between offenders and non-offenders on any IRI subscale. First order theory of mind (ToM) In line with existing research (Charman & Campbell, 2002; Frith, Happe , & Siddons, 1994; Jervis & Baker, 2004), since both tests are thought to measure first order theory of mind, results of the Deceptive Box (Smarties) and Location Change (Sally-Anne) tasks were combined such that participants were judged to have passed only if they passed both tests. One participant from the offender group and three from the non-offender group failed to understand one or other of the tasks and were excluded. No significant differences were found between the groups. However, 82% of non-offenders and 96% of offenders passed both tests, suggesting a ceiling effect. Second order theory of mind (ToM) One offender and two non-offenders were excluded due to being unable to follow the Second Order Location Change (Ice Cream Van) task. A significant difference was found between the two groups, with offenders performing significantly better on the Ice Cream Van task than non-offenders (x2(1)517.08, p5.001). In order to examine the possible role of IQ in this difference, the IQ scores of those who passed and failed were compared using another Mann-Whitney test. There was no significant difference between the IQ scores of the two groups (U5181, p5.06), however this approached significance, so to clarify the relative contributions of IQ and offender status, sequential logistic regression was carried out. The predictor added in block 1 was IQ, which showed a significant effect (x2(1)58.591, p5.003). Offender status was added in block 2 and also showed a significant effect over and above that of IQ (x2(1)514.971, p50.001), indicating that offender status had a significant effect on performance for the Ice Cream Van task, even when the effect of IQ was taken into account. Sample size is slightly small here, but the significant result suggests power was adequate.

Empathy and theory of mind in offenders with ID

Table 2. TEP mean scores (SD) Type of emotion Variable Mean number of correctly identified emotions (SD) Mean number of correctly selected emotional responses (SD) Mean response time (SD) Group Offenders Non-offenders Offenders Non-offenders Offenders Non-offenders Mean number of prompts required (SD) Offenders Non-offenders Happy 2.00 (0.00) 1.75 (0.61) 1.58 (0.70) 1.33 (0.76) 19.35 (16.80) 15.75 (8.82) 1.08 (1.60) 2.38 (2.83) Sad 1.65 (0.63) 1.33 (0.70) 0.85 (0.88) 0.54 (0.66) 24.23 (14.33) 20.96 (17.58) 1.85 (2.15) 2.83 (2.96) Angry 1.38 (0.75) 1.13 (0.85) 0.96 (0.72) 0.79 (0.72) 24.77 (14.11) 15.54 (10.05) 2.12 (2.53) 2.12 (2.96) RANOVA results

89

Offender status: F(1,49)56.17, p5.015 Type of emotion: F(1,49)512.28, p5.001 Offender status: F(1,49)53.70, p5.08 Type of emotion: F(1,49)514.53, p5.001 Offender status: F(1,49)55.01, p5.029 Type of emotion: F(1,49)53.19, p5.049 Offender status: F(1,49)55.01, p5.03 Type of emotion: F(1,49)53.19, p5.049

than non-offenders (p5.496 for sad and p5.389 for angry respectively). Empathy of theory of mind passers and failers T-tests were performed to compare the IRI empathy scores of those who passed and failed the second order theory of mind test. A significant difference was found for the IRI fantasy subscale only. Those who passed the second order theory of mind test scored significantly lower on the fantasy subscale of the IRI than those who failed (t(1, 42)52.080, p5.044). Similar comparisons were carried out on the TEP scores. Here, those who passed the second order theory of mind test selected the correct picture for angry significantly more often than those who failed (U5153.5, p5.042). When asked how they might feel in response to anothers sadness, those who failed the second order theory of mind test responded with concern significantly more often than those who passed (U5162.0, p5.023). All findings are summarised in Table 3. Discussion The simplest first order theory of mind (Smarties Box and Sally-Anne) tasks found a ceiling effect for both offenders and non-offenders with ID. This was somewhat surprising given that a meta-analysis suggested that adults with ID often fail these tasks (Yirmiya et al., 1998). The discrepancy may be explained by the new video presentation of the SallyAnne task, which was designed to reduce failures

due to lack of general understanding rather than a lack of theory of mind. The second order theory of mind (Ice Cream Van) task found that offenders with ID performed significantly better than nonoffenders with ID, even when the effect of IQ was taken into account. This is contrary to the hypothesis that predicts non-offenders would be better at this task. Taken together with the first order theory of mind findings, these results suggest that although both groups were able to understand simple situations requiring theory of mind, the non-offenders could not follow a more complex situation, and this was not only related to IQ. These findings appear to clarify the existing literature (which has found that people with ID often fail theory of mind tasks) by making this finding more specific: when theory of mind tasks are presented in a more user-friendly way, adults with ID can successfully perform them. It remains to be seen whether the same can be achieved with an alternative presentation of the more complex second order tasks. The more objective of the two empathy measures, the Test of Emotional Perception (Moffatt et al., 1995) did find some differences between the two groups. This measure appears particularly suitable for use with people with ID, since it uses videotapes of emotion being expressed in a natural way rather than photographs, pictures or words. Offenders with ID performed significantly better than non-offenders with ID on emotion recognition, and required significantly fewer prompts than non-offenders to provide an emotion descriptor. This fits with the above finding that offenders were able to recognise and note the importance of emotions significantly

90

T. Proctor & N. Beail

Table 3. Summary of offender versus non-offender findings Test First order theory of mind (Smarties Box and Sally-Anne tasks) Second order theory of mind (Ice Cream Van Task) Interpersonal Reactivity Index (IRI) Perspective taking Fantasy Empathic concern Personal distress Test of Emotional Perception (TEP) Emotion recognition Emotional response prediction Response prediction (time) Number of prompts required to elicit emotion Own emotional response Finding Ceiling effect (no group differences) Offenders significantly better than non-offenders

No No No No

group group group group

differences differences differences differences

(Note: Bonferroni correction applied) Offenders significantly better than non-offenders No group differences Offenders took significantly longer to select a picture Offenders required significantly fewer prompts Offenders gave an empathic response significantly more often than non-offenders for the happy condition

more successfully than non-offenders. These findings strongly suggest that offenders have superior abilities in this area. Although the above differences between groups were not supported by the results on the Interpersonal Reactivity Index (Davis, 1983), the format of this questionnaire had been adapted for use with people with ID and therefore no reliability or validity data were available. It is therefore possible that this adapted version failed to measure the desired constructs. Further research is required to investigate the properties of this modified version before these findings can be fully interpreted. In addition, the questionnaire attempts to assess its subcomponents by verbally asking participants how they would react in certain situations, and therefore does not provide a concrete measure. While not ideal, such adaptations to existing measures are necessary to facilitate research with people with ID, and the resulting data provide valuable clues about the issues under scrutiny, albeit requiring further clarification once the tools have been validated. When participants were asked to predict the emotional response of the other person in the TEP vignettes by selecting an appropriate photograph, offenders and non-offenders with ID did not differ, although the offenders took significantly longer to make their choice. This pattern of results is difficult to explain, as it apparently suggests that the offenders processed the information more slowly but just as accurately, or possibly that the nonoffenders, unable to carry out the necessary reasoning, made quick intuitive judgements. Before attempting to interpret this finding, more research

is needed to clarify whether it is a robust finding and if so, whether it holds true that offenders process all aspects of emotional stimuli more slowly than non-offenders, despite matching or beating them on accuracy. In addition, the a priori power calculation suggested that 30 participants would be needed per group for this subtest, from which it must be concluded that with only 25 per group, there may have been insufficient power to detect a difference here. For the moment, this particular finding therefore sheds little light on the research question. Finally, when the participants were asked how they would feel if one of their friends received a letter or phone call like the one they saw in the video scenario, offenders gave an empathic response significantly more often than non-offenders for the happy condition only, with no other group differences. This finding is also difficult to interpret without further research. A single group difference in this comparison is unconvincing, but certainly warrants further research. For now, all that can be said is that offenders and non-offenders with ID do not appear to differ in the way they see themselves as responding to emotional situations. It should be noted that this is the least objective part of the test of emotional perception, since it asks participants how they might react if a similar hypothetical situation happened to one of their friends. A person with an ID might be expected to find this very difficult to comprehend without being given concrete details about the situation. The complexity of these results has been deliberately retained in order to draw out issues of oversimplification in the existing literature, which

Empathy and theory of mind in offenders with ID has often treated the concept of empathy as a unitary model. Where differences have been found between offenders and non-offenders with ID, the offenders appear to be the more skilled in certain areas of theory of mind and empathy, such as emotion recognition and salience of emotions (i.e., they required fewer prompts to mention these), which correspond to the perceptual and cognitive aspects of empathy. When an attempt was made to consider the relationship between theory of mind task performance and empathy task performance, it was found that those who passed the second order theory of mind tests scored lower on the fantasy subscale of the IRI. This is a counterintuitive finding which suggests that those with good theory of mind abilities are less skilled at the fantasy or inner representational aspects of empathy. However, as this finding only just reached significance among a large number of insignificant results, it could represent a type I error and it would therefore be difficult to use it to draw conclusions. The large number of insignificant results strongly suggests that theory of mind performance does not affect empathic ability as measured by the IRI. Further research is required to clarify this issue, and future studies should be specifically designed to do this. In the spirit of thorough investigation, results for the different emotions were compared, as well as scores for offenders versus non-offenders. This is because it is feasible that some emotions might be recognised or otherwise processed more easily than others. In fact, the only differences found were in emotion recognition and selection of an appropriate emotional response for another (which in itself involves emotion recognition), for which both groups managed happiness better than sadness or anger. This supports the idea that not only should empathy and theory of mind be measured by their individual components, but also that skills may vary across emotions, meaning that performance for different emotions should be measured separately where possible, as has been the case in this study. Given the dearth of literature on the topic of empathy for individuals with ID, it is not possible to directly compare the findings of the current study with previous research. A meta-analysis by Jolliffe & Farrington (2004) found that when intelligence was controlled for, the otherwise significant effect of affective empathy on offending disappeared, but the significant effect of cognitive empathy remained. This might theoretically be part of a wider pattern of certain aspects of empathy decreasing with intelligence to a certain point, before increasing again. However, the studies included in that meta-analysis would have been based on participants without ID,

91

making any direct comparisons impossible. Specific research findings comparing empathy in offenders and non-offenders without ID has suggested that emotion recognition may be hindered in sex offenders (Covell & Scalora, 2002; Marshall et al., 1995), which would appear to contradict the present results. However it is possible that both groups of participants in the current study would demonstrate poorer emotion recognition than non-offenders without ID, which would therefore not oppose the previous finding. In this respect, it would be useful to conduct further research including offenders and non-offenders both with and without ID, which was beyond the scope of this study. Another possibility is that it is only sex offenders who have inferior emotion recognition skills; however the current study does not have sufficient data to determine whether there were different abilities in various types of offender. To conclude, doubt has been cast on the merit of empathy training in offenders with intellectual disability. It has been shown that evidence for the majority of interventions for offenders with ID is based on small-scale studies rather than properly designed and controlled research (Barron, Hassiotis, & Banes, 2002). Those studies that do exist appear to support the efficacy of behavioural interventions (although these have drawbacks relating to their intrusiveness, and may not be useful for offenders with milder ID) and cognitive behavioural interventions (Taylor, 2002). In fact, although empathy training appears to be commonly prescribed for offenders without ID, the literature does not appear to support the idea that this training should be applied to those who do have intellectual disabilities (Barron et al., 2002). It is possible that, as in other services for people with ID, services for these offenders tend to focus on anger and aggression. If so, this may for once have worked in their favour, since in this instance there is little evidence for the efficacy of empathy-based treatments. In general, these findings underline the need for thorough assessment and formulation of each individuals treatment needs (Taylor, 2002), to avoid inappropriate and possibly detrimental treatments. It is clear that the intellectual disability of these offenders sets them apart from other groups of offenders, and that there is an obligation to investigate the needs of this vulnerable group separately and sensitively. Appropriate measures, which are derived from or in keeping with a modern componential definition of empathy and/or theory of mind, are therefore required for both research and clinical purposes. This must be the starting point for future work in this under-researched area.

92

T. Proctor & N. Beail

non-spectrum disorders on emotion recognition tests? European Child and Adolescent Psychiatry, 10, 105116. Eisenberg, N., & Strayer, J. (Eds.). (1987). Empathy and its development. Cambridge: Cambridge University Press. Frankish, P. (Guest Ed.). (2001). People with learning disabilities who offend [Special issue]. British Journal of Forensic Practice, 3(1). Fraser, W. I., & Taylor, J. L. (Guest Eds.) (2002). Forensic learning difficulties: The evidence base [Special issue]. Journal of Intellectual Disability Research, 46 (Suppl.1). Frith, U., Happe , F. G. E., & Siddons, F. (1994). Autism and theory of mind in everyday life. Social Development, 3, 107124. Froman, R. D., & Peloquin, S. M. (2001). Rethinking the use of the Hogan Empathy Scale: A critical psychometric analysis. American Journal of Occupational Therapy, 55, 566572. Geer, J. H., Estupinan, L. A., & Manguno-Mire, G. M. (2000). Empathy, social skills and other relevant cognitive processes in rapists and child molesters. Aggression and Violent Behaviour, 5, 99126. Goldstein, H., & Higgins DAlessandro, A. (2001). Empathy and attachment in relation to violent vs. non-violent offence history among jail inmates. Journal of Offender Rehabilitation, 32, 3153. Gopnick, A., & Astington, J. W. (1988). Childrens understanding of representational change and its relation to the understanding of false belief and the appearance-reality distinction. Child Development, 59, 2637. Hogan, R. (1969). Development of an empathy scale. Journal of Consulting and Clinical Psychology, 33, 307316. Hughes, C., Adlam, A., Happe , F. G. E., Jackson, J., Taylor, A., & Caspi, A. (2000). Good test-retest reliability for standard and advanced false-belief tasks across a wide range of abilities. Journal of Child Psychology and Psychiatry, 41, 483490. Jervis, N., & Baker, M. (2004). Clinical and research implications of an investigation into theory of mind (ToM) task performance in children and adults with non-specific intellectual disabilities. Journal of Applied Research in Intellectual Disabilities, 17, 4957. Johnson, J. A., Cheek, J. M., & Smither, R. (1983). The structure of empathy. Journal of Personality and Social Psychology, 45, 12991312. Jolliffe, D., & Farrington, D. P. (2004). Empathy and offending: A systematic review and meta-analysis. Aggression and Violent Behaviour, 9, 441476. Kasari, C., Freeman, S. F. N., & Bass, W. (2003). Empathy and response to distress in children with Down syndrome. Journal of Child Psychology and Psychiatry and Allied Disciplines, 44, 424431. Keenan, T., & Ward, T. (2000). A theory of mind perspective on cognitive, affective, and intimacy deficits in child sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 12, 4960. Knopp, F. H., Freeman-Longo, R. E., & Stevenson, W. (1992). Nationwide survey of juvenile and adult sex-offender treatment programs. Orwell, VT: Safer Society Press. Lindsay, W. R. (Guest Ed.) (2002a). Offenders with intellectual disability [Special issue]. Journal of Applied Research in Intellectual Disabilities, 15(2). Lindsay, W. R. (2002b). Research and literature on sex offenders with intellectual and developmental disabilities. Journal of Intellectual Disability Research, 46(Suppl. 1), 7485. Marshall, W. L., Hudson, S. M., Jones, R., & Fernandez, Y. M. (1995). Empathy in sex offenders. Clinical Psychology Review, 15, 99113.

Author note This study was prepared as a submission for the Doctorate in Clinical Psychology at the University of Sheffield, UK. Note
1 Due to the nature of the criminal justice system, offenders with ID are likely to be incarcerated outside the normal prison system and without a formal conviction. Not all residents in these environments have committed criminal offences, which is why the stated check was carried out. This should result in a group that approximates as closely as possible an offender group with ID, comparable to convicted offenders without ID.

References
Baron Cohen, S. (1989). The autistic childs theory of mind: A case of specific developmental delay. Journal of Child Psychology and Psychiatry, 30, 285298. Barron, P., Hassiotis, A., & Banes, J. (2002). Offenders with intellectual disability: The size of the problem and therapeutic outcomes. Journal of Intellectual Disability Research, 46, 454463. Bovasso, G. B., Alterman, A. I., Cacciola, J. S., & Rutherford, M. J. (2002). The prediction of violent and non-violent criminal behavior in a methadone maintenance population. Journal of Personality Disorders, 16, 360373. Bowler, D. M., & Strom, E. (1998). Elicitation of firstorder theory of mind in children with autism. Autism, 2, 3344. Burke, D. M. (2001). Empathy in sexually offending and nonoffending adolescent males. Journal of Interpersonal Violence, 16, 222233. Bush, C. A., Mullis, R. L., & Mullis, A. K. (2000). Differences in empathy between offender and non-offender youth. Journal of Youth and Adolescence, 29, 467478. Charman, T., & Baron Cohen, S. (1995). Understanding photos, models and beliefs: A test of the modularity thesis of theory of mind. Cognitive Development, 10, 287298. Charman, T., & Campbell, A. (1997). Reliability of theory of mind task performance by individuals with a learning disability: A research note. Journal of Child Psychology and Psychiatry and Allied Disciplines, 38, 725730. Charman, T., & Campbell, A. (2002). Theory of mind and social competence in individuals with a mental handicap. Journal of Developmental and Physical Disability, 14, 263276. Covell, C. N., & Scalora, M. J. (2002). Empathic deficits in sexual offenders: An integration of affective, social, and cognitive constructs. Aggression and Violent Behavior, 7, 251270. Dahlgren, S., Dahlgren Sandberg, A., & Hjelmquist, E. (2003). The non-specificity of theory of mind deficits: Evidence from children with communicative disabilities. European Journal of Cognitive Psychology, 15, 129155. Davis, M. (1983). Measuring individual differences in empathy: Evidence for a multidimensional approach. Journal of Personality and Social Psychology, 44, 113126. Doody, G. A., Gotz, M., Johnstone, E. C., Frith, C. D., & Owens, D. G. C. (1998). Theory of mind and psychoses. Psychological Medicine, 28, 397405. Dyck, M. J., Ferguson, K., & Shochet, I. M. (2001). Do autism spectrum disorders differ from each other and from

Empathy and theory of mind in offenders with ID

McFall, R. M. (1990). The enhancement of social skills: An information-processing analysis. In W. L. Marshall (Ed.), Handbook of sexual assault: Issues, theories, and treatment of the offender (pp. 311330). New York, NY: Plenum Press. Mehrabian, A., & Epstein, N. (1972). A measure of emotional empathy. Journal of Personality, 40, 525543. Moffatt, C. W., Hanley-Maxwell, C., & Donnellan, A. M. (1995). Discrimination of emotion, affective perspective-taking and empathy in individuals with intellectual disability. Education and Training in Intellectual Disability and Developmental Disabilities, 30, 7685. Perner, J., Frith, U., Leslie, A., & Leekam, S. (1989). Exploration of the autistic childs theory of mind: Knowledge, belief and communication. Child Development, 60, 689700. Perner, J., & Wimmer, H. (1985). John thinks that Mary thinks that: Attribution of second-order beliefs by 5- to 10-year-old children. Journal of Experimental Child Psychology, 39, 437471. Prior, M., Dahlstrom, B., & Squires, T. (1990). Autistic childrens knowledge of thinking and feeling states in other people. Journal of Child Psychology and Psychiatry, 31, 587601. Reed, T. (1994). Performance of autistic and control participants on three cognitive perspective-taking tasks. Journal of Autism and Developmental Disorders, 24, 5366. Rose, J., Jenkins, R., OConnor, C., Jones, C., & Felce, D. (2002). A group treatment for men with intellectual disabilities who sexually offend or abuse. Journal of Applied Research in Intellectual Disabilities, 15, 138150. Smallbone, S. W., Wheaton, J., & Hourigan, D. (2003). Trait empathy and criminal versatility in sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 15, 4960. Sullivan, K., & Tager-Flusberg, H. (1999). Second-order belief attribution in Williams syndrome: Intact or impaired? American Journal on Mental Retardation, 104, 523532.

93

Tager-Flusberg, H., & Sullivan, K. (2000). A componential view of theory of mind: Evidence from Williams syndrome. Cognition, 76, 5989. Taylor, J. L. (2002). A review of the assessment and treatment of anger and aggression in offenders with intellectual disability. Journal of Intellectual Disability Research, 46(Suppl. 1), 5773. Ward, T., Keenan, T., & Hudson, S. M. (2000). Understanding cognitive, affective and intimacy deficits in sexual offenders: A developmental perspective. Aggression & Violent Behavior, 5, 4162. Wechsler, D. (1999). Wechsler Abbreviated Scale of Intelligence (WASI). The Psychological Corporation. Wimmer, H., & Perner, J. (1983). Beliefs about belief: Representation and constraining function of wrong beliefs in young childrens understanding of deception. Cognition, 13, 103128. Yirmiya, N., Erel, O., Shaked, M., & Solomonica-Levi, D. (1998). Meta-analyses comparing theory of mind abilities of individuals with autism, individuals with intellectual disability and normally developing individuals. Psychological Bulletin, 124, 283307. Yirmiya, N., & Shulman, C. (1996). Seriation, conservation, and theory of mind abilities in individuals with autism, individuals with intellectual disability, and normally developing children. Child Development, 67, 20452059. Yirmiya, N., Solomonica-Levi, D., Shulman, C., & Pilowsky, T. (1996). Theory of mind abilities in individuals with autism, Down syndrome, and intellectual disability of unknown etiology: The role of age and intelligence. Journal of Child Psychology and Psychiatry and Allied Disciplines, 37, 10031014. Zelazo, P. D., Burack, J. A., Benedetto, E., & Frye, D. (1996). Theory of mind and rule use in individuals with Downs syndrome: A test of the uniqueness and specificity claims. Journal of Child Psychology and Psychiatry, 37, 479484.