Attentional bias toward negative and positive pictorial stimuli and its relationship with distorted cognitions, empathy, and moral reasoning among men with intellectual disabilities who have committed crimes

The aims of this study were to examine: (a) whether men with intellectual disabilities who have a history of criminal offending attend to affective pictorial stimuli in a biased manner, and (b) whether there is a relationship between an affective attentional bias and offense ‐ supportive cognitions, empathy, and moral reasoning. Forty ‐ six men with intellectual disabilities who had a documented history of criminal offending, and 51 men who also had intellectual disabilities, but no such history, were recruited and asked to complete a computer ‐ based dot ‐ probe task using affective pictorial stimuli with randomization, along with measures of distorted cognitions, empathy, and moral reasoning. Those with a history of criminal offending endorsed significantly more offense ‐ supportive cognitions, had significantly lower general empathy, and more “ mature ” moral reasoning, as well as a significant attentional bias toward affective pictorial stimuli. Attentional bias significantly predicted offense ‐ supportive cognitions, and vice versa, having controlled for offense history, and Full ‐ Scale IQ, but this was not the case for empathy or moral reasoning. While the findings require replication, interventions that aim to modify attention bias with this population should be tested.

, affective empathy (Gibbs, 2013;Hoffman, 2000;Kohlberg, 1984), and somatic markers (Baird, 2008;Taber-Thomas & Tranel, 2012). Garrigan et al. (2018) combined these and other factors together within a six-step framework that has social information processing at its core, but incorporated developmental factors that change over time as a consequence of maturation and developmental experiences, inclusive of socialization and related factors (e.g., socioeconomic status and culture) as well as psychological theories about moral development, social-perspective taking and empathy. The aim of this framework was to attempt to provide an explanation as to how moral decisions are made, while further considering the processes that impact decision-making and develop over time via socialization and maturation. The first step within their framework was labeled the "encoding of cues," which relies upon the allocation of attention and recognition of situational cues, which is affected by the ability to recognize emotions and empathic responsiveness. Attentional ability is considered an important construct related to moral reasoning and development (Gibbs, 2013). Garrigan et al. (2018) hypothesized that shifting attention and related executive functions, as well as attentional bias, will influence what information is encoded within a social situation, thus influencing subsequent affective, cognitive, and behavioral responses within the latter steps within their framework.
The remaining steps in the framework are related to those outlined by Dodge (1994, 1996) and are interpretation of cues, clarification of goals, response access or construction, moral response decision, and behavioral enactment. These latter steps involve processing, which is triggered by the encoding of cues.
However, encoding is also impacted by top-down processing as those with a moral developmental delay and/or difficulty with cognitive or affective empathy may encode social information differently. Biased encoding may develop as a consequence of socialization and maturation, inclusive of the maturation of brain regions, which may be different for those who have conditions, which are inherently characterized by atypical development, such as intellectual disabilities (Garrigan et al., 2018).
Attentional bias and its relationship to other psychological constructs and behavior has been investigated within numerous studies, especially with those who have anxiety disorders. Van Bockstaele et al. (2014) reviewed studies examining the role of attentional bias in our understanding of anxiety and fear and concluded that bias is associated with anxiety and fear, and the relationship may be bidirectional. They also noted that changing attentional bias changes fear and anxiety. While the findings have not always been consistent, attentional bias has also been shown to be related to eating disorders (Faunce, 2002), addiction (Field & Cox, 2008), depression (Peckham, McHugh, & Otto, 2010), psychosis (Moritz & Laudan, 2007), and sleep disorders (Harris et al., 2015).
For those with a history of committing crimes, there is evidence that those with antisocial personality disorder have a bias toward violent words (Domes, Mense, Vohs, & Habermeyer, 2013); men and women with either a history of violent or non-violent crimes have an attentional bias toward violent words, with those with the history of violent crime having the most marked bias (Smith & Waterman, 2003), sexual offenders have a bias toward sexual words (Price & Karl Hanson, 2007;Smith & Waterman, 2004), and images (Ciardha & Gormley, 2012); men with a history of perpetrating domestic violence have a bias toward aggressive words (Chan, Raine, & Lee, 2010), and teenage fire-setters have an attentional bias toward fire-related pictures (Gallagher-Duffy, MacKay, Duffy, Sullivan- Thomas, & Peterson-Badali, 2009). Kimonis, Graham, and Cauffman (2018) demonstrated that attention to affective pictures among teenage boys with a history of violent crime was predicted by callous and unemotional or uncaring traits, which was moderated by the severity of aggression, a finding reported in previous studies (Kimonis, Frick, Fazekas, & Loney, 2006;Kimonis, Frick, Munoz, & Aucoin, 2008). Edalati, Walsh, and Kosson (2016) reported that convicted offenders scoring low on a measure of psychopathy had an attentional bias away from affective faces compared to those scoring high, but the difference was not statistically significant; these studies suggest that attention bias may be affected by personality, affect, and other related constructs as outlined by Garrigan et al. (2018).
To further investigate attentional bias among those with a history of criminal offending, and whether such a bias is related to additional psychological constructs, we recruited men with intellectual disabilities who either did or did not have a history of criminal offending. We specifically recruited a sample of men with intellectual disabilities because they are experiencing atypical development and are a markedly vulnerable population. We invited participants to take part in a dot-probe task using affective images to examine attentional bias and to complete questionnaires about empathy, offense-supportive beliefs, and moral reasoning. The aims of this study were twofold: (a) to investigate whether men with intellectual disabilities who have a history of criminal offending attend to affective pictorial stimuli in a biased manner, and (b) whether there is a relationship between an affective attentional bias and offense-supportive cognitions, empathy, and moral reasoning. We specifically hypothesized that (a) men with a history of criminal offenses will attend to affective images more rapidly in comparison to men with no such history, and (b) such attention biases will predict offense-supportive cognitions, empathy, and moral reasoning and vice versa.

| Participants
Forty-six men with intellectual disabilities who had a documented history of criminal offending behavior, and 51 men with intellectual disabilities and no known history of engaging in criminal offending behavior were recruited and invited to take part in this study. Those with a history of criminal offending were recruited from secure inpatient services within eastern England, while those without such a history were recruited from the community. Those with a history of crime were significantly younger, 95% bias-corrected-and-accelerated confidence intervals-BC a CI [−12.98, −2.67], Adj. R 2 = .07, and had a significantly higher Full-Scale IQ, 95% BC a CI [0.59, 4.40], Adj. R 2 = .05.
Data collected about mental and physical health comorbidity are found in Table 1. Those with a history of committing criminal offenses were more likely to have a recorded diagnosis of schizophrenia, χ 2 (1) = 5.98, p = .01, or personality disorder χ 2 (1) = 5.98, p = .01, while those with no such history were more likely to have difficulties with hypertension,4.65, p = .03.

| Eligibility criteria
Participants were considered eligible to take part in this study if they were (a) a man aged between 18 and 65 years, (b) who had a mild intellectual disability, (c) with the capacity to give or withhold informed consent to take part in this study study, and (d) successfully completed a practice block on the dot-probe task defined as making no more than one error within 10 trials. Participants with a history of offending behaviors were only included if they had a history of committing an indictable offense, rather than a summary offense, or an "either-way" offense (NB: an either-way offense is one which can be heard in either a Magistrate's of Crown Court). This means that the participants included within this study had committed serious offenses that can only be tried by a Crown court within England and Wales. This includes offenses such as those involving violence (e.g., murder, manslaughter, wounding), sexual offenses, burglary, robbery, theft, criminal damage (e.g., arson), drug offenses, and kidnapping, among others. A full breakdown of the most recent indictable offense for the men with intellectual disabilities who had a history of criminal offending is found in Table 2a and was generated through both selfreport and checking health records. Participants were excluded if they (a) had a known history of acquired brain injury or a diagnosis of dementia, (b) were a woman, or (c) were unable to speak English.
Women were excluded from this study for two reasons: (a) there is some evidence that women and men may score differently on measures of related constructs, such as empathy (Baron-Cohen & Wheelright, 2004); and (b) the population of offenders with intellectual disabilities within secure services in the United Kingdom are predominantly men. The sample of participants in this study have taken part in a previous study (Daniel, Sadek, & Langdon, 2018).

| Design and procedure
Utilizing a simple between-groups design, participants were initially invited to complete an assessment of their Full-Scale IQ using a twosubtest Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler, 1999) when no estimate of Full-Scale IQ was available within their records that had been completed within the last 3 years. Where a previous estimate was available, this data was collected. Seventythree percent of participants completed the WASI. Following this, participants were invited to complete a dot-probe task, following by measures of empathy, distorted cognitions, and moral reasoning that were read aloud. For the measures of empathy and distorted cognitions, which required a forced-choice response, an analogue scale was used to help participants respond.

| Measures
Dot-probe task A dot-probe task using pictorial stimuli was used to examine attentional bias. Following the presentation of a fixation cross on a computer screen, pairs of images were presented followed by the presentation of a dot. Participants were asked to make a timed response by pressing a button to indicate the location of the dot. Pictorial stimuli were chosen for this task as people with intellectual disabilities are likely to have difficulty with reading words. Each single-trial lasted for 11,000 ms and contained (a) a fixation cross presented for 1,000 ms, (b) a fixation cross and two pictures, which were presented directly left and right of the fixation cross for 500 ms, (c) the fixation cross, and a small circle (dot-probe) located either to the left or right of the fixation cross. This remained until a participant made a response, and following this, a new trial started.
Pictures were presented side by side to ensure that they were congruent with the orientation of the buttons on the response box. Neutral-neutral trials are neither congruent nor incongruent.
All participants took part in a short explanation and demonstration of the task with the researcher before taking part in three practice blocks of 10 trials using neutral images that were different from those used within the main task. To progress to the main task, participants had to respond to 9 out of 10 trials correctly in either the first, second, or third practice block. If a participant made two or more errors within a practice block, they completed another practice block. Any participant who did not successfully complete the practice blocks were excluded from the study. Stimuli Dominance ratings were not used in the selection process because of evidence to suggest that they explain a relatively smaller proportion of the variance in the judgments of these stimuli (Coan & Allen, 2007). Considering that some of the participants had a history of sexual offenses, images of children or those of a sexual nature, were excluded. Further, participants had committed offenses against both persons (e.g., wounding) and objects/property (e.g., arson), and therefore images were chosen to ensure that each group (positive, negative, or neutral) of eight pictures contained four pictures of people and four pictures of objects.

| Offense-supportive cognitions
Distorted cognitions, or offense supported cognitions, were measured using the How I Think Questionnaire (HIT; Barriga & Gibbs, 1996;Barriga, Gibbs, Potter, & Liau, 2001) which was adapted for use with adults with intellectual disabilities by Daniel et al. (2018) where it was read to participants. A thorough description of the adaptations can be found elsewhere, where the authors also reported that this measure has excellent internal consistency and good test-retest reliability (Daniel et al., 2018). Hockley & Langdon, 2015;Langdon & Hockley, 2012). In one study, the internal consistency was reported as α = .70 (Daniel et al., 2018).

| Moral reasoning
The Sociomoral Reflection Measure-Short Form (Gibbs, Basinger, & Fuller, 1992) was used to measure moral reasoning. This is a production instrument which has been used with men and women with intellectual disabilities, where it was read to participants, including those with a history of criminal offending (Langdon, Murphy, Clare, & Palmer, 2010;Langdon, Murphy, Clare, Steverson, & Palmer, 2011;McDermott & Langdon, 2016) and it has been shown to possess substantial internal consistent and good test-retest reliability . The measure is comprised of 11 questions pertaining to 7 constructs including Contract, Truth, Affiliation, Life, Law, and Legal Justice.
The instrument was administered as a semistructured interview.
Higher scores are associated with a more "mature" moral develop- Verbatim answers on the Sociomoral Reflection Measure-Short Form were scored, and a second-rater scored 25% of the completed questionnaires; the inter-rater reliability was r i = .96.

| Data preparation and analysis
Data generated using the dot-probe task were inspected and incorrect responses were removed along with reaction time data that were more than two standard deviations above each participant's mean (i.e., outliers). Considering the number of practice trials needed by participants to meet the criterion necessary to complete the dotprobe task, 2.9% of those with an offense history needed two practice trials, while the remaining required only one trial. For those without a history of offending behavior, 15.6% required two practice trials, while the remaining required one trial. The reaction time data generated during the dot-probe task were used to calculate an attention bias score for positive and negative images, as well as overall.
This was done by calculating the mean reaction time for congruent trials (where the dot replaces the affective image), and incongruent trials (where the dot replaces the neutral image) for both positive and negative images. To calculate a bias score, the mean reaction time to congruent trials was subjected from incongruent trials for positive images and then negative images. The sum of mean reaction times to both positive and negative congruent images was subtracted from the sum of mean reaction times to both positive and negative congruent images to calculate a total bias score. A negative score was indicative of a bias away from affective images, while a positive score was indicative of a bias toward affective images.
In terms of statistical analysis, initially, hierarchical linear regression was used to compare those with and without an offending history. Excluding the group comparisons for Full-Scale IQ and age,

| DISCUSSION
The findings from this study indicated that men with intellectual disabilities who have a history of committing crimes have an attentional bias toward negative affective stimuli, while those without this history had an attention bias away from negative affective stimuli.
Considering positive images, those with an offense history had a significantly smaller bias away from positive images than those without this history. Combining bias scores to positive and negative images revealed that those with an offense history had a bias toward affective images. These findings are similar to that reported by others using samples of offenders who do not have intellectual disabilities (Chan et al., 2010;Ciardha & Gormley, 2012;Domes et al., 2013;Gallagher-Duffy et al., 2009;Smith & Waterman, 2003. In addition, and as reported by others, men with intellectual disabilities and a history of offending endorsed more distorted cognitions (Daniel et al., 2018;Langdon & Talbot, 2006;Lindsay & Michie, 2004;Lindsay et al., 2006), and reported less general empathy (Hockley & Langdon, 2015; (Beail & Proctor, 2004;Proctor & Beail, 2007;Ralfs & Beail, 2012) than men without such a history, which is likely related to measurement and sampling.
Further, men with intellectual disabilities who had a history of committing crimes had more "mature" moral reasoning than those without such a history. This finding has been reported in other studies McDermott & Langdon, 2016), and the literature in this area has been previously reviewed , but the finding is inconsistent with T A B L E 3 Distorted cognitions, empathy, and moral reasoning were entered into separate hierarchical regression models to determine whether each predicted attentional bias having controlled for Full-Scale IQ and whether participants did or did not have an offending history increasingly "mature" moral reasoning is associated with less criminal behavior. This is likely to be associated with the lack of individuals within these studies who present with moral reasoning characterized by the earliest developmental stages, which can be seen within some adults with intellectual disabilities. Wilson and Herrnstein (1985) hypothesized that the relationship between crime and intelligence is curvilinear, and within the current study, while those with a history of committing crimes had more "mature" moral reasoning than those without, both groups had a moral developmental stage that has been associated with criminal offending behavior in samples of adolescents (Blasi, 1983;Gregg, Gibbs, & Basinger, 1994;Nelson, Smith, & Dodd, 1990).  previously hypothesized that the relationship between moral development and illegal behavior is curvilinear and this relationship is actually moderated by general intellectual functioning, with those evidencing the highest and lowest moral development stages being the least likely to engage in illegal behavior, while they will also have higher and lower levels of general intellectual functioning. Examining the moral developmental stage of participants across the constructs assessed by the Sociomoral Reflection Measure-Short Form, for those with no history of criminal offending, moral reasoning about the Law was more "immature" and associated with reasoning likely to characterize an earlier stage, and based upon adherence to rules and avoidance of punishment, while this group also had a significantly lower Full-Scale IQ; this may partially explain why these participants did not have a history of criminal offending. This finding has been previously reported McDermott & Langdon, 2016),  discussed this further where they argued that many people with intellectual disabilities may be at lower risk of committing criminal offenses because they present with "immature" moral reasoning. Further evidence of a curvilinear relationship between crime and intelligence was reported by Mears and Cochran (2013) (Garrigan et al., 2018), attributions and beliefs within Step 2: the interpretation of cues, are thought to be affected by encoding within Step 1, and vice versa, and our findings appear consistent with this hypothesized relationship.
Third, Garrigan et al. (2018) suggested the encoding of cues within Step 1 of the framework is hypothesized to be affected by empathic responsiveness, but we did not find a relationship between the measure of empathy used and attentional bias, or vice versa, controlling for offense history. A possible reason for this is that the empathy quotient is a measure of cognitive and affective empathy, and associated social skills (Lawrence, Shaw, Baron-Cohen, & David, 2004), but it may not index empathy responsiveness appropriately as other laboratory-based methods where affect is aroused (Robinson, Roberts, Strayer, & Koopman, 2007). As such, the measure of empathy used may be more distally related to the information processing that occurred during the dot-probe task. Fourth, moral judgments within Step 2 are expected to influence encoding within Step 1, and again, vice versa. However, within the current study moral reasoning did not predict attention bias, nor did attention bias predict moral reasoning after controlling for offense history. As a measure of moral reasoning was used to allow participants to be characterized into stages, rather than make moral judgments about the images they were viewing, a relationship may not have been found. As such, the measure of moral reasoning may also index reasoning that is more distal to attentional bias, while distorted cognitions, are more proximal.
There are some strengths and weaknesses associated with the current study. A convenience sample of participants was used within the study, but men with an offense history had a documented history of serious offending that necessitated detention within hospital, indicating that they had substantial conduct-related problems, which is a strength. We did make use of robust eligibility criteria, and as is the case with the dot-probe task, the order of presentation of pairs of images was randomized. Also, the use of pictures, rather than words, has clear advantages for people with intellectual disabilities who are likely to have difficulties with reading rapidly.
Unfortunately, as this is a sample that is difficult to recruit into research studies, our sample size was too small to allow for more complex analysis. As such, our findings are merely correlational, which is a substantial weakness. Further, it may be the case that difficulties with empathy, distorted cognitions, and moral reasoning may vary with offense type and level of intellectual disability. For example, sexual offenders with intellectual disabilities may have more difficulties with empathy in some contexts, but not other contexts (Hockley & Langdon, 2015;Langdon & Hockley, 2012).
While there is some evidence that affective empathy may relate to violent crime in teenagers (Jolliffe & Farrington, 2007), there is further evidence that cognitive empathy is more likely related to offending than affective empathy (van Langen, Wissink, van Vugt, Van der Stouwe, & Stams, 2014). Jolliffe and Farrington (2004) demonstrated that differences in empathy between offenders and nonoffenders could be accounted for by intelligence and socioeconomic status. Within this study, we made use of a mixed sample of offenders, which varied according to offense type (e.g., violent offenders and sexual offenders), which may have had an impact upon our findings, and all had intellectual disabilities with likely atypical development within the affective domain.
Finally, considering the finding that men with intellectual disabilities who have a history of criminal offending behavior have an attentional bias that is different from those without such a history, the question as to whether procedures to modify such a bias would help address forensic risk needs to be addressed. While there is evidence that attention bias modification training is associated with a small effect size for symptoms of anxiety (Mogoase, David, & Koster, 2014), such procedures may offer an opportunity for engaging people with intellectual disabilities within psychological therapies who may not be able to take part in traditional talking-based psychological interventions to target criminogenic risk Health had no role in the study design, collection, analysis or interpretation of the data, writing the manuscript, or the decision to submit the paper for publication.