Introduction
A cognitive psychological model of psychosis suggests that responses to unusual psychotic-like experiences are cognitively mediated by maladaptive self-schemas and appraisals and also by self-beliefs (Garety, Kuipers, Fowler, Freeman and Bebbington, Reference Garety, Kuipers, Fowler, Freeman and Bebbington2001). Research suggests that core schemas are important in (1) the development and the maintenance of psychosis and (2) the distress associated with it (Beck and Rector, Reference Beck and Rector2003; Garety et al., Reference Garety, Kuipers, Fowler, Freeman and Bebbington2001; Morrison, Reference Morrison2001). Krabbendam et al. Reference Krabbendam, Janssen, Bak, Bijl, de Graaf and van Os(2002) argue that low self-esteem is a risk factor for psychosis, and other researchers implicate it in the development of persecutory delusions (Bentall and Kaney, Reference Bentall and Kaney1996; Bentall, Kinderman and Kaney, Reference Bentall, Kinderman and Kaney1994). Several other studies have also found strong relationships between poor self-esteem and serious mental health problems (Freeman et al., Reference Freeman, Garety, Fowler, Kuipers, Dunn and Bebbington1998; Silverstone and Salsali, Reference Silverstone and Salsali2003; Warner, Taylor, Powers and Hyman, Reference Warner, Taylor, Powers and Hyman1989) and suggest that low self-esteem is related to poorer outcome in people experiencing a first episode of psychosis (FEP; Vracotas, Iyer and Malla, Reference Vracotas, Iyer and Malla2008). In examining self-esteem in psychosis, researchers have used measures like Rosenberg's (Reference Rosenberg1965), which was developed for the general population (e.g. Fowler et al., Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006; Krabbendam et al., Reference Krabbendam, Janssen, Bak, Bijl, de Graaf and van Os2002; Silverstone and Salsali, Reference Silverstone and Salsali2003), and Robson's (Reference Robson1988, Reference Robson1989), which was created with psychiatric populations in mind (e.g. Freeman et al., Reference Freeman, Garety, Fowler, Kuipers, Dunn and Bebbington1998; Hall and Tarrier, Reference Hall and Tarrier2003; Jackson et al., Reference Jackson, Trower, Reid, Smith, Hall and Townend2009). However, these measures do not offer a direct assessment of “the negative self-evaluation construct consistent with contemporary schema constructs as applied to psychosis” (Fowler et al., Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006, pp. 750). In other words, many contemporary models of psychosis symptoms describe a role for the accumulation of ongoing, moment-to-moment negative self-evaluations into negative self-schemas that further impact upon a person's interpretation of events and interactions in social situations; and it is these important negative self-schemas that are not measured by typical self-esteem scales. In addition, it has been argued that some of Rosenberg's self-esteem schedule is outdated and is psychometrically inadequate (Fowler et al., Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006; Keith and Bracken, Reference Keith, Bracken and Bracken1996).
To address the fact that existing self-esteem questionnaires are unable to measure core schemas, Fowler et al. Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth(2006) developed the Brief Core Schema Scales (BCSS). The items on the BCSS operationalize core schemas by specifically addressing a person's positive and negative beliefs about self and others. The researchers found that people who experience chronic psychosis report high levels of negative beliefs about the self and others. However, levels of positive beliefs about the self and others in the psychosis sample were similar to that in a student population (Fowler et al., Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006). A number of subsequent studies have used this measure with both clinical and non-clinical groups (Addington and Tran, Reference Addington and Tran2009; Oliver, O’Connor, Jose, McLachlan and Peters, Reference Oliver, O’Connor, Jose, McLachlan and Peters2011; Stowkowy and Addington, Reference Stowkowy and Addington2012). Specifically, Addington and Tran Reference Addington and Tran(2009) found that the BCSS is appropriate for individuals experiencing an at-risk mental state (ARMS) and that such individuals appear to have high levels of negative schemas (see also Stowkowy and Addington, Reference Stowkowy and Addington2012). As yet, no published studies have compared core schemas in an ARMS group and a psychosis group. In non-clinical populations, negative schemas have been found to predict higher rates of delusional thinking (Oliver et al., Reference Oliver, O’Connor, Jose, McLachlan and Peters2011).
There is a strong emphasis in the current literature in understanding both the development and the maintenance of psychosis, and some researchers suggest that a cognitive style characterized by low self-esteem, neuroticism, worry or depression may increase the risk for developing psychosis (Krabbendam, Myin-Germeys, Bak and van Os, Reference Krabbendam, Myin-Germeys, Bak and van Os2005). Similarly, recent research offers some support to the theory that maladaptive schemas play a role in the onset of psychosis (Stowkowy and Addington, Reference Stowkowy and Addington2012). Therefore, it is important for us to understand what cognitive mechanisms may be shared by or distinguish the ARMS and the psychosis populations in order to improve our understanding of the development of psychosis and to discover targets for psychological interventions.
Aims
The aim of the current study is to investigate and compare core schemas in individuals experiencing FEP, individuals with ARMS, and a help-seeking clinical group who do not have ARMS (HSC), with a non-help-seeking (NH) group who endorse some psychotic-like experiences. We will also examine relationships among psychotic symptoms and core schemas.
We predicted that participants in all three clinical groups would score significantly higher on the negative-other (NO) and negative-self (NS) subscales of the BCSS compared to the NH group. Furthermore, participants in the NH group would score significantly higher than the clinical groups on the positive-other (PO) subscale and the positive-self (PS) subscales of the BCSS. We made no a priori predictions about differences amongst the clinical groups, but these were investigated in exploratory post hoc analyses. Furthermore, we predicted that, in general, the NO and NS subscales would be positively related to psychotic symptoms and the PS and PO subscales would be negatively related to psychotic symptoms.
Method
Participants
FEP group
Participants consisted of 20 help-seeking individuals who were referred to the Early Detection and Intervention Evaluation for people at high-risk of psychosis-2 trial, a multi-site randomized controlled trial of cognitive therapy for the prevention of psychosis (EDIE-2; Morrison et al., Reference Morrison, Stewart, French, Bentall, Birchwood and Byrne2011, Reference Morrison, French, Stewart, Birchwood, Fowler and Gumley2012) and were assessed as being above threshold for ARMS on the Comprehensive Assessment of the At-Risk Mental State (CAARMS; Yung et al., Reference Yung, Yuen, McGorry, Phillips, Kelly and Dell’Olio2005). They had no prior history of psychosis.
ARMS group
This group consisted of 113 help-seeking individuals with no history of psychosis who were referred to EDIE-2 and met the criteria for ARMS on the CAARMS. Of these, 98 participants met the criteria for attenuated psychotic symptoms group, 8 met criteria for the family history group, and 7 met criteria for both attenuated symptoms and family history. No participants met criteria for the brief limited intermittent psychotic symptoms group.
HSC group
Participants consisted of 28 help-seeking individuals with no history of psychosis who were referred to EDIE-2 but were assessed as being below the threshold for ARMS on the CAARMS.
NH group
Thirty student participants who had endorsed schizotypy experiences as operationalized as a score of two on any item of the Community Assessment of Psychic Experiences (CAPE; Stefanis et al., Reference Stefanis, Hanssen, Smirnis, Avramopoulos, Evdokimidis and Stefanis2002) as part of another postgraduate study were asked to participate in the present research. All NH participants were interviewed on the CAARMS, the results of which showed that, theoretically, 22 NH participants were subthreshold for ARMS while 8 met ARMS criteria (for attenuated symptoms). This sample represents a population who have psychotic-like experiences (PLEs) but who do not seek help for those experiences, in contrast to our other participant groups. As this group is similar to the HSC and, to some extent, the ARMS group in terms of their CAARMS data, differences that exist between this group and the others on the measures studied here should hopefully help to explain why some individuals who experience PLEs seek help and others do not.
The groups vary greatly in participant numbers as the ARMS group was recruited through participation in EDIE-2. Participants for the FEP and HSC groups were recruited by convenience sampling individuals who were referred to but did not meet assessment criteria for EDIE-2; also recruitment of these participants began much later. For this reason, as well as resource constraints, the sizes of the FEP, HSC and NH groups are much smaller than the ARMS group.
Measures
The Comprehensive Assessment for At Risk Mental States (CAARMS; Yung et al., Reference Yung, Yuen, McGorry, Phillips, Kelly and Dell’Olio2005)
The CAARMS is a standardized clinical interview that has been developed (1) to determine if an individual meets criteria for having ARMS and (2) to assess psychopathology thought to indicate imminent development of psychosis. The CAARMS has seven categories, each of which consists of multiple sub-scales. For the purpose of this study and of determining if someone meets the ARMS criteria, only the first category, Positive Symptoms, and its four subscales (Unusual Thought Content (e.g. delusional mood), Non-Bizarre Ideas (e.g. specific delusional ideas), Perceptual Abnormalities, and Disorganized Speech) were used (see also Morrison et al., Reference Morrison, Stewart, French, Bentall, Birchwood and Byrne2011, Reference Morrison, French, Stewart, Birchwood, Fowler and Gumley2012). For each subscale, scores for severity of experiences, frequency of experiences, influence of substances on experiences, and distress at symptoms are given. In the current study, scores for severity and distress were used when examining relationships between the CAARMS and the BCSS subscales. Testing of the instrument to date has shown good to excellent validity and reliability and, specifically, good interrater reliability (ICC of overall CAARMS score = .85; Yung et al., Reference Yung, Yuen, McGorry, Phillips, Kelly and Dell’Olio2005).
Brief Core Schema Scales (BCSS; Fowler et al., Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006)
The BCSS is a 24-item self-report assessment that aims to measure beliefs about the self and others in psychosis. Items are rated on a 5-point rating scale (0–4). Four scores, each with six items, are obtained: negative-self (NS), positive-self (PS), negative-other (NO) and positive-other (PO). The BCSS has been described as having good internal consistency (Cronbach's alpha = 0.78–0.88; Fowler et al., Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006).
Procedure
All participants were interviewed on the CAARMS by a trained research assistant working for the EDIE-2 trial. Data from the trial used here were collected over a 2.5 year period across five sites in the UK by 17 different research assistants (including HT). Interrater reliability for the CAARMS was assessed at eight time points during the trial, and the intraclass correlation coefficient (0.90, SD = 0.03) showed good reliability (see Morrison et al., Reference Morrison, Stewart, French, Bentall, Birchwood and Byrne2011, Reference Morrison, French, Stewart, Birchwood, Fowler and Gumley2012). Information on age, gender, years of full-time education, and ethnicity was collected. Participants then completed the BCSS.
Analysis
All analyses were performed in SPSS 19.0 (IBM Corporation, 2010). Non-parametric tests were used where they were appropriate. To test for differences in the distribution of gender and ethnicity, chi-square tests were used. For differences in age and education, the Kruskal-Wallis test was used. For the main hypotheses, a one-way ANOVA was employed for the positive-other scale, while the Kruskal-Wallis test was used for negative-self, positive-self, and negative other. We adjusted for multiple hypothesis testing by applying a Bonferroni correction for the four tests of the BCSS subscales (α = .0125). We followed the Kruskal-Wallis test with pairwise comparisons and used the adjusted significance values. Effect sizes can be interpreted as follows: η2 = 0.01 represents a small effect, η2 = 0.06 represents a medium effect, and η2 = 0.14 represents a large effect. Also, for both Cramér's V and r, an effect size of .1 is small, an effect size of .3 is medium, and an effect size of .5 is large.
For analysing the relationships among psychotic symptoms and core schemas, we correlated the severity and distress scores of each of the four CAARMS subscales with each of the four BCSS subscales. As this was 32 correlations, we present results both at α = .05 and a Bonferroni corrected value of α = .0016. With each CAARMS subscale, we used only those participants whose score was higher than zero on the severity scale, as the distress score is only completed when the severity score is higher than zero.
The participants in the current research participated simultaneously in research presented in Taylor et al. (in press, Reference Taylor, Gumley, Dunn, Stewart, Parker and Morrison2012). In the former they were compared on severity and distress for the subscales of the CAARMS as well as the Beck Depression Inventory for Primary Care (Winter, Steer, Jones-Hicks and Beck, Reference Winter, Steer, Jones-Hicks and Beck1999) and the Social Interaction Anxiety Scale (Mattick and Clarke, Reference Taylor, Stewart, Dunn, Parker, Bentall and Birchwood1998) and in the latter they were compared on the Metacognitions Questionnaire-Revised (Cartwright-Hatton and Wells, Reference Cartwright-Hatton and Wells2004), Interpretations of Voices Inventory (Morrison, Nothard, Bowe and Wells, Reference Morrison, Nothard, Bowe and Wells2004), and the Beliefs about Paranoia Scale (Gumley, Gillan, Morrison and Schwannauer, Reference Gumley, Gillan, Morrison and Schwannauer2011). We attempted to control for family-wise error within this study but did not factor in the error that may result from multiple comparisons reported in the studies mentioned above. Readers should take note of the other comparisons when evaluating our results.
Results
Comparisons on demographic variables
Descriptive statistics for the demographic variables can be found in Table 1. Pearson's chi-square showed that there was a difference in the distribution of gender among the groups (χ2 = 20.854, df = 3, p < .001; Cramér's V = .331). The three clinical groups all had more males than females, while the NH group had more females than males. Because of the very small numbers of some minority ethnic groups, we compared the distribution of White versus Minority Ethnic individuals and found no difference among our participant groups using Fisher's Exact Test (p = .238; Cramér's V = .142). The Kruskal-Wallis test showed a significant difference for age (χ2 = 11.867, df = 3, p < .01; η2 = 0.063). Pairwise comparisons revealed that the NH group was significantly older than the ARMS group (p < .01; r = .267). For education, the Kruskal-Wallis test showed a significant difference (χ2 = 34.380, df = 3, p < .001; η2 = 0.203). The NH group had more years of education than the FEP (p < .001; r = .583), the ARMS (p < .001; r = .475), and the HSC groups (p < .015; r = .411).
Table 1. Demographic information for the four participant groups
Hypothesis testing
Descriptive statistics and a summary of the pairwise comparison results for the BCSS can be found in Table 2. For negative-self, the Kruskal-Wallis test was significant at our corrected level (χ2 = 24.951, df = 3, p < .001; η2 = 0.152). The NH group scored significantly lower than the FEP (p = .005; r = .495), ARMS (p < .001; r = .438), and HSC (p < .05; r = .380) groups. There were no significant differences between the FEP and ARMS (p = 1.000; r = .005), the FEP and HSC (p = 1.000; r = .134), and the ARMS and HSC (p = 1.000; r = .105) groups.
Table 2. Means, standard deviations and medians for the BCSS subscales and pairwise comparison results
An asterisk (*) by the median indicates that the variable was non-normally distributed for that group.
There were also differences for positive-self (χ2 = 26.677, df = 3, p < .001; η2 = 0.164) in which the NH scored significantly higher than the ARMS (p < .001; r = .456) and HSC (p < .005; r = .474) groups. There were no significant differences between the FEP and ARMS (p = .719; r = .148), the FEP and HSC (p = 1.000; r = .152), the FEP and NH (p = .226; r = .306), and the ARMS and HSC (p = 1.000; r = .045) groups.
For negative-other, the Kruskal-Wallis test was significant (χ2 = 28.168, df = 3, p < .001; η2 = 0.176), and pairwise comparisons showed that the NH group scored significantly lower than the FEP (p < .001; r = .650), ARMS (p < .001; r = .433), and HSC (p < .05; r = .375) groups. There were non-significant differences between the FEP and ARMS (p = 1.000; r = .018), the FEP and HSC (p = .373; r = .010), and between the ARMS and HSC (p = 1.000; r = .027) groups. The ANOVA for positive-other was non-significant (F (3, 158) = .202, p = .895; ηp 2 = 0.004).
Descriptive statistics for each of the groups on the CAARMS severity and frequency can be found in Table 3, and results of the correlations between the CAARMS and the BCSS subscales can be found in Table 4. At p < .05, the following relationships were significant: UTC severity with NS (rs = .276), NO (rs = .318); NBI severity with NS (rs = .196), NO (r = .264); NBI distress with NO (rs = .213); PA severity with NO (rs = .209); and DS severity with NS (rs = .196) and NO (rs = .283). Two correlations were significant at our Bonferroni corrected level: NBI distress with NS (rs = .299) and PA distress with NO (rs = .320). There were no relationships between UTC distress and core schemas or between DS distress and core schemas. There were also no relationships between psychotic symptoms and PS or between psychotic symptoms and PO.
Table 3. Means, standard deviations and medians for the participant groups on CAARMS severity and distress
Table 4. Results of correlations among CAARMS and BCSS subscales
* = significant at p < .05; ** = significant at p < .0016
Discussion
It has been suggested that the psychosis prodrome is characterized by low levels of self-esteem and that a cognitive style that includes low self-esteem, worry, depression or neuroticism may increase the risk of developing psychosis (Krabbendam, et al., Reference Krabbendam, Janssen, Bak, Bijl, de Graaf and van Os2002; Krabbendam, Myin-Germeys, Bak et al., Reference Krabbendam, Myin-Germeys, Hanssen, de Graaf, Vollebergh and Bak2005; Krabbendam, Myin-Germeys, Hanssen et al., Reference Krabbendam, Myin-Germeys, Bak and van Os2005). The fact that our clinical groups all scored higher than the NH group on negative-self and that the ARMS and HSC groups (but not the FEP group) scored lower than the NH group on positive-self supports this idea. Further support comes from our previous findings that the clinical groups used in this study were significantly more depressed than the NH group (Taylor et al., in press). Therefore, the current study supports the hypothesis that low self-esteem, negative beliefs about the self (and possibly fewer positive beliefs about the self), and depression are characteristic of prodromal and first-episode psychosis.
Interestingly, the scores for positive-self for the FEP group did not significantly differ from either the other two clinical groups or the NH group. The results of our correlational analysis were consistent; also, there were no relationships between positive-self and the CAARMS subscales. This sort of “middling” score is difficult to interpret but it does seem to be in contrast to their dysfunctional negative views of the self and depression (Taylor et al., in press). Our results suggest there was a small effect for the difference between the FEP and ARMS groups for this variable, with the FEP group experiencing higher positive-self scores Thus, it is possible that some of the FEP sample may have been experiencing some grandiosity, which may be reflected in the higher positive-self scores for that group. This is speculative, but the effect sizes suggest that future studies with greater power may find some interesting differences for positive-self.
Furthermore, our results demonstrate that higher levels of negative beliefs about others can distinguish clinical from non-clinical groups, which supports the idea that negative evaluations and mistrust of others can feed into the development of paranoia or suspiciousness on their own or in combination with negative evaluations of the self (Fowler, Reference Taylor, Stewart, Dunn, Parker, Bentall and Birchwood2000; Trower and Chadwick, Reference Trower and Chadwick1995). Future research could examine the specific relationships between paranoia and negative beliefs about others in both psychotic and ARMS populations, as Fowler et al. Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth(2006) found strong links between negative-other schemas and paranoia in a NH population.
No significant differences were found between the groups for positive beliefs about others, which is in stark contrast to the differences found on the other three subscales. It may be that the other types of schemas measured by the BCSS are better discriminators between groups along the continuum of psychosis as positive-other seems to be weakly related to psychotic phenomena (Addington and Tran, Reference Addington and Tran2009; Fowler et al., Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006). This idea is supported by the lack of significant relationships between positive-other and the CAARMS subscales.
The group differences we found for negative-self and negative-other were reinforced by the discovery of several significant relationships among the CAARMS subscales and the negative-self and negative-other subscales (though only the relationships between distress on non-bizarre ideas (i.e. specific delusional ideas) and negative-self, and between distress on perceptual abnormalities and negative-other remained significant after correcting for multiple testing). It seems that negative schemas, in particular, are associated with a range of positive psychotic symptoms and especially so with distress associated with non-bizarre ideas and perceptual abnormalities. Again, these findings support the idea that prodromal psychosis is characterized by low self-esteem and negative schemas (Krabbendam et al., Reference Krabbendam, Janssen, Bak, Bijl, de Graaf and van Os2002).
When interpreting the results, it is important to remember that there were also differences among the groups in terms of demographics: the clinical groups had proportionately more males than the NH group, the NH group had more years of education than the clinical groups, and the NH group was older than the ARMS group. However, a tendency towards maleness and towards fewer years of education (Kampman et al., Reference Kampman, Kiviniemi, Koivisto, Väänänen, Kilkku and Leinonen2004) is typical of individuals who suffer from or who are at risk of psychosis. Such naturally occurring group differences are often found in clinical research where participants cannot be randomized (Miller and Chapman, Reference Miller and Chapman2001).
Our study attempted to contextualize the experiences of help-seeking clinical groups by comparing them to a non-help-seeking group experiencing PLEs. However, it is possible that recruiting the NH group from a student population meant that our sample was not representative of the population of individuals who experience PLEs but who do not seek help, particularly as we did not inquire about current or past mental health difficulties (including psychosis). Nonetheless, including NH samples like ours is beneficial to the evidence base for the continuum theory of psychosis and to improving our understanding and treatment of psychosis (van ‘t Wout, Aleman, Kessels, Larøi and Kahn, Reference van ‘t Wout, Aleman, Kessels, Larøi and Kahn2004).
Furthermore, although a fairly large sample was recruited for the ARMS group, the other groups were much smaller due to practical resource constraints, which meant that we were underpowered to detect small effects. Future research should endeavour to recruit more similar sample sizes as this is likely to result in more powerful analyses. In the future, studies of this kind may like to include other groups along the continuum of psychosis, such as a chronic psychosis sample and a non-clinical, non-help seeking group that endorses no PLEs, as well as examine the differences in schemas between ARMS individuals who transition to psychosis and those who do not.
This study has some clinical implications. The identification of elevated core schemas in the FEP and ARMS groups suggests that this may be an important target for CT in the ARMS and FEP populations, particularly given the high levels of negative beliefs about the self in both the ARMS and FEP groups; and previous research on self-esteem in psychosis supports this idea (Hall and Tarrier, Reference Hall and Tarrier2003, Reference Hall and Tarrier2004; Laithwaite et al., Reference Laithwaite, Gumley, Benn, Scott, Downey and Black2007; Vracotas et al., Reference Vracotas, Iyer and Malla2008). Clinicians could easily integrate this approach into the CBT treatment package for the ARMS population as it is based upon the same cognitive model often used to treat people with psychosis (Morrison, Reference Morrison2001). Additionally, clinicians should be aware that some mental health procedures, like involuntary treatment, as well as the stigma attached to mental health difficulties, may damage self-esteem and may prevent people from seeking help (Link, Struening, Neese-Todd, Asmussen and Phelan, Reference Link, Struening, Neese-Todd, Asmussen and Phelan2001; Sartorius, Reference Sartorius2007; Swartz and Monahan, Reference Swartz and Monahan2001). This may be particularly important for people experiencing prodromal or early psychosis who are likely to present with dysfunctional core schemas from the outset.
Our results also give insight into why some people seek help for mental health difficulties and others do not. Our HSC and NH groups were similar in terms of their CAARMS data (100% of the HSC group and 73.3% of the NH group were subthreshold for ARMS). However, the HSC participants sought help for their mental health difficulties and also were significantly different from the NH group in having more negative beliefs about the self and others and fewer positive beliefs about the self. Core schemas may an important factor in discriminating individuals who seek help for mental health difficulties versus those who do not, and they should be evaluated by clinicians.
To summarize, these results give us insight into core schemas across the psychosis continuum. The findings suggest that elevated levels of negative beliefs about the self and others are prominent in the FEP and ARMS populations and are associated with a range of positive psychotic experiences and the distress that results from those experiences. As psychological interventions are seen as more ethical over medication for the ARMS population (Bentall and Morrison, Reference Bentall and Morrison2002), core schemas are likely to be an important target for such interventions.
Acknowledgements
The authors would like to thank all participants who took part in the study. We would also like to thank Carolyn Crane, Melissa Wardle, and the other EDIE-2 research assistants for their help with recruitment and data collection and Jane Owens for her help with the assessments. We also thank the Mental Health Research Network, the Scottish Mental Health Research Network, and the OpenCDMS team for their support. This research was partially supported by funding for the EDIE-2 trial from the Medical Research Council (G0500264) and the Department of Health.
Suzanne L. K. Stewart is now at the Department of Psychology, University of Chester, Parkgate Road, Chester CH1 4BJ, UK.
Comments
No Comments have been published for this article.