- Email: [email protected]
Denial of illness in schizophrenia as a disturbance of self-reflection, self-perception and insight
Schizophrenia Research 152 (2014) 89–96
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Denial of illness in schizophrenia as a disturbance of self-reflection, self-perception and insight Nicholas J. Bedford ⁎, Anthony S. David Section of Cognitive Neuropsychiatry, Department of Psychosis Studies, Institute of Psychiatry, King's College London, Denmark Hill, London, SE5 8AF, UK
a r t i c l e
i n f o
Article history: Received 13 April 2013 Received in revised form 28 June 2013 Accepted 2 July 2013 Available online 23 July 2013 Keywords: Insight Memory Schizophrenia Self-evaluation Self-reference effect Self-serving bias
a b s t r a c t Background: A substantial proportion of schizophrenia patients deny aspects of their illness to others, which may indicate a deeper disturbance of ‘insight’ and a self-reflection deficit. The present study used a ‘levels-of-processing’ mnemonic paradigm to examine whether such patients engage in particularly brief and shallow self-reflection during mental illness-related self-evaluation. Methods: 26 schizophrenia patients with either an overall acceptance or denial of their illness and 25 healthy controls made timed decisions about the self-descriptiveness, other-person-descriptiveness and phonological properties of mental illness traits, negative traits and positive traits, before completing surprise tests of retrieval for these traits. Results: The acceptance patients and denial patients were particularly slow in their mental illness-related self-evaluation, indicating that they both found this exercise particularly difficult. Both patient groups displayed intact recognition but particularly reduced recall for self-evaluated traits in general, possibly indicating poor organisational processing during self-reflection. Lower recall for self-evaluated mental illness traits significantly correlated with higher denial of illness and higher illness-severity. Whilst explicit and implicit measures of self-perception corresponded in the healthy controls (who displayed an intact positive N negative ‘self-positivity bias’) and acceptance patients (who displayed a reduced self-positivity bias), the denial patients' self-positivity bias was explicitly intact but implicitly reduced. Conclusions: Schizophrenia patients, regardless of their illness-attitudes, have a particular deficit in recalling new self-related information that worsens with increasing denial of illness. This deficit may contribute towards rigid self-perception and disturbed self-awareness and insight in patients with denial of illness. © 2013 Elsevier B.V. All rights reserved.
1. Introduction
represents an important way in which self-awareness can be disturbed in schizophrenia.
1.1. Self-awareness and self-evaluation in schizophrenia 1.2. Self-evaluation, denial of illness and insight in schizophrenia A disturbance of self-awareness has long been viewed as a core deficit of schizophrenia. For example, it has been argued that an overawareness of one's normally subconscious perceptions and thoughts forms the basis of hallucinations, delusions and incoherent speech (Frith, 1979), and that an under-awareness of one's intentions to move, think and imagine can result in delusions of control, delusions of thought insertion and hallucinations, respectively (Frith, 1992). Yet these disturbances of self-awareness do not fully account for the formation of such psychotic symptoms. Along with the experience of failing to feel control for self-initiated activity, and its resulting attribution to an external agent, there must arguably be the concurrent evaluation that such erroneous perceptions and beliefs are veridical, and not the result of abnormal self-functioning (Bedford, 2010). Thus, a problem of self-evaluation during the experience of psychotic symptoms ⁎ Corresponding author. Tel.: +44 207 848 0138. E-mail address: [email protected] (N.J. Bedford). 0920-9964/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.schres.2013.07.006
When patients evaluate themselves during a clinical interview, most do appear to at least partially accept that their psychotic episodes indicate abnormal self-functioning, stem from having a mental illness, and require treatment, although a substantial proportion do not. For example, of the 221 schizophrenia patients studied by Amador et al. (1994), 40% and 58% completely rejected having their hallucinations and delusions, respectively, 32% completely rejected having a mental illness, and 22% completely rejected the efficacy of their medication. Such rejection of illness-related aspects is clinically referred to as ‘poor insight’ (David, 1990) or ‘unawareness of illness’ (Amador et al., 1991), and may indicate the presence of disturbed self-awareness and a self-reflection deficit.1 Alternatively, such a patient may have intact self-reflection and good privately-held self-awareness of illness, yet a conscious motivation 1 The term ‘self-reflection’ will be used to refer to a theoretical cognitive mechanism of self-directed thinking that underlies self-evaluative decision-making.
90
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
to withhold this awareness from others (e.g., because of a need to avoid stigma or hospitalisation). The present study sought to address this issue by examining self-reflection ability in schizophrenia patients with either an overall acceptance or denial of illness. To avoid confusion, the terms ‘acceptance’ and ‘denial’ will be used to refer to patients' explicitly-stated attitudes, and ‘insight’ to refer to their privately-held self-awareness of illness.2 1.3. Neuroimaging studies of self-evaluation in schizophrenia A handful of studies have examined whether self-reflection is impaired in schizophrenia. Using a functional neuroimaging paradigm, Bedford et al. (2012) found reduced dorsomedial prefrontal cortex (PFC) activity during the self-evaluation (but not otherperson-evaluation) of positive and negative personality traits in schizophrenia patients (compared to healthy controls), indicating underuse of a key region of the cortical midline brain network that is normally highly active during self-reflection (Northoff et al., 2006; van der Meer et al., 2010). However, when self-evaluating mental illness traits, both the dorsomedial PFC and dorsolateral PFC (a region normally associated with executive function rather than self-reflection) were overactive in the patients, with higher denial of having a mental illness moderately correlating with lower activity in the former region but higher activity in the latter. These findings indicate that, rather than being underused, the self-reflection brain network of schizophrenia patients is under increased effort during mental illness-related self-evaluation, and that the inappropriate use of executive regions during this exercise is related to denial of illness. Whilst other neuroimaging studies have found either reduced or increased cortical midline activity during personality-related self-evaluation in schizophrenia (Holt et al., 2011; Shad et al., 2012; van der Meer et al., 2012), they did not examine illness-related self-evaluation. 1.4. Memory studies of self-evaluation in schizophrenia The present study used a similar evaluative task to that of Bedford et al. (2012), but with episodic memory performance rather than neural activity as the index of self-reflective processing. The task is based on the ‘levels-of-processing’ theory that the ‘depth’ to which stimuli are originally processed can be measured by the extent to which they are subsequently remembered (Craik and Lockhart, 1972), as exemplified by the finding that higher retrieval of words results from deeper (e.g., semantic) processing than shallower (e.g., phonological) processing (Craik and Tulving, 1975). In line with this theory, the finding that self-evaluation results in higher retrieval of trait-adjectives than semantic-evaluation (the ‘self-reference effect’; SRE; Rogers et al., 1977) indicates that self-reflection normally involves a particularly deep level of processing, and offers a paradigm for gauging self-reflection depth. Three previous studies have used this paradigm in schizophrenia. Harvey et al. (2011) found reduced recognition for a mixture of self-evaluated (but not semantically-evaluated or orthographicallyevaluated) positive and negative personality traits in schizophrenia outpatients (compared to healthy controls), indicative of particularly shallow self-reflection. However, in a study by Pauly et al. (2011), there appeared to be reductions in the schizophrenia patients' recognition for other-person-evaluated as well as self-evaluated positive and negative personality traits (compared to healthy controls), indicative of shallow person-reflection in general (although the statistical significance of these reductions was not reported). Finally, Clifford and Hemsley (1987) observed an intact SRE and increased recall for self-evaluated (but not semantically-evaluated or orthographically-evaluated) positive 2
The term ‘denial’ will be used in line with its neuropsychological definition as ‘the statement that some state of affairs does not exist’ (Beaumont et al., 1996), and does not imply that the denier privately accepts what is being explicitly denied.
and negative personality traits in schizophrenia patients (compared to patients with severe depression), indicative of deep self-reflection in schizophrenia and shallow self-reflection in depression (in line with other studies that also found reduced retrieval for self-evaluated traits in depression, e.g., Derry and Kuiper, 1981; Kuiper and Derry, 1982). In conclusion, it is unclear whether schizophrenia patients engage in particularly shallow self-reflection (Harvey et al., 2011) or personreflection (Pauly et al., 2011), and, given that both these studies compared schizophrenia patients with mild depression to healthy controls, it is possible that the patients' shallow self-reflection was due to their depression rather than their schizophrenia.
1.5. Present study's aims The present study used a similar mnemonic paradigm to examine self-reflection duration and depth in schizophrenia patients with either an overall acceptance or denial of illness and healthy controls. Mental illness traits were employed as well as personality traits in order to examine whether the denial patients might display particular selfreflection problems during mental illness-related self-evaluation. It was predicted that, if denial of illness in schizophrenia is related to particularly (1) brief and (2) shallow self-reflection during mental illness-related self-evaluation, the denial patients would (1) make faster evaluative decisions and (2) have lower recall and recognition for self-evaluated mental illness traits than the comparison groups. Furthermore, it was predicted that there would be correlations in the total patient sample between higher denial of illness and (1) faster evaluative decisions and (2) lower recall and recognition for self-evaluated mental illness traits. Finally, the experimental paradigm offered an opportunity to examine self-perception and the ‘self-positivity bias’ (the normal tendency to regard oneself as having more positive than negative traits; Mezulis et al., 2004) in each group, both at an explicitlystated level (encoding performance) and at an arguably implicit level (retrieval performance).
2. Methods 2.1. Participants Schizophrenia patients were recruited from the South London & Maudsley NHS Foundation Trust (SLAM) and were eligible if they were aged between 18 and 65 years, were proficient speakers and readers of English, were aware of the current British Prime Minister (Tony Blair), and had a diagnosis of schizophrenia with hallucinations and delusions (DSM-IV-TR criteria; American Psychiatric Association, 2000), a stable dose of antipsychotic medication and no other major medical condition. Using a modified version of the expanded Schedule for the Assessment of Insight (a clinical interview for rating illness-related attitudes; SAI-E; Kemp and David, 1997),3 26 patients were divided into two groups: 14 who displayed an overall acceptance of their illness (SAI-E total score ranging from 15 to 28) and 12 who displayed an overall denial of their illness (SAI-E total score ranging from 3 to 9). 25 healthy controls with similar demographic characteristics to the patients were recruited according to the same inclusion criteria, but with the exceptions that they did not have a major medical condition and that they did not score highly (i.e., more than 21 out of 37) on the Schizotypal Personality Scale (STA; Mason et al., 1995). All participants gave their informed consent to take part and were remunerated for their assistance. The study was approved by the SLAM research ethics committee (NHS Research Ethics 3 Patients' SAI-E ‘treatment-acceptance’ scores were measured on the basis of their own (rather than informants') reports for the two items of (1) behavioural acceptance and (2) unprompted request for treatment, in line with the study's focus on patients' own explicitly-stated attitudes towards their illness.
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
Database study 04/Q0706/114) and was carried out in accordance with the World Medical Association's (2008) Declaration of Helsinki. 2.2. Stimuli A computer program running on DMDX software (Forster and Forster, 2003) presented the stimuli and recorded participants' responses. 72 encoding questions were presented, each containing a different trait: 24 asked whether the trait was self-descriptive (‘Do you think you are trait?’), 24 asked whether the trait was other-person-descriptive (‘Do you think Tony Blair is trait?’) and 24 asked whether the trait and a comparison word had similar-sounding endings (‘Does comparator rhyme with trait?’). Mental illness traits (e.g., ‘mad’), negative personality traits (e.g., ‘rude’) and positive personality traits (e.g., ‘wise’) were used, with each trait-type divided into three ‘trait-groups’ of eight traits that were matched for word-length, orthographic similarity4 and semantic similarity,5 and the three trait-groups from each trait-type divided equally between the three encoding-levels (resulting in nine conditions of eight traits each; Table S3, Appendix A). The mental illness traits were all related to schizophrenia, the negative and positive traits were all of extremely negative and positive valence, respectively (in an attempt to match the relatively low prevalence of the mental illness traits), and none of the negative traits were synonyms of any mental illness trait or antonyms of any positive trait. Whilst the three trait-types were matched for word-length, the mental illness traits appeared to have a higher number of common prefixes and suffixes and lower linguistic frequency than the other trait-types. Each rhyme-evaluative question used a unique rhyme-comparator of neutral valence that was semantically unrelated to any of the traits used in the study, with half of the rhymecomparators in each trait-type rhyming with their partnered trait. The encoding questions were presented in one of three different presentation orders, with each trait-group assigned to a different encoding-level for each order, but with the traits themselves always presented in the same order. No more than two questions of the same encoding-level were presented consecutively, and no two traits of the same trait-type, or that began with the same letter, or that had similar-sounding endings, were presented consecutively. 2.3. Procedure The encoding stage began following a practice session, with participants responding either ‘Yes / A bit’ (full or partial endorsement) or ‘No, not at all’ (complete rejection) to each question, a response choice devised to encourage more careful and representative responding about trait-ownership than a more traditional ‘Yes’ / ‘No’ dichotomy might. A surprise recall stage followed the encoding stage, with participants instructed to tell the researcher as many of the previously seen ‘words in red’ as they could. This was followed by a surprise recognition stage, in which participants decided whether the presented trait (either a target-trait or a synonymous distractor-trait; Table S3, Appendix A) had been shown at the encoding stage (‘Was this word shown before?’) by responding either ‘Yes’ or ‘No’, with the 74 target-traits and 74 intermixed distractor-traits presented in the same pseudo-randomised order to all participants. Each encoding stage and recognition stage question appeared in the centre of a black computer screen in white lettering and the trait italicised in red, remained on-screen until a response was made, and was succeeded by the next question after a 333 ms interval. Participants were instructed to respond to each question via one of two
4 Within each trait-type, the three trait-groups had similar numbers of traits beginning and ending with common syllables (e.g., ‘de-’ and ‘-al’). 5 Each trait-type was divided into eight ‘trait-elements’ of three particularly synonymous traits each (e.g., ‘ill’, ‘poorly’, ‘unwell’), with each trait-element divided equally between the three trait-groups.
91
clearly-labelled buttons as soon as they had decided which response most closely matched their honest, personal opinion about the answer. Following completion of the recognition stage, all participants were debriefed about the experiment, and all indicated that there had been no unfamiliar traits and that the retrieval stages had been unexpected. 2.4. Statistical analysis Each experimental variable (encoding response-type, encoding speed, recall, recognition sensitivity) was analysed using a 3 × 3 × 3 mixed analysis of variance and follow-up pairwise comparisons (PC) or t-tests, with Spearman's correlations also employed. The significance level of p ≤ 0.05 was Bonferroni-adjusted (i.e., divided by the number of tests involved) for hypothesis-testing but not exploratory analysis involving multiple t-tests and correlations. Any response of 100 ms or less was discarded, and all individual response times were logarithmically transformed to remove distribution skewness prior to analysis (anti-logarithmically transformed means are presented). For the recognition stage, d′ recognition sensitivity scores were calculated from hit rates and false alarm rates to measure participants' ability to correctly recognise target-traits whilst correctly rejecting distractor-traits (Stanislaw and Todorov, 1999). The term ‘SRE’ will refer to the retrieval advantage for self-evaluated traits over Blair-evaluated traits. 3. Results 3.1. Participants Data are presented in Table 1. All groups were matched for age, gender, nationality and awareness of Tony Blair, and the patient groups were also matched for duration of illness, depression (Beck Depression Inventory; Beck et al., 1979), verbal fluency (animal naming semantic subtest, Controlled Oral Word Association Test; Benton and Hamsher, 1976), verbal intelligence (vocabulary definition subtest, Wechsler Abbreviated Scale of Intelligence; Wechsler, 1999) and working memory (forward and backward digit-span subtests, Wechsler Adult Intelligence Scale; Wechsler, 1997). The healthy controls had significantly lower depression than the patient groups, significantly higher verbal fluency than the denial patients, and significantly higher verbal intelligence and working memory than the acceptance patients. The acceptance patients had significantly lower clinician-rated denial of illness, lower illness-severity (Positive and Negative Syndrome Scale; Kay et al., 1987) and a lower proportion of inpatients than the denial patients. 3.2. Encoding response-type Data are presented in Fig. 1.1 and Table S1.1 (Appendix A). There was no significant effect of participant-group or encoding-level on the number of endorsements made and no significant interaction between the two, i.e., the three groups made similar 3:5 ratios of ‘Yes / A bit’ to ‘No, not at all’ responses at each of the three encoding-levels (Fig. 1.1d). There was a significant three-way interaction (F (5.4, 128.5) = 4.4, p b 0.01; Fig. 1.1a,b,c), and exploratory t-tests revealed that the acceptance patients self-endorsed significantly more mental illness traits than the healthy controls (t (18.8) = 4.33, p b 0.001) and denial patients (t (24) = 2.43, p b 0.05), with no significant difference between the healthy controls and denial patients, and that the acceptance patients self-endorsed significantly fewer positive traits than the healthy controls (t (37) = 2.56, p b 0.01; no further significant between-group differences found). 3.3. Encoding speed Data are presented in Fig. 1.2 and Table S1.2 (Appendix A). Overall encoding speed significantly differed between the groups (F (2,
92
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
Table 1 Characteristics of the participant groups. Healthy controls (HC; N = 25)
Acceptance patients (AP; N = 14)
Denial patients (DP; N = 12)
Group comparisons
Mean age (In years)
35.4 (±2.5)
37.9 (±2.9)
32.8 (±2.3)
HC ≈ AP ≈ DP
Gender (Male / female)
13 / 12 (52% / 48%)
7/7 (50% / 50%)
9/3 (75% / 25%)
HC ≈ AP ≈ DP
Nationality (British / non-British)
17 / 8 (68% / 32%)
12 / 2 (76% / 14%)
12 / 0 (100% / 0%)
HC ≈ AP ≈ DP
Mean duration of schizophrenia (In years)
–
10.0 (±2.4)
11.6 (±2.2)
AP ≈ DP
Patient status (Inpatients / outpatients)
–
2 / 12 (10% / 90%)
10 / 2 (90% / 10%)
AP ≠ DP (χ2 (1) = 11.0, p b 0.001)
PANSS (30–210)
–
78.9 (±3.1)
102.3 (±2.5)
AP b DP (t (24) = 6.0, p b 0.001)
SAI-E (0–28)
–
19.2 (±1.6)
5.0 (±0.6)
AP N DP (t (24) = 8.45, p b 0.001)
BDI-IA (0–63)
5.9 (±1.1)
14.6 (±3.3)
13.0 (±2.5)
HC b AP (t (37) = 2.77, p b 0.01) HC b DP (t (35) = 2.81, p b 0.01) AP ≈ DP
COWAT verbal fluency (N animals named in 1 minute)
23.7 (±1.2)
20.1 (±1.7)
17.2 (±1.6)
HC b AP (t (37) = 1.72, p b 0.10) HC b DP (t (35) = 3.19, p b 0.01) AP ≈ DP
WASI vocabulary definition (T-score 20–80)
57.9 (±1.9)
46.1 (±3.1)
52.3 (±3.7)
HC N AP (t (37) = 3.48, p = 0.001) HC ≈ DP AP ≈ DP
WAIS-III digit-span (0–30)
19.8 (±0.9)
15.7 (±1.0)
17.8 (±0.9)
HC N AP (t (37) = 2.76, p b 0.01) HC ≈ DP AP ≈ DP
Tony Blair awareness (0–4)
2.4 (±0.3)
2.1 (±0.2)
1.8 (±0.2)
HC ≈ AP ≈ DP
± = standard error; PANSS = Positive and Negative Syndrome Scale total score; SAI-E = expanded Schedule for the Assessment of Insight total score; BDI-IA = revised Beck Depression Inventory total score; COWAT = Controlled Oral Word Association Test; WASI = Weschler Abbreviated Scale of Intelligence; WAIS-III = Wechsler Adult Intelligence Scale, third edition. Participants rated their awareness of the public persona of Tony Blair on a five-point scale ranging from ‘complete unfamiliarity’ (0) to ‘complete familiarity’ (4).
48) = 10.0, p b 0.001), with the healthy controls (M = 2614 ms, SE = 277) faster than the acceptance patients (M = 4629 ms, SE = 370; significant PC: p b 0.001) and denial patients (M = 3528 ms, SE = 438; near-significant PC: p b 0.10), and no significant difference between the patient groups. There was a significant interaction between encoding-level and participant-group (F (4, 96) = 7.16, p b 0.001), with Fig. 1.2d indicating that both patient groups were particularly slow during self-evaluation (contrary to the study's hypothesis). In support of this, exploratory t-tests revealed that, whilst the healthy controls made significantly faster responses at self-evaluation than Blair-evaluation (t (24) = 4.08, p b 0.001), this was non-significantly reversed in both patient groups. There was a significant three-way interaction (F (8, 192) = 1.99, p b 0.05), with Fig. 1.2a,b,c indicating that both patient groups were particularly slow during the self-evaluation of mental illness traits. In support of this, exploratory t-tests revealed that, whilst the healthy controls made significantly faster responses to self-evaluated than Blair-evaluated mental illness traits (t (24) =2.21, p b 0.05), this was near-significantly reversed in the acceptance patients (t (13) = 1.55, p = 0.07) and denial patients (t (11) = 1.58, p = 0.07).
3.4. Recall Data are presented in Fig. 2.1 and Table S2.1 (Appendix A). Overall numbers of correctly recalled traits significantly differed between the groups (F (2, 48) = 3.91, p b 0.05), with the healthy controls (M = 11.8, SE = 1.00) recalling significantly more traits than the denial patients (M = 7.3, SE = 1.11; PC: p b 0.05) but not the acceptance patients (M = 8.9, SE = 1.45), and no significant difference between the patient groups. There was a significant interaction between encodinglevel and participant-group (F (4, 96) = 2.51, p b 0.05; Fig. 2.1d) and no further interaction with trait-type (Fig. 2.1a,b,c). Exploratory t-tests revealed that the healthy controls recalled significantly more self-evaluated traits than the acceptance patients (t (37) = 2.12, p b 0.05) and denial patients (t (35) = 2.71, p b 0.01), with no significant difference between the patient groups, and that, whilst the healthy controls demonstrated a significant SRE overall (t (24) =2.53, p b 0.01), both patient groups did not. Three hypothesis-testing, Bonferroni-adjusted t-tests revealed that the denial patients recalled fewer self-evaluated mental illness traits than the acceptance patients (non-significant: t (24) = 1.11, p = 0.14) and healthy controls (near-significant: t (35) = 1.87,
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
‘Yes / A Bit’ Responses (0–8)
1.1
7
Self
Blair
93
Rhyme
6 5 4 3 2 1 0
Mental Negative Positive (a) Healthy Controls (HC)
Response Time (in milliseconds)
1.2
7000
Mental Negative Positive (b) Acceptance Patients (AP)
Self
Mental Negative Positive (c) Denial Patients (DP)
Blair
Combined Traits (d) HC AP DP
Rhyme
6000 5000 4000 3000 2000 1000 0
Mental Negative Positive (a) Healthy Controls (HC)
Mental Negative Positive (b) Acceptance Patients (AP)
Mental Negative Positive (c) Denial Patients (DP)
Combined Traits (d) HC AP DP
Fig. 1. Mean numbers of ‘Yes / A Bit’ responses (1.1) and mean response times (1.2) at encoding. Figure 1.1: graphs a,b,c depict a significant interaction between trait-type, encoding-level and participant-group, whereas graph d depicts a lack of significant interaction between encoding-level and participant-group over all trait-types. Figure 1.2: graphs a,b,c depict a significant interaction between trait-type, encoding-level and participant-group, and graph d depicts a significant interaction between encoding-level and participant-group over all trait-types. Error bar = standard error; mental = mental illness; corresponding data presented in Tables S1.1 and S1.2 (Appendix A).
p = 0.035), with no significant difference between the acceptance patients and healthy controls. 3.5. Recognition sensitivity Data are presented in Fig. 2.2 and Table S2.2 (Appendix A). Overall recognition sensitivity did not significantly differ between the healthy controls (M = 1.35, SE = 0.08), acceptance patients (M = 1.26, SE = 0.11) and denial patients (M = 1.29, SE = 0.12). A significant effect of encoding-level was found (F (2, 96) = 35.8, p b 0.001), with a SRE demonstrated over all participants (PC: p b 0.05). Whilst there was no significant interaction with participant-group (Fig. 2.2d), and whilst exploratory t-tests revealed that the groups did not significantly differ in their recognition for self-evaluated traits, further exploratory t-tests revealed that the acceptance patients failed to demonstrate the significant SRE displayed by the healthy controls (t (24) = 3.07, p b 0.05) and denial patients (t (11) = 2.70, p b 0.05). There was a significant three-way interaction (F (8, 192) = 3.49, p b 0.001; Fig. 2.2a,b, c). Exploratory t-tests revealed that, whilst the healthy controls displayed a significant SRE for mental illness traits (t (24) = 1.71, p = 0.05), this was reduced for the acceptance patients (t (13) = 0.76, p = 0.23) and significantly reversed for the denial patients (t (11) = 1.81, p b 0.05), and whilst the healthy controls (t (24) = 3.30, p b 0.01) and denial patients (t (11) = 3.87, p b 0.01) displayed a
significant SRE for positive traits, this was near-significantly reversed for the acceptance patients (t (13) = 1.35, p = 0.10). Three hypothesis-testing, Bonferroni-adjusted t-tests revealed no significant differences between the three groups' recognition for self-evaluated mental illness traits.
3.6. Correlational analysis Three pairs of hypothesis-testing correlations were conducted in the total patient sample, each using a Bonferroni-adjusted significance level of p ≤ 0.016. Firstly, small / very small, non-significant correlations of inconsistent direction were found between self-reflection speed and denial of illness, both in terms of patients' self-endorsement of mental illness traits at encoding and SAI-E clinician ratings.6 Secondly, large / medium, significant correlations were found between lower recall for self-evaluated mental illness traits and higher denial of illness, both self-rated (r s = .56, p = 0.002) and clinician-rated (r s = .45, p = 0.01). Thirdly, small, non-significant correlations were found between lower recognition for self-evaluated mental illness traits and higher denial of illness, both self-rated and clinician-rated. 6 With regard to r correlation sizes,‘very small’ = .01 – .09, ‘small’ = .10 – .29, ‘medium’ = .30 – .49, ‘large’ = .50 – .69, and ‘very large’ = .70 – 1.00.
94
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
2.1 Self
Recalled Traits (0–8)
2.5
Blair
Rhyme
2
1.5
1
0.5
0
Mental Negative Positive (a) Healthy Controls (HC)
2.2
Self
2.5
Recognition Sensitivity (in d’)
Mental Negative Positive (b) Acceptance Patients (AP)
Mental Negative Positive (c) Denial Patients (DP)
Blair
Combined Traits (d) HC AP DP
Rhyme
2
1.5
1
0.5
0
Mental Negative Positive (a) Healthy Controls (HC)
Mental Negative Positive (b) Acceptance Patients (AP)
Mental Negative Positive (c) Denial Patients (DP)
Combined Traits (d) HC AP DP
Fig. 2. Mean numbers of recalled traits (2.1) and mean recognition sensitivity scores (2.2) at retrieval. Figure 2.1: graphs a,b,c depict a lack of significant interaction between trait-type, encoding-level and participant-group, whereas graph d depicts a significant interaction between encoding-level and participant-group over all trait-types. Figure 2.2: graphs a,b,c depict a significant interaction between trait-type, encoding-level and participant-group, whereas graph d depicts a lack of significant interaction between encoding-level and participant-group over all trait-types. Error bar = standard error; mental = mental illness; corresponding data presented in Tables S1.1 and S1.2 (Appendix A).
In an exploratory examination of potential confounding factors, no significant correlations were found in the total patient sample between recall for self-evaluated mental illness traits and either depression, verbal fluency, verbal intelligence, working memory or self-reflection speed. However, a medium, significant correlation was found between lower recall for self-evaluated mental illness traits and higher illness-severity (r s = .45, p b 0.05).
3.7. Explicit and implicit self-perception In an examination of the self-positivity bias at an explicitlystated level, exploratory t-tests revealed that, whilst the healthy controls (t (24) = 6.94, p b 0.001; Fig. 1.1a) and denial patients (t (11) = 2.68, p b 0.05; Fig. 1.1c) self-endorsed significantly more positive than negative traits, the acceptance patients did not (Fig. 1.1b). Regarding the self-positivity bias at an arguably implicit level, exploratory t-tests revealed that, whilst the healthy controls recalled (near-significantly: t (24) = 1.33, p b 0.10; Fig. 2.1a) and recognised (significantly: t (24) = 1.86, p b 0.05; Fig. 2.2a) more self-evaluated positive than negative traits, the acceptance patients (Figs. 2.1b and 2.2b) and denial patients (Figs. 2.1c and 2.2c) did not; indeed, the acceptance patients recognised significantly fewer self-evaluated positive than negative traits (t (24) = 2.41, p b 0.05).
4. Discussion 4.1. Self-reflection speed and denial of illness Both patient groups were generally slower than the healthy controls at the encoding stage and particularly slow when self-evaluating mental illness traits (but not personality-related traits). This finding corresponds with the observation that the brain regions involved in self-reflection and executive function are unusually overactive during mental illness-related (but not personality-related) self-evaluation in schizophrenia (Bedford et al., 2012). There was no evidence that denial of illness is related to speed of self-reflection, with the present findings indicating that schizophrenia patients, regardless of their illness-attitudes, find mental illness-related self-evaluation particularly difficult, and that it may provoke increased uncertainty, presentational concern and emotional discomfort in acceptance patients and denial patients alike. This slowness also indicates that patients with denial of illness have an implicit awareness of the potential self-relevance of mental illness, despite their explicit rejection of it, although it is unclear whether this awareness is consciously or subconsciously held. 4.2. Self-perception and denial of illness The acceptance patients displayed a high self-endorsement of mental illness traits, a low self-endorsement of positive traits, and a
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
reduced self-positivity bias at encoding. These findings were perhaps to be expected, as studies tend to show a relationship between higher acceptance of illness and higher depression in schizophrenia (Mintz et al., 2003), and between higher depression and a lower self-positivity bias more generally (Watson et al., 2008). However, whilst both patient groups reported similarly higher levels of depression than the healthy controls, the denial patients still displayed a significant self-positivity bias at encoding, raising the intriguing speculation that, at an explicitly-stated level, the denial patients' personality-related self-perception was more resistant to depression than the acceptance patients'. The acceptance patients had particularly poor recognition for self-evaluated positive traits, in line with other studies that have shown reduced retrieval for self-evaluated positive traits in mildly (Kuiper and Derry, 1982) and severely (Derry and Kuiper, 1981) depressed people with a reduced self-positivity bias. The acceptance patients' failure to retrieve more self-evaluated positive than negative traits perhaps indicated that they held a reduced self-positivity bias at an implicit level, and that their implicitly-held and explicitlystated self-perception were congruent. Similarly, the healthy controls' tendency to retrieve more self-evaluated positive than negative traits indicated a possible correspondence between their implicit and explicit self-perception. However, there appeared to be a disparity between implicit and explicit self-perception in the denial patients, who shared the acceptance patients' reduced self-positivity bias at retrieval, but the healthy controls' intact self-positivity bias at encoding. This could be interpreted as indicating that patients with denial of illness implicitly hold a far less positive self-perception than they explicitly display. However, it is uncertain whether implicit selfschemas can be gauged by retrieval performance in this way, and if so, whether they are consciously or subconsciously held. 4.3. Self-reflection depth and denial of illness Positive traits excepted, both patient groups' recognition for self-evaluated traits appeared to be intact. Although the denial patients' recognition-SRE for mental illness traits was significantly reversed, supporting the hypothesis that they engaged in particularly shallow mental illness-related self-reflection, this may have been due to their unusually high recognition for Blair-evaluated mental illness traits (Fig. 2.2c). In contrast to their generally intact recognition, both patient groups displayed particularly reduced recall for self-evaluated traits in general. Whilst the denial patients did display lower recall for self-evaluated traits (mental illness-related and overall) than the acceptance patients, there was arguably insufficient statistical power for these differences to achieve significance because of the low group numbers involved. Indeed, when more adequately-powered analyses were performed over the total patient sample, significant correlations emerged between lower recall for self-evaluated mental illness traits and (1) higher denial of illness and (2) higher illness-severity. These correlations support the hypothesis that denial of illness in schizophrenia is related to a self-related recall deficit that worsens as the illness becomes more severe, and is in line with other studies' consistent demonstration of a relationship between higher denial of illness and higher illness-severity in schizophrenia (Mintz et al., 2003). However, it is worth noting that the acceptance patients also displayed poor self-related recall relative to the healthy controls, indicating that this deficit alone is not sufficient for denial of illness to occur. The underlying nature of the SRE is uncertain, with theories usually postulating that it results from the self-concept being a particularly well-organised, emotion-arousing and often-used schema (Symons and Johnson, 1997). The present recall-recognition dissociation indicates that the patients did benefit from the deep processing that normally accompanies self-evaluation and leads to the SRE, but only when presented with recognition cues that they did not have to
95
generate themselves. According to a theory posited by Klein and Kihlstrom (1986), this may have been because, during encoding, the patients had intact ‘item-specific’ self-processing of individual traits, but poor ‘organisational processing’ of the different self-evaluated traits around the self-concept, i.e., poor implicit formation of inter-trait connections during self-evaluation. Such a problem would theoretically impact on the ability of one recalled self-evaluated trait to trigger the recall of another, impairing recall more than recognition because of the former's greater reliance on the self-generation of cues. This explanation of the recall-recognition dissociation is in line with evidence from other studies that poor recall in schizophrenia reflects a disturbance of organisational processing, with schizophrenia patients producing less clustered recall of semantically-related words than healthy controls (Paulsen et al., 1995), but improving when the semantic relatedness of the words is emphasised at encoding (Gold et al., 1992). The present findings are also congruent with results from a meta-analysis that showed recall to be generally more impaired in schizophrenia than recognition (Aleman et al., 1999).
4.4. Conclusions The present study found evidence for the following conclusions. Firstly, schizophrenia patients, regardless of their illness-attitudes, are particularly slow at making decisions about their mental health, and may experience particular difficulty during this exercise. Patients with denial of illness are implicitly aware of the potential self-relevance of mental illness, despite their explicit rejection of it. Secondly, patients with denial of illness explicitly present themselves in a generally more positive than negative light, a normal self-serving bias that may however be at odds with a more negative implicitly-held self-schema and an elevated level of depression. This may indicate rigid, inflexible self-perception in patients with denial of illness. Thirdly, schizophrenia patients, regardless of their illness-attitudes, have a particular deficit in recalling new self-related episodes that worsens with increasing denial of illness and illness-severity, and that may be due to impaired organisational processing during self-reflection. It could be speculated that this deficit contributes to rigid self-perception and an entrenched sense of being mentally healthy at a privately-held level of belief. If so, then denial of illness in schizophrenia may be regarded less as a conscious impression-management strategy and more as a genuine disturbance of self-awareness and insight.
4.5. Limitations The present findings were derived from the use of small samples and several exploratory analyses, and require replication with larger samples. Analyses involving the patient subgroups had low statistical power, increasing the likelihood of differences not achieving statistical significance. The healthy controls were not on antipsychotic medication and had lower depression, higher verbal fluency, higher verbal intelligence and higher working memory than the patients, although there was no evidence that the last four variables were related to the selfrelated recall deficit. Whilst the difference in illness-severity and inpatient proportion between the two patient groups represented a potential confound, it was perhaps unavoidable given the aforementioned relationship that appears to exist between denial of illness and illness-severity in schizophrenia.
Role of the funding source The study was funded by a Medical Research Council PhD studentship to Dr. Bedford. Prof. David acknowledges support from the National Institute For Health Research Biomedical Research Centre at SLAM and the Institute of Psychiatry, King's College London. Neither funding source played a further role in the study.
96
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
Contributors Prof. David conceived the original idea for the study, supervised the project and edited the manuscript. Dr. Bedford conceived the idea to examine mental illness-related selfevaluation, designed the study, wrote the computer program, tested the participants, performed the analysis and wrote the manuscript. Conflict of interest The authors have no conflicts of interest to declare. Acknowledgements We thank Dr. Simon Surguladze for help with patient recruitment, Mr. Christopher Maltby for computer programming assistance, Dr. Nicholas Sharples for general computer-related wizardry, and the Schizophrenia Research reviewers for their insightful comments.
Appendix A. Supplementary data Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.schres.2013.07.006. References Aleman, A., Hijman, R., de Haan, E.H.F., Kahn, R.S., 1999. Memory impairment in schizophrenia: a meta-analysis. Am. J. Psychiatry 156, 1358–1366. Amador, X.F., Strauss, D.H., Yale, S.A., Gorman, J.M., 1991. Awareness of illness in schizophrenia. Schizophr. Bull. 17, 113–132. Amador, X.F., Flaum, M., Andreasen, N.C., Strauss, D.H., Yale, S.A., Clark, S.C., Gorman, J.M., 1994. Awareness of illness in schizophrenia and schizoaffective and mood disorders. Arch. Gen. Psychiatry 51, 826–836. American Psychiatric Association, 2000. Diagnostic and Statistical Manual of Mental Disorders: DSM-IV-TR. text revision 4th ed. American Psychiatric Association, Washington. Beaumont, J.G., Rogers, M.J.C., Kenealy, P.M. (Eds.), 1996. The Blackwell Dictionary of Neuropsychology. Blackwell, Oxford. Beck, A.T., Rush, A.J., Shaw, B.F., Emery, G., 1979. Cognitive Therapy of Depression. Guildford Press, New York. Bedford, N.J., 2010. Denial of Illness in Schizophrenia as a Disturbance of Insight and Self-Awareness. PhD Thesis Institute of Psychiatry, King's College London Library. Bedford, N.J., Surguladze, S., Giampietro, V., Brammer, M.J., David, A.S., 2012. Self-evaluation in schizophrenia: an fMRI study with implications for the understanding of insight. BMC Psychiatry 12, 106. Benton, A.L., Hamsher, K., 1976. Multilingual Aphasia Examination. University of Iowa, Iowa City. Clifford, P.I., Hemsley, D.R., 1987. The influence of depression on the processing of personal attributes. Br. J. Psychiatry 150, 98–103. Craik, F.I.M., Lockhart, R.S., 1972. Levels of processing: a framework for memory research. J. Verbal Learn. Verbal Behav. 11, 671–684. Craik, F.I.M., Tulving, E., 1975. Depth of processing and the retention of words in episodic memory. J. Exp. Psychol. 104, 268–294. David, A.S., 1990. Insight and psychosis. Br. J. Psychiatry 156, 798–808. Derry, P., Kuiper, N., 1981. Schematic processing and self-reference in clinical depression. J. Abnorm. Psychol. 90, 286–297. Forster, K.I., Forster, J.C., 2003. DMDX: a Windows display program with millisecond accuracy. Behav. Res. Methods Instrum. Comput. 35, 116–124. Frith, C.D., 1979. Consciousness, information processing and schizophrenia. Br. J. Psychiatry 134, 225–235.
Frith, C.D., 1992. The Cognitive Neuropsychology of Schizophrenia. Psychology Press, Hove, East Sussex. Gold, J.M., Randolph, C., Carpenter, C.J., Goldberg, T.E., Weinberger, D.R., 1992. Forms of memory failure in schizophrenia. J. Abnorm. Psychol. 101, 487–494. Harvey, P., Lee, J., Horan, W.P., Ochsner, K., Green, M.F., 2011. Do patients with schizophrenia benefit from a self-referential memory bias? Schizophr. Res. 127, 171–177. Holt, D.J., Cassidy, B.S., Andrews-Hanna, J.R., Lee, S.M., Coombs, G., Goff, D.C., Gabrieli, J.D., Moran, J.M., 2011. An anterior-to-posterior shift in midline cortical activity in schizophrenia during self-reflection. Biol. Psychiatry 69, 415–423. Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The Positive and Negative Syndrome Scale (PANSS) for schizophrenia. Schizophr. Bull. 13, 261–276. Kemp, R., David, A.S., 1997. Insight and compliance. In: Blackwell, B. (Ed.), Treatment Compliance and the Therapeutic Alliance in Serious Mental Illness. Harwood Academic Publishers, Amsterdam, pp. 61–84. Klein, S.B., Kihlstrom, J.F., 1986. Elaboration, organization, and the self-reference effect in memory. J. Exp. Psychol. Gen. 115, 26–38. Kuiper, N.A., Derry, P.A., 1982. Depressed and nondepressed content self-reference in mild depressives. J. Pers. 50, 67–80. Mason, O., Claridge, G., Jackson, M., 1995. New scales for the assessment of schizotypy. Personal. Individ. Differ. 18, 7–13. Mezulis, A.H., Abramson, L.Y., Hyde, J.S., Hankin, B.L., 2004. Is there a universal positivity bias in attributions? A meta-analytic review of individual, developmental, and cultural differences in the self-serving attributional bias. Psychol. Bull. 130, 711–747. Mintz, A.R., Dobson, K.S., Romney, D.M., 2003. Insight in schizophrenia: a meta-analysis. Schizophr. Res. 61, 75–88. Northoff, G., Heinzel, A., de Greck, M., Bermpohl, F., Dobrowolny, H., Panksepp, J., 2006. Self-referential processing in our brain: a meta-analysis of imaging studies on the self. Neuroimage 31, 440–457. Paulsen, J.S., Heaton, R.K., Sadek, J.R., Perry, W., Delis, D.C., Braff, D., Kuck, J., Zisook, S., Jeste, D.V., 1995. The nature of learning and memory impairments in schizophrenia. J. Int. Neuropsychol. Soc. 1, 88–99. Pauly, K., Kircher, T., Weber, J., Schneider, F., Habel, U., 2011. Self-concept, emotion and memory performance in schizophrenia. Psychiatry Res. 186, 11–17. Rogers, T.B., Kuiper, N.A., Kirker, W.S., 1977. Self-reference and the encoding of personal information. J. Pers. Soc. Psychol. 35, 677–688. Shad, M.U., Keshavan, M.S., Steinberg, J.L., Mihalakos, P., Thomas, B.P., Motes, M.A., Soares, J.C., Tamminga, C.A., 2012. Neurobiology of self-awareness in schizophrenia: an fMRI study. Schizophr. Res. 138, 113–119. Stanislaw, H., Todorov, N., 1999. Calculation of signal detection theory measures. Behav. Res. Methods Instrum. Comput. 31, 137–149. Symons, C.S., Johnson, B.T., 1997. The self-reference effect in memory: a meta-analysis. Psychol. Bull. 121, 371–394. Van der Meer, L., Costafreda, S.C., Aleman, A., David, A.S., 2010. Self-reflection and the brain: a theoretical review and meta-analysis of neuroimaging studies with implications for schizophrenia. Neurosci. Biobehav. Rev. 34, 935–946. Van der Meer, L., de Vos, A.E., Stiekema, A.P.M., Pijnenborg, G.H.M., van Tol, M., Nolen, W.A., David, A.S., Aleman, A., 2012. Insight in schizophrenia: involvement of selfreflection networks? Schizophr. Bull. http://dx.doi.org/10.1093/schbul/sbs122 (Retrieved 2013). Watson, L.A., Dritschel, B., Jentzsch, I., Obonsawin, M.C., 2008. Changes in the relationship between self-reference and emotional valence as a function of dysphoria. Br. J. Psychol. 99, 143–152. Wechsler, D., 1997. Wechsler Adult Intelligence Scale, third ed. The Psychological Corporation, San Antonio, Texas. Wechsler, D., 1999. Wechsler Abbreviated Scale of Intelligence (WASI). The Psychological Corporation, San Antonio, Texas. World Medical Association, 2008. Declaration of Helsinki — ethical principles for medical research involving human subjects, 6th revision. www.wma.net/en/30publications/ 10policies/b3 (Retrieved 2013).
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Denial of illness in schizophrenia as a disturbance of self-reflection, self-perception and insight Nicholas J. Bedford ⁎, Anthony S. David Section of Cognitive Neuropsychiatry, Department of Psychosis Studies, Institute of Psychiatry, King's College London, Denmark Hill, London, SE5 8AF, UK
a r t i c l e
i n f o
Article history: Received 13 April 2013 Received in revised form 28 June 2013 Accepted 2 July 2013 Available online 23 July 2013 Keywords: Insight Memory Schizophrenia Self-evaluation Self-reference effect Self-serving bias
a b s t r a c t Background: A substantial proportion of schizophrenia patients deny aspects of their illness to others, which may indicate a deeper disturbance of ‘insight’ and a self-reflection deficit. The present study used a ‘levels-of-processing’ mnemonic paradigm to examine whether such patients engage in particularly brief and shallow self-reflection during mental illness-related self-evaluation. Methods: 26 schizophrenia patients with either an overall acceptance or denial of their illness and 25 healthy controls made timed decisions about the self-descriptiveness, other-person-descriptiveness and phonological properties of mental illness traits, negative traits and positive traits, before completing surprise tests of retrieval for these traits. Results: The acceptance patients and denial patients were particularly slow in their mental illness-related self-evaluation, indicating that they both found this exercise particularly difficult. Both patient groups displayed intact recognition but particularly reduced recall for self-evaluated traits in general, possibly indicating poor organisational processing during self-reflection. Lower recall for self-evaluated mental illness traits significantly correlated with higher denial of illness and higher illness-severity. Whilst explicit and implicit measures of self-perception corresponded in the healthy controls (who displayed an intact positive N negative ‘self-positivity bias’) and acceptance patients (who displayed a reduced self-positivity bias), the denial patients' self-positivity bias was explicitly intact but implicitly reduced. Conclusions: Schizophrenia patients, regardless of their illness-attitudes, have a particular deficit in recalling new self-related information that worsens with increasing denial of illness. This deficit may contribute towards rigid self-perception and disturbed self-awareness and insight in patients with denial of illness. © 2013 Elsevier B.V. All rights reserved.
1. Introduction
represents an important way in which self-awareness can be disturbed in schizophrenia.
1.1. Self-awareness and self-evaluation in schizophrenia 1.2. Self-evaluation, denial of illness and insight in schizophrenia A disturbance of self-awareness has long been viewed as a core deficit of schizophrenia. For example, it has been argued that an overawareness of one's normally subconscious perceptions and thoughts forms the basis of hallucinations, delusions and incoherent speech (Frith, 1979), and that an under-awareness of one's intentions to move, think and imagine can result in delusions of control, delusions of thought insertion and hallucinations, respectively (Frith, 1992). Yet these disturbances of self-awareness do not fully account for the formation of such psychotic symptoms. Along with the experience of failing to feel control for self-initiated activity, and its resulting attribution to an external agent, there must arguably be the concurrent evaluation that such erroneous perceptions and beliefs are veridical, and not the result of abnormal self-functioning (Bedford, 2010). Thus, a problem of self-evaluation during the experience of psychotic symptoms ⁎ Corresponding author. Tel.: +44 207 848 0138. E-mail address: [email protected] (N.J. Bedford). 0920-9964/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.schres.2013.07.006
When patients evaluate themselves during a clinical interview, most do appear to at least partially accept that their psychotic episodes indicate abnormal self-functioning, stem from having a mental illness, and require treatment, although a substantial proportion do not. For example, of the 221 schizophrenia patients studied by Amador et al. (1994), 40% and 58% completely rejected having their hallucinations and delusions, respectively, 32% completely rejected having a mental illness, and 22% completely rejected the efficacy of their medication. Such rejection of illness-related aspects is clinically referred to as ‘poor insight’ (David, 1990) or ‘unawareness of illness’ (Amador et al., 1991), and may indicate the presence of disturbed self-awareness and a self-reflection deficit.1 Alternatively, such a patient may have intact self-reflection and good privately-held self-awareness of illness, yet a conscious motivation 1 The term ‘self-reflection’ will be used to refer to a theoretical cognitive mechanism of self-directed thinking that underlies self-evaluative decision-making.
90
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
to withhold this awareness from others (e.g., because of a need to avoid stigma or hospitalisation). The present study sought to address this issue by examining self-reflection ability in schizophrenia patients with either an overall acceptance or denial of illness. To avoid confusion, the terms ‘acceptance’ and ‘denial’ will be used to refer to patients' explicitly-stated attitudes, and ‘insight’ to refer to their privately-held self-awareness of illness.2 1.3. Neuroimaging studies of self-evaluation in schizophrenia A handful of studies have examined whether self-reflection is impaired in schizophrenia. Using a functional neuroimaging paradigm, Bedford et al. (2012) found reduced dorsomedial prefrontal cortex (PFC) activity during the self-evaluation (but not otherperson-evaluation) of positive and negative personality traits in schizophrenia patients (compared to healthy controls), indicating underuse of a key region of the cortical midline brain network that is normally highly active during self-reflection (Northoff et al., 2006; van der Meer et al., 2010). However, when self-evaluating mental illness traits, both the dorsomedial PFC and dorsolateral PFC (a region normally associated with executive function rather than self-reflection) were overactive in the patients, with higher denial of having a mental illness moderately correlating with lower activity in the former region but higher activity in the latter. These findings indicate that, rather than being underused, the self-reflection brain network of schizophrenia patients is under increased effort during mental illness-related self-evaluation, and that the inappropriate use of executive regions during this exercise is related to denial of illness. Whilst other neuroimaging studies have found either reduced or increased cortical midline activity during personality-related self-evaluation in schizophrenia (Holt et al., 2011; Shad et al., 2012; van der Meer et al., 2012), they did not examine illness-related self-evaluation. 1.4. Memory studies of self-evaluation in schizophrenia The present study used a similar evaluative task to that of Bedford et al. (2012), but with episodic memory performance rather than neural activity as the index of self-reflective processing. The task is based on the ‘levels-of-processing’ theory that the ‘depth’ to which stimuli are originally processed can be measured by the extent to which they are subsequently remembered (Craik and Lockhart, 1972), as exemplified by the finding that higher retrieval of words results from deeper (e.g., semantic) processing than shallower (e.g., phonological) processing (Craik and Tulving, 1975). In line with this theory, the finding that self-evaluation results in higher retrieval of trait-adjectives than semantic-evaluation (the ‘self-reference effect’; SRE; Rogers et al., 1977) indicates that self-reflection normally involves a particularly deep level of processing, and offers a paradigm for gauging self-reflection depth. Three previous studies have used this paradigm in schizophrenia. Harvey et al. (2011) found reduced recognition for a mixture of self-evaluated (but not semantically-evaluated or orthographicallyevaluated) positive and negative personality traits in schizophrenia outpatients (compared to healthy controls), indicative of particularly shallow self-reflection. However, in a study by Pauly et al. (2011), there appeared to be reductions in the schizophrenia patients' recognition for other-person-evaluated as well as self-evaluated positive and negative personality traits (compared to healthy controls), indicative of shallow person-reflection in general (although the statistical significance of these reductions was not reported). Finally, Clifford and Hemsley (1987) observed an intact SRE and increased recall for self-evaluated (but not semantically-evaluated or orthographically-evaluated) positive 2
The term ‘denial’ will be used in line with its neuropsychological definition as ‘the statement that some state of affairs does not exist’ (Beaumont et al., 1996), and does not imply that the denier privately accepts what is being explicitly denied.
and negative personality traits in schizophrenia patients (compared to patients with severe depression), indicative of deep self-reflection in schizophrenia and shallow self-reflection in depression (in line with other studies that also found reduced retrieval for self-evaluated traits in depression, e.g., Derry and Kuiper, 1981; Kuiper and Derry, 1982). In conclusion, it is unclear whether schizophrenia patients engage in particularly shallow self-reflection (Harvey et al., 2011) or personreflection (Pauly et al., 2011), and, given that both these studies compared schizophrenia patients with mild depression to healthy controls, it is possible that the patients' shallow self-reflection was due to their depression rather than their schizophrenia.
1.5. Present study's aims The present study used a similar mnemonic paradigm to examine self-reflection duration and depth in schizophrenia patients with either an overall acceptance or denial of illness and healthy controls. Mental illness traits were employed as well as personality traits in order to examine whether the denial patients might display particular selfreflection problems during mental illness-related self-evaluation. It was predicted that, if denial of illness in schizophrenia is related to particularly (1) brief and (2) shallow self-reflection during mental illness-related self-evaluation, the denial patients would (1) make faster evaluative decisions and (2) have lower recall and recognition for self-evaluated mental illness traits than the comparison groups. Furthermore, it was predicted that there would be correlations in the total patient sample between higher denial of illness and (1) faster evaluative decisions and (2) lower recall and recognition for self-evaluated mental illness traits. Finally, the experimental paradigm offered an opportunity to examine self-perception and the ‘self-positivity bias’ (the normal tendency to regard oneself as having more positive than negative traits; Mezulis et al., 2004) in each group, both at an explicitlystated level (encoding performance) and at an arguably implicit level (retrieval performance).
2. Methods 2.1. Participants Schizophrenia patients were recruited from the South London & Maudsley NHS Foundation Trust (SLAM) and were eligible if they were aged between 18 and 65 years, were proficient speakers and readers of English, were aware of the current British Prime Minister (Tony Blair), and had a diagnosis of schizophrenia with hallucinations and delusions (DSM-IV-TR criteria; American Psychiatric Association, 2000), a stable dose of antipsychotic medication and no other major medical condition. Using a modified version of the expanded Schedule for the Assessment of Insight (a clinical interview for rating illness-related attitudes; SAI-E; Kemp and David, 1997),3 26 patients were divided into two groups: 14 who displayed an overall acceptance of their illness (SAI-E total score ranging from 15 to 28) and 12 who displayed an overall denial of their illness (SAI-E total score ranging from 3 to 9). 25 healthy controls with similar demographic characteristics to the patients were recruited according to the same inclusion criteria, but with the exceptions that they did not have a major medical condition and that they did not score highly (i.e., more than 21 out of 37) on the Schizotypal Personality Scale (STA; Mason et al., 1995). All participants gave their informed consent to take part and were remunerated for their assistance. The study was approved by the SLAM research ethics committee (NHS Research Ethics 3 Patients' SAI-E ‘treatment-acceptance’ scores were measured on the basis of their own (rather than informants') reports for the two items of (1) behavioural acceptance and (2) unprompted request for treatment, in line with the study's focus on patients' own explicitly-stated attitudes towards their illness.
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
Database study 04/Q0706/114) and was carried out in accordance with the World Medical Association's (2008) Declaration of Helsinki. 2.2. Stimuli A computer program running on DMDX software (Forster and Forster, 2003) presented the stimuli and recorded participants' responses. 72 encoding questions were presented, each containing a different trait: 24 asked whether the trait was self-descriptive (‘Do you think you are trait?’), 24 asked whether the trait was other-person-descriptive (‘Do you think Tony Blair is trait?’) and 24 asked whether the trait and a comparison word had similar-sounding endings (‘Does comparator rhyme with trait?’). Mental illness traits (e.g., ‘mad’), negative personality traits (e.g., ‘rude’) and positive personality traits (e.g., ‘wise’) were used, with each trait-type divided into three ‘trait-groups’ of eight traits that were matched for word-length, orthographic similarity4 and semantic similarity,5 and the three trait-groups from each trait-type divided equally between the three encoding-levels (resulting in nine conditions of eight traits each; Table S3, Appendix A). The mental illness traits were all related to schizophrenia, the negative and positive traits were all of extremely negative and positive valence, respectively (in an attempt to match the relatively low prevalence of the mental illness traits), and none of the negative traits were synonyms of any mental illness trait or antonyms of any positive trait. Whilst the three trait-types were matched for word-length, the mental illness traits appeared to have a higher number of common prefixes and suffixes and lower linguistic frequency than the other trait-types. Each rhyme-evaluative question used a unique rhyme-comparator of neutral valence that was semantically unrelated to any of the traits used in the study, with half of the rhymecomparators in each trait-type rhyming with their partnered trait. The encoding questions were presented in one of three different presentation orders, with each trait-group assigned to a different encoding-level for each order, but with the traits themselves always presented in the same order. No more than two questions of the same encoding-level were presented consecutively, and no two traits of the same trait-type, or that began with the same letter, or that had similar-sounding endings, were presented consecutively. 2.3. Procedure The encoding stage began following a practice session, with participants responding either ‘Yes / A bit’ (full or partial endorsement) or ‘No, not at all’ (complete rejection) to each question, a response choice devised to encourage more careful and representative responding about trait-ownership than a more traditional ‘Yes’ / ‘No’ dichotomy might. A surprise recall stage followed the encoding stage, with participants instructed to tell the researcher as many of the previously seen ‘words in red’ as they could. This was followed by a surprise recognition stage, in which participants decided whether the presented trait (either a target-trait or a synonymous distractor-trait; Table S3, Appendix A) had been shown at the encoding stage (‘Was this word shown before?’) by responding either ‘Yes’ or ‘No’, with the 74 target-traits and 74 intermixed distractor-traits presented in the same pseudo-randomised order to all participants. Each encoding stage and recognition stage question appeared in the centre of a black computer screen in white lettering and the trait italicised in red, remained on-screen until a response was made, and was succeeded by the next question after a 333 ms interval. Participants were instructed to respond to each question via one of two
4 Within each trait-type, the three trait-groups had similar numbers of traits beginning and ending with common syllables (e.g., ‘de-’ and ‘-al’). 5 Each trait-type was divided into eight ‘trait-elements’ of three particularly synonymous traits each (e.g., ‘ill’, ‘poorly’, ‘unwell’), with each trait-element divided equally between the three trait-groups.
91
clearly-labelled buttons as soon as they had decided which response most closely matched their honest, personal opinion about the answer. Following completion of the recognition stage, all participants were debriefed about the experiment, and all indicated that there had been no unfamiliar traits and that the retrieval stages had been unexpected. 2.4. Statistical analysis Each experimental variable (encoding response-type, encoding speed, recall, recognition sensitivity) was analysed using a 3 × 3 × 3 mixed analysis of variance and follow-up pairwise comparisons (PC) or t-tests, with Spearman's correlations also employed. The significance level of p ≤ 0.05 was Bonferroni-adjusted (i.e., divided by the number of tests involved) for hypothesis-testing but not exploratory analysis involving multiple t-tests and correlations. Any response of 100 ms or less was discarded, and all individual response times were logarithmically transformed to remove distribution skewness prior to analysis (anti-logarithmically transformed means are presented). For the recognition stage, d′ recognition sensitivity scores were calculated from hit rates and false alarm rates to measure participants' ability to correctly recognise target-traits whilst correctly rejecting distractor-traits (Stanislaw and Todorov, 1999). The term ‘SRE’ will refer to the retrieval advantage for self-evaluated traits over Blair-evaluated traits. 3. Results 3.1. Participants Data are presented in Table 1. All groups were matched for age, gender, nationality and awareness of Tony Blair, and the patient groups were also matched for duration of illness, depression (Beck Depression Inventory; Beck et al., 1979), verbal fluency (animal naming semantic subtest, Controlled Oral Word Association Test; Benton and Hamsher, 1976), verbal intelligence (vocabulary definition subtest, Wechsler Abbreviated Scale of Intelligence; Wechsler, 1999) and working memory (forward and backward digit-span subtests, Wechsler Adult Intelligence Scale; Wechsler, 1997). The healthy controls had significantly lower depression than the patient groups, significantly higher verbal fluency than the denial patients, and significantly higher verbal intelligence and working memory than the acceptance patients. The acceptance patients had significantly lower clinician-rated denial of illness, lower illness-severity (Positive and Negative Syndrome Scale; Kay et al., 1987) and a lower proportion of inpatients than the denial patients. 3.2. Encoding response-type Data are presented in Fig. 1.1 and Table S1.1 (Appendix A). There was no significant effect of participant-group or encoding-level on the number of endorsements made and no significant interaction between the two, i.e., the three groups made similar 3:5 ratios of ‘Yes / A bit’ to ‘No, not at all’ responses at each of the three encoding-levels (Fig. 1.1d). There was a significant three-way interaction (F (5.4, 128.5) = 4.4, p b 0.01; Fig. 1.1a,b,c), and exploratory t-tests revealed that the acceptance patients self-endorsed significantly more mental illness traits than the healthy controls (t (18.8) = 4.33, p b 0.001) and denial patients (t (24) = 2.43, p b 0.05), with no significant difference between the healthy controls and denial patients, and that the acceptance patients self-endorsed significantly fewer positive traits than the healthy controls (t (37) = 2.56, p b 0.01; no further significant between-group differences found). 3.3. Encoding speed Data are presented in Fig. 1.2 and Table S1.2 (Appendix A). Overall encoding speed significantly differed between the groups (F (2,
92
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
Table 1 Characteristics of the participant groups. Healthy controls (HC; N = 25)
Acceptance patients (AP; N = 14)
Denial patients (DP; N = 12)
Group comparisons
Mean age (In years)
35.4 (±2.5)
37.9 (±2.9)
32.8 (±2.3)
HC ≈ AP ≈ DP
Gender (Male / female)
13 / 12 (52% / 48%)
7/7 (50% / 50%)
9/3 (75% / 25%)
HC ≈ AP ≈ DP
Nationality (British / non-British)
17 / 8 (68% / 32%)
12 / 2 (76% / 14%)
12 / 0 (100% / 0%)
HC ≈ AP ≈ DP
Mean duration of schizophrenia (In years)
–
10.0 (±2.4)
11.6 (±2.2)
AP ≈ DP
Patient status (Inpatients / outpatients)
–
2 / 12 (10% / 90%)
10 / 2 (90% / 10%)
AP ≠ DP (χ2 (1) = 11.0, p b 0.001)
PANSS (30–210)
–
78.9 (±3.1)
102.3 (±2.5)
AP b DP (t (24) = 6.0, p b 0.001)
SAI-E (0–28)
–
19.2 (±1.6)
5.0 (±0.6)
AP N DP (t (24) = 8.45, p b 0.001)
BDI-IA (0–63)
5.9 (±1.1)
14.6 (±3.3)
13.0 (±2.5)
HC b AP (t (37) = 2.77, p b 0.01) HC b DP (t (35) = 2.81, p b 0.01) AP ≈ DP
COWAT verbal fluency (N animals named in 1 minute)
23.7 (±1.2)
20.1 (±1.7)
17.2 (±1.6)
HC b AP (t (37) = 1.72, p b 0.10) HC b DP (t (35) = 3.19, p b 0.01) AP ≈ DP
WASI vocabulary definition (T-score 20–80)
57.9 (±1.9)
46.1 (±3.1)
52.3 (±3.7)
HC N AP (t (37) = 3.48, p = 0.001) HC ≈ DP AP ≈ DP
WAIS-III digit-span (0–30)
19.8 (±0.9)
15.7 (±1.0)
17.8 (±0.9)
HC N AP (t (37) = 2.76, p b 0.01) HC ≈ DP AP ≈ DP
Tony Blair awareness (0–4)
2.4 (±0.3)
2.1 (±0.2)
1.8 (±0.2)
HC ≈ AP ≈ DP
± = standard error; PANSS = Positive and Negative Syndrome Scale total score; SAI-E = expanded Schedule for the Assessment of Insight total score; BDI-IA = revised Beck Depression Inventory total score; COWAT = Controlled Oral Word Association Test; WASI = Weschler Abbreviated Scale of Intelligence; WAIS-III = Wechsler Adult Intelligence Scale, third edition. Participants rated their awareness of the public persona of Tony Blair on a five-point scale ranging from ‘complete unfamiliarity’ (0) to ‘complete familiarity’ (4).
48) = 10.0, p b 0.001), with the healthy controls (M = 2614 ms, SE = 277) faster than the acceptance patients (M = 4629 ms, SE = 370; significant PC: p b 0.001) and denial patients (M = 3528 ms, SE = 438; near-significant PC: p b 0.10), and no significant difference between the patient groups. There was a significant interaction between encoding-level and participant-group (F (4, 96) = 7.16, p b 0.001), with Fig. 1.2d indicating that both patient groups were particularly slow during self-evaluation (contrary to the study's hypothesis). In support of this, exploratory t-tests revealed that, whilst the healthy controls made significantly faster responses at self-evaluation than Blair-evaluation (t (24) = 4.08, p b 0.001), this was non-significantly reversed in both patient groups. There was a significant three-way interaction (F (8, 192) = 1.99, p b 0.05), with Fig. 1.2a,b,c indicating that both patient groups were particularly slow during the self-evaluation of mental illness traits. In support of this, exploratory t-tests revealed that, whilst the healthy controls made significantly faster responses to self-evaluated than Blair-evaluated mental illness traits (t (24) =2.21, p b 0.05), this was near-significantly reversed in the acceptance patients (t (13) = 1.55, p = 0.07) and denial patients (t (11) = 1.58, p = 0.07).
3.4. Recall Data are presented in Fig. 2.1 and Table S2.1 (Appendix A). Overall numbers of correctly recalled traits significantly differed between the groups (F (2, 48) = 3.91, p b 0.05), with the healthy controls (M = 11.8, SE = 1.00) recalling significantly more traits than the denial patients (M = 7.3, SE = 1.11; PC: p b 0.05) but not the acceptance patients (M = 8.9, SE = 1.45), and no significant difference between the patient groups. There was a significant interaction between encodinglevel and participant-group (F (4, 96) = 2.51, p b 0.05; Fig. 2.1d) and no further interaction with trait-type (Fig. 2.1a,b,c). Exploratory t-tests revealed that the healthy controls recalled significantly more self-evaluated traits than the acceptance patients (t (37) = 2.12, p b 0.05) and denial patients (t (35) = 2.71, p b 0.01), with no significant difference between the patient groups, and that, whilst the healthy controls demonstrated a significant SRE overall (t (24) =2.53, p b 0.01), both patient groups did not. Three hypothesis-testing, Bonferroni-adjusted t-tests revealed that the denial patients recalled fewer self-evaluated mental illness traits than the acceptance patients (non-significant: t (24) = 1.11, p = 0.14) and healthy controls (near-significant: t (35) = 1.87,
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
‘Yes / A Bit’ Responses (0–8)
1.1
7
Self
Blair
93
Rhyme
6 5 4 3 2 1 0
Mental Negative Positive (a) Healthy Controls (HC)
Response Time (in milliseconds)
1.2
7000
Mental Negative Positive (b) Acceptance Patients (AP)
Self
Mental Negative Positive (c) Denial Patients (DP)
Blair
Combined Traits (d) HC AP DP
Rhyme
6000 5000 4000 3000 2000 1000 0
Mental Negative Positive (a) Healthy Controls (HC)
Mental Negative Positive (b) Acceptance Patients (AP)
Mental Negative Positive (c) Denial Patients (DP)
Combined Traits (d) HC AP DP
Fig. 1. Mean numbers of ‘Yes / A Bit’ responses (1.1) and mean response times (1.2) at encoding. Figure 1.1: graphs a,b,c depict a significant interaction between trait-type, encoding-level and participant-group, whereas graph d depicts a lack of significant interaction between encoding-level and participant-group over all trait-types. Figure 1.2: graphs a,b,c depict a significant interaction between trait-type, encoding-level and participant-group, and graph d depicts a significant interaction between encoding-level and participant-group over all trait-types. Error bar = standard error; mental = mental illness; corresponding data presented in Tables S1.1 and S1.2 (Appendix A).
p = 0.035), with no significant difference between the acceptance patients and healthy controls. 3.5. Recognition sensitivity Data are presented in Fig. 2.2 and Table S2.2 (Appendix A). Overall recognition sensitivity did not significantly differ between the healthy controls (M = 1.35, SE = 0.08), acceptance patients (M = 1.26, SE = 0.11) and denial patients (M = 1.29, SE = 0.12). A significant effect of encoding-level was found (F (2, 96) = 35.8, p b 0.001), with a SRE demonstrated over all participants (PC: p b 0.05). Whilst there was no significant interaction with participant-group (Fig. 2.2d), and whilst exploratory t-tests revealed that the groups did not significantly differ in their recognition for self-evaluated traits, further exploratory t-tests revealed that the acceptance patients failed to demonstrate the significant SRE displayed by the healthy controls (t (24) = 3.07, p b 0.05) and denial patients (t (11) = 2.70, p b 0.05). There was a significant three-way interaction (F (8, 192) = 3.49, p b 0.001; Fig. 2.2a,b, c). Exploratory t-tests revealed that, whilst the healthy controls displayed a significant SRE for mental illness traits (t (24) = 1.71, p = 0.05), this was reduced for the acceptance patients (t (13) = 0.76, p = 0.23) and significantly reversed for the denial patients (t (11) = 1.81, p b 0.05), and whilst the healthy controls (t (24) = 3.30, p b 0.01) and denial patients (t (11) = 3.87, p b 0.01) displayed a
significant SRE for positive traits, this was near-significantly reversed for the acceptance patients (t (13) = 1.35, p = 0.10). Three hypothesis-testing, Bonferroni-adjusted t-tests revealed no significant differences between the three groups' recognition for self-evaluated mental illness traits.
3.6. Correlational analysis Three pairs of hypothesis-testing correlations were conducted in the total patient sample, each using a Bonferroni-adjusted significance level of p ≤ 0.016. Firstly, small / very small, non-significant correlations of inconsistent direction were found between self-reflection speed and denial of illness, both in terms of patients' self-endorsement of mental illness traits at encoding and SAI-E clinician ratings.6 Secondly, large / medium, significant correlations were found between lower recall for self-evaluated mental illness traits and higher denial of illness, both self-rated (r s = .56, p = 0.002) and clinician-rated (r s = .45, p = 0.01). Thirdly, small, non-significant correlations were found between lower recognition for self-evaluated mental illness traits and higher denial of illness, both self-rated and clinician-rated. 6 With regard to r correlation sizes,‘very small’ = .01 – .09, ‘small’ = .10 – .29, ‘medium’ = .30 – .49, ‘large’ = .50 – .69, and ‘very large’ = .70 – 1.00.
94
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
2.1 Self
Recalled Traits (0–8)
2.5
Blair
Rhyme
2
1.5
1
0.5
0
Mental Negative Positive (a) Healthy Controls (HC)
2.2
Self
2.5
Recognition Sensitivity (in d’)
Mental Negative Positive (b) Acceptance Patients (AP)
Mental Negative Positive (c) Denial Patients (DP)
Blair
Combined Traits (d) HC AP DP
Rhyme
2
1.5
1
0.5
0
Mental Negative Positive (a) Healthy Controls (HC)
Mental Negative Positive (b) Acceptance Patients (AP)
Mental Negative Positive (c) Denial Patients (DP)
Combined Traits (d) HC AP DP
Fig. 2. Mean numbers of recalled traits (2.1) and mean recognition sensitivity scores (2.2) at retrieval. Figure 2.1: graphs a,b,c depict a lack of significant interaction between trait-type, encoding-level and participant-group, whereas graph d depicts a significant interaction between encoding-level and participant-group over all trait-types. Figure 2.2: graphs a,b,c depict a significant interaction between trait-type, encoding-level and participant-group, whereas graph d depicts a lack of significant interaction between encoding-level and participant-group over all trait-types. Error bar = standard error; mental = mental illness; corresponding data presented in Tables S1.1 and S1.2 (Appendix A).
In an exploratory examination of potential confounding factors, no significant correlations were found in the total patient sample between recall for self-evaluated mental illness traits and either depression, verbal fluency, verbal intelligence, working memory or self-reflection speed. However, a medium, significant correlation was found between lower recall for self-evaluated mental illness traits and higher illness-severity (r s = .45, p b 0.05).
3.7. Explicit and implicit self-perception In an examination of the self-positivity bias at an explicitlystated level, exploratory t-tests revealed that, whilst the healthy controls (t (24) = 6.94, p b 0.001; Fig. 1.1a) and denial patients (t (11) = 2.68, p b 0.05; Fig. 1.1c) self-endorsed significantly more positive than negative traits, the acceptance patients did not (Fig. 1.1b). Regarding the self-positivity bias at an arguably implicit level, exploratory t-tests revealed that, whilst the healthy controls recalled (near-significantly: t (24) = 1.33, p b 0.10; Fig. 2.1a) and recognised (significantly: t (24) = 1.86, p b 0.05; Fig. 2.2a) more self-evaluated positive than negative traits, the acceptance patients (Figs. 2.1b and 2.2b) and denial patients (Figs. 2.1c and 2.2c) did not; indeed, the acceptance patients recognised significantly fewer self-evaluated positive than negative traits (t (24) = 2.41, p b 0.05).
4. Discussion 4.1. Self-reflection speed and denial of illness Both patient groups were generally slower than the healthy controls at the encoding stage and particularly slow when self-evaluating mental illness traits (but not personality-related traits). This finding corresponds with the observation that the brain regions involved in self-reflection and executive function are unusually overactive during mental illness-related (but not personality-related) self-evaluation in schizophrenia (Bedford et al., 2012). There was no evidence that denial of illness is related to speed of self-reflection, with the present findings indicating that schizophrenia patients, regardless of their illness-attitudes, find mental illness-related self-evaluation particularly difficult, and that it may provoke increased uncertainty, presentational concern and emotional discomfort in acceptance patients and denial patients alike. This slowness also indicates that patients with denial of illness have an implicit awareness of the potential self-relevance of mental illness, despite their explicit rejection of it, although it is unclear whether this awareness is consciously or subconsciously held. 4.2. Self-perception and denial of illness The acceptance patients displayed a high self-endorsement of mental illness traits, a low self-endorsement of positive traits, and a
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
reduced self-positivity bias at encoding. These findings were perhaps to be expected, as studies tend to show a relationship between higher acceptance of illness and higher depression in schizophrenia (Mintz et al., 2003), and between higher depression and a lower self-positivity bias more generally (Watson et al., 2008). However, whilst both patient groups reported similarly higher levels of depression than the healthy controls, the denial patients still displayed a significant self-positivity bias at encoding, raising the intriguing speculation that, at an explicitly-stated level, the denial patients' personality-related self-perception was more resistant to depression than the acceptance patients'. The acceptance patients had particularly poor recognition for self-evaluated positive traits, in line with other studies that have shown reduced retrieval for self-evaluated positive traits in mildly (Kuiper and Derry, 1982) and severely (Derry and Kuiper, 1981) depressed people with a reduced self-positivity bias. The acceptance patients' failure to retrieve more self-evaluated positive than negative traits perhaps indicated that they held a reduced self-positivity bias at an implicit level, and that their implicitly-held and explicitlystated self-perception were congruent. Similarly, the healthy controls' tendency to retrieve more self-evaluated positive than negative traits indicated a possible correspondence between their implicit and explicit self-perception. However, there appeared to be a disparity between implicit and explicit self-perception in the denial patients, who shared the acceptance patients' reduced self-positivity bias at retrieval, but the healthy controls' intact self-positivity bias at encoding. This could be interpreted as indicating that patients with denial of illness implicitly hold a far less positive self-perception than they explicitly display. However, it is uncertain whether implicit selfschemas can be gauged by retrieval performance in this way, and if so, whether they are consciously or subconsciously held. 4.3. Self-reflection depth and denial of illness Positive traits excepted, both patient groups' recognition for self-evaluated traits appeared to be intact. Although the denial patients' recognition-SRE for mental illness traits was significantly reversed, supporting the hypothesis that they engaged in particularly shallow mental illness-related self-reflection, this may have been due to their unusually high recognition for Blair-evaluated mental illness traits (Fig. 2.2c). In contrast to their generally intact recognition, both patient groups displayed particularly reduced recall for self-evaluated traits in general. Whilst the denial patients did display lower recall for self-evaluated traits (mental illness-related and overall) than the acceptance patients, there was arguably insufficient statistical power for these differences to achieve significance because of the low group numbers involved. Indeed, when more adequately-powered analyses were performed over the total patient sample, significant correlations emerged between lower recall for self-evaluated mental illness traits and (1) higher denial of illness and (2) higher illness-severity. These correlations support the hypothesis that denial of illness in schizophrenia is related to a self-related recall deficit that worsens as the illness becomes more severe, and is in line with other studies' consistent demonstration of a relationship between higher denial of illness and higher illness-severity in schizophrenia (Mintz et al., 2003). However, it is worth noting that the acceptance patients also displayed poor self-related recall relative to the healthy controls, indicating that this deficit alone is not sufficient for denial of illness to occur. The underlying nature of the SRE is uncertain, with theories usually postulating that it results from the self-concept being a particularly well-organised, emotion-arousing and often-used schema (Symons and Johnson, 1997). The present recall-recognition dissociation indicates that the patients did benefit from the deep processing that normally accompanies self-evaluation and leads to the SRE, but only when presented with recognition cues that they did not have to
95
generate themselves. According to a theory posited by Klein and Kihlstrom (1986), this may have been because, during encoding, the patients had intact ‘item-specific’ self-processing of individual traits, but poor ‘organisational processing’ of the different self-evaluated traits around the self-concept, i.e., poor implicit formation of inter-trait connections during self-evaluation. Such a problem would theoretically impact on the ability of one recalled self-evaluated trait to trigger the recall of another, impairing recall more than recognition because of the former's greater reliance on the self-generation of cues. This explanation of the recall-recognition dissociation is in line with evidence from other studies that poor recall in schizophrenia reflects a disturbance of organisational processing, with schizophrenia patients producing less clustered recall of semantically-related words than healthy controls (Paulsen et al., 1995), but improving when the semantic relatedness of the words is emphasised at encoding (Gold et al., 1992). The present findings are also congruent with results from a meta-analysis that showed recall to be generally more impaired in schizophrenia than recognition (Aleman et al., 1999).
4.4. Conclusions The present study found evidence for the following conclusions. Firstly, schizophrenia patients, regardless of their illness-attitudes, are particularly slow at making decisions about their mental health, and may experience particular difficulty during this exercise. Patients with denial of illness are implicitly aware of the potential self-relevance of mental illness, despite their explicit rejection of it. Secondly, patients with denial of illness explicitly present themselves in a generally more positive than negative light, a normal self-serving bias that may however be at odds with a more negative implicitly-held self-schema and an elevated level of depression. This may indicate rigid, inflexible self-perception in patients with denial of illness. Thirdly, schizophrenia patients, regardless of their illness-attitudes, have a particular deficit in recalling new self-related episodes that worsens with increasing denial of illness and illness-severity, and that may be due to impaired organisational processing during self-reflection. It could be speculated that this deficit contributes to rigid self-perception and an entrenched sense of being mentally healthy at a privately-held level of belief. If so, then denial of illness in schizophrenia may be regarded less as a conscious impression-management strategy and more as a genuine disturbance of self-awareness and insight.
4.5. Limitations The present findings were derived from the use of small samples and several exploratory analyses, and require replication with larger samples. Analyses involving the patient subgroups had low statistical power, increasing the likelihood of differences not achieving statistical significance. The healthy controls were not on antipsychotic medication and had lower depression, higher verbal fluency, higher verbal intelligence and higher working memory than the patients, although there was no evidence that the last four variables were related to the selfrelated recall deficit. Whilst the difference in illness-severity and inpatient proportion between the two patient groups represented a potential confound, it was perhaps unavoidable given the aforementioned relationship that appears to exist between denial of illness and illness-severity in schizophrenia.
Role of the funding source The study was funded by a Medical Research Council PhD studentship to Dr. Bedford. Prof. David acknowledges support from the National Institute For Health Research Biomedical Research Centre at SLAM and the Institute of Psychiatry, King's College London. Neither funding source played a further role in the study.
96
N.J. Bedford, A.S. David / Schizophrenia Research 152 (2014) 89–96
Contributors Prof. David conceived the original idea for the study, supervised the project and edited the manuscript. Dr. Bedford conceived the idea to examine mental illness-related selfevaluation, designed the study, wrote the computer program, tested the participants, performed the analysis and wrote the manuscript. Conflict of interest The authors have no conflicts of interest to declare. Acknowledgements We thank Dr. Simon Surguladze for help with patient recruitment, Mr. Christopher Maltby for computer programming assistance, Dr. Nicholas Sharples for general computer-related wizardry, and the Schizophrenia Research reviewers for their insightful comments.
Appendix A. Supplementary data Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.schres.2013.07.006. References Aleman, A., Hijman, R., de Haan, E.H.F., Kahn, R.S., 1999. Memory impairment in schizophrenia: a meta-analysis. Am. J. Psychiatry 156, 1358–1366. Amador, X.F., Strauss, D.H., Yale, S.A., Gorman, J.M., 1991. Awareness of illness in schizophrenia. Schizophr. Bull. 17, 113–132. Amador, X.F., Flaum, M., Andreasen, N.C., Strauss, D.H., Yale, S.A., Clark, S.C., Gorman, J.M., 1994. Awareness of illness in schizophrenia and schizoaffective and mood disorders. Arch. Gen. Psychiatry 51, 826–836. American Psychiatric Association, 2000. Diagnostic and Statistical Manual of Mental Disorders: DSM-IV-TR. text revision 4th ed. American Psychiatric Association, Washington. Beaumont, J.G., Rogers, M.J.C., Kenealy, P.M. (Eds.), 1996. The Blackwell Dictionary of Neuropsychology. Blackwell, Oxford. Beck, A.T., Rush, A.J., Shaw, B.F., Emery, G., 1979. Cognitive Therapy of Depression. Guildford Press, New York. Bedford, N.J., 2010. Denial of Illness in Schizophrenia as a Disturbance of Insight and Self-Awareness. PhD Thesis Institute of Psychiatry, King's College London Library. Bedford, N.J., Surguladze, S., Giampietro, V., Brammer, M.J., David, A.S., 2012. Self-evaluation in schizophrenia: an fMRI study with implications for the understanding of insight. BMC Psychiatry 12, 106. Benton, A.L., Hamsher, K., 1976. Multilingual Aphasia Examination. University of Iowa, Iowa City. Clifford, P.I., Hemsley, D.R., 1987. The influence of depression on the processing of personal attributes. Br. J. Psychiatry 150, 98–103. Craik, F.I.M., Lockhart, R.S., 1972. Levels of processing: a framework for memory research. J. Verbal Learn. Verbal Behav. 11, 671–684. Craik, F.I.M., Tulving, E., 1975. Depth of processing and the retention of words in episodic memory. J. Exp. Psychol. 104, 268–294. David, A.S., 1990. Insight and psychosis. Br. J. Psychiatry 156, 798–808. Derry, P., Kuiper, N., 1981. Schematic processing and self-reference in clinical depression. J. Abnorm. Psychol. 90, 286–297. Forster, K.I., Forster, J.C., 2003. DMDX: a Windows display program with millisecond accuracy. Behav. Res. Methods Instrum. Comput. 35, 116–124. Frith, C.D., 1979. Consciousness, information processing and schizophrenia. Br. J. Psychiatry 134, 225–235.
Frith, C.D., 1992. The Cognitive Neuropsychology of Schizophrenia. Psychology Press, Hove, East Sussex. Gold, J.M., Randolph, C., Carpenter, C.J., Goldberg, T.E., Weinberger, D.R., 1992. Forms of memory failure in schizophrenia. J. Abnorm. Psychol. 101, 487–494. Harvey, P., Lee, J., Horan, W.P., Ochsner, K., Green, M.F., 2011. Do patients with schizophrenia benefit from a self-referential memory bias? Schizophr. Res. 127, 171–177. Holt, D.J., Cassidy, B.S., Andrews-Hanna, J.R., Lee, S.M., Coombs, G., Goff, D.C., Gabrieli, J.D., Moran, J.M., 2011. An anterior-to-posterior shift in midline cortical activity in schizophrenia during self-reflection. Biol. Psychiatry 69, 415–423. Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The Positive and Negative Syndrome Scale (PANSS) for schizophrenia. Schizophr. Bull. 13, 261–276. Kemp, R., David, A.S., 1997. Insight and compliance. In: Blackwell, B. (Ed.), Treatment Compliance and the Therapeutic Alliance in Serious Mental Illness. Harwood Academic Publishers, Amsterdam, pp. 61–84. Klein, S.B., Kihlstrom, J.F., 1986. Elaboration, organization, and the self-reference effect in memory. J. Exp. Psychol. Gen. 115, 26–38. Kuiper, N.A., Derry, P.A., 1982. Depressed and nondepressed content self-reference in mild depressives. J. Pers. 50, 67–80. Mason, O., Claridge, G., Jackson, M., 1995. New scales for the assessment of schizotypy. Personal. Individ. Differ. 18, 7–13. Mezulis, A.H., Abramson, L.Y., Hyde, J.S., Hankin, B.L., 2004. Is there a universal positivity bias in attributions? A meta-analytic review of individual, developmental, and cultural differences in the self-serving attributional bias. Psychol. Bull. 130, 711–747. Mintz, A.R., Dobson, K.S., Romney, D.M., 2003. Insight in schizophrenia: a meta-analysis. Schizophr. Res. 61, 75–88. Northoff, G., Heinzel, A., de Greck, M., Bermpohl, F., Dobrowolny, H., Panksepp, J., 2006. Self-referential processing in our brain: a meta-analysis of imaging studies on the self. Neuroimage 31, 440–457. Paulsen, J.S., Heaton, R.K., Sadek, J.R., Perry, W., Delis, D.C., Braff, D., Kuck, J., Zisook, S., Jeste, D.V., 1995. The nature of learning and memory impairments in schizophrenia. J. Int. Neuropsychol. Soc. 1, 88–99. Pauly, K., Kircher, T., Weber, J., Schneider, F., Habel, U., 2011. Self-concept, emotion and memory performance in schizophrenia. Psychiatry Res. 186, 11–17. Rogers, T.B., Kuiper, N.A., Kirker, W.S., 1977. Self-reference and the encoding of personal information. J. Pers. Soc. Psychol. 35, 677–688. Shad, M.U., Keshavan, M.S., Steinberg, J.L., Mihalakos, P., Thomas, B.P., Motes, M.A., Soares, J.C., Tamminga, C.A., 2012. Neurobiology of self-awareness in schizophrenia: an fMRI study. Schizophr. Res. 138, 113–119. Stanislaw, H., Todorov, N., 1999. Calculation of signal detection theory measures. Behav. Res. Methods Instrum. Comput. 31, 137–149. Symons, C.S., Johnson, B.T., 1997. The self-reference effect in memory: a meta-analysis. Psychol. Bull. 121, 371–394. Van der Meer, L., Costafreda, S.C., Aleman, A., David, A.S., 2010. Self-reflection and the brain: a theoretical review and meta-analysis of neuroimaging studies with implications for schizophrenia. Neurosci. Biobehav. Rev. 34, 935–946. Van der Meer, L., de Vos, A.E., Stiekema, A.P.M., Pijnenborg, G.H.M., van Tol, M., Nolen, W.A., David, A.S., Aleman, A., 2012. Insight in schizophrenia: involvement of selfreflection networks? Schizophr. Bull. http://dx.doi.org/10.1093/schbul/sbs122 (Retrieved 2013). Watson, L.A., Dritschel, B., Jentzsch, I., Obonsawin, M.C., 2008. Changes in the relationship between self-reference and emotional valence as a function of dysphoria. Br. J. Psychol. 99, 143–152. Wechsler, D., 1997. Wechsler Adult Intelligence Scale, third ed. The Psychological Corporation, San Antonio, Texas. Wechsler, D., 1999. Wechsler Abbreviated Scale of Intelligence (WASI). The Psychological Corporation, San Antonio, Texas. World Medical Association, 2008. Declaration of Helsinki — ethical principles for medical research involving human subjects, 6th revision. www.wma.net/en/30publications/ 10policies/b3 (Retrieved 2013).