INTRODUCTION
Hallucinations constitute a very common and distressing symptom in patients with schizophrenia. Recent research has attempted to determine the cognitive mechanisms that underlie them to enable development of more effective therapies. Hallucinations have most often been unrelated to cognitive efficiency. However, they were repeatedly found associated with deficit in a specific function, namely source monitoring. Source monitoring refers to the ability to remember not the information itself, but the origin of this information (Johnson et al., 1993). One type of source monitoring is the ability to make a distinction between self-generated information and information produced by an external source (self-monitoring). Several studies of schizophrenic patients have revealed an association between auditory hallucinations and tendency to misattribute one's own thoughts or speech to an external source (Ditman & Kuperberg, 2005, for review). Source monitoring may also refer to the ability to distinguish between an event that occurred in the real world and an imaginary event (reality-monitoring). This function has also been associated with hallucinations in schizophrenia. In particular, hallucinations were associated with an abnormality in a certain type of response bias, reflecting the tendency to detect events that have not occurred. Bentall and Slade (1985) observed that, in nonpsychiatric individuals predisposed to hallucinations as well as in patients with schizophrenia, hallucinations were related to an increased tendency to report hearing a signal in a noise background. Rankin and O'Carroll (1995) also reported this association in a nonpsychiatric sample. These findings were derived from signal detection tasks. We attempted to extend them to memory processes, using word recognition tasks. In two previous studies of patients with schizophrenia (Brébion et al., 1998, 2005), we observed that higher global hallucination scores were associated with increased response bias toward false recognition of words which were not presented in the target lists. However, in the data of Ragland et al. (2003), response bias in word recognition was related to delusions rather than to hallucinations.
Another type of source monitoring is memory for the context in which the information was presented. Context refers to the spatiotemporal features associated with the stimuli. Considering the well-established link between source monitoring deficit and hallucinations, one could expect that hallucinations would also be linked with context memory deficit, as already suggested (Hemsley, 2005; Servan-Schreiber et al., 1996). Accordingly, in our previous work on source memory errors in schizophrenia, we observed that a global hallucination score was correlated with errors in remembering the temporal context of word presentation (Brébion et al., 2002). In subsequent research, we replicated this correlation and established that it was specific to auditory hallucinations (Brébion et al., 2007). Waters et al. (2006) reported that auditory hallucinations were associated with confusion in remembering the temporal context of actions.
Spatial context has been the object of fewer studies than temporal context. Impairment in remembering the spatial location of words or objects has been reported in patients with schizophrenia (Rizzo et al., 1996; van't Wout et al., 2006). However, whether auditory hallucinations are also associated with such impairment has not been tested, as far as we know.
Most often, the association with source memory deficit pertains to auditory hallucinations, which have been extensively studied due to their high prevalence in schizophrenia (David, 2004). No cognitive underpinnings of visual hallucinations in patients with schizophrenia have been established so far. One reason for this may be that previous cognitive research on hallucinations has mostly used verbal or auditory stimuli. In this study, we investigated the relationships between source memory deficit and both auditory and visual hallucinations using visual material, on the assumption that potential relationships with visual hallucinations were more likely to emerge with visual material. A picture recognition task was administered to a schizophrenic sample, followed by a spatial context memory task. We studied the associations of hallucinations and delusions with response bias toward false recognition of pictures, and with errors in remembering the spatial context of picture presentation. Our hypotheses were as follows: Auditory and visual hallucination scores would be positively correlated with the response bias, which would suggest a supramodal dysfunction in source monitoring. Auditory and visual hallucination scores would also be positively correlated with the number of spatial context errors, thereby extending the correlation with temporal context errors. On the other hand, no correlation with the delusion score was expected, which would illustrate the specificity of the associations with hallucinations. Half of the target pictures were presented in color and half in black-and-white, to vary encoding conditions. In exploratory analyses, we investigated whether this factor had some influence on the expected associations with hallucinations.
Studies of source memory impairment in schizophrenia are most often confined to the associations with positive symptoms. We observed in previous research that various types of memory error, which were increased by hallucinations, were also unexpectedly decreased by certain negative symptoms. In particular, response bias toward fewer false recognitions of words was associated significantly with anhedonia and, at a trend level, with affective flattening in two independent studies (Brébion et al., 1999, 2005). However, this inverse association between response bias and certain negative symptoms was not observed by Ragland et al. (2003). Heinrichs and McDermid Vaz (2004) reported that a global negative symptom score was associated with fewer intrusions in free recall in a sample of schizophrenic patients. They suggested that negative symptoms may involve the intensification of inhibitory mechanisms. Other types of error, mostly stemming from source memory failure, were found to be reduced along with negative symptom increment in patients with schizophrenia. Stirling et al. (2001) reported that higher negative symptom scores were associated with greater performance in a self-monitoring task. Errors in remembering the temporal context of word presentation were inversely associated with emotional and social withdrawal in one of our studies (Brébion et al., 2002) and with affective flattening and anhedonia in a subsequent study (Brébion et al., 2007). These findings suggest that cognitive mechanisms other than inhibition may also be involved in the inverse association between certain negative symptoms and memory errors.
In the current study, negative symptoms were assessed with the Scale for the Assessment of Negative Symptoms (SANS), which includes five symptoms. We hypothesized that affective flattening and anhedonia would be associated with decreased response bias toward false recognitions of pictures, and with fewer errors in remembering the spatial context of picture presentation. On the other hand, the other three negative symptoms of the scale were expected to be unrelated to either response bias or spatial context errors.
METHOD
Subjects
A total of 41 mostly outpatients with schizophrenia were recruited at the Maudsley Hospital, London (26 males, 15 females; 21 Caucasian, 16 Black, 4 Indian/Asiatics; age: mean = 34.2 years, SD = 8.0 years; education level: mean = 12.8; SD = 2.5; National Adult Reading Test: mean = 101.5; SD = 13.1). The diagnosis was made on basis of Diagnostic and Statistical Manual of Mental Disorders-Fourth Edition, criteria by two experienced psychiatrists who used clinical interview, patient history, and chart review, and reached a consensus. They also conducted the clinical assessment. Videotape training in clinical rating was undertaken to achieve over 90% interrater agreement in assigning subscale score. Exclusion criteria for the study were any evidence of alcohol or drug abuse, organic mental illness or mental impairment, history of brain injury, and current severe physical illness. All patients except three were on daily antipsychotic medication (olanzapine, quetiapine, amisulpiride, clozapine, risperidone, haloperidol). Ethical approval for the study was obtained from the South London and Maudsley NHS Trust Research Ethics Committee. After a full explanation of the study, subjects provided written informed consent to participate.
Clinical Ratings
Positive symptoms were assessed with the Scale for the Assessment of Positive Symptoms (SAPS). Twenty-four patients had significant hallucinations of some kind as reflected by a score ≥ 2 at the “global rating of hallucinations” item of the scale. Scores obtained at the auditory and visual hallucination items (scale range 0–5 for each item) were tallied: 17 patients had significant auditory hallucinations (score ≥ 2 on the “auditory hallucinations” item), and 8 patients had significant visual hallucinations (score ≥ 2 on the “visual hallucinations” item). Delusion score was computed by summing the subscores for various types of delusion. Negative symptoms were assessed with the SANS. A score for each specific negative symptom was computed by summing all the subscale scores. The clinical symptom assessment and cognitive testing were always conducted within the same week, in most cases the same day. The psychiatrists were blind to cognitive results during clinical assessment. The experimenter (G.B.) was blind to patients' symptoms during cognitive testing.
Materials
The visual stimuli were 32 black-and-white (b&w) and colored pictures from art museums, which did not readily lend themselves to verbal labeling. All the paintings were representational, and attempts were made to select those that would not be familiar to the subjects. Of the 32, 16 pictures (8 b&w and 8 colored) were to be presented as targets, and the remaining 16 (8 b&w and 8 colored) were to be used as distractors. The use of each picture as target or distractor was counterbalanced, so that pictures 1–16 were targets for half of the subjects, whereas pictures 17–32 were targets for the other half.
To counterbalance the effect of color, we prepared two equivalent sets of the 32 pictures (set A and set B). The 16 pictures that were b&w in one set were colored in the other set. Set A was used for half of the subjects and set B for the other half. This strategy resulted in four picture presentation conditions (set A/targets 1–16, set A/targets 17–32, set B/targets 1–16, set B/targets 17–32).
Four small cards were laid on the table at four locations (top left, top right, bottom left, bottom right) to indicate spatial context. Each picture was to be associated with a specific spatial context. The assignment of pictures to spatial context was counterbalanced among subjects according to four conditions, so that each picture was alternately presented in association with each of the four cards. Combined with the 4 above-mentioned picture presentation conditions, the design led to 16 experimental conditions. The subjects were randomly assigned to one of them.
Procedure
The subjects were instructed to watch the presented pictures and were informed that they would have to recognize them afterward. They were also required to pay attention to the spatial context of each picture. The pictures were laid on the table by groups of four mixed b&w and colored pictures over 20 seconds. Each picture was placed next to a small card that constituted the spatial context.
After the 16 pictures had been presented, the subjects were administered another task for a delay of approximately 5 min. The distracting task consisted of paper and pencil tests of visuomotor speed. Then the 16 target pictures mixed with the 16 distractor pictures were presented one by one in random order. The subjects were required to indicate for each of them whether it had been presented during the acquisition phase, or was new. The number of recognized target pictures, as well as the number of false recognitions of new pictures, was tallied for b&w and for colored pictures.
The recognition task was followed by a spatial context memory task. The four small cards that constituted spatial context were laid on the table again at the same locations as previously presented. The subjects were shown the 16 target pictures one by one in random order. They were informed that all of them had been presented and were required to remember which card each picture had been associated with. The number of spatial context errors for b&w and for colored pictures was tallied.
The numbers of correct recognitions of target pictures (m = 11.80) and false recognitions of nontarget pictures (m = 1.46) were used to compute the Pr and Br recognition indices according to Corwin's (1994) formula. The recognition index Pr, reflecting the accuracy in discriminating target pictures from distractors (rate of correct recognitions of target pictures minus rate of false recognitions of nontarget pictures), was computed for b&w and for colored pictures. The response bias index Br, reflecting the tendency to make false recognitions of nontarget pictures (rate of false recognitions of nontarget pictures/[1-Pr]), was also computed for both types of picture. Average Pr and Br indices were derived. All the measures obtained from the task (recognition indices and spatial context errors) and all the used clinical measures followed a normal distribution in the whole patient sample as well as in the subsample of 24 patients with any significant hallucinations.
RESULTS
The average Pr and Br indices were not significantly intercorrelated in the sample (r = .28; p < .08). The scores obtained on the recognition task, as well as the number of spatial context errors, are presented in Table 1.
Scores on the picture recognition and spatial context tasks (mean and standard deviation) for the 41 patients
Recognition Accuracy
Correlational studies between recognition accuracy indices and hallucination and delusion scores were conducted. The auditory hallucination score was not significantly correlated with Pr-b&w (r = −.16). However, a significant negative correlation between auditory hallucination score and Pr-color emerged (r = −.38; p < .015). This finding indicates that higher ratings of auditory hallucinations were associated with poorer recognition of the colored pictures. The strength of the correlation was much increased when only the subsample of 24 patients with definite hallucinations was considered (r = −.62; p < .001), whereas the correlation with Pr-b&w remained insignificant in this subsample (r = −.16). To deal with the overlap between auditory and visual hallucinations, we recomputed the correlation between auditory hallucination score and Pr-color after partialling out the visual hallucination score. The result was not altered (r = −.61). Visual hallucination score was not associated with Pr-b&w or Pr-color in either the whole sample or the subsample of 24 patients with definite hallucinations (all correlations near zero). In the whole sample, the delusion score was not significantly associated with either Pr-b&w (r = −.20) or Pr-color (r = −.19).
Response Bias
Correlations of response bias with hallucination and delusion scores were computed after partialling out the Pr indices. Considering that affective flattening and anhedonia were expected to present associations with the response bias opposite to those of hallucinations/delusions, the sum of affective flattening and anhedonia scores was also partialled out. In the subsample of 24 hallucinators, the average Br was significantly correlated with the auditory hallucination score (r = .45; p < .05). Adding the visual hallucination score to the covariates did not affect the size of this correlation (r = .45). The direction of the correlation indicates that higher ratings of auditory hallucinations were associated with increased tendency to make false recognition of nontarget pictures, as expected. It should be noted that auditory hallucinations and response bias were not significantly intercorrelated when a bivariate correlation was computed, which shows the influence of the controlled factors on the association.
In view of the differential association of auditory hallucinations with the recognition of black-and-white and colored pictures, we studied the correlation with the response bias separately for each type of picture. The correlation with auditory hallucination score was r = .43, p = .054 for Br-b&w, and r = .37, p = .10 for Br-color. Contrary to our expectations, the visual hallucination score was unrelated to the average Br in the subsample of 24 hallucinators (r near zero). The delusion score was unrelated to the average Br in the whole patient sample (r near zero).
Then, the correlations between response bias and negative symptoms were computed in the whole sample, to test the hypothesis that certain negative symptoms were associated with a more conservative response bias. The Pr indices were partialled out as was the auditory hallucination score. Contrary to our expectations, affective flattening was not associated with the average Br (r = −.14). A significant negative correlation between anhedonia and average Br was revealed (r = −.46; p < .005). It indicates that anhedonia was associated with reduced tendency to make false recognition of nontarget pictures, as expected. After controlling for auditory hallucination score, anhedonia was not correlated with either Pr-b&w (r = −.16) or Pr-color (r = −.15), which demonstrates the specificity of its association with the response bias. As expected, the average response bias was unrelated to alogia, avolition, or attention disorders (the three correlations were near zero).
Considering that the presence of hallucinations might influence potential inverse associations with negative symptoms, we recomputed the correlations of response bias with affective flattening and anhedonia in the subsample of 17 patients who did not present any significant hallucinations. The auditory hallucination score was removed from the covariates. The correlation between Br and anhedonia was strengthened (r = −.64; p < .01). A trend for a negative correlation between Br and affective flattening emerged (r = −.48; p < .07).
Spatial Context Errors
Correlational analyses between the total number of spatial context errors and hallucination and delusion scores were conducted. The Pr indices were partialled out, as well as the sum of affective flattening and anhedonia scores. In the subsample of 24 hallucinators, the total number of spatial context errors was significantly correlated with the auditory hallucination score (r = .49; p < .025). Thus, higher ratings of auditory hallucinations were associated with more errors in remembering the spatial context of picture presentation, as expected. The correlation was unchanged when the visual hallucination score was added to the covariates (r = .49). Again, auditory hallucinations and spatial context errors were not significantly associated when a mere bivariate correlation was used.
Considering that auditory hallucinations were restrictively associated with the recognition of the colored pictures, we tested whether they were similarly restrictively associated with errors in remembering the spatial context of the colored pictures. Separate analyses showed that the auditory hallucination score was in fact significantly correlated with the number of spatial context errors for the b&w pictures (r = .49; p < .025), whereas the correlation with the number of spatial context errors for the colored pictures did not reach statistical significance (r = .36; p < .11).
Contrary to our expectations, no correlation between total number of spatial context errors and visual hallucination score was revealed in the 24 hallucinating patients (r = −.14). The correlation with the delusion score in the whole sample was near zero.
Then, the correlations between negative symptoms and total number of spatial context errors were computed in the whole sample, after partialling out the Pr indices and the auditory hallucination score. Contrary to our expectations, the number of spatial context errors was not associated with either affective flattening (r = −.17) or anhedonia (r near zero).
DISCUSSION
In this study, we used visual material to investigate novel hypotheses about hallucinations. Clear associates of auditory hallucinations emerged. First, the intensity of auditory hallucinations was strongly correlated with deficit in the recognition of the colored pictures. On the other hand, auditory hallucinations were not associated with poor recognition of black-and-white pictures, which suggests that the patients who suffered from this symptom did not present a deficit in visual recognition per se. It is difficult to propose an explanation for this finding. We had no reason to expect a correlation between auditory hallucinations and picture recognition efficiency, and the dissociated pattern of colored versus noncolored pictures is puzzling. Patients suffering from auditory hallucinations might be unable to integrate the shape and color attributes of visual stimuli. The effect of colored versus achromatic stimuli could be investigated with other paradigms to better understand why color should have a negative effect on memory efficiency, and possibly other cognitive functions, in patients suffering from auditory hallucinations. The mixing of black-and-white and colored pictures at the acquisition phase is likely to have induced differential processing of each type of picture. It would be interesting to investigate whether the negative correlation between auditory hallucinations and recognition of colored pictures is replicated when only colored pictures are presented. Other tasks could explicitly address patients' ability to remember whether stimuli were presented in black-and-white or in color.
Higher ratings of auditory hallucinations were also associated with increased response bias toward false recognition of nontarget pictures, provided that the influence of recognition efficiency and certain negative symptoms was controlled. Thus, the association observed with false detection of auditory signals (Bentall & Slade, 1985; Rankin & O'Carroll, 1995) and false recognition of words (Brébion et al., 1998, 2005) also holds true for pictures. Even though auditory hallucinations were strongly related to poor recognition of colored pictures, their association with the response bias does not appear to stem from this recognition deficit, because the recognition accuracy index for colored pictures was partialled out. Furthermore, the correlation with the response bias seems to concern the black-and-white pictures at least as much as the colored pictures. These findings suggest a widespread tendency for people with auditory hallucinations to make false positive errors in different modalities. Given the correlational nature of the study, there is no way of determining whether the false recognitions are a consequence of hallucinations, or rather a causal mechanism involved in the production of hallucinations.
Lastly, auditory hallucinations were associated with failure to remember the spatial context of presentation of the pictures. Again, this was observed only when the effect of recognition efficiency and negative symptoms was controlled. Auditory hallucinations were significantly associated with errors in remembering the spatial context for the black-and-white pictures, whereas they were not associated with ability to recognize these pictures. This pattern of results demonstrates the dissociation between memory for the stimuli themselves and memory for their presentation context (Sullivan et al., 1997). The association of auditory hallucinations with failure to remember spatial context extends the previously observed association with failure to remember the temporal context of word presentation (Brébion et al., 2007). These findings give support to the theory that auditory hallucinations are linked to deficit in context processing (Hemsley, 2005), more specifically in the ability to bind the contextual components of memory together (Waters et al., 2006). Further research should contrast immediate versus delayed recall of spatial information to determine whether the association with auditory hallucinations pertains to the processing of context information, or to its maintenance in memory. It would also be interesting to investigate whether the association with spatial context errors more specifically concerns the spatial location of the stimuli and/or the visual association with other stimuli, because both factors were confounded in this study. Lastly, the role of emotional valence of pictorial stimuli on the association between auditory hallucinations and spatial context errors could be investigated, because this factor was shown to affect memory for spatial context (Van't Wout et al., 2006).
The study of visual hallucinations yielded entirely negative results. No correlation between visual hallucination score and any memory measure was revealed, even though the range of this variable was similar to that of the auditory hallucination score. The absence of findings may seem surprising, because the use of visual stimuli might have facilitated the emergence of associations with visual hallucinations. However, few patients presented significant visual hallucinations; therefore, these specific analyses might not have been powerful enough to enable any effect to be detected. In addition, the overlap between patients presenting auditory and visual hallucinations might have influenced the results. Our conclusions relative to visual hallucinations should be regarded as very tentative, until they are corroborated by further studies conducted on more sizeable samples.
With regard to negative symptoms, we had previously reported in two independent studies that anhedonia was significantly associated with reduced tendency to make false recognition of words (Brébion et al., 1999, 2005). In the current study, we observed that this symptom was also associated with reduced tendency to make false recognition of pictures. On the other hand, anhedonia was unrelated to accuracy in recognizing the target pictures, as it was unrelated to accuracy in recognizing the target words in the above-mentioned studies. Affective flattening might also be related to decreased rate in false recognitions. Certain negative symptoms might thus enhance cognitive inhibition, thereby resulting in fewer commission errors.
Contrary to our expectations, neither affective flattening nor anhedonia was associated with fewer errors in remembering the spatial context of picture presentation. Our previous finding that these specific negative symptoms were related to fewer temporal context errors in lists of words is therefore not extended to spatial context. It may be either that spatial context is not involved in this relationship or that verbal material is required for such inverse association with contextual memory to emerge. One way of testing this question would be to assess memory for temporal context using pictures rather than words, and determine whether affective flattening and anhedonia are associated with fewer errors in remembering temporal context of presentation for these stimuli.
In summary, this study reveals that auditory hallucinations are associated with response bias toward false recognitions of pictures, and with failure to process or remember the spatial context of picture presentation. The association with response bias corroborates that previously observed with acoustic and verbal material. The association with spatial context errors has never been reported before, and extends the previously observed association with temporal context errors. These findings add to the studies showing an association between hallucinations and various types of source memory deficit and indicate that the impairment extends across modalities.
ACKNOWLEDGMENTS
The first author was funded by grants from the Bial Foundation, the British Academy, NARSAD, and the Wodecroft Foundation. We gratefully acknowledge awards from MedicAlert and Eli Lilly.
