Disruption of consciousness depends on insight in obsessive-compulsive disorder and on positive symptoms in schizophrenia

Abstract

Disruption of conscious access contributes to the advent of psychotic symptoms in schizophrenia and could also explain lack of insight in other psychiatric disorders. In this study, we explored how insight and psychotic symptoms related to disruption of consciousness in obsessive-compulsive disorder (OCD) and schizophrenia, respectively. Patients with schizophrenia, and patients with OCD with good versus poor insight and matched controls underwent clinical assessments and performed a visual masking task. We used a principal component analysis to reduce symptom dimensionality. We found that clinical dimensions could be isolated by principal components that correlated with consciousness measures. More specifically, positive symptoms were associated with impaired conscious access in patients with schizophrenia, whereas the level of insight delineated two subtypes of OCD patients, those with poor insight who had consciousness impairments similar to patients with schizophrenia, and those with good insight who resemble healthy controls. Our study provides new insights about consciousness disruption in psychiatric disorders, showing that it relates to positive symptoms in schizophrenia and with insight in OCD. In OCD, it revealed a distinct subgroup sharing neuropathological features with schizophrenia. Our findings refine the mapping between symptoms and cognition and confirm that consciousness disruption can be observed in various psychiatric disorders.

Keywords: cognitive neuroscience; insight; obsessive-compulsive disorder; schizophrenia; visual masking; consciousness.

Figure 1

Visual backward masking paradigm: a fixation cross was displayed in the center of the screen during ~1 s (randomly jittered across trials). A target digit (2, 3, 7, or 8) was then presented for a fixed duration of 16.7 ms at a random position among four (1.4 degrees above or below and 1.4 degrees right or left of the fixation cross). In 20% of the trials, the target digit was replaced by a blank of the same duration (catch trials) to measure participants’ ability to detect the presence of the digit. After a variable delay (stimulus onset asynchrony), a metacontrast mask appeared at the target location for 200 ms. The mask was composed of four letters (two horizontally aligned M and two vertically aligned E) surrounding the target stimulus location without superimposing or touching it. Target digit visibility was parametrically manipulated by using eight possible target–mask delays (16.7, 33.3, 50, 66.7, 83.3, 100, 116.7, or 166.7 ms) that were randomly intermixed across trials. Participants had at most 10 s to determine whether the number was smaller or greater than 5, by pressing ‘S’ or ‘L’, respectively, on the keyboard. Then, they had to rate the subjective visibility of the digit. The response words ‘seen’ and ‘unseen’ randomly appeared on the right and the left sides of the fixation cross and participants responded by pressing the button (‘S’ or ‘L’) corresponding to the side of the response they wanted to select. The two alternatives remained on screen until a response was made. Each participant performed a training block of 20 trials, followed by two blocks of 160 trials each, separated by a break. Our experiment was coded with MATLAB and Psychtoolbox (Brainard 1997; Kleiner et al. 2007).

Figure 2

Impairments of conscious perception across groups: A. Left, discrimination accuracy (i.e. comparing the masked digit with 5); middle, visibility (i.e. proportion of seen trials) for target (thick lines) and catch trials (thin lines) separately; right, proportion of conscious trials (i.e. trials seen and correctly compared with 5); all as a function of target–mask delays for patients with OCD and good insight, patients with OCD and poor insight, patients with schizophrenia and healthy controls. The inset in the middle plane depicts detection d-prime, which is the ability of visibility ratings to distinguish between digit and catch trials. The thick lines correspond to fit of the logistic model. The error bars represent standard errors of the mean. B. Left, schematic of the logistic model and its different free parameters. The consciousness threshold is determined by finding the target–mask delay for which 50% of the trials are consciously perceived (i.e. accurate and seen). Middle, model parameters (i.e. upper asymptote, inflection point, lower asymptote, and slope) for each participant (dots) in each group. The horizontal lines represent the group averages. Overall, patients with OCD and good insight have similar parameter values as healthy controls, whereas patients with OCD and poor insight have similar parameter values as patients with schizophrenia. Right, consciousness threshold measured on fitted logistic models in the different groups. C. Left, the logistic model was simulated with different upper asymptote values (between 50 and 100%, with 5% increments), all other parameters being held constant. Measured consciousness threshold (i.e. duration, in milliseconds, for which the logistic model equals 50%) decreases with increasing values of the upper asymptote. Middle, proportion of conscious trials (as in panel A) before (dotted thin line) and after (thick line) correcting for non-one upper asymptotes on a participant basis. Right, consciousness threshold measured before (dotted lines) and after (thick lines) setting the upper asymptote to 1 in all groups. Group differences survive the correction for non-one upper asymptotes.

Figure 3

Reaction times and relationships between visibility and discrimination across groups: A. Top, reaction time distributions corresponding to the discrimination and visibility questions. Insets depict the cumulative distributions. The vertical dotted line shows the lower cutoff for exclusion of abnormal reaction times. Bottom, average reaction times corresponding to the discrimination and visibility questions; and discrimination reaction times as a function of target–mask delay. Patients with schizophrenia are significantly slower than any other group for both the discrimination and visibility tasks, thus precluding that differences in performance between groups are due to a shifted speed-accuracy tradeoff. B. Left, discrimination accuracy is shown separately in target trials rated as seen versus unseen. All participants have a discrimination accuracy at chance level for unseen trials and a near-ceiling accuracy for seen trials. Right, sensitivity of visibility ratings to discrimination accuracy is measured by the difference in accuracy in unseen versus seen trials. Patients with schizophrenia have seen versus unseen difference in discrimination accuracy compared with the other groups (healthy controls, patients with OCD and good insight, and with OCD and poor insight). Each dot represents a participant and the horizontal lines represent the group average.

Figure 4

Clinical dimensions yielded by PCA: A. In the group of patients with OCD, PCA yielded nine significant PCs accounting for the 55% of the symptoms. In PC1, accounting for 31.4% of the variance, almost all items have positive loadings except one item of the DOCS (avoidance related to unpleasant thoughts), some physical items of the HAM-D (hypochondriasis, loss of weight) and of the HAM-A (cardiovascular and respiratory symptoms). PC2 accounting for 12.9% of the variance was dominated by the OVIS items and also shows high loadings for Y-BOCS, the contamination items of the DOCS and suicide, but negative loadings for the thought items of the DOCS and many items of HAM-A and HAM-D (HAM-A: somatic symptoms, cardiovascular, genitourinary, autonomic symptoms, HAM-D: retardation, insomnia, feeling of guilt). PC3, accounting for 10.9% of the variance, shows an inverse relationship between contamination concerns, on the one hand, and the responsibility concerns (and to a lesser extent, symmetry concerns) on the other hand, as measured by the DOCS. B. In the group of patients with schizophrenia, PCA yielded four significant PC accounting for 66% of the symptoms. In PC1, accounting for 32.4% of the variance, all SANS items have high loadings, as well as several items of the CDS (‘depression,’ ‘morning depression,’ ‘observed depression’). PC2, accounting for 20.2% of the variance, is highly dominated by positive symptoms (all items of hallucinations, all items of thought disorder, except circumstantiality, many delusional items apart from influence syndrome) and few depressive symptoms (pathological guilt and suicide). Regarding negative symptoms, there was a negative loading for several alogia symptoms. PC3, accounting for 11.6% of the variance, has positive and negative loadings both in the positive symptoms and the depression dimension. More specifically, it shows an anticorrelation between, on the one hand, delusion of reference, most of the items related to thought disorder in the SAPS and guilty ideas of reference and self-depreciation in the CDS and, on the other hand, hallucinations (visual, olfactory and voices), delusions (religious, guilt, grandiose), and specific symptoms of depression (suicide, pathological guilt and morning depression). PC4, accounting for 10.0% of the variance, captures mostly bizarre behavior, and a mixture of positive and negative symptoms (religious and grandiose delusions, avolition including hygiene and apathy).

Figure 5

Correlation between consciousness threshold and clinical measures: A. In the group of patients with OCD, consciousness thresholds (in ms) significantly correlate with OVIS and PC2 scores (which included insight items). The vertical gray line represents the OVIS threshold used to distinguish patients with good (OVIS < 6) and poor insight (OVIS ≥6). B. In the group of patients with schizophrenia, consciousness thresholds significantly correlate with SAPS, PC2 scores (which included hallucinations and delusions items), and PC3 scores (which includes both positive and depressive symptoms, including delusion of reference, thought disorder, guilty ideas of reference, and self-depreciation). Each dot represents a participant, the lines represent linear regressions, and the shaded areas represent 95% confidence intervals. The bar plots represent null distributions of the correlation coefficient between each clinical measure and consciousness threshold (10 000 permutations), significance thresholds (vertical lines), and the observed correlation value (arrow).