Executive functions and insight in OCD: a comparative study

Obsessive-compulsive disorder (OCD) affects 2% of adults worldwide and 10% of the adult clinical population [1]. Cognitive-behavioural psychotherapy is ineffective in 20% of cases and selective serotonin reuptake inhibitors (SSRIs) in 40% [2], although other studies suggest higher percentages for both treatments, around 40–60% [3]. Three factors can account for these low success rates: comorbidity, intensity of the symptoms and insight [4], with a prevalence of the latter [5‐10]. Around 15–36% of OCD patients have poor insight [5, 11‐16], which makes insight one of the most promising topics for future research in OCD [3, 17, 18].

Poor insight has been linked to many different factors: lower education levels [9], earlier age of disease onset, greater duration of disease [11, 15], chronic evolution and OCD family history [13] were the most cited. It has also been associated with more severe symptoms [5, 11, 12, 14] and a greater psychiatric comorbidity [9, 11‐15].

The notion of insight has been the object of controversy and is viewed since 1990 as a multidimensional construct [19, 20]. On a psychopathological level, the connection between a lower insight in OCD and obsessive-compulsive symptoms has yet to be clarified [21]. Recent meta-analyses suggest that individuals with OCD have poorer verbal memory/fluency and visuospatial abilities/memory than healthy individuals and fare worse across several executive domains, such as sustained attention, processing speed and working memory [22]. However, OCD is a heterogeneous disorder, and these analyses do not specify which subgroup of OCD patients best fits this neuropsychological profile.

Executive functions are subserved by a neural network which includes regions of the frontal lobes [23, 24]. Executive functioning leads us to a set of functions or cognitive skills that, as defined in the dictionary of the International Neuropsychological Society, are “necessary to perform complex behaviors aimed at a certain objective” and determine our “adaptive capacity to different demands and environmental changes” [25]. These functions thus include a variety of adaptive skills and processes that allow the individual to “analyze what they want, how they might get it (ie form a plan, based often on recollections of past experience), and then carry that plan out”. Executive functions play an important and determining role in the individual’s cognitive, emotional and social regulation and, therefore, in the adoption of an effective, creative and socially acceptable conduct [26]. Given its nature and its role, it is not surprising that a commitment at this level can have a devastating impact on the person’s life, both in terms of the effectiveness of its daily functioning and in terms of the relationships it establishes with others. For this reason, it is important to have neuropsychological instruments to assess these functions. Executive functions include a wide range of cognitive processes, such as strategic working memory, cognitive flexibility, cognitive control of behaviour, planning and problem solving. Attention, the ability to anticipate, the establishment of objectives, the sequencing of activities, self-regulation and monitoring of behaviors, initiative, abstraction and spontaneity.

Since the prefrontal cortex has been identified by the specialized literature as the main neuroanatomical substrate for executive functions [26], much scientific knowledge has been produced in relation to the topic. Nevertheless, there are debates about issues related to the true nature of this construct, as well as the skills that integrate them. In this sense, several models of brain organization have been developed in order to explain the complex nature of these functions [25].

The present study aims to clarify whether poor-insight OCD patients have a different neuropsychological profile than their good-insight counterparts. To do so, and regardless of other clinical variables—such as age, sex, education, OCD symptoms (Yale–Brown Obsessive-Compulsive Scale, Y-BOCS), medication and insight (Brown Assessment of Beliefs Scale, BABS)—different aspects of neurocognition were analysed, namely: visual and verbal memory (Rey’s figure test, California verbal learning test, subjective memory questionnaire), visual attention and task switching (trail-making test), attention, working memory, visual processing, abstract reasoning, and the ability to change problem-solving strategies (Wisconsin Card Sorting Test), attention and perception (Toulouse–Piéron test).

The neuropsychological performance of OCD patients has drawn increasing attention [27, 28]. Since OCD patients were historically viewed as having average or above average intelligence [29], their performance deficits were not fully understood. However, the existing research has shown these patients have an average intelligence quotient (IQ), especially if one considers their verbal IQ but not their performance IQ.

It is widely accepted that OCD patients have a poorer performance in several neuropsychological dimensions compared to healthy controls, but the corresponding findings are heterogeneous, a sign that these patients are clinically diverse. Of the most significant neuropsychological findings observed to be compromised among OCD patients, we would like to highlight: attention, executive function, memory, visuospatial ablilities, processing speed and working memory and subdomains, as sustained attention, planning, response inhibition, set shifting/cognitive flexibility, verbal memory, non-verbal memory and spatial working memory [22]. There is a long debate concerning the reasons for these differences, and the type and severity of symptoms (assessed by the total Y-BOCS score) have been mentioned as possible explanations. In the last decades, however, despite the publication of different meta-analyses [30], no definite conclusion has been reached.

Poor insight may result from cognitive deficits and neuropsychological factors [31]: OCD patients with poor insight exhibited poorer speech learning and memory [32], fared worse in the trail-making test (TMT) and the Auditory-Verbal Learning Test (AVLT) than the group with better insight [33] and exhibited more severe neuropsychological deficits in executive functions [33], conflict resolution/response inhibition and verbal memory [28]. Nevertheless, very few studies have directly examined the specific neuropsychological alterations of OCD patients with poor insight [34]. Although the worse performance of patients with poor insight has yet to be conclusively explained, three reasons were advanced for their worse performance regarding specific cognitive deficits: 1) difficulties in the inhibition of the response-resolution conflict may prevent them from solving, in an adaptive manner, conflicts between their belief system and the corrective information they receive from the outside; 2) memory impairments may prevent from adequately updating their belief system; 3) fluency impairments may render the access to pre-existing memories more difficult [35].

Understanding the differences between insight groups in OCD can be very useful in the future. If we find, for instance, that neuropsychological profiles display significant contrasts depending on insight, which can be quickly evaluated in a routine psychiatric consultation, this discovery will allow us to determine which OCD patients merit a more detailed neuropsychological evaluation and to direct cognititive-behavioural therapy and psycho-education specifically towards the cognitive deficits mentioned in the previous paragraph.

In this study, we hypothesized that OCD patients with poor insight would have greater impairments in tests of executive functions, such as the WCST and TMT, compared with both OCD patients with good insight and healthy control subjects. As for memory tests, which require the use of a very different neural network, involving the hippocampus, we hypothesized that OCD patients with both good and poor insight would be equally impaired and show greater impairment as compared with the healthy control group.

The key results of this study can be divided into two parts: the first one comprises the results relating to the executive functions, summarized in Table 3, and the second part comprises those relating to memory functions (subserved by the hippocampus/temporal lobes), shown in Table 4. All results are based on the evaluation of 110 participants, divided into the control group (n = 53), the good-insight group (n = 37) and the poor-insight group (n = 20), please see Table 2.

As shown in Table 2, the total number of patients (good-insight + poor-insight) is 57, which is higher than that of the control group [53]. With a false discovery rate of 0.029, the three groups did not differ significantly with regard to age, sex, Y-BOCS total score and compulsion and obsession scores. The p value for sex, 0.055, although close to the cut-off point (0.029), does not allow us to consider this variable as a relevant factor differentiating the three samples.

With p = 0.009, education (years) has proved a significant factor, and was therefore controlled afterwards with an ANCOVA (see Tables 3 and 4).

Comorbidity was assessed using the MINI (based on DSM-IV-TR), as indicated in the Measurements section. Moreover, various disorders constituted exclusion criteria, including schizophrenia and bipolar affective disorder. Among those that were not excluded, three were present in the sample, namely mild depression, agoraphobia and schizoid personality disorder. Considering that the cut-off point established according to FDR was 0.029, two disorders did not exhibit significant differences in the two groups and were therefore excluded: agoraphobia (p = 0.580) and schizoid personality (p = 0.103). Mild depression (p = 0.029) was significantly more prevalent in the poor-insight group.

Table 3 shows the results of the tests on executive functions and attention. After the correction for multiple comparisons mentioned above, only the results with very low p remained: the “time” criterion of the TMT, which was more significant in TMT-B (time and errors), and the WCST’s different evaluation criteria. These results are relevant due to the magnitude of the differences found. See, for example, the “errors” variable in WCST: the errors of patients with poor insight are considerably higher than those of the other two groups. This fact, along with the increase in the TMT’s “time” criterion, is crucial for the debate on the neuropsychological profile of poor-insight patients, discussed in detail in the next section. For the ANCOVA tests presented in Table 3, the control group and the poor insight OCD group exhibited statistically significant differences in TMT A (time) and TMT B (time), in total WCST errors, WCST perseverance and in the WCST perseverance errors. The control group and the good insight OCD group showed statistically significant differences in the TBTP-RT (work-efficiency), TMT A (time) and TMT B (time). The good insight OCD group and the poor insight OCD group exhibited statistically significant differences in TBTP-RT (work-efficiency), total WCST errors, WCST perseverance and WCST perseverance errors.

Table 4 shows the variables used for assessing memory, i.e. the Rey’s figure and CVLT results controlled for education years. The differences concerning the Rey’s figure test (copy and 5-min recall) were the most significant. However, since the analysis was based on a multiple comparison, we resorted to the correction prescribed by Benjamini and Hochberg, as mentioned earlier. Following the correction, only the subscores with p ≤ 0.029 remained, i.e. FRey time (immediate recall), FRey time (late recall) and FRey points (late recall), which concern visual memory. None of the CVLT subscores were significant. For the ANCOVA tests presented in Table 4, the control group and the poor insight OCD exhibited statistically significant differences in FRey time (copy) and in FRey time (recall). The control group and the good insight OCD group showed statistically significant differences in the FRey time (copy) and FRey points (recall). The good insight and the poor insight OCD group exhibited statistically significant differences in FRey time (copy), FRey points (recall) and in FRey time (recall).

The main objective of this study was to determine whether there are relevant neuropsychological differences between good-insight and poor-insight OCD patients. It is already known that OCD patients have various neuropsychological deficits. They were found to have worse performances in many different studies [22], particularly concerning sustained attention, planning, response inhibition, set shifting, cognitive flexibility, verbal and non-verbal memory, visuospatial abilities, processing speed, working memory and special working memory.

We hypothesized that OCD patients with poor insight would have a worse performance on executive function tests (e.g. WCST and TMT) compared with OCD patients with good insight. The comparison between different neuropsychological studies is very difficult, since the test batteries are not the same and other study design characteristics also differ significantly, like the total number of patients or the methods used to collect the sample (studies using internet questionnaires and auto-filled tests tend to have larger samples).

As regards sustained attention, accuracy and fatigue resistance (TBTP work-efficiency (RT) score, see Table 3), poor-insight OCD patients showed a worse performance compared to the good-insight group. Given that poor insight corresponds to a more severe form of OCD, and that obsessions lead to a decrease in the patients’ reasoning speed, this finding demonstrates the cognitive difficulties faced by OCD patients, and particularly by those with poor insight. Previous studies have already considered the decrease in OCD patients’ task performance speed, rejecting the hypothesis that such a decrease is due to intrusive thoughts or meticulousness [54], and linking it instead to an inadequate precocious inhibition of competing ideas. However, it should be noted that these studies have not taken insight into account [54]. Still with regard to task switching and processing speed (TMT’s time criterion), we found clear differences between the three groups. Similar differences were also found in a meta-analysis including 115 studies [22], showing that OCD patients are indeed slower than controls.

Since TMT-B is more demanding than TMT-A, the former presented an even greater difference between the two groups (p = 0.002 vs p = 0.003), highlighting the cognitive limitations of the poor-insight group, in particular regarding executive functions and not motor slowing. Given that poor-insight patients are more severely affected by OCD [55], it is plausible that their TMT performance is worse. Indeed, their difficulty in performing a task that requires continuous response inhibition and attentional set shifting is linked to the deficits of patients who do not respond to, or resist, treatment, namely poor-insight patients. The cognitive processes involved in TMT-B are varied, which implies that the cerebral regions involved are also varied [56]. In studies using voxel-wise lesion symptom mapping (VLSM), which investigates the neural correlates of a given lesion, it was concluded that the left rostral anterior cingulate was not only not exclusive to TMT, but also indicative of poorer WCST performance. This conclusion is particularly important in the present context, since the patients with poor insight we have analysed also fared worse in WCST (see Table 3). In sum, the regions involved in poor TMT-B performance were all located on the left side. Nevertheless, the conclusions to be drawn from TMT must take into account that the increase in the execution time is not linked to a specific pathology, as it was also recorded for schizophrenia, for example [57].

In the literature published in the last 10 years on the executive functions of OCD patients, which does not consider insight, researchers agree that the WCST is one of the tests in which OCD patients fare worst [58]. In our sample, the most common errors in the poor-insight group concerned perseverance (“perseverance errors”—WCST) and the patients’ difficulty in adapting their decisions to the results obtained during the test. Cognitive flexibility and set-shifting were clearly compromised among poor-insight patients (wcst total errors, wsct perseverance and perseverance errors) when compared to the two other groups (see Table 3), what is specifically due to impairments in cognitive flexibility but not to poor learning. Given the cognitive rigidity for which obsessive patients are known, this explanation is plausible and allows us to account for their insight from a neuroanatomical perspective, involving the pre-frontal and dorso-lateral cortices. From an anatomical perspective, these differences are in accordance with the neuroanatomical basis of obsessions and compulsions (orbito-frontal cortex, anterior cingulate cortex and caudate nucleus) proposed by many authors [4, 59‐61]. Another study, with a smaller sample (14 patients), crossed the WCST results with the observation of the regional blood flow and concluded that the perseverance errors were related with flow changes in the right thalamus [62]. However, this study did not focus specifically on insight.

Although the neuropsychological characteristics of OCD have been regarded as a phenotypic marker [63], most results are inconsistent [58]. An important study comprising 150 OCD patients found significant changes in their WCST results when compared with those of the control group [27]. It was proposed that the increased checking behaviour, due to an effort to avoid making mistakes and an inability to spontaneously generate alternative solutions and organizational strategies, or simply indecision when evaluating and choosing between alternatives, may explain these findings [58]. Deficits in the executive functions (e.g. decision-making, response inhibition and cognitive flexibility) may be associated with insensitivity to the future consequences of patients’ choices and to defective planning in daily life [27, 64]. According to the authors who proposed this association, these differences can be explained by the patients’ organizational strategies. However, the reason for this has yet to be clarified, and not all studies point in the same direction [23, 53, 55, 56]. Finally, in a research completed in 2009, WCST scores were similar in different insight groups [33], but not in schizophrenic patients. This study proposed cognitive flexibility and set-shifting as “candidate” characteristics to what the authors call an “endophenotype of early-onset OCD” [65, 66] based on the WCST and TMT (among other) scores. In light of Tumkaya et al.’s conclusions and our own results, OCD patients with poor insight show some similarities in cognitive dysfunction to patients with schizophrenia [33].

As for visuospatial constructional abilities and visual memory (Rey’s complex figure test), these skills (associated with frontal lobe function [28, 59, 67]) are very important for daily life and for many different aspects of cognitive and visual capabilities [68]. The differences observed in the studied groups were statistically significant (see Table 4) and require a detailed analysis. If, on the one hand, patients with poor insight scored higher than those with good insight (which means that their copy was more accurate), the opposite is true when it comes to the time spent, measured by the F.Rey time copy and the F.Rey recall items. It is possible that patients with poor insight, although having provided better copies, spent more time drawing them. This result may be seen as a consequence of the dysfunction and the deficit in planning and organizational strategies exhibited by OCD patients [48, 63].

In what concerns verbal memory, the existing results have always been controversial, and our study is no exception [22]. While no differences were found on the CVLT, some links between verbal memory and poor insight in OCD were recorded [28]. To the best of our knowledge, no real consensus has been reached as to what these findings mean [32]. It was also proposed that verbal memory is not affected by OCD [27, 58, 63, 69]. These studies argue that verbal memory is not compromised, but rather the patients’ “organizational strategies” [63]. Deckersbach et al. proposed that OCD patients underutilize organizational strategies rather than lacking verbal memory per se [70]. The fact that verbal memory was not significantly different in the good-insight and poor-insight groups suggests that what differentiates OCD patients is not memory, which is preserved in this disease [55], but more complex functions, such as the executive functions discussed above.

In order to secure the homogeneity of our sample and avoid biases, several comorbities were excluded from our study, such as schizophrenia, bipolar affective disorder and mental retardation. Of those that were included, according to the MINI’s criteria, three were found in the sample: schizotypal personality disorder, agoraphobia and mild depression (see Table 2). Only one case of schizotypy was found, in the poor-insight group. This fact, although lacking statistical relevance, is plausible, given the link between schizotypy and insight [71]. Indeed, several bibliographic sources point to a correlation between OCD with poor insight, and hence with worse prognosis, and schizotypal disorders [5]. However, we have not used a specific scale to measure schizotypy, which might have yielded different results. Agoraphobia, a fairly frequent disorder, was also statistically insignificant, with only four cases in the whole sample—two in the good-insight group and two in the control group. Although many authors claim that patients with poor insight display more comorbidities, recent studies have also failed to identify statistically relevant values of agoraphobia in OCD groups with good and poor insight [72], which suggests, for now, that the comorbidity of OCD with agoraphobia is not a research priority. As for depression, the percentage of patients suffering from this condition was higher in the poor-insight group (with p = 0.029, which coincides with the cut-off value, and a low effect size (0.25)). In our view, since the absolute quantity of individuals with depression was less than 5, any conclusions based on a statistical analysis of these data would be biased. Therefore, we leave the investigation of this aspect to future studies with larger subsamples.

Finally, it is worth mentioning the main limitations of this research. However, insight assessment changes over time, limiting the research of this issue. Our study is cross-sectional, which means we have not taken into account changes in symptomatology and insight in the course of time, unlike other authors [73]. This limits the depth of our analysis. Moreover, although the scale we have used to measure insight, BABS, enables good comparisons with other studies on insight, it is based only on patients’ pre-existing and explicitly held beliefs [35, 74], which limits its results. As regards the sample size, although it is similar to that of other studies, future research should aim for larger samples so as to reach broader conclusions with stronger p values.

The overall direction of our results leads us to conclude that OCD patients with poor insight have a significantly worse neuropsychological performance than the good-insight and control groups. In summary, poor-insight patients have greater neuropsychological dysfunction than those with good insight, especially regarding the executive functions. On the contrary, verbal memory does not seem to differ in the three groups. Moreover, we believe that a neuropsychological evaluation of poor-insight patients is crucial for their future and clinical outcome, particularly if it leads to the development of targeted neurocognitive therapies.

Future work in this field will require larger sample sizes, the inclusion of functional imaging examination and a longitudinal clinical evaluation.