Attention and cognition in patients with obsessive–compulsive disorder


  • FEMKE DE GEUS msc,

    1. Rudolf Magnus Institute of Neuroscience, Department of Psychiatry, University Medical Center Utrecht, Utrecht, the Netherlands
    Search for more papers by this author
  • DAMIAAN A. J. P. DENYS md, phd,

    1. Rudolf Magnus Institute of Neuroscience, Department of Psychiatry, University Medical Center Utrecht, Utrecht, the Netherlands
    Search for more papers by this author

    1. Rudolf Magnus Institute of Neuroscience, Department of Psychiatry, University Medical Center Utrecht, Utrecht, the Netherlands
    Search for more papers by this author

    1. Rudolf Magnus Institute of Neuroscience, Department of Psychiatry, University Medical Center Utrecht, Utrecht, the Netherlands
    Search for more papers by this author

Femke de Geus, MSc, UMC Utrecht, B.01.206, PO Box 85500, 3508 GA Utrecht, the Netherlands. Email:


Abstract  Although a dysfunctional prefrontal-striatal system is presupposed in obsessive–compulsive disorder (OCD), this is not sustained by neuropsychological studies. The aim of this study was twofold: (i) to investigate the cognitive deficits in patients with OCD compared to matched healthy controls; and (ii) to relate cognitive performance to clinical characteristics in patients with OCD. In this study, 39 patients with primary OCD according to Diagnostic and Statistical Manual, fourth edition criteria were compared to 26 healthy control subjects on a battery measuring verbal memory and executive functioning. Patients with OCD showed slowed learning on the verbal memory task and made more errors on the Wisconsin Card Sorting Test. Errors were failures to maintain set, which were related to severity of OCD symptomatology. The results show that patients with OCD have cognitive deficits. The authors hypothesize that these deficits may be interpreted by attentional deficits caused by a dysfunctional anterior cingulate cortex.


Patients with obsessive–compulsive disorder (OCD) suffer from recurrent anxiety-provoking thoughts (obsessions), and ritualized behaviors directed at reducing this anxiety (compulsions).

Most neurobiological studies in OCD point toward an underlying dysfunctional prefrontal-striatal system. For instance, several structural neuroimaging studies showed reduced orbitofrontal cortex and basal ganglia volumes.1–3 In addition, functional neuroimaging studies showed hyperactivity in these same regions during rest-state, symptom provocation, and cognitive activity.4–7 This hyperactivity normalizes after successful treatment with serotonin reuptake inhibitors (SRI) or behavioral therapy.8–10 Moreover, OCD symptoms might decrease following neurosurgical disruption of prefrontal-striatal circuits.11

If the neurobiological underpinning of OCD is the prefrontal-striatal system, one would expect deficits in domains such as set shifting, spatial working memory, focused attention and verbal fluency.12–14 However, neuropsychological findings are inconsistent, with some studies finding these deficits,15–17 some studies finding no deficits,18–20 and other studies finding no deficits other than slowed performance.21–23 These conflicting findings can be explained in part by methodological factors such as inadequate matching of patients to controls or lack of control for medication status and the presence of comorbid disorders.

The aim of this study was twofold: (i) to investigate the cognitive deficits in patients with OCD compared to matched healthy controls; and (ii) to examine the relationship between clinical characteristics and cognitive performance in patients with OCD. In this study, patients and controls were carefully matched for age, gender and IQ. All patients with OCD were on stable antidepressant medication and free of major comorbid disorders. The authors hypothesized that patients with OCD would perform worse than healthy controls on the cognitive tests used and the authors expected negative correlations between performance and OCD severity.



The study population consisted of 39 severely ill, therapy resistant patients with OCD and 26 healthy controls. The patients with OCD were recruited at the department of psychiatry of the University Medical Center Utrecht, Utrecht, the Netherlands, the controls were recruited by newspaper advertisements. All patients and controls gave written informed consent. This study is part of a larger study, in which the effect of quetiapine addition was investigated. A full description of inclusion and exclusion criteria can be found in the article by Denys et al.24

Inclusion criteria for patients were: aged between 18 and 65 years, primary OCD according to Diagnostic and Statistical Manual, fourth edition (DSM-IV) criteria, and minimal Yale–Brown Obsessive–Compulsive Scale (Y-BOCS25,26) score of 18 or 12 if only obsessions or compulsions were present. All patients showed significant symptoms, despite previous and current treatments. Exclusion criteria were: significant depressive symptoms (defined as a score of 15 or more on the Hamilton Depression rating scale [HAM-D]27), bipolar disorder, anxiety disorder, schizophrenia or any other psychotic condition, current psychotherapeutic treatment, substance abuse within the past 6 months, primary personality disorder, organic mental disorders, stroke within the last year, epilepsy or other central nervous system disorders. The Mini-International Neuropsychiatric Interview (M.I.N.I.)28 was given to assess DMS-IV disorders. Patients with OCD were divided into subtypes based on the Y-BOCS checklist, as described by Denys et al.29 Clinical characteristics of the OCD subjects are shown in Table 1.

Table 1.  Clinical characteristics of patients with obsessive–compulsive disorder (n = 39) and controls (n = 26)
Mean ± SD
Mean ± SD
  1. CBT, Cognitive Behavioral Therapy; HAM-A, Hamilton Anxiety rating scale; HAM-D, Hamilton Depression rating scale; NA, not applicable; NART IQ, National Adult Reading Test Intelligence Quotient; OCD, obsessive–compulsive disorder; SD, standard deviation; SRI, Serotonin Reuptake Inhibitor; Y-BOCS, Yale–Brown Obsessive–Compulsive Scale.

Gender distribution (male : female)10:2910:16χ2 = 1.200.289
Age 36.1 ± 12.2 34.0 ± 11.50.510.477
NART IQ102.0 ± 10.2106.2 ± 6.33.550.064
 Range 81–11894–118  
Medication at time of testing
 Paroxetine n = 14 dose 47 ± 15NA  
 Citalopram n = 11 dose 52 ± 13NA  
 Venlafaxine n = 5 dose300 ± 0NA  
 Fluoxetine n = 4 dose 35 ± 19NA  
 Fluvoxamine n = 3 dose150 ± 86NA  
 Imipramine n = 1 dose150NA  
 Clomipramine n = 1 dose 75NA  
Duration of illness (years) 21.6 ± 10.9NA  
 Obsessions13.4 ± 4.0NA  
 Compulsions13.9 ± 3.4NA  
 Total27.3 ± 5.4NA  
HAM-D12.0 ± 6.0NA  
HAM-A13.7 ± 5.9NA  
Number of previous SRI treatments 2.8 ± 1.0NA  
Number of previous CBT treatments 1.1 ± 1.0NA  
OCD subtype
 Contamination and cleaningn = 7NA  
 Aggressive, sexual and religious obsessions n = 6NA  
 Somatic obsessions and checkingn = 5NA  
 Symmetry and counting/arranging n = 10NA  
 High risk-assessment and checking n = 11NA  

Exclusion criteria for the healthy control group were: neurological illness or other disorders of the central nervous system and substance abuse within the past 6 months. Further exclusion criteria were history of psychiatric disease (assessed by Comprehensive Assessment of Symptoms and History [CASH)30), history of personality disorder (assessed by Structured Interview for DSM-IV Personality Disorders [SIDP-IV])31 and psychiatric disease in first or second degree relatives (assessed by Family Interview for Genetic Study [FigS]).32 Healthy controls and patients with OCD were matched groupwise for age, gender, handedness and IQ (measured by the Dutch version of the National Adult Reading Test33,34).

As can be seen in Table 1, patients with OCD and controls did not differ with respect to demographic and clinical characteristics.

Cognitive measures

National Adult Reading Test 33,34

In this test, known to be a good indication for verbal IQ, subjects are asked to read aloud a list of phonetically irregular words. The authors transformed the raw scores into IQ scores, which were then used to match the two groups.

Verbal fluency

The authors measured verbal fluency with a word-generation tasks in which subjects are given 1 min to retrieve as many words as possible in response to letter (‘N’, ‘A’) and semantic (‘animals’, ‘professions’) cues. These four trials result in four scores: number of unique words. The scores for the two phonemic trials are added and so are the scores for the two semantic trials. Verbal fluency is considered to be a measure of executive function.35

Trail Making Test 36

In this task the time required to track a number sequence (Trail Making Test [TMT] A) and a sequence of alternating numbers and letters (TMT B) is measured. Time needed for part A and part B can both be considered a measure of visual scanning and mental speed, where part B focuses more on alternated attention. Since performances in both parts exhibit a close linear relationship, an outcome measure for set shifting measured by the TMT is time B divided by time A.37

California Verbal Learning Test 38

The California Verbal Learning Test (CVLT) is a verbal memory task that consists of five presentations of an identical list of words. Subjects are asked to reproduce as many words as they can. By repeating the same list five times, the examiner can study the learning curve (i.e. the amount of words remembered for each presentation). The test also yields information on the subject’s memory capacity, encoding strategy, recall after a long and short delay and recognition. The first presentation of the list is generally regarded as a test of immediate word span, more closely related to attention than to memory per se.39 The measures used in analysis were: total recall in five trials, number of words learned over five trials (trial 5 minus trial 1), semantic clustering index, retrieval after short delay, retrieval after long delay, number of hits on recognition trial. To prevent practice effects due to the two administrations, the authors used the parallel version of this test in a counterbalanced design.40 A Dutch version of the CVLT was used.41

Wisconsin Card Sorting Test 42

The Wisconsin Card Sorting Test (WCST) is one of the most widely used tasks in the assessment of neurocognitive function, it measures so called ‘frontal’ functions, such as set-shifting, category formation and set maintenance.39 In the WCST, the subject has to discover criteria by which to sort cards (according to three ‘sets’: color, shape, or number). After 10 consecutive sorts, the examiner changes the set unbeknown to the subject (this is only indicated by feedback from the examiner). The authors used the version as described by Milner43 in which two packs of 64 cards are used and the test is discontinued when the subject reaches six categories. For each subject, the authors calculated the number of categories completed (10 consecutive correct sorts), percentage of errors made, percentage of perseverative errors and failure to maintain set (five to nine consecutive correct sorts followed by an error) as described in the manual.

Clinical characteristics

The clinical characteristics included in the analyses were: age, duration of illness, number of previous treatments, OCD severity measured by the Y-BOCS (obsessions and compulsions subscales, total), anxiety symptoms measured by the Hamilton Anxiety Rating Scale (HAM-A),44 depressive symptoms measured by the HAM-D, OCD symptom subtype.

Statistical analyses

Data were analyzed using the Statistical Package for the Social Sciences (SPSS) version 11.5 (SPSS Inc., Chicago, IL, USA). The authors used multivariate anova (factor is Group with two levels: OCD and healthy controls) to compare patients with OCD and healthy controls, and used Pearson’s coefficients to calculate correlations between clinical characteristics and neuropsychological performance. P-values under 0.05 were considered significant.


Patients with obsessive–compulsive disorder versus healthy controls

Table 2 summarizes the group mean performance and statistical comparisons for each task.

Table 2.  Performance on neuropsychological measures
 OCD (n = 39)
Mean ± SD
Controls (n = 26)
Mean ± SD
  1. CVLT, California Verbal Learning Test; OCD, obsessive–compulsive disorder; SD, standard deviation; TMT, Trail Making Test; WCST, Wisconsin Card Sorting Test.

Verbal fluency
 Phonemic24.46 ± 7.9324.68 ±
 Semantic42.92 ± 9.5444.04 ±
 Time A30.72 ± 10.0528.39 ± 9.720.860.357
 Time B65.49 ± 21.5963.45 ± 23.420.130.720
 Time B/time A2.24 ± 0.772.28 ± 0.510.570.811
 Total recall 5 trials53.90 ± 13.0856.38 ± 9.370.700.406
 Recall 1st trial6.74 ± 2.078.15 ± 1.877.790.007
 Recall 2nd trial10.15 ± 2.9811.04 ± 2.691.490.228
 Recall 3rd trial11.51 ± 3.3211.96 ± 1.820.400.532
 Recall 4th trial12.56 ± 2.9212.38 ±
 Recall 5th trial12.90 ± 2.8812.85 ± 2.330.010.940
 Number of words learned over 5 trials6.15 ± 2.134.69 ± 2.007.700.007
 Semantic clustering index2.28 ± 1.002.31 ± 0.870.010.924
 Retrieval after short delay0.62 ± 1.481.46 ± 1.333.340.074
 Retrieval after long delay0.38 ± 1.180.77 ±
 Number of hits on recognition trial15.00 ± 1.5914.81 ± 1.770.210.649
 Number of categories completed4.64 ± 1.875.54 ± 1.274.560.037
 % errors31.64 ± 14.5422.32 ± 14.086.570.013
 % perseverative errors15.95 ± 7.2912.88 ± 9.772.100.152
 Failure to maintain set1.03 ± 1.160.42 ± 0.815.300.025

The patients with OCD and healthy controls did not differ in the two timed tasks, verbal fluency and TMT. The ratio between part A and part B of the TMT, which is a measure of set shifting, showed no abnormalities either.

Patients with OCD showed greater learning on the CVLT (measured by the ‘number of words learned over five trials’ variable), this was caused by a significantly lower performance on the first trial (see Table 2: recall 1st trial). The performance on the other four trials did not differ.

The patients with OCD performed worse on the WCST compared to the controls: they completed significantly fewer categories; the percentage of errors was higher as was the number of failures to maintain set. The percentage of perseverative errors was not elevated in the OCD group.

Clinical characteristics and neuropsychological performance

The authors performed exploratory correlational analyses to examine the relationships between the clinical measures. These analyses showed that HAM-A and HAM-D scores correlated significantly (r = 0.784, P = 0.000), HAM-D was also related to age (r = 0.376, P = 0.018). Duration of illness was related to age (r = 0.819, P = 0.000) and HAM-A score (r = 0.399, P = 0.013). Number of previous SRI and CBT treatments correlated significantly (r = 0.433, P = 0.006). These relationships were taken into account in further analyses, the authors corrected for factors by means of partial correlations.

Correlations between clinical and neuropsychological variables are shown in Table 3.

Table 3.  Pearson correlation coefficients between clinical and neuropsychological variables in the obsessive–compulsive disorder group (n = 39)
 AgeDuration of illnessNo. previous SRI treatmentsNo. previous CBT treatmentsY-BOCSHAM-AHAM-D
  • Correlation is significant at the 0.05 level (two-tailed).

  • ** 

    Correlation is significant at the 0.00055 level (Bonferonni corrected α).

  • CBT, Cognitive Behavioral Therapy; CVLT, California Verbal Learning Test; HAM-A, Hamilton Anxiety rating scale; HAM-D, Hamilton Depression rating scale; SRI, Serotonin Reuptake Inhibitor; TMT, Trail Making Test; WCST, Wisconsin Card Sorting Test; Y-BOCS, Yale–Brown Obsessive–Compulsive Scale.

Verbal fluency
 Time B/time A0.04−−0.05−0.10
 Total recall 5 trials−0.54**−0.53**0.170.07−0.30−0.38*−0.33*
 Number of words learned over 5 trials0.−0.09−0.15−0.10
 Semantic clustering index−0.29−0.41*0.210.14−0.30−0.39*−0.24
 Retrieval after short delay0.
 Retrieval after long delay0.280.32*0.04−*0.27
 Number of hits on recognition trial−0.44**−0.43**−0.030.00−0.20−0.13−0.06
 Number of categories completed−0.31−0.32−0.34*0.07−0.27−0.03−0.07
 % errors0.200.160.31−0.130.09−0.080.01
 % perseverative errors0.210.090.25−0.210.18−0.110.03
 Failure to maintain set0.38*0.28−**0.280.38*

Severity of OCD symptomatology, measured by the Y-BOCS, correlated significantly with failure to maintain set on the WCST. Depressive symptomatology, measured with HAM-D was also related to failure to maintain set. When corrected for age (which correlates with both failure to maintain set and HAM-D), this effect disappeared (r = 0.270, P = 0.101). WCST number of categories completed was negatively related to number of previous SRI treatments.

Duration of illness correlated significantly with several CVLT measures: total recall, semantic clustering index, retrieval after a long delay and number of hits on recognition. After correction for age (since duration of illness is closely linked to age), the significant results disappeared (total recall: r = −0.130, P = 0.443; semantic clustering index: r = −0.277, P = 0.097; retrieval after long delay: r = 0.133, P = 0.327; number of hits on recognition: r = −0.111, P = 0.521). The authors interpreted these results as an artifact, because verbal memory is known to be related to age, as can also be seen in Table 3.39

Severity of anxiety symptoms, measured by HAM-A, correlated significantly with several CVLT measures: total recall, semantic clustering index and long-term retrieval. Again, these effects could be explained by duration of illness and age. After correction for duration of illness, all significant results disappeared (total recall: r = −0.277, P = 0.093; semantic clustering index: r = 0.271, P = 0.099; retrieval after long delay: r = −0.329, P = 0.044). Because of the direct relationship between duration of illness and age, the authors interpreted the results as an artifact.

When HAM-D scores were corrected for age, semantic verbal fluency showed a significant relationship with depressive symptoms (r = −0.328, P = 0.045), but the relationship with CVLT total recall disappeared (r = −0.167, P = 0.316).

No other effects achieved significance. In particular, OCD severity did not have any effect on cognitive measures other than the WCST. Depressive symptomatology and anxiety symptoms did not relate to any of the cognitive measures. There were no differences between the five symptom subtypes for any of the cognitive measures.


The main finding in the present study is that patients with OCD showed slowed learning on the CVLT and deficits in set maintenance as measured by the WCST, the latter being related to severity of OCD symptoms.

In contrast to many studies;16,17,45 the present study failed to find slower performance on timed tests (TMT, verbal fluency). The slower performance of patients in other studies might be explained by depressive comorbidity, approximately 50% to 80% of patients with OCD suffer from comorbid depression, a disorder renowned for its slowness of thought and movement.46 Schmidtke et al.,17 for instance, found deficits in patients with OCD on timed tests. Eight of the 29 patients (28%) had a lifetime diagnosis of major depressive disorder. In 2001, Basso et al.16 showed that abnormalities in executive function were related to comorbid depressive severity and argued that conflicting findings in past studies regarding executive functioning are due to comorbid depression. The current sample consisted of depression-free patients with OCD, with a mean Hamilton–Depression score of 12, which is indicative of mild depressive symptoms. This could explain their normal performance on timed tests.

Regarding the CVLT, the authors did not find any deficits in verbal memory, other than a slightly slower learning curve. The patients in the current sample showed a lower performance on the first trial, probably reflecting a lack of attention.39 They made up for this in later trials, where they performed at the same level as controls, therefore, the authors conclude that these patients do not suffer from a verbal memory deficit. Almost all studies investigating verbal memory tasks (like the CVLT) show normal performance in patients with OCD,47–49 confirming that there is no mnemonic deficit. When the material is non-verbal (e.g. Rey complex figure test), most studies do find differences in recall scores between patients with OCD and healthy controls, but this is most often explained in terms of a failure to use organizational strategies.17,50–55

In patients with OCD, one would expect to find a profile of deficits associated with prefrontal-striatal dysfunction.1–11 To be more precise, the structures that are most consistently shown to be involved in OCD are the orbitofrontal cortex and basal ganglia structures (especially caudate nucleus). A cognitive profile matching dysfunction in these structures, might include executive deficits leading to, namely; errors in set shifting, focused attention and verbal fluency.12–14 The most prominent feature of executive/frontal dysfunction is perseverative behavior. The deficits found in the current patient sample did not fit this profile. First of all, the authors did not find deficits in tasks such as verbal fluency and TMT (set shifting, attention). Furthermore, the authors did find a deficit in WCST performance (set shifting), but failed to find the expected elevated level of perseverative errors. In summary, the cognitive deficits found in patients with OCD do not fit the profile of deficits associated with prefrontal-striatal dysfunction.

An alternative line of explanation is that the current results on the CVLT and the WCST can be explained by a deficit in the attention system. The deficit found on the WCST is a failure to maintain set. Failure to maintain set occurs when the subject makes an error after five or more consecutive correct responses (the subject ‘loses’ the set) and is often explained as reflecting difficulty in sustaining attention and remaining ‘on task’.56,57 As mentioned earlier, the lower performance on the first presentation of the CVLT can also be explained in terms of an attention deficit.39 The authors did not find deficits on the TMT, a task sometimes considered to be a measure of attention. However, several authors argue that the TMT should be seen as a measure of cognitive flexibility, visual scanning and simple motor skills.58 Therefore, it is plausible to assume that the CVLT and WCST data indicate a deficit in the attention system of the subjects with OCD.

This attention deficit could be due to abnormal functioning of the anterior cingulate cortex (ACC), a structure involved in attention functions and conflict monitoring in information processing (for an overview see References59,60). Studies in OCD patients consistently showed hypermetabolism of the ACC during symptom provocation,4,6,7,61 at rest,8,62 and during the execution of neuropsychological tasks.63–65 Structural studies found abnormalities in the ACC as well: more total gray matter66,67 and white matter abnormalities.68 In line with this, event-related potential studies demonstrated an increased error-related negativity (a negative waveform time-locked to incorrect responses) in patients with OCD.69–71 This error-related negativity is attributed to the action-monitoring function of the ACC. Many authors argue that an overactive action-monitoring system leads to constant feelings of erroneous performance, which in its turn leads to the doubt and checking behavior characteristic for OCD.

Monchi et al.72 showed that receiving negative feedback during the WCST leads to activity in the dorsal part of the ACC of healthy subjects. This is another argument for the hypothesis that the deficits in WCST performance in the current sample might be due to ACC abnormalities. Furthermore, failure to maintain set correlated significantly with the Y-BOCS, suggesting that patients with more severe symptoms showed more attentional dysfunction.

In summary, in a sample of 39 patients with OCD the authors found deficits in set maintenance and slightly slower learning on a verbal memory task. These deficits seem to be indicative of attention dysfunction in patients with OCD, possibly related to abnormal ACC functioning. The ACC is involved in action-monitoring, which could be linked to doubt and checking behavior. Therefore, it would be interesting to investigate the ACC functioning in different OCD-subgroups, for example ‘checkers’, ‘washers’, and patients with ‘pure obsessions’.