Schizophrenia Research
Volume 100, Issue 1 , Pages 161-171, March 2008

Automatization and working memory capacity in schizophrenia

  • Tamar R. van Raalten

      Affiliations

    • Rudolf Magnus Institute of Neuroscience, Department of Psychiatry, University Medical Center Utrecht, The Netherlands
    • Corresponding Author InformationCorresponding author. Department of Psychiatry A01.126, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands. Tel.: +31 250 8352; fax: +31 250 5443.
    • ,
  • Nick F. Ramsey

      Affiliations

    • Rudolf Magnus Institute of Neuroscience, Department of Neurology and Neurosurgery, University Medical Center Utrecht, The Netherlands
    • ,
  • J. Martijn Jansma

      Affiliations

    • Laboratory of Functional and Molecular Imaging (LFMI), National Institute of Neurological Disorders and Stroke (NINDS), National Institutes of Health (NIH), Bethesda, MD, USA
    • ,
  • Gerry Jager

      Affiliations

    • Rudolf Magnus Institute of Neuroscience, Department of Psychiatry, University Medical Center Utrecht, The Netherlands
    • ,
  • René S. Kahn

      Affiliations

    • Rudolf Magnus Institute of Neuroscience, Department of Psychiatry, University Medical Center Utrecht, The Netherlands

Received 8 May 2007; received in revised form 17 October 2007; accepted 29 October 2007. published online 03 January 2008.

Article Outline

Abstract 

Working memory (WM) dysfunction in schizophrenia is characterized by inefficient WM recruitment and reduced capacity, but it is not yet clear how these relate to one another. In controls practice of certain cognitive tasks induces automatization, which is associated with reduced WM recruitment and increased capacity of concurrent task performance. We therefore investigated whether inefficient function and reduced capacity in schizophrenia was associated with a failure in automatization. FMRI data was acquired with a verbal WM task with novel and practiced stimuli in 18 schizophrenia patients and 18 controls. Participants performed a dual-task outside the scanner to test WM capacity. Patients showed intact performance on the WM task, which was paralleled by excessive WM activity. Practice improved performance and reduced WM activity in both groups. The difference in WM activity after practice predicted performance cost in controls but not in patients. In addition, patients showed disproportionately poor dual-task performance compared to controls, especially when processing information that required continuous adjustment in WM. Our findings support the notion of inefficient WM function and reduced capacity in schizophrenia. This was not related to a failure in automatization, but was evident when processing continuously changing information. This suggests that inefficient WM function and reduced capacity may be related to an inability to process information requiring frequent updating.

Keywords: Functional MRI, Schizophrenia, Working memory, Automatization, Dual-task

 

Back to Article Outline

1. Introduction 

Cognitive dysfunction is a core characteristic of schizophrenia (Elvevag and Goldberg, 2000) and associated with deficits in working memory (WM). WM refers to the temporary maintenance and utilization of information (Baddeley, 1986) and is considered important for complex cognitive performance (Miller and Cohen, 2001). WM dysfunction in schizophrenia is characterized by inefficient prefrontal function as most patients exhibit excessive activity when performing a moderately difficult WM task (Callicott et al., 2003, Callicott et al., 2000a, Callicott et al., 2000a, Callicott et al., 2003, Manoach et al., 1999, Manoach et al., 2000, Manoach, 2003, Jansma et al., 2004, Perlstein et al., 2001). When performing a more difficult WM task (i.e. with more information that has to be memorized and processed), they generally exhibit poor performance and lower levels of WM activity than controls (Berman et al., 1992, Andreasen et al., 1992b, Karlsgodt et al., 2007, Van Snellenberg et al., 2006), indicating that their WM capacity is quite limited (Callicott et al., 1999, Callicott et al., 2000a, Callicott et al., 2003, Jansma et al., 2004). In spite of structural abnormalities in the prefrontal cortex (Callicott et al., 2000b, Bertolino et al., 2000), and of genetic variation contributing to prefrontal activation levels (Egan et al., 2001, Bertolino et al., 2006a, Bertolino et al., 2006b, Bertolino et al., 2004) it is not clear what causes inefficient WM function and reduced WM capacity in schizophrenia.

Recently we demonstrated that the ability to reduce demands on WM with practice is closely related to the capacity to perform an additional task simultaneously in controls (Ramsey et al., 2004). If a cognitive task involves a constant relationship between stimuli and responses, practice can induce a shift from demanding to effortless processing, which is referred to as automatization. In a previous study we demonstrated that automatization is associated with improved performance and reduced activity in WM. In a related study we compared the difference in WM activity after practice to the ability to perform two tasks simultaneously. It was found that subjects with a larger reduction in WM activity after practice were better at performing two tasks concurrently (Ramsey et al., 2004). This suggests that the drop in WM activity induced by practice may reflect an increase of available WM capacity to accommodate concurrent task performance.

Although behavioral tests of automatization indicate that schizophrenic patients improve performance with practice (Harvey et al., 2000, Serper et al., 1990) several studies have shown that patients are either unable to process a second task concurrently (Granholm et al., 1991, Granholm et al., 1996) or need significantly more practice to achieve normal dual-task performance (Gold and Harvey, 1993, Serper et al., 1990, Granholm et al., 1991, Granholm et al., 1996, Harvey et al., 2000). This suggests that schizophrenic patients may fail to reduce neural activity in spite of improved performance after practice and are therefore unable to liberate sufficient neural resources to perform additional tasks simultaneously.

This raises the question whether inefficient WM function and limited capacity in schizophrenia could be associated with a failure in automatization. To test this, brain activity was examined with fMRI using a verbal WM task with novel and practiced stimulus sets (Sternberg, 1966). Subjects subsequently participated in a behavioral dual-task paradigm to measure their ability of concurrent task performance.

Back to Article Outline

2. Methods and materials 

2.1. Subjects 

Patients were recruited from the Department of Psychiatry of the University Medical Center Utrecht. DSM-IV diagnosis of schizophrenia was confirmed using The Comprehensive Assessment of Symptoms and History (CASH) (Andreasen et al., 1992a) and severity of symptoms was assessed with The Positive And Negative Syndrome Scale (PANSS) interview (Kay et al., 1987). Controls were recruited through advertisement and were rewarded for participation. The Mini International Neuropsychiatric Interview (M.I.N.I.) (Sheehan et al., 1998) was used to exclude controls with a history of neurological illness, psychiatric disorders or substance abuse. All participants were tested for right-handedness using the Edinburgh Handedness Index (EHI) (Oldfield, 1971). Initially, 18 controls and 22 patients were scanned after signing an informed consent. In total four patients were excluded from the study; one could not complete the scan procedure due to technical problems with the scanner, three were excluded from analysis because their performance was at chance level. Presented results are based on the remaining cohort with 18 controls and 18 patients matched for age and gender. Demographic and illness-related details are shown in Table 1.

Table 1. Demographic and illness-related variables in healthy controls and subjects with schizophrenia
Patients with schizophreniaHealthy controls
NMeanSDRangeNMeanSDRangep-valuea
Male14 13 –
Female4 5 –
Age (yr) 28.47.419.6–41.8 26.05.818.9–43.8n.s.
EHI index 0.800.190.42–1.00 0.860.200.33–1.00n.s.
Education (yr) 13.11.949–15 13.41.5010–16n.s.
PANSS (item mean)
Positive scale 1.770.571.00–2.83
Negative scale 2.100.721.00–3.29
General scale 1.590.321.06–2.19
Length of Illness (yr) 5.234.360.77–17.53
Age of Onset (yr) 23.175.0415–33
Medication (mg/day)
Clozapine7145.361790–400
Quetiapine1700–700
Risperidon31.330.581–2
Olanzapine610.757.642–20

N: number of subjects, SD: standard deviation. PANSS:Positive and Negative Syndrom Scale (Kay et al., 1987), EHI: Edinburgh Handedness Index (Oldfield, 1971).

aSignificance of differences calculated by using t-tests and nonparametric Kolmogorov–Smirnov Z-test, two-tailed.

2.2. Assessment of automatization 

The fMRI experiment to test automatization involved a modified version of the Sternberg paradigm (STERN) (see also (Jansma et al., 2001, Ramsey et al., 2004, Jager et al., 2006). Subjects were instructed to memorize a set of five letters and subsequently respond to matching probes (targets) (Fig. 1). A novel (NT) and a practiced task (PT) were administered. In PT a fixed set of target and non-target stimuli was selected from an array of ten preset letters (Czerwinski et al., 1992). In NT, target and non-target letters were randomly chosen from the ten remaining consonants that were not used in PT. In addition, a control task (CT) was included during which subjects made a button press when the symbol ‘< >’ appeared. Prior to fMRI subjects performed five series of 100 PT stimuli (25 min) to induce automatization. During scanning an epoch (memory set and ten probes) lasted 32 s. The four tasks (NT, PT, CT and rest of equal epoch length) were presented eight times in a pseudo-randomized and counterbalanced order. Reaction times (RT) of all correctly identified targets and percentage correct responses for all stimuli were recorded.

  • View full-size image.
  • Fig. 1. 

    The STERN task. Each epoch starts with presentation of the target set and is followed by ten probes. Subjects have to press a button as quickly as possible if the probe letter belongs to the set of targets.

2.3. Assessment of capacity: dual-task paradigm 

After fMRI, STERN (five NT and five PT blocks) was performed concurrently with a selective attention task (SAT) outside the scanner. SAT involved detection of tones with a higher or lower pitch than a baseline tone (see also (Jager et al., 2006, Ramsey et al., 2004)). Task difficulty of SAT was standardized for each subject prior to the experiment, by adjusting pitch difference until the subject detected 80% of the deviant tones. The 200 ms tones (16% deviants randomly distributed) were presented in blocks of 25 s. STERN intertrial interval in the dual-task was 2500 ms with stimulus duration of 1500 ms. Although tones and letters frequently coincided, only on three out of 64 occasions a tone and STERN target overlapped. To prevent interference at the response level, subjects silently counted target tones and verbally reported the number after each series of 25 stimuli. STERN and SAT were also administered separately. STERN and SAT “performance cost” was defined as the difference in performance on the dual versus the single tasks and was also computed for NT and PT blocks separately.

2.4. FMRI procedure and acquisition 

FMRI was performed on a Philips 1.5 T Intera scanner. Subjects were positioned supinely in the scanner. Head movement was reduced by using a strap around the forehead and foam padding. A mirror attached to the head coil enabled subjects to see a through-projection screen positioned near the feet. A video projector located outside the scanner room projected the tasks on the screen. A pneumatic push-button box with air pressure cables was used to record responses. To measure BOLD signal changes we used a three-dimensional navigated PRESTO pulse-sequence (Ramsey et al., 1998). A single run of 384 scans was acquired over a period of 17 min. This was followed by one high-contrast scan for registration purposes and a 3-D anatomical scan for spatial localization (for scan parameters see Jager et al., 2006, Jansma et al., 2001).

2.5. Data analysis: behavioral data 

Individual mean and variance of RT calculated over all correct responses and percentage of correct responses of STERN during fMRI were tested with a general linear model (GLM) with repeated measurements, with practice (NT and PT) as within-subject factors and group (patients and controls) as between-subject factor. Dual-task performance cost was separately tested with a GLM with task (STERN and SAT) and practice (NT and PT) as within-subject factors and group (patients and controls) as between-subject factor.

2.6. Data analysis: imaging data 

After motion correction, fMRI signals were analyzed voxelwise, using multiple regression (Worsley and Friston, 1995, Worsley et al., 1992). This resulted in individual activation t-maps for the three tasks contrasted with rest. Individual t-maps (spatially smoothed with a Gaussian kernel with a FWHM of 8 mm) and anatomical volumes were normalized into standard MNI-305 space (Collins et al., 1994). A group t-map was calculated for NT-CT by testing the difference values in each voxel against zero. A value of 4.51 was used as an activity threshold, corresponding to a p-value of <0.05 (one-sided) Bonferroni-corrected for total brain volume. For all significantly active regions in the group map with a cluster size of at least ten voxels, a mean activity score (b-values obtained from the regression analysis) was obtained for NT, PT and CT for each subject. These final variables were entered into repeated measurements GLM analysis with practice (NT and PT) and region (listed in Table 2) as within-subject factors and group (patients and controls) as between-subjects factor.

Table 2. Regions identified in the t-map of the STERN task (NT versus CT contrast, for the two groups combined) and used for further analyses
RegionBrodmann areaNumber of voxelsXYZMaximum t-value
1 Left fusiform gyrus (LFG)371445−59−1110.53
2 Left dorsolateral prefrontal cortex (LPFC)9/4619346112914.36
3 Left superior parietal cortex (LSPC)713333−564110.31
4 Right superior parietal cortex (RSPC)756−32−60428.77
5 Anterior cingulate cortex (ACC)6/24574235318.52

These regions constitute the WM network associated with STERN (p<0.05 Bonferroni-corrected, minimum clustersize 10) The MNI-coordinates of the voxel with the highest statistical value (group-map t-score) within each region are listed under X, Y and Z (Collins et al., 1994). The numbers of the regions correspond to those shown in Fig. 3.

Back to Article Outline

3. Results 

3.1. Practice and working memory 

3.1.1. Effects of practice on working memory performance 

The performance results are shown in Fig. 2. NT reaction times were significantly longer than in PT in both groups (main effect of task [F(1,32)=57.1; p<0.0001]). Patients were overall slower than controls (main effect of group [F(1,32)=13.9; p=0.001]). There was no significant group by task interaction [F(1,32)=0.12; p=0.65], indicating that patients improved RT with practice to the same degree as controls. Although RT were more variable in schizophrenia across conditions (main effect of group [F(1,26)=9.29, p=0.005]), RT variability was larger in NT than in PT in both groups (main effect of task [F1,26)=19.01, p<0.0001]). There was no significant group by task interaction [F(1,26)=0.76, p=0.39], suggesting that the effect of practice on RT variability was similar in both groups. Patients and controls did not significantly differ in percentage of correct responses [F(1,32)=2.45, p=0.13]. The main effect of task shows that performance was more accurate after practice [F(1,32)=13.98, p=0.001]. The group by task interaction was not significant [F(1,32)=0.10, p=0.75]. This suggests that the effect of practice on accuracy was similar in both groups. In summary, although RT were generally slower and more variable in patients, practice improved performance to the same degree in both groups.

  • View full-size image.
  • Fig. 2. 

    Graphs show performance on the STERN task for both groups, during fMRI scans. A: reaction times of correct target responses. B: number of correct responses as percent of all trials (considering both misses and false alarms). C: Variance of reaction time of all correct target responses. Errorbars denote standard error of the mean. Abbreviations: NT novel task, PT practised task, CT control task.

3.1.2. Overview of regions of interest 

Regions in frontal (left dorsolateral prefrontal cortex (LPFC) and anterior cingulate cortex (ACC)) parietal (left superior parietal cortex (LSPC) and right superior parietal cortex (RSPC)) and visual cortex (left fusiform gyrus (LFG) reached significance in both groups and were used for further analyses (Table 2 and Fig. 3). These regions are the same in the previous studies with this task (Jansma et al., 2001, Ramsey et al., 2004, Jager et al., 2006). Patients did not activate additional regions.

  • View full-size image.
  • Fig. 3. 

    Activity map of 18 patients and 18 controls combined, of the STERN task (NT versus CT contrast), showing WM regions: 1. left fusiform gyrus LFG, 2. left prefrontal cortex LPFC, 3. left superior parietal cortex LSPC, 4. right superior parietal cortex RSPC, 5. anterior cingulate cortex ACC. The numbers in the slices correspond to MNI z-coordinates (Collins et al., 1994). Threshold for significance corresponded to p<0.05 Bonferroni-corrected, with a minimum clustersize of 10 voxels. Slices are in radiological orientation (left side is right hemisphere and vice versa).

3.1.3. Effects of practice on working memory function 

To assess the effect of practice on WM activity mean b-values from individual NT and PT maps of the five ROI's were used in the analysis. Practice significantly reduced brain activity in all WM regions in both groups (main effect of practice [F(1,32)=142.10; p<0.001] (Fig. 2, Fig. 3). Practice induced a larger drop in brain activity in schizophrenics (group by practice interaction [F(1,32)=8.79; p=0.005]). Post-hoc ANOVA indicates that this was due to excessive activity in schizophrenia during NT, especially in LPFC [F(1,32)=5.04; p=0.03] and LSPC [F(1,32)=6.96; p=0.012] (Fig. 4). These results indicate that patients excessively activated WM, especially left prefrontal and left parietal regions during NT, but were capable of normalizing levels of brain activity after practice.

  • View full-size image.
  • Fig. 4. 

    Brain activity for the STERN tasks in WM regions depicted in Fig. 3. Results are displayed for each of the tasks (NT, PT and CT) for controls (A) and patients (B) separately. Activity levels (y-axis) are represented by b-values obtained from the GLM analyses of fMRI data. Groups were analysed together (see methods section). Activity was reduced with practise (NT versus PT) in all regions, for both groups. A significant group difference was found in regions 2 (left prefrontal cortex) and 3 (left superior parietal cortex), in the magnitude of activity decline following practise (NT versus PT). Asterisks denote a significant interaction between group and practise, which was due to elevated activity during NT in patients (see methods for details).

3.2. Practice and processing capacity 

3.2.1. Dual task performance 

STERN and SAT performance cost scores (Fig. 5) were calculated to assess WM capacity. Patients showed disproportionate performance cost for both SAT and STERN (main effect of group [F(1,32)=10.53; p=0.003)]). SAT performance cost was higher than STERN (main effect of task [F(1,32)=21.42; p<0.001]), indicating that SAT performance was more influenced by simultaneous processing demands than STERN. The difference in STERN and SAT performance cost was more pronounced in schizophrenia [task x group interaction F(1,32)=5.64; p=0.02)]. Practice reduced performance cost in both tasks as was shown by a main effect of condition [F(1,32)=9.03; p=0.005]. Thus for both groups performance cost was larger for SAT than STERN, but practice reduced performance cost in both tasks. In addition, patients showed excessive performance cost for both tasks and especially for SAT.

  • View full-size image.
  • Fig. 5. 

    Effect of performing two tasks simultaneously (dual task) on performance for the STERN and SAT tasks (dual task performance cost, outside the scanner). On the y-axis the difference in accuracy is shown between single and dual-task performance of STERN and SAT; a high value corresponds to a greater loss of performance. Patients show higher performance cost than controls, especially on SAT. Asterisk denotes a significant difference between patients and controls, on the NT and PT during SAT combined (p<0.05).

3.2.2. Practice-related activity changes in WM and performance cost 

The difference in WM activity after practice predicted STERN performance cost in controls (r=0.58, p<0.02) (Fig. 6). Thus, subjects with a larger drop in activity were better at maintaining STERN performance when a second task was added (Ramsey et al., 2004). There was no such relationship in schizophrenia (r=−0.16, p=0.54) (Fig. 6). The difference between the correlations was significant: (Fisher z-transform, z=2.256, p=0.024 (Snedecor and Cochran, 1980)).

  • View full-size image.
  • Fig. 6. 

    Correlation between WM efficiency and capacity. On the y-axis, the difference in activity between NT and PT is shown as a measure of practise-induced neurophysiological efficiency of WM function. A larger value indicates a larger reduction in brain activity, and therefore a greater efficiency. On the x-axis performance cost is displayed, which is computed as the difference in errors between single task performance and dual-task performance, and represents the capacity of the WM system. A larger value corresponds with larger performance decrement in the dual-task, and therefore a smaller capacity. The fitted lines and the corresponding rho values reflect the correlation between efficiency and capacity, computed for patients and controls separately. There was a correlation only in controls. The difference between correlations for the groups was significant (see Methods and materials).

Back to Article Outline

4. Discussion 

To investigate whether inefficient WM function and reduced capacity in schizophrenia were associated with a failure in automatization, schizophrenia and healthy subjects performed a WM task with novel and practiced stimuli during fMRI and a dual-task outside fMRI.

Although RT were generally longer and more variable in schizophrenia, patients displayed normal accuracy on the novel WM task which was accompanied by excessive brain activity in WM regions. Practice improved performance in patients to the same extent as in controls and normalized excessive levels of activity. While practice reduced performance cost in both groups, patients exhibited overall disproportionate performance cost in the dual-task, especially during tone-counting. The difference in WM activity after practice predicted STERN performance cost in controls but not in patients.

Excessive WM activity during novel WM performance in patients confirms the notion of inefficient WM function in schizophrenia (Callicott et al., 2000a, Callicott et al., 2003, Manoach et al., 2000, Manoach, 2003, Jansma et al., 2004, Karlsgodt et al., 2007). Patients were able to improve performance and reduce inefficient activity down to normal levels with practice. This is in line with most behavioral studies (Harvey et al., 2000, Serper et al., 1990), however not many studies have yet investigated the effect of practice on brain function in schizophrenia. Most studies investigated procedural learning (Kumari et al., 2002, Zedkova et al., 2006); a rule-based type of skill acquisition that is implicitly acquired through practice (Kumari et al., 2002), and have reported abnormal patterns of brain activity associated with either intact (Zedkova et al., 2006) or impaired procedural learning (Kumari et al., 2002). A major difference with procedural learning however is that automatization does not involve implicit acquisition as memory sets were explicitly practiced during training.

Excessive performance cost in patients complements other studies reporting impaired dual-task performance in patients with schizophrenia (Granholm et al., 1991, Granholm et al., 1996, Harvey et al., 2000, Serper et al., 1990). A study by Oram et al. (Oram et al., 2005) however reported that a tone-counting task did not significantly affect performance on a simple visual search task in patients with schizophrenia. In our study the impact of the visual task (STERN) on tone-counting was larger than the reciprocal effect of tone-counting on STERN performance. In Oram et al.'s study the effect of the visual search task on tone-counting performance was not reported. This does not exclude the possibility that visual search may have affected tone-counting performance as was the case in our study.

The drop in WM activity after practice predicted performance cost in controls. Thus subjects with a larger difference in WM activity after practice were better capable to utilize WM capacity to resolve interference between tasks when executed simultaneously (Ramsey et al., 2004). The lack of such a relationship with performance cost in schizophrenia suggests that the dual-task was too difficult and that patients may have failed to recruit WM to accommodate concurrent task performance. Our data thus shows that patients were able to normalize initial inefficient WM function with practice. The question however remains why concurrent performance results in greater performance cost in schizophrenia.

SAT and STERN both activate WM, which is most likely involved in storage of verbal and auditory information and executive processes associated with response selection (STERN) or updating the count of target tones (SAT). Executive processes in a dual-task cannot be performed on more than one task at the time (Pashler, 1994a, Pashler, 1994b, Ruthruff et al., 2001). When SAT and STERN are performed concurrently processing conflicts may thus arise as they compete for common WM resources (Bunge et al., 2000, Klingberg, 2000, Adcock et al., 2000). This may induce errors as the interfering task temporarily disrupts and delays ongoing processing (Pashler, 1994a, Pashler, 1994b, Ruthruff et al., 2001). Excessive performance cost in schizophrenia may therefore reflect aggravated disruption of ongoing processing in a dual-task and may be associated with greater competition for WM to resolve processing conflicts when simultaneously performing STERN and SAT.

Performance cost was most pronounced for SAT. This task requires continuous adjustment of temporarily stored information each time an oddball is detected. Disproportionate SAT performance cost in schizophrenia further suggests that patients are most susceptible to the disruptive effect of a concurrent task when information needs to be frequently updated. Together with our finding of inefficient brain function during WM performance with continuously changing information, this may indicate that schizophrenia is associated with a failure to properly recruit WM when information changes frequently and thus requires continuous updating.

In spite of overall impaired dual-task performance, practice reduced performance cost in patients as well. This is in agreement with other behavioral studies reporting that patients are able to improve dual-task performance to some extent with practice on a single task (Harvey et al., 2000, Serper et al., 1990). The question may rise whether practicing the dual-task would have reduced performance cost in patients. Dual-task practice may reorganize two tasks into a single task, which may eliminate processing conflicts (Ruthruff et al., 2006). A recent study (Ruthruff et al., 2006) compared the effects of single-task and dual-task practice on subsequent dual-task performance, but did not find evidence that dual-task practice induced more efficient task integration than single-task practice. Although it is likely that patients would improve dual-task performance with dual-task practice we do not expect that this would eliminate the difference in performance cost with controls.

The training session in the current study to induce automatization was relatively short. It has been argued that automaticity is not completed until performance has reached plateau level (Shiffrin and Schneider, 1977). On the other hand it has been suggested that automaticity is induced after only a few trials of practice (Logan, 2002). Although we cannot draw conclusions about potential long-term changes in brain function and performance associated with automatization, our current results suggests that during the early phase of practice patients were able to improve task performance and reduce demands on WM.

RT may be a more precise measure for performance cost (Ruthruff et al., 2001). In the current study however we used accuracy to calculate performance cost, as in our previous work this was found to be correlated with the difference in WM activity after practice in controls. In addition, in order to minimize interference at the level of response selection between the two tasks, only one task in this paradigm required a manual response, RT measures were therefore not available for both tasks.

All patients in the present study received pharmacological treatment with atypical antipsychotics. Studies have indicated that antipsychotics facilitate automatization in schizophrenia (Serper et al., 1990). Particularly patients taking atypical antipsychotics became more efficient than patients on typical antipsychotics in performing a practiced task simultaneously with an additional cognitive task (Harvey et al., 2000). It is therefore possible that atypical antipsychotic medication may have positively affected automatization in schizophrenia. However this cannot explain the disproportional performance loss in the dual-task in our patients.

Several other potential limitations could be relevant to the interpretation and implications of the present study. For one, general intelligence may have been different for both groups. Although IQ was not assessed in the present study, we did find that both groups had equal numbers of years of education (Table 1). Moreover, elevated activity during NT is not likely to be the result of potentially lower IQ levels in patients: previous studies in healthy volunteers have reported a positive correlation between intelligence and magnitude of brain activity (eg Gray et al., 2003), which would predict patients to exhibit less rather than more activity in the WM network. As with other studies, the implications for schizophrenia as a whole should be considered with several limitations in mind. Given the fact that only 18 patients were included, who all were capable of performing the task, that only a few female patients were assessed, and that the duration of illness was relatively short (i.e. 5 yr), the results may not generalize across the schizophrenia spectrum. It may well be that more severe, chronic patients would perform more poorly and exhibit different brain activity levels. Inclusion of such patients, however, requires a different approach to data analysis and interpretation because poor task performance has complex effects on brain activity (eg Jansma et al., 2004).

To conclude, the ability to reduce inefficient WM function with practice does not support the notion that automatization is impaired in schizophrenia. We also did not find a relationship between the difference in WM activity after practice and performance cost in schizophrenia. Together, this may suggest that inefficient WM function and reduced capacity in schizophrenia are associated with a failure to properly engage WM when task demands are increased (i.e. during novel WM performance and when performing an additional task concurrently). In addition, this WM failure may be specifically related to an inability to process continuously changing information requiring frequent updating.

Back to Article Outline

Role of funding source 

This research was sponsored by Netherlands Organisation for Health Research and Development (ZonMW) and Utrecht University, Grant 016.036.401 as a personal grant for N.F.Ramsey (title: ‘The neural basis of automatization of cognitive functions’). Both ZonMW and Utrecht University had no further role in the study design; in the collection, analyses and interpretation of the data; in the writing of the article; and the decision to submit the article for publication.

Back to Article Outline

Contributors 

Martijn Jansma and Nick Ramsey designed the study. Tamar van Raalten collected the data, performed the statistical analyses and wrote the first draft of the article. Gerry Jager participated in the data collection of the controls. All authors contributed to and have approved the final manuscript.

Back to Article Outline

Conflict of interest 

All authors disclose no conflict of interests, including any financial, personal or other relationships with other people or organizations that could have inappropriately influenced or could be perceived to influence the work presented in this paper.

Back to Article Outline

Acknowledgements 

We wish to thank Martijn van den Heuvel and Matthijs Vink for their support with the data analysis.

Back to Article Outline

References 

  1. Adcock RA, Constable RT, Gore JC, Goldman-Rakic PS. Functional neuroanatomy of executive processes involved in dual-task performance. Proc. Natl. Acad. Sci. U S A. 2000;97:3567–3572
  2. Andreasen NC, Flaum M, Arndt S. The Comprehensive Assessment of Symptoms and History (CASH). An instrument for assessing diagnosis and psychopathology. Arch. Gen. Psychiatry. 1992;49:615–623
  3. Andreasen NC, Rezai K, Alliger R, Swayze VW, Flaum M, Kirchner P, et al. Hypofrontality in neuroleptic-naive patients and in patients with chronic schizophrenia. Assessment with xenon 133 single-photon emission computed tomography and the Tower of London. Arch. Gen. Psychiatry. 1992;49:943–958
  4. Baddeley A. Working Memory. Oxford: Claredon Press; 1986;
  5. Berman KF, Torrey EF, Daniel DG, Weinberger DR. Regional cerebral blood flow in monozygotic twins discordant and concordant for schizophrenia. Arch. Gen. Psychiatry. 1992;49:927–934
  6. Bertolino A, Esposito G, Callicott JH, Mattay VS, Van Horn JD, Frank JA, et al. Specific relationship between prefrontal neuronal N-acetylaspartate and activation of the working memory cortical network in schizophrenia. Am. J. Psychiatry. 2000;157:26–33
  7. Bertolino A, Caforio G, Blasi G, De Candia M, Latorre V, Petruzzella V, et al. Interaction of COMT Val108/158 Met genotype and olanzapine treatment on prefrontal cortical function in patients with schizophrenia. Am. J. Psychiatry. 2004;161:1798–1805
  8. Bertolino A, Blasi G, Latorre V, Rubino V, Rampino A, Sinibaldi L, et al. Additive effects of genetic variation in dopamine regulating genes on working memory cortical activity in human brain. J. Neurosci. 2006;26:3918–3922
  9. Bertolino A, Caforio G, Petruzzella V, Latorre V, Rubino V, Dimalta S, et al. Prefrontal dysfunction in schizophrenia controlling for COMT Val158Met genotype and working memory performance. Psychiatry Res.: Neuroimaging. 2006;147:221–226
  10. Bunge SA, Klingberg T, Jacobsen RB, Gabrieli JD. A resource model of the neural basis of executive working memory. Proc. Natl. Acad. Sci. U S A. 2000;97:3573–3578
  11. Callicott JH, Mattay VS, Bertolino A, Finn K, Coppola R, Frank JA, et al. Physiological characteristics of capacity constraints in working memory as revealed by functional MRI. Cereb. Cortex. 1999;9:20–26
  12. Callicott JH, Bertolino A, Mattay VS, Langheim FJ, Duyn J, Coppola R, et al. Physiological dysfunction of the dorsolateral prefrontal cortex in schizophrenia revisited. Cereb. Cortex. 2000;10:1078–1092
  13. Callicott JH, Bertolino A, Egan MF, Mattay VS, Langheim FJP, Weinberger DR. Selective relationship between prefrontal N-acetylaspartate measures and negative symptoms in schizophrenia. Am. J. Psychiatry. 2000;157:1646–1651
  14. Callicott JH, Mattay VS, Verchinski BA, Marenco S, Egan MF, Weinberger DR. Complexity of prefrontal cortical dysfunction in schizophrenia: more than up or down. Am. J. Psychiatry. 2003;160:2209–2215
  15. Collins DL, Neelin P, Peters TM, Evans AC. Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space. J. Comput. Assist. Tomogr. 1994;18:192–205
  16. Czerwinski M, Lightfoot N, Shiffrin RM. Automatization and training in visual search. Am. J. Psychol. 1992;105:271–315
  17. Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, Straub RE, et al. Effect of COMT Val108/158 Met genotype on frontal lobe function and risk for schizophrenia. Proc. Natl. Acad. Sci. U S A. 2001;98:6917–6922
  18. Elvevag B, Goldberg TE. Cognitive impairment in schizophrenia is the core of the disorder. Crit. Rev. Neurobiol. 2000;14:1–21
  19. Gold JM, Harvey PD. Cognitive deficits in schizophrenia. Psychiatr. Clin. North Am. 1993;16:295–312
  20. Granholm E, Asarnow RF, Marder SR. Controlled information processing resources and the development of automatic detection responses in schizophrenia. J. Abnorm. Psychology. 1991;100:22–30
  21. Granholm E, Asarnow RF, Marder SR. Dual-task performance operating characteristics, resource limitations, and automatic processing in schizophrenia. Neuropsychology. 1996;10:11–21
  22. Gray JR, Chabris CF, Braver TS. Neural mechanisms of general fluid intelligence. Nat. Neurosci. 2003;6(3):316–322
  23. Harvey PD, Moriarty PJ, Serper MR, Schnur E, Lieber D. Practice-related improvement in information processing with novel antipsychotic treatment. Schizophr. Res. 2000;46:139–148
  24. Jager G, Kahn R, Van Den Brink W, Van Ree J, Ramsey N. Long-term effects of frequent cannabis use on working memory and attention: an fMRI study. Psychopharmacology. 2006;185:358–368
  25. Jansma JM, Ramsey NF, Slagter HA, Kahn RS. Functional anatomical correlates of controlled and automatic processing. J. Cogn. Neurosci. 2001;13:730–743
  26. Jansma JM, Ramsey NF, van der Wee NJA, Kahn RS. Working memory capacity in schizophrenia: a parametric fMRI study. Schizophr. Res. 2004;68:159–171
  27. Karlsgodt KH, Glahn DC, van Erp TG, Therman S, Huttunen M, Manninen M, et al. The relationship between performance and fMRI signal during working memory in patients with schizophrenia, unaffected co-twins, and control subjects. Schizophr. Res. 2007;68(1–3):191–197
  28. Kay SR, Fiszbein A, Opler LA. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr. Bull. 1987;13:261–276
  29. Klingberg T. Limitations in information processing in the human brain: neuroimaging of dual task performance and working memory tasks. Prog. Brain Res. 2000;126:95–102
  30. Kumari V, Gray JA, Honey GD, Soni W, Bullmore ET, Williams SCR, et al. Procedural learning in schizophrenia: a functional magnetic resonance imaging investigation. Schizophr. Res. 2002;57:97–107
  31. Logan GD. An instance theory of attention and memory. Psychol. Rev. 2002;109:376–400
  32. Manoach DS. Prefrontal cortex dysfunction during working memory performance in schizophrenia: reconciling discrepant findings. Psychol. Rev. 2003;60:285–298
  33. Manoach DS, Press DZ, Thangaraj V, Searl MM, Goff DC, Halpern E, et al. Schizophrenic subjects activate dorsolateral prefrontal cortex during a working memory task, as measured by fMRI. Biol. Psychiatry. 1999;45:1128–1137
  34. Manoach DS, Gollub RL, Benson ES, Searl MM, Goff DC, Halpern E, et al. Schizophrenic subjects show aberrant fMRI activation of dorsolateral prefrontal cortex and basal ganglia during working memory performance. Biol. Psychiatry. 2000;48:99–109
  35. Miller EK, Cohen JD. An integrative theory of prefrontal function. Annu. Rev. Neurosci. 2001;24:167–202
  36. Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia. 1971;9:97–113
  37. Oram J, Geffen GM, Geffen LB, Kavanagh DJ, McGrath JJ. Executive control of working memory in schizophrenia. Psychiatry Res. 2005;135:81–90
  38. Pashler H. Overlapping mental operations in serial performance with preview. Q. J. Exp. Psychol. A. 1994;47:161–191
  39. Pashler H. Dual-task interference in simple tasks: data and theory. Psychol. Bull. 1994;116:220–244
  40. Perlstein WM, Carter CS, Noll DC, Cohen JD. Relation of prefrontal cortex dysfunction to working memory and symptoms in schizophrenia. Am. J. Psychiatry. 2001;158:1105–1113
  41. Ramsey NF, van den Brink JS, van Muiswinkel AMC, Folkers PJM, Moonen CTW, Jansma JM, et al. Phase navigator correction in 3d fmri improves detection of brain activation: quantitative assessment with a graded motor activation procedure. NeuroImage. 1998;8:240–248
  42. Ramsey NF, Jansma JM, Jager G, Van Raalten T, Kahn RS. Neurophysiological factors in human information processing capacity. Brain. 2004;127:517–525
  43. Ruthruff E, Pashler HE, Klaassen A. Processing bottlenecks in dual-task performance: structural limitation or strategic postponement?. Psychon. Bull. Rev. 2001;8:73–80
  44. Ruthruff E, Van Selst M, Johnston JC, Remington R. How does practice reduce dual-task interference: integration, automatization, or just stage-shortening?. Psychol. Res. 2006;70:125–142
  45. Serper MR, Bergman RL, Harvey PD. Medication may be required for the development of automatic information processing in schizophrenia. Psychiatry Res. 1990;32:281–288
  46. Sheehan DV, Lecrubier Y, Sheehan KH, Amorim P, Janavs J, Weiller E, et al. The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J. Clin. Psychiatry. 1998;20(59 Suppl):22–33
  47. Shiffrin RM, Schneider W. Controlled and automatic human information processing: II. Perceptual learning, automatic attending and a general theory. Psychol. Rev. 1977;84:127–190
  48. Snedecor , Cochran . Statistical Methods. Iowa State University Press; 1980;
  49. Sternberg S. High-speed scanning in human memory. Science. 1966;153:652–654
  50. Van Snellenberg JX, Torres IJ, Thornton AE. Functional neuroimaging of working memory in schizophrenia: task performance as a moderating variable. Neuropsychology. 2006;20:497–510
  51. Worsley KJ, Friston KJ. Analysis of fMRI Time-Series Revisited-Again. NeuroImage. 1995;2:173–181
  52. Worsley KJ, Evans AC, Marrett S, Neelin P. A three-dimensional statistical analysis for CBF activation studies in human brain. J. Cereb. Blood Flow Metab. 1992;12:900–918
  53. Zedkova L, Woodward ND, Harding I, Tibbo PG, Purdon SE. Procedural learning in schizophrenia investigated with functional magnetic resonance imaging. Schizophr. Res. 2006;88:198–207

PII: S0920-9964(07)00499-9

doi:10.1016/j.schres.2007.10.035

Schizophrenia Research
Volume 100, Issue 1 , Pages 161-171, March 2008

Article Tools

* Image Usage & Resolution