Elsevier

Biological Psychiatry

Volume 64, Issue 12, 15 December 2008, Pages 1035-1041
Biological Psychiatry

Archival Report
Multivariate Pattern Analysis of Functional Magnetic Resonance Imaging Data Reveals Deficits in Distributed Representations in Schizophrenia

https://doi.org/10.1016/j.biopsych.2008.07.025Get rights and content

Background

Multivariate pattern analysis is an alternative method of analyzing functional magnetic resonance imaging (fMRI) data, which is capable of decoding distributed neural representations. We applied this method to test the hypothesis of the impairment in distributed representations in schizophrenia. We also compared the results of this method with traditional general linear model (GLM)-based univariate analysis.

Methods

Nineteen schizophrenia and 15 control subjects viewed two runs of stimuli—exemplars of faces, scenes, objects, and scrambled images. To verify engagement with stimuli, subjects completed a 1-back matching task. A multivoxel pattern classifier was trained to identify category-specific activity patterns on one run of fMRI data. Classification testing was conducted on the remaining run. Correlation of voxelwise activity across runs evaluated variance over time in activity patterns.

Results

Patients performed the task less accurately. This group difference was reflected in the pattern analysis results with diminished classification accuracy in patients compared with control subjects, 59% and 72%, respectively. In contrast, there was no group difference in GLM-based univariate measures. In both groups, classification accuracy was significantly correlated with behavioral measures. Both groups showed highly significant correlation between interrun correlations and classification accuracy.

Conclusions

Distributed representations of visual objects are impaired in schizophrenia. This impairment is correlated with diminished task performance, suggesting that decreased integrity of cortical activity patterns is reflected in impaired behavior. Comparisons with univariate results suggest greater sensitivity of pattern analysis in detecting group differences in neural activity and reduced likelihood of nonspecific factors driving these results.

Section snippets

Subjects

Nineteen individuals with schizophrenia (SZ) and 15 healthy control subjects (C) were studied. Demographic and clinical data are displayed in Table 1. Data from a subset of these subjects have been published in a separate study (18). Patients were clinically stable and outpatients. Diagnostic status was evaluated with the Structured Clinical Interview for DSM-IV-Text Revision (SCID-I) conducted by masters- or doctoral-level clinicians and confirmed by consensus conference. Symptoms were

Results

Both groups performed the task well, but control subjects showed significantly higher accuracy than patients, 93% versus 81% (t = −5.5, p < .001). For all categories, control subject performance was higher, and for all but the everyday objects, the differences were significant. Control subjects responded faster, but RT differences were not significant (MC= 565 msec, SDC= 96 msec, MSZ= 617 msec, SDSZ = 117 msec, p = .17). For faces, there was a nonsignificant difference in reaction time (t =

Discussion

The accuracy of a multivariate pattern classifier was high for healthy subjects and in close agreement with other studies (15, 25). Classification accuracy was significantly lower in subjects with schizophrenia. In both groups, there was an inverse correlation between classification accuracy and reaction time, and in the schizophrenia group, there was a positive correlation between classification accuracy and behavioral accuracy. Diminished voxelwise correlation in activity across runs in

References (29)

  • D.M. Barch et al.

    Selective deficits in prefrontal cortex function in medication-naive patients with schizophrenia

    Arch Gen Psychiatry

    (2001)
  • K.F. Berman et al.

    Physiologic dysfunction of dorsolateral prefrontal cortex in schizophreniaII. Role of neuroleptic treatment, attention, and mental effort

    Arch Gen Psychiatry

    (1986)
  • J.H. Callicott et al.

    Physiological dysfunction of the dorsolateral prefrontal cortex in schizophrenia revisited

    Cereb Cortex

    (2000)
  • S. Heckers et al.

    Impaired recruitment of the hippocampus during conscious recollection in schizophrenia

    Nat Neurosci

    (1998)
  • Cited by (37)

    • Neural Representation in Visual Word Form Area during Word Reading

      2021, Neuroscience
      Citation Excerpt :

      Finally, among studies investigating whether the VWFA represents case information, most of them analyzed their neuroimaging data by performing univariate activation analysis, in which each voxel is treated independently and multi-voxel pattern information is missed (Haxby, 2012; Haynes, 2015). The multi-voxel pattern information should be taken into account because the neural pattern information has been repeatedly found to be associated with information processing (Norman et al., 2006; Haxby, 2012; Haynes, 2015; Heilbron et al., 2020), and because there is evidence that the neural pattern information differs across different conditions even in brain regions showing common activations in several domains such as language and cognitive control (Yoon et al., 2008; Raizada et al., 2010; Xu et al., 2017; Carlos et al., 2019). In addition, multivariate methods (e.g., multivariate pattern analysis, MVPA) are able to examine stimulus-independent representation by training the classifier to distinguish one type of stimuli (e.g., lowercase words and consonant strings) and testing the classifier’s ability in predicting another type of stimuli (e.g., uppercase words and consonant strings) which own the same abstract representation as the stimuli used for training (Wurm et al., 2016; Van de Putte et al., 2017; Heilbron et al., 2020).

    View all citing articles on Scopus
    View full text