A multiple case study examining neuropsychological properties of the Cognitive Assessment System

The purpose of this exploratory study was to arrive at a first approximation of neuropsychological properties of the Cognitive Assessment System (CAS) in a sample of adult brain-damaged patients (N = 32) each with a single localized infarct in the cortex and subcortical regions. Thirty-five robust double dissociations were found between individual CAS subtests and 12 patterns were evident. Matching Numbers was doubly dissociated from Figure Memory (n = 1). Number Detection was doubly dissociated from Word Series (n = 3), Sentence Repetition (n = 6), and Sentence Questions (n = 2). Nonverbal Matrices was doubly dissociated from Word Series (n = 4) and Sentence Repetition (n = 4). Verbal Spatial Relations was doubly dissociated from Word Series (n = 2) and Sentence Repetition (n = 4). Figure Memory was doubly dissociated from Word Series (n = 3), Sentence Repetition (n = 3) and Sentence Questions (n = 2). Finally Sentence Repetition was doubly dissociated from Sentence Questions (n = 1). Each case was analyzed separately in terms of the average z-score index of impairment and in the context of previously published neuropsychological studies examining similarly constructed tasks. Group lesion analysis revealed that damage to the basal ganglia predicted impairment on Planned Codes (p < 0.001) right anterior cingulate damage predicted impairment on Expressive Attention (p < 0.001) left posterior perisylvian region damage predicted impairment on Verbal-Spatial Relations (p = 0.02) and left temporoparietal or right frontal lobe damage predicted impairment on Word Series (p = 0.01). ANOVA at the level of PASS scales with lesions divided into four quadrants of the brain (anterior-posterior vs. left-right) revealed contradiction of 5 of 7 statements defining the neural correlates of PASS theory. Right hemisphere patients were more impaired on the planning scale (p < 0.01) while posterior patients were more impaired on the attention (p = 0.03) and the simultaneous scales (p = 0.02). Right hemisphere patients were more impaired on the simultaneous scale (p = 0.02) and there was a location by hemisphere interaction such that right posterior patients performed exceedingly poor (p = 0.04). The data seriously challenged the assertion that code content is independent of code type as specified in the simultaneous-successive theory. In sum there was little evidence that the PASS model as currently conceived has the necessary specificity to be used in neuropsychological practice. However at the level of the CAS subtests reliable patterns of double dissociations and specificity for regional damage to distributed neural networks underlying performance were demonstrated.