Background: Collins and Quillian ( Acta Psychol 1970;33:304-314) proposed that semantic representations in the human brain could have a "networklike" theoretical construct. Thought disorders in schizophrenia have been described as disturbances in the spread of activation within semantic networks. Semantic networks are typically evaluated indirectly via reaction times of priming tasks. Medications may interfere with the reaction time of patients, thus, we sought to investigate semantic networks, independent of time, by having patients and controls rate textual associations in sentences organized to various degrees.
Methods: Twenty-eight schizophrenic patients (17 non-thought-disordered and 11 thought-disordered) and 27 healthy controls performed a rating of textual associations task in which they were asked to rate the associative relationship between concepts in sentences on a scale from 1 (totally dissociated) to 10 (completely associated). The task contained 3 sets of sentences; organized meaningful sentences, vague sentences (intermediately disorganized), and completely disorganized sentences. To avoid order effects, sentences were randomly mixed at presentation.
Results: Analysis of variance calculations indicated significant differences among the 3 groups (controls, thought-disordered, and non-thought-disordered). The differences were greater for the vague sentences. Compared with controls, schizophrenic patients demonstrated increased SDs in rating associative values between concepts in the sentences, which is higher in disorganized sentences. Inadequate ability to identify and rate associations in disorganized sentences is discussed in the context of disordered semantic networks of schizophrenic patients.