Cognitive substrates in semantic memory of formal thought disorder in schizophrenia

Abstract

Formal thought disorder (FTD) has been associated with abnormalities in the semantic memory system. However, it is still unclear whether these abnormalities are related to the organization of the semantic system, or to the automatic spread of activation-inhibition in semantic networks. In this paper these alternative proposals are examined. Schizophrenic patients and healthy matched controls were given two semantic memory tasks. In the first task, participants were required to judge the similarity between pairs of natural concepts. These ratings are assumed to reflect the underlying knowledge organization. In the second task, participants were required to name pictures that were preceded by related or unrelated word primes. Interference is typically observed when pictures are preceded by semantically related primes, and it is explained as due to inhibitory processes from the word prime to the related picture target. The results showed that the semantic structures derived from the similarity ratings were similar for patients with and without FTD and for control participants. However, results from the picture-naming task indicated that both non-FTD and control participants showed the normal interference/inhibition effects from the related prime words, whereas the patients with FTD showed similar performance for pictures preceded by related words than for pictures preceded by unrelated words. These findings support the hypothesis that abnormalities in inhibitory processes in semantic memory underlie FTD.

This research was supported by Grant BSO2002–0159 from the Ministerio de Educación y Ciencia of the Spanish Government to the last author.

Notes

¹According to this account, inhibitory and excitatory connections within and between levels of representation are responsible for lexical selection. (Cutting & Ferreira, 1999; Schade & Berg, 1992). Between-level links (e.g., from conceptual to lexical representations) are supposed to be excitatory, whereas links within the lexical level of representation are supposed to be inhibitory. In the primed picture-naming task, the lexical node for the word prime (e.g., “EYE”) would receive (a) direct activation from the word prime and (b) activation descending from the related picture target (e.g., “EAR”). Thus, the prime's lexical representation would receive extra activation, and this extra-activated lexical representation would inhibit the related lexical node (the picture target lexical node). Hence, interference would be the result of inhibitory processes acting on the target's lexical representation.

²Since this null result might be due to the large variability in response times within the groups, and also because RTs were not normally distributed, we performed an additional analysis. First we computed an interference score: (no related RT – related RT)/no related RT. Second, we submitted the interference scores to one-way ANOVA with groups (FTD patients, non-FTD patients, and controls) as independent variable. The results of this analysis still showed no effect of group, F(2, 46) = 1.08, MSE = 0.02, p >.05, indicating that response times were not sensitive to possible group differences.

³To make sure that the absence of significant correlations was not due to the large between-subjects variability or to the interference scores not being normally distributed we performed an additional analysis on the log transformation of the interference scores. These correlations were still nonsignificant (all ps > .05). The correlation between the interference scores and the PANSS total was r = −.14. Similarly, the correlation between the interference scores and global clinical scale was score r = −.11. Length of illness, r = −.03, number of hospitalizations, r = .03, cognitive impairment, r = .19, were nonsignificant (all ps > .05).