Horn et al ^{Reference Horn, Federspiel, Wirth, Müller, Wiest and Wang1} investigate a very important, somewhat underexplored area of neural correlates of schizophrenic speech disturbance. Given the probability of underlying deficits in contextual integration and theory of mind, formal thought disorder yields a fertile ground for structural and functional connectivity analysis in schizophrenia. Although the use of hitherto unused techniques such as resting perfusion scan to study formal thought disorder must be lauded, the results of this preliminary study must be treated with caution for various reasons.

The composite score of the Scale for the Assessment of Thought, Language and Communication (TLC) has been used as a measure of severity of formal thought disorder. The authors have administered the scale 45 min before the scanning procedure for each participant. It is widely perceived that uncontrolled generation of thought is required to reliably measure formal thought disorder in schizophrenia. The TLC itself lacks a standardised practical method of eliciting such thought flow in contrast to some recently developed instruments. ^{Reference Liddle, Ngan, Caissie, Anderson, Bates and Quested2} The cross-sectional use of the TLC to measure formal thought disorder severity must be treated with prudence.

The authors extract components from the Positive and Negative Syndrome Scale (PANSS) using factor analysis and demonstrate that none of these components correlate with formal thought disorder severity as measured by the TLC total score. The validity of factor analysis in such a small sample is questionable and not in synchrony with available factorial structures of PANSS. ^{Reference Fitzgerald, de Castella, Brewer, Filia, Collins and Davey3} As a result, all principal components extracted were from negative symptoms in PANSS (except the conceptual disorganisation item, which was rightly excluded from further analysis). Consequently, the results only show a lack of correlation between severity of formal thought disorder and negative symptoms as measured by PANSS. Findings from the magnetic resonance imaging may still be explained by positive symptoms alone and not by formal thought disorder. Lastly, the pervasive issue of sample size in neuroimaging studies becomes more prominent when correlation analyses are attempted in whole brain analyses.