25 toolbox (www.vislab.ucl.ac.uk/cogent) Selleck PCI32765 on a notebook computer. Music stimuli were presented in free-field at a comfortable listening level for each subject (at least 70 dB). Subjects were

first familiarised with the paradigm using musical examples not subsequently presented in the actual test. Twenty test trials were administered in each condition; conditions were presented in fixed order (non-mentalising followed by mentalising). Combinations of words and pictures (high quality colour images) were simultaneously presented on the computer monitor. Trials were presented in a fixed randomised order, and the relative screen positions of targets and foils were randomised from trial to trial. Subject selections were recorded and stored for offline selleck inhibitor analysis. In addition, on each trial the subject was asked if they were familiar with the piece, and this information was also recorded. Each piece was presented once; a single repeat of a trial was allowed if the examiner considered that the subject had been distracted during the original presentation. No time limit was imposed and no feedback about performance was given during the test. Behavioural data were analysed using

STATA 12©. Experimental data were analysed using analysis of variance (ANOVA) regression models incorporating subject scores in the mentalising or non-mentalising condition as a within-subject variable, group (bvFTD or control) as a between-subjects variable; and subject age, gender, and scores on the colour-word inhibition Stroop task, the British Picture Vocabulary Scale (BPVS; Lloyd et al., 1982), and the National Adult Reading Test (NART) as covariates of no interest (to adjust for possible performance effects of demographic bias, general executive capacity, single-word comprehension, and premorbid IQ, respectively). Imageability and lexical frequency of the words presented in both conditions were calculated using the N-Watch psycholinguistic research database

(http://www.pc.rhul.ac.uk/staff/c.davis/Utilities/), in order to examine whether such characteristics could be contributing to the results. Population averaged models for repeated measures were used to examine the group by task interaction, with and without adjustment for word imageability and lexical frequency. In order to assess how well mentalising and non-mentalising Ergoloid conditions were able to discriminate bvFTD patients from healthy controls we constructed receiver operating characteristic (ROC) curves whereby the discriminatory ability of each task was quantified using the area under the curve (AUC). The AUC is the probability that in a randomly selected patient/control pair, the patient has a lower score than the control (Hanley and McNeil, 1982); perfect discrimination between patient and control groups would correspond to an AUC of 1, whilst the same distribution of scores in patients and controls would correspond to an AUC of .5.