The effect of sensory modality (visual or auditory) on semantic relatedness was tested using a 2��2 within subjects factorial design. Factor 1 was semantic relatedness (SR versus SU). Factor 2 was sensory input. The statistical analysis tested for the main effect of relatedness (SR>SU), the main effect of sensory input (auditory versus visual) and the interaction between these factors. A within-subjects one-way ANOVA was performed on the three naming conditions: Sentence production (Se), verb production (Vb) and object naming (ON). Three statistical contrasts compared (1) Se>ON, (2) Se>Vb and (3) ON>Vb. Activation related to sentence processing was expected <a href="http://www.selleckchem.com/products/cx-5461.html">CX
5461</a> to show an effect of Se>ON and Se>Vb. Activation related to lexical retrieval was expected to show an effect of Se>Vb and ON>Vb. The behavioural data were analysed with SPSS (IBM SPSS, NY, US), with three sets of analyses to compare accuracy and RTs from the conditions involved in the three different <a href="https://en.wikipedia.org/wiki/Dipivefrine">Dipivefrine</a>
fMRI analyses. One-way ANOVAs were used to compare the three conditions in Analysis 1 (i.e. ON, SU and SR) and Analysis 3 (i.e. Se, Vb and ON). A 2��2 repeated-measures ANOVA with relatedness and sensory input as factors was used to compare the conditions included in Analysis 2 (i.e. visual and auditory SU and SR conditions). For summary of the accuracy and RTs for each condition, see Table 3. There were no significant differences in either accuracy or RTs for the three different conditions. This was predicted on the basis of our pilot study that attempted to select items with the most consistently accurate responses. Indeed, accuracy was above 89% for all 20 semantic matching trials and above 80% for each of the 35 participants in all the conditions. There were <a href="http://www.selleckchem.com/products/forskolin.html">Forskolin
nmr</a> no significant differences in accuracy for the four different conditions. Accuracy was above 89% for all trials with the exception of one unrelated trial in the auditory condition (��clown and gloves��) that was classed as semantically unrelated by 74% of participants, and semantically related by the remaining 26% of participants. As we mentioned before, only activation related to the correct trials was included in any of the imaging analyses. Response times were slower for unrelated compared to related conditions (F [1, 34]=4.1; p=0.05). This is expected given that participants may spend more time searching for semantic associations when there is no obvious semantic link between two objects (in unrelated trials) than when there is a semantic link (in related trials). Responses were also slower for the auditory compared to visual conditions (F [1, 34]=1971.8; p<0.05). This was expected because the auditory stimuli were presented sequentially (one object name after another) whereas the visual stimuli presented two pictures of objects simultaneously. Matching semantic associations was therefore delayed relative to stimulus onsets in the auditory modality.