On the basis of the significantly higher detection rate and cluster sizes documented here using SCN, and assuming that these results generalize to patient populations, we conclude that SCN is a better baseline for speech in clinical setups. This advantage may be enhanced when scanner noise increases. If we attribute the responses to reversed speech as unsuccessful attempts to parse it, we predict that the difference
between baselines will be even more pronounced as scanner noise increases, that is, using Inhibitors,research,lifescience,medical high-field MRI and lower audio/headphone quality. Under such conditions, it could take longer to recognize that reversed speech is not speech, which will lengthen the overlap period between these responses. Importantly, providing a quiet epoch for stimulus Inhibitors,research,lifescience,medical presentation using sparse sampling or clustered acquisition
is expected to improve the quality of the auditory stimulation and may thus reduce the advantage of SCN over reversed speech. Yet, sparse sampling requires prolonged acquisition time, and is typically used with event-related designs. Inhibitors,research,lifescience,medical These choices have their own disadvantages in the context of a functional localizer, particularly reduced power at the individual subject level and less efficient use of scan time (Dale 1999). Finally, SCN is preferred over a rest baseline if one aims to calculate lateralization indices in temporal speech processing regions, Inhibitors,research,lifescience,medical which are difficult to disentangle adequately from bilateral primary auditory responses without an active auditory baseline. In basic research designs, functional localizers provide a tool for isolating language regions in individual participants, followed by an in depth analysis of the responses for well matched conditions in independent experiments within these ROIs. We have argued in the introduction that such a localizer should satisfy several constraints: efficiency, sensitivity, specificity, and independence (see also Inhibitors,research,lifescience,medical Fedorenko et al. 2010). On the basis of our results,
we can now determine that reversed speech fails on sensitivity at the individual subject level. Low sensitivity at the individual level can be overcome in group analysis. Indeed, some fMRI studies report significant group activation maps for Speech versus out Reversed (Crinion and Price 2005; Balsamo et al. 2006; Leff et al. 2008), though other group analyses have failed to do so (Binder et al. 2000; Ahmad et al. 2003). In a group analysis of the data reported here we still failed to detect activation for speech compared with reversed speech in the IFG (see Fig. S4). We consider two alternative explanations for the inconsistency in group analyses of Speech versus Reversed: in terms of statistical power or in terms of the task manipulation. In our study, which targets individual PR-171 cell line localization of speech-related cortex, the small sample size (N = 12) may well have contributed to the null result achieved at the group level.