Fig. 4: Spectrotemporal encoding models are better than semantic encoding models for short-onset latency sites in the frontal lobe and STG.
From: Parallel encoding of speech in human frontal and temporal lobes
A Short-onset latency sites in each area encode low-level spectrotemporal information. Sites with onset latencies less than 200 ms show significantly higher STRF model predictions than semantic model predictions consistent with dominant spectrotemporal over semantic coding at these short-onset latency sites. This is consistent with short-onset latency sites encoding spectrotemporal speech information in parallel (p < 0.001, Wilcoxon rank-sum test, two-sided). Longer onset latency sites in the same areas show no difference between spectrotemporal or semantic encoding models. All data are presented as mean values ± SEM. n site CC prediction values for early (<200 ms)/late (>200 ms) onset latencies = 46/31 IFG, 115/68 mPreCG, 110/43 vSMC, 389/72 STG. CC Pearson correlation coefficient; STRF spectrotemporal receptive field.
