Figure 2.

Behavioral results and MEG signals. (A) Results of the forced-choice recognition task. The discriminability of stimuli (d′) is shown against various intervals of the first verb. The SEMs and 95% confidence intervals (Bonferroni-corrected) are shown in the lighter and lightest shades, respectively (n = 10). The results showed that subliminal verbs of 34 ms were too short to be seen. (B) Interference from a subliminal verb for the SV sentences in the syntactic decision task. The histograms show the differences in RTs obtained by subtracting RTs for the OV sentences from those for the SV sentences (mean ± s.e.m., n = 15), averaged under both Cong and Incong conditions. A significantly increased difference in RTs was observed for a subliminal verb, but not for a subliminal NP. An asterisk denotes a significant difference (P = 0.05, paired t-test). (C) MEG signals for the NPs from artifact-free and correct trials, averaged for all of the four conditions and across 15 participants, are shown for each sensor before normalization. (D) MEG signals for the target verbs are shown for each sensor. The black bars (a: 110–140; b: 140–160; c: 280–300 ms) denote all of the temporal bins that showed any significant differences between the tested conditions (Figures 3A–C). (E) MEG topographies on the scalp averaged under the OV-Cong condition, at each temporal bin of a-c (D). The upper and lateral scalp surfaces are shown.