3.4. MRCPs by RT
To understand the relationship between MRCPs and RT (whether a particular response elicited a prompt or a slow response), individual participant data trials for each individual were binned by RT quartile into the fastest 25% of trials, middle 250-50%, middle 50-75%, and slowest 25%. Group averages were constructed by averaging individual participant traces for each RT quartile (Figure 4 ). Omnibus GLMM analyses indicated a three-way interaction between participant group (ASD vs. TD), RT quartile, and electrode site (Fz vs. P3) for all three time windows of interest (Late BP: F(6,3818) = 2.284, p = 0.034; MP: F(6,3818) = 9.161, p < 0.001; RAP: F(6,3818) = 3.466, p = 0.002) (Table 3 ). Visualizations of these three-way interactions are given in Figure S2D-F . For each RT bin at each electrode site (Fz and P3), at each timewindow of interest (BP, MP, RAP), individuals with ASD show lower-amplitude cortical responses relative to TD. For the MP, an additional 3-way interaction between group, RT quartile, and age was identified, indicating that RT differences in MP amplitude were most prominent in the youngest ASD participants and are most clearly visible in the slowest 25% of trials.
To visualize MRCP modulation by RT, group average traces for ASD and TD participants were constructed by averaging individual participant traces for each RT quartile (Figure 4 ). Statistical cluster plots depicting within-group differences as a function of RT bin are shown separately for TD (4A ) and ASD (4B ) participants. Representative electrodes are shown from right frontal (TD: 4C , ASD: 4D ) and left parietal ( TD: 4E, ASD:4F ). In general, faster RTs were associated with larger amplitude responses for both participant groups.
For each RT quartile, SCPs were computed to contrast ASD and TD participants (4I ), and scalp topographies were rendered to visualize results (4G, 4H, 4J ). For the fastest 25% of trials, significant differences between ASD and TD participants were evident between -75 and -25 ms over fronto-central, parietal, and parieto-occipital scalp regions, as well as between 50 and 100 ms after the response over fronto-central, left temporal, and left parietal scalp regions (4I.1 ). For trials showing an intermediate RT between 25-50th percentile, significant differences emerged between -150 and -50 ms over fronto-central, left parietal, and right parietal scalp regions, and after the response between 0 and 50 ms primarily over the fronto-central scalp (4I.2 ). For trials with an intermediate RT between the 50th-75th percentile, between-group differences emerged primary between -200 and -100 ms prior to the button press over left parietal and bilateral occipital scalp sites (4I.3 ). Minimal differences were observed between groups for the slowest 25% of trials (4I.4 ). Topographic maps are presented for each RT quartile and group separately from -150 to -100 ms (4G ), -25 to -75 ms (4H ) and 0 to 50 ms (4J ) to visualize group effects.