Cue removal gives smaller residuals than recorded delays in 70 of 84 ds000030 control schedules with task and response terms.

In a model-based calculation on 126 OpenNeuro ds000030 control schedules, we compared a cue-removal model (C) with a recorded-delay model (L), using a cue model (H) as the directional target. After adding cue amplitude, a cue temporal derivative, shared task terms, and shared response terms, C had a smaller residual in 70 of 84 flagged schedules. Median residuals were 1.607017% for C and 1.738042% for L of the same H norm. In contrast, C had the larger raw cue-column change in all 84 schedules. Across ten specified settings, C had the smaller full-model residual in 69 to 72 schedules. These are calculations on recorded schedules, not population estimates or results from measured brain activity. The acoustic cue time is unknown, and visual and response times are model assumptions. The result depends on the included terms, unrestricted coefficients, assumed event times, and H as the directional target.

stop-signal task · functional MRI · cue timing · residual analysis · model comparison · temporal derivative · OpenNeuro · least-squares fit