Predicted Sensory Modality Determines the Timing and Topographies of Omitted Stimulus Potentials

Abstract

It is thought that our brains actively predict what will happen next in the environment, but it remains unclear how specific the prediction of an upcoming event is. This study investigated whether the prediction about the sensory modality of the upcoming stimulus modulates neural responses to unexpected omissions of stimuli. Previous research has reported that the peak latencies of omitted stimulus potentials (OSPs) are shorter in the auditory modality than in the visual modality when tested in separate blocks. In this study, we presented auditory and visual stimuli in a fixed alternating pattern to examine whether modality-specific OSPs occur even within a single block. Participants (N = 33) were asked to press a mouse button at a constant interval of 1 s. Each button press triggered either an auditory or visual stimulus, and these were presented twice in an alternating pattern (A, A, V, V, A, A, etc.). The stimuli were omitted in 12% of the trials. This method ensured each type of omission (of either auditory or visual stimuli) to be preceded equally often by either an auditory or a visual stimulus, thereby controlling for late event-related potential components of the preceding stimulus, if any. The results showed that auditory OSPs had shorter peak latencies than visual OSPs and that their scalp topographies differed; auditory OSPs had more anterior and central distributions than visual OSPs. These findings suggest that OSPs occur in a modality-specific manner according to the predicted sensory modality of the upcoming stimulus.

Keywords: event‐related potential; omission; prediction; predictive coding; self‐generation; sensory modality.

FIGURE 1

Schematic representation of the experimental design. Participants pressed a mouse button at intervals of approximately 1 s. Immediately after each button press, an auditory stimulus (1000 Hz pure tone) or a visual stimulus (LED light) was presented. The auditory and visual stimuli were presented twice each in an alternating sequence, with one stimulus presented per button press. Stimuli were omitted in 12% of the trials. Participants were instructed to respond with the opposite hand during catch trials, in which a tone or light was presented twice following a single button press. In the motor control condition, no stimuli followed button presses.

FIGURE 2

Grand mean ERP waveforms and difference waveforms (motor‐corrected) of omission trials. (A) Grand mean ERP waveforms of the auditory omission trials (red), visual omission trials (blue), and motor control condition (green), averaged across left temporal electrodes (FC3, FC5, FT7, C3, C5, T7, CP3, CP5, and TP7), frontocentral electrodes (F1, Fz, F2, FC1, FCz, FC2, C1, Cz, and C2), and right temporal electrodes (FC4, FC6, FT8, C4, C6, T8, CP4, CP6, and TP8). (B) Motor‐subtracted difference waveforms of the auditory and visual omission trials, each plotted with 95% confidence intervals. Scalp topographies, derived from motor‐subtracted difference waveforms of the auditory and visual omission trials in oN1, oN2, and oP3 time windows, are overlaid on the waveforms.

FIGURE 3

Comparisons of oN1 and oN2 peak latencies between the auditory and visual omission trials. In both modalities, the peak latencies of oN1 were identified within a 50–150 ms time window, and the peak latencies of oN2 were identified within a 150–250 ms time window. Above the raincloud plots, the mean peak latencies are shown with their 95% confidence intervals.

FIGURE 4

Comparisons of topographies between the auditory and visual omission trials. (A) The results of the TANOVA comparing scalp topographies for the auditory and visual omission trials. The line depicts the TANOVA's point‐by‐point p‐values, with the green line indicating p = 0.05. The shaded area highlights the time window of reliable differences between the modalities. (B) Scalp voltage and CSD maps plotted from motor‐subtracted difference waveforms for the auditory and visual omission trials in the time window of reliable differences (89–155 ms).