Abstract
The perception of near-threshold tones varies strongly across trials, likely because of fluctuations in sustained attention or arousal. We used parallel fMRI and pupillometry to study the role of attention networks for the detection of near-threshold tones in three phases: (1) passive listening, (2) active detection of salient tones, and (3) active detection of near-threshold tones. Results confirmed previous findings from magnetoencephalography that auditory cortex activity and pupil-dilation responses for near-threshold tones were only observed when task-relevant, stronger for hit trials, but also present for miss trials. We then sought which attention-related areas show a similar response pattern, and found it in insular cortex, anterior midcingulate cortex, and inferior precentral sulcus. Moreover, activity in the insula was already stronger for hit than miss trials in the prestimulus interval. Activity for hit trials was also observed in a number of subcortical nuclei, including thalamus, periaqueductal gray, locus coeruleus, and the colliculi. Like insula, activity in the intralaminar nuclei of the thalamus additionally showed activity for miss trials and stronger activity for hit trials in the baseline. Finally, BOLD activity correlated to spontaneous pupil fluctuations was evaluated and revealed biphasic activation and deactivation in a widespread cortical network, with a maximum 3 sec and minimum 7 sec after pupil dilation. The cortical networks included insula, anterior midcingulate cortex, retro-splenial, and sensory cortex. Overall, these data identify the cingulo-insular network and the intralaminar thalamic nuclei as potential sources of fluctuations in auditory cortex activity in the context of near-threshold tone detection.