β- And γ-band EEG power predicts illusory auditory continuity perception

Abstract

Because acoustic landscapes are complex and rapidly changing, auditory systems have evolved mechanisms that permit rapid detection of novel sounds, sound source segregation, and perceptual restoration of sounds obscured by noise. Perceptual restoration is particularly important in noisy environments because it allows organisms to track sounds over time even when they are masked. The continuity illusion is a striking example of perceptual restoration with sounds perceived as intact even when parts of them have been replaced by gaps and rendered inaudible by being masked by an extraneous sound. The mechanisms of auditory filling-in are complex and are currently not well-understood. The present study used the high temporal resolution of EEG to examine brain activity related to continuity illusion perception. Masking noise loudness was adjusted individually for each subject so that physically identical sounds on some trials elicited a continuity illusion (failure to detect a gap in a sound) and on other trials resulted in correct gap detection. This design ensured that any measurable differences in brain activity would be due to perceptual differences rather than physical differences among stimuli. We found that baseline activity recorded immediately before presentation of the stimulus significantly predicted the occurrence of the continuity illusion in 10 out of 14 participants based on power differences in γ-band EEG (34-80 Hz). Across all participants, power in the β and γ (12- to 80-Hz range) was informative about the subsequent perceptual decision. These data suggest that a subject's baseline brain state influences the strength of continuity illusions.

Female túngara frogs do not experience the continuity illusion.

“…Recent EEG and fMRI studies of humans confirm a role for auditory cortex in the perceptual restoration of obscured sounds (Heinrich, Carlyon, Davis, & Johnsrude, 2008, 2011; Micheyl et al, 2003; Riecke, van Opstal, Goebel, & Formisano; Riecke et al, 2009, Riecke et al, 2012; Vinnik, Itskov, & Balaban, 2012). Single-unit recordings from the mammalian auditory cortex in cats (Sugita, 1997; but see Micheyl et al, 2003) and rhesus macaques (Petkov, O'Connor, & Sutter, 2007) have identified neural correlates of auditory induction, consistent with the purported contribution of cortical processing (see also Kubota, Miyamoto, Hosokawa, Sugimoto, & Horikawa, 2012).…”

Dynamic cortical representations of perceptual filling-in for missing acoustic rhythm

“…Samuel's paradigm 22 ). Our method subscribes to the latter approach, also referred to as 'filling-in', which emphasizes the signal detection strategy followed in cases where a listener classification is inconsistent with the token absence in a gap 3,[23][24][25][26][27][28][29] . As has been noted 30 , from the listener's utilitarian perspective, this effect of induction in a challenging environment is not aimed at the production of decision errors (or illusions) but to assist against masking.…”

“…Depending on stimulus, neural correlates have been localized to different areas, including Heschl's gyrus for missing AM noise 36 , the posterior aspect of superior temporal gyrus for disrupted vowels 28 , and wider brain networks including the superior temporal lobe in the case of missed phonemes 24,37 . In addition, mixed evidence points to a basis for restoration in terms of endogenous modulations to boundary encoding: on the one hand, the search for differential onset responses to noise when under restoration, indexing alternative encoding, has yielded negative results so far 27,28 ; on the other, induced narrow-band (3-4 Hz) desynchronizations that are restoration-specific, and occur after gap onset, have been suggested by results from EEG 28,38 .…”

“…18, 2017; In this study the differential temporal boundary encoding under restoration was not specifically addressed 38 8,9 . An cortical EEG study by Vinnik and colleagues 27 showed no change to neural spectral power sustained along noise gaps embedded in a 40 Hz AM context stimulus during restoration; on the other hand, it has been shown that changes to neural spectral power in brainstem responses may occur during restored pitch of a missing 800 Hz carrier tone 29 . It is possible that while gamma-rate acoustic modulations can be represented cortically with a temporal code 13,40 , they are also at rates that involve pitch quality -a representation of which implies substantially distinct cortical coding modes 41 assisting restoration.…”

“…Correspondingly, the data here show that with perceptual restoration of masked rhythm, endogenous representational differences may emerge as early as 0.6 s post masking, at the target rhythm. There is also activity at the first harmonic, 10 Hz, but there one cannot entirely rule out yet alternative explanations involving enhanced alpha activity 27 , since with increased alertness at some trials over others, a systematic differential in spontaneous alpha activity might be responsible 51 . For filling-in and rhythm-missed trials related to inattention, reduced vigilance might be expected to effectively increase alpha activity.…”