Release from informational masking by auditory stream segregation: perception and its neural correlate

Abstract

In the analysis of acoustic scenes, we easily miss sounds or are insensitive to sound features that are salient if presented in isolation. This insensitivity that is not due to interference in the inner ear is termed informational masking (IM). So far, the cellular mechanisms underlying IM remained elusive. Here, we apply a sequential IM paradigm to humans and gerbils using a sound level increment detection task determining the sensitivity to target tones in a background of standard (same frequency) and distracting tones (varying in level and frequency). The amount of IM that was indicated by the level increment thresholds depended on the frequency separation between the distracting and the standard and target tones. In humans and gerbils, we observed similar perceptual thresholds. A release from IM of more than 20 dB was observed in both species if the distracting tones were well segregated in frequency from the other tones. Neuronal rate responses elicited by similar sequences in gerbil inferior colliculus and auditory cortex were recorded. At both levels of the auditory pathway, the neuronal thresholds obtained with a signal-detection-theoretic approach deducing the sensitivity from the analysis of the neurons' receiver operating characteristics matched the psychophysical thresholds revealing that IM already emerges at midbrain level. By applying objective response measures in physiology and psychophysics, we demonstrated that the population of neurons has a sufficient sensitivity for explaining the perceptual level increment thresholds indicating IM. There was a good correspondence between the neuronal and perceptual release from IM being related to auditory stream segregation.

Interaction of spatial source separation, fundamental frequency, and vowel pairing in a sequential informational masking paradigm in Mongolian gerbils.

“…Here, we applied a sequential IM paradigm to investigate the release from IM induced by stream segregation due to differences in vowel type, fundamental frequency, both cues interacting, and spatial separation of sequentially presented vowels. The paradigm was already successfully used in previous human and animal studies to investigate the release from IM by frequency separation and binaural cues (Dolležal et al, 2017; Tolnai et al, 2015; Winkler et al, 2003). In these studies, using tones, standards/deviants and distracting sounds either shared a similar frequency range and were perceived as a single stream, or they were presented in different frequency ranges and thus perceived in different auditory streams (Dolležal et al, 2017; Winkler et al, 2003).…”

“…Gerbils were trained in a sequential IM paradigm (Dolležal et al, 2017) and performed an intensity discrimination task. We presented synthetic vowels that were based on the vowel stimuli presented in previous experiments (Peterson & Barney, 1952; Sinnott & Kreiter, 1991; Sinnott & Mosteller, 2001).…”

“…In complex acoustic environments, the challenge is for listeners to focus on a single sound source and its specific sound features to perceptually analyze an auditory scene (Bregman, 1990). Signal changes that are salient if perceived in isolation might be missed in a complex mixture of sounds (e.g., Arbogast, Mason, & Kidd, 2002; Cherry, 1953; Ding & Simon, 2012; Dolležal, Tolnai, Beutelmann, & Klump, 2017; Elhilali, Xiang, Shamma, & Simon, 2009; Gutschalk, Micheyl, & Oxenham, 2008; Rimmele, Sussman, & Poeppel, 2015; Tolnai, Dolležal, & Klump, 2015; Wiegand & Gutschalk, 2012). Informational masking (IM; Pollack, 1975) is defined as the compromised ability to perceive and analyze signals from a single source in a clutter of other sounds even if there is no interference of these signals’ excitation patterns in the inner ear (e.g., Durlach et al, 2003; Kidd, Mason, Richards, Gallun, & Durlach, 2008; Lutfi, Chang, Stamas, & Gilbertson, 2012; Watson, 2005).…”

“…Informational masking (IM; Pollack, 1975) is defined as the compromised ability to perceive and analyze signals from a single source in a clutter of other sounds even if there is no interference of these signals’ excitation patterns in the inner ear (e.g., Durlach et al, 2003; Kidd, Mason, Richards, Gallun, & Durlach, 2008; Lutfi, Chang, Stamas, & Gilbertson, 2012; Watson, 2005). Presenting stimuli as a sequence of temporally separated sounds makes it possible to investigate IM, avoiding simultaneous masking effects (Dolležal et al, 2017; Eipert, Selle, & Klump, 2019; Tolnai et al, 2015; Watson, 2005; Winkler et al, 2003). IM is affected by the similarity between target and masker (similarity-based IM) and the variation of stimulus features from trial to trial (uncertainty-based IM), both modulating discrimination thresholds (e.g., Lutfi, 1993; Watson, 2005).…”

“…Mongolian gerbils are a proven model for investigating the perception of vowels (Sinnott & Mosteller, 2001). The sequential IM paradigm presented here was adapted from an infant-EEG study (Winkler et al, 2003) and offers to determine the effect of perceptual stream segregation in humans and animal models on IM (Dolležal et al, 2017; Tolnai et al, 2015). Animals were tested to detect level increments between constant-level standard (reference) and deviant (target) vowels in a background of level-varying distracting (masker) vowels.…”