Optimization of the Operant Silent Gap-in-Noise Detection Paradigm in Humans

Abstract

Background: In the auditory domain, temporal resolution is the ability to respond to rapid changes in the envelope of a sound over time. Silent gap-in-noise detection tests assess temporal resolution. Whether temporal resolution is impaired in tinnitus and whether those tests are useful for identifying the condition is still debated. We have revisited these questions by assessing the silent gap-in-noise detection performance of human participants.

Methods: Participants were seventy-one young adults with normal hearing, separated into preliminary, tinnitus and matched-control groups. A preliminary group (n = 18) was used to optimise the silent gap-in-noise detection two-alternative forced-choice paradigm by examining the effect of the position and the salience of the gap. Temporal resolution was tested in case-control observational study of tinnitus (n = 20) and matched-control (n = 33) groups using the previously optimized silent gap-in-noise behavioral paradigm. These two groups were also tested using silent gap prepulse inhibition of the auditory startle reflex (GPIAS) and Auditory Brain Responses (ABRs).

Results: In the preliminary group, reducing the predictability and saliency of the silent gap increased detection thresholds and reduced gap detection sensitivity (slope of the psychometric function). In the case-control study, tinnitus participants had higher gap detection thresholds than controls for narrowband noise stimuli centred at 2 and 8 kHz, with no differences in GPIAS or ABRs. In addition, ABR data showed latency differences across the different tinnitus subgroups stratified by subject severity.

Conclusions: Operant silent gap-in-noise detection is impaired in tinnitus when the paradigm is optimized to reduce the predictability and saliency of the silent gap and to avoid the ceiling effect. Our behavioral paradigm can distinguish tinnitus and control groups suggesting that temporal resolution is impaired in tinnitus. However, in young adults with normal hearing, the paradigm is unable to objectively identify tinnitus at the individual level. The GPIAS paradigm was unable to differentiate the tinnitus and control groups, suggesting that operant, as opposed to reflexive, silent gap-in-noise detection is a more sensitive measure for objectively identifying tinnitus.

Keywords: auditory system; humans; no hearing loss; operant behavior; subjective tinnitus; young adults.

Fig. 1. Silent gap in noise detection as a measure of temporal resolution.

(A) Possible mechanisms of impairment in silent gap detection produced by tinnitus. This could be caused by the tinnitus phantom sound percept masking the silent gap, impaired temporal resolution causing the edges of the gap to become blurred, and/or tinnitus-associated hearing loss resulting in the silent gap being less salient compared to the continuous sound. (B) Diagram showing changes in gap position, duration, and depth modulation to decrease the predictability and the saliency of the salient gap. Stimuli used were 30 kHz low pass filtered broadband noise (BBN), and one octave wide narrowband noises (NBN) either centred at 1, 2, 4, 8 or 16 kHz (400 ms duration, 75 dB sound pressure level (SPL). See Materials and Methods).

Fig. 2. Experimental design.

(A) The observational study comprised three groups: a preliminary group, where predictability and saliency of the silent gap were modified, and a tinnitus and matched-control groups, where the silent gap-in-noise detection paradigm, GPIAS, and ABRs were compared. (B) The operant behavior task was a two-alternative forced-choice (2AFC) operant paradigm with two identical sound stimuli with or without a silent gap on it. Abreviations: GPIAS, gap prepulse inhibition of the auditory startle reflex; ABRs, auditory brainstem recordings.

Fig. 3. Examples of gap detection psychometric functions.

(A) Silent gap durations against the sensitivity index (d’) for the preliminary experiment. (B) Psychometric curve plotting left button press probability against gap duration for the case control observational study (control participant C22.10), equivalent to the hit rate for silent gap trials and the false alarm rate for no gap trials. Abbreviations: ipGDT, gap detection threshold corresponding to the inflection point; 65% GDT, gap detection threshold defined at 65% probability; mod 0, 2, 4 gap modulation depth 100, 80, 60% respectively.

Fig. 4. Variation of gap position, gap intensity modulation and sound stimulus type changes the gap detection threshold, slope and false alarm.

Changes in threshold (A), slope (B) and False Alarm rate (C) produced by varying the gap intensity modulation for 16 kHz narrowband noises (NBN) are depicted. Changes across stimulus types (different spectral composition) without modulation (mod = 0) on threshold (D), slope (E) and False Alarm rate (F). 1–16 centre frequency in kHz for 1-octave wide narrowband noises. Abbreviations: BBN, broadband noise.

Fig. 5. Gap detection thresholds in control and tinnitus.

(A) Changes in gap detection threshold (GDT) in the control group using the 65% hit rate (65% GDT, solid line) and the inflection point (ipGDT, dashed line) values for the different type of stimulus used. The average gap detection thresholds at 65% performance. (B) Changes in gap detection threshold in the tinnitus group using the 65% hit rate (solid line) and the inflection point (dashed line) values for the different type of stimulus used. (C,D) show the comparison between control and tinnitus groups for the different stimulus used with 65% GDT (C) and ipGDT (D). The silent gap detection inflection threshold was lower overall for control than tinnitus participants (2-way ANOVA, F_(1,24) = 12.3, p < 0.01). Post hoc analysis revealed this difference was significant at 2 and 8 kHz NBN (Bonferroni test, p < 0.05). Solid and dashed lines represent the averages of 65% GDT and ipGDT, respectively, with the shadows representing the standard error of the mean (sem). In black the controls and in red the tinnitus group. 65% GDT for broadband noise (BBN) was not represented for controls in (A) or for controls and tinnitus in (B) because only 3 datapoints in the controls present with lower than 65% hit rates whilst all other participants had higher hit rates.

Fig. 6. Correlation of Tinnitus questionnaires.

(A) Tinnitus functional index (TFI) and tinnitus handicap inventory (THI) scores were highly correlated (r = 0.937) and the combined tinnitus index (CTI) was calculated. (B) CTI scores and tinnitus duration were negatively correlated (r = –0.555, p = 0.007). (C) Tinnitus participants were stratified into groups of tinnitus severity according to their CTI scores.

Fig. 7. Comparing 65% GDT (A) and ipGDT (B) gap detection thresholds between controls and tinnitus cohorts stratified by CTI grades.

Those with slight tinnitus (CTI 1) had elevated GDTs relative to controls, for 4 kHz (7.89 vs. 4.58 ms, p = 0.007, 65% GDT; 7.22 vs. 3.8 ms, p = 0.002, ip GDT; Mann-Whitney U (MWU)) and 8 kHz (5.73 vs. 3.464 ms, p = 0.002, 65% GDT; 4.8 vs. 3.11 ms, p = 0.001, ipGDT; MWU) stimuli and to those with mild tinnitus (CTI2), for 4 kHz (p = 0.007, 65% GDT; MWU) and 8 kHz (p = 0.004, 65% GDT; p = 0.007, ipGDT; MWU) stimuli. Statistical significance is indicated by asterisks, * p < 0.05, ** p < 0.01, *** p < 0.001.

Fig. 8. Exemplars of the recorded blink reflex in response to loud BBN presented in noise and preceded (gap trials) or not by a silent gap (no gap trials).

(A) Control subject showing prepulse inhibition and (B) for a tinnitus participant who does not show prepulse inhibition. Lines represent the mean across trials and shadowed areas the standard error of the mean. (C) Comparison of bootstrapped means (log(ASR) showing no differences between control (n = 27) and tinnitus (n = 12) groups (p > 0.05, t-test)). ASR, auditory startle response; ns, no significance.

Fig. 9. Auditory Brainstem Responses (ABRs).

(A) Exemplary ABRs for low intensity hearing level clicks presented on the left ear, participant (2021C01). (B) Exemplary ABRs for high intensity hearing level clicks presented on the right ear, participant (2022C16). (C) Wave V amplitudes and latencies for the left and right ears. Post hoc analysis demonstrated that there were no significant differences between control (n = 16) and tinnitus (n = 12) groups (median ± interquartile range (IQR). Mann-Whitney U -test, p > 0.05). dB HL, decibels hearing level.

Fig. 10. Latencies of different ABR waves (I, III, and IV) according to stimulus type.

There were statistically significant differences between tinnitus subgroups classified according to CTI grade. CTI 1 had an elevated mean wave latency for waves I and III at 90 Hz and 80 Hz intensities, respectively (A). This difference in the CTI 1 group compared to other tinnitus CTI subgroups was also observed for 4 kHz tone bursts stimuli (B) (One-way ANOVAs and post-hoc Bonferroni multiple comparison analysis, with an adjusted p-value of p = 0.0083).