1 of 1

Investigating the perceptual consequences of low-frequency hearing loss

Elaea E. Purmalietis, Andrew J. Oxenham

Department of Psychology, University of Minnesota, Minneapolis MN 55455

Aim 1 methods & preliminary results

Background

Low-frequency sensorineural hearing loss (LFHL) characterized by elevated thresholds below ~1-2 kHz with intact high-frequency sensitivity (Parving & Bak-Pedersen, 1978).

LFHL is associated with heterogeneous etiologies, Including genetic, congenital, and acquired cases (Konigsmark et al., 1971; Parving et al., 2000; Oishi et al., 2010).

Converging evidence suggests that listeners with LFHL may detect low-frequency tones via upward spread of excitation to healthier basal regions.

Off-place listening rather than true apical encoding (Humes et al., 1984; Mair & Laukali, 1986; Moore, 2001).
Standard audiometry may overestimate apical integrity.
Speech perception deficits may be disproportionate to the degree of audiometric loss (Thornton & Abbas, 1980; Goldstein et al., 1983).

These mechanisms complicate clinical management :

Low-frequency amplification may exacerbate upward spread of masking, obscuring critical mid- and high-frequency speech cues, and degrading speech comprehension rather than improving it (Schum & Collins, 1992).

Improved perceptual characterization of LFHL may provide critical support for evidence-based counseling about tradeoffs between restoring audibility and preserving intelligibility for listeners with LFHL.

Beyond clinical relevance, LFHL may also provide a natural model of how selective apical dysfunction alters auditory coding in humans.

Experiment 1.1: Psychophysical Tuning Curves (PTCs)

3AFC task, listener asked to identify which interval contained a 200ms probe tone in the presence of a simultaneous masker (continuous throughout the trial)
Masker level varied adaptively (2-down, 1-up tracking procedure; Levitt, 1971)
Target level fixed at 10 dB SL

Selected frequencies. A, B, and C (Fig.2)

Masker presented at interval steps [0.5:1.5]*probe frequency

Aim 1 → targeted psychoacoustic battery to characterize perceptual consequences of LFHL beyond standard clinical diagnostics.

The present work focuses on preliminary results from Aim 1 tasks

Aim 2 → Findings from Aim 1 will inform Aim 2, which will focus on developing abbreviated, clinically feasible measures to improve diagnostic precision and treatment decisions.

0.2: DPOAEs

Objective / noninvasive assessment of outer hair cell (OHC) function (Lonsbury-Martin and Martin, 1990)

Test frequency selection informed by audiometric profile:

Fig. 2. Schematic representation of three frequencies selected based on audiometric thresholds (dB HL).

Used as test frequencies in Aim 1 experiments 1.1-1.4

A: Frequency at edge of low-frequency region of loss

B: Frequency near the center of the rising slope

C: Selected from region of intact high-frequency hearing, typically listener’s “best” (lowest) threshold

Initial measures to ensure that participants meet eligibility criteria, identify suspected cochlear dead regions, and determine relative contributions of cochlear (OHC) and neural (IHC) loss

0.3: Forced choice detection thresholds

Pure-tone detection thresholds: 3AFC task, 2-down, 1-up adaptive tracking applied to the level of 200ms probe

In quiet → thresholds for selected frequencies (Fig.1) used to determine presentation levels of each in Aim 1 tasks 1.1-1.4.
In high-pass noise → compared to quiet thresholds at each frequency; higher thresholds in the presence of HP noise may indicate abnormal tuning.

70 dB / ERB; ½ octave above test frequency – 20,000 Hz

0.4: EHF thresholds

Highest audible frequency and behavioral detection thresholds

0.1: Audiometric testing

Air and bone thresholds at standard frequencies (125-8,000 Hz)

Eligibility criteria:

2 or more thresholds ≥ 30 dB HL below 2 kHz.
1 or more threshold ≤ 20 dB HL (or normal equivalent for age group).
In at least 1 ear.

Baseline measures

HI 01

Will threshold shifts ≥ 10 dB in HP TEN vs. quiet predict abnormally tuned PTCs?

Selected examples from pilot data

Threshold shift at frequency B corresponds to abnormal basal PTC slope.

Large threshold shifts (~ 30 dB) → PTC tips for selected frequencies A and B shifted towards 2200 Hz.
Small threshold shift at 500 Hz (11 dB) → Broad PTC, shallow basal slope

HI 04

Only moderate threshold shift (~12 dB) at Freq. A (500 Hz) → Broad PTC, shallow basal slope

HI 02

No threshold shifts observed, normal PTCs

Experiment 0.3: Pure-tone detection thresholds

Quiet vs. HP noise

Experiment 1.4: Complex pitch

F0 DLs with harmonic complex tones

3AFC, 2down-1up adaptive tracking

Estimate the smallest change in F0

Overall level = 72 dB SPL
Resolved harmonics should produce better pitch discrimination if than conditions with unresolved harmonics

Conditions:

200 Hz resolved

Filter range [200 1200]
TEN lc = ½ octave above upper filter cutoff
TEN hc = 20,000 Hz

200 Hz unresolved

Filter range [2400 6000]
Two bands of TEN:

Band 1 lc = 20; hc = 2000 Hz
Band 2 lc = 8000; hc = 20000

600 Hz resolved

Filter range [600 3600]
TEN lc = ½ octave above upper filter cutoff;
TEN hc = 20,000 Hz

Experiment 1.2: AM and FM

Envelope / TFS processing fidelity

Pure-tone carriers at 3 selected frequencies (fig.1)

20 dB SL

4 Hz modulation rate

Envelope rate in speech (Varnet et al.,

3AFC task – 1 interval modulated
Modulation varied adaptively

AM: Temporal envelope processing – can envelope encoding be achieved via off-place listening?

FM: At low modulation rates, TFS may be more dependent on place cues unavailable to LFHL listeners.

LFHL listeners do not appear to derive benefit from resolved harmonics

Fig. 6. AM and FM detection thresholds for selected PT carriers.

Blue shaded region represents estimated range of NH thresholds from NH control data and Whiteford & Oxenham (2017).

Experiment 1.3: Pure-tone pitch discrimination

Pitch DLs with pure tones

3AFC, 2down-1up adaptive tracking
Selected frequencies (Fig. 2) presented at 20 dB SL in quiet
“Which interval has a different pitch?”

Thresholds expected to be closer to normal in higher frequency conditions.

Fig. 7. Frequency DL (%) for pure tones. Thresholds shown for individual LFHL participants. Thresholds for frequency A are generally higher (worse). Blue region represents range of NH scores in pilot data collection.

QR CODE

NEXT STEPS IN AIM 1

Continue both LFHL (HI) and NH data collection; increase sample size

Experiment 1.5: Speech perception

Speech perception tasks will be run under audio-only and audiovisual contitions in quiet and speech shaped noise

STeVI nonsense sentence corpus

Do LFHL listeners rely disproportionately on high-frequency speech cues?

Fig. 8. F0DL (%) shown for individual LFHL participants and NH average. It appears that LFHL listeners did not elicit additional information from resolved harmonics available in the stimuli.

Participants

7 LFHL participants have been enrolled to date (Fig. 1)

Ages 24-83
Recruitment goal = 20

Normal-Hearing (NH) controls age-matched to LFHL participants

Fig. 1. LFHL test ear audiometric thresholds (dB HL). The shaded blue region represents range of clinically normal thresholds

Fig. 3. Pure tone thresholds in quiet and high-pass TEN and PTCs at selected frequencies for participant HI01

Fig. 4. Pure tone thresholds in quiet and high-pass TEN and PTCs at selected frequencies for participant HI04

Fig. 5. Pure tone thresholds in quiet and high-pass TEN and PTCs at selected frequencies for participant HI02