Acuity of Lateralizing Transients

Acuity of Lateralizing Transients
- Acuity = clarity/precision with which the auditory system discriminates small differences.
- Lateralization = perceived location of a sound inside the head when listening over headphones (contrasts with localization, which is external in free-field).
- Transient = any stimulus having a rapid onset and/or offset (e.g., splash, poof, click).
- Goal: Determine how precisely the auditory system uses tiny interaural time differences (ITDs) and interaural intensity differences (IIDs) to place such brief sounds to the left or right.

All real-world sounds possess onsets and offsets; many also change intensity and spectrum over time.
Transient segments (especially onsets) often dominate localization cues.
Head-phone studies isolate binaural cues without room reflections ⇒ termed lateralization for analytical control.

Stevens & Newman ➜ noise-like sounds localized more accurately than pure tones because noise contains many rapid fluctuations (transients) that provide extra timing cues.
Adding artificial onsets/offsets to tones improves lateralization performance.

Band-limited noise 150\text{–}1700\ \text{Hz} (rich, continuous transients)
1000\ \text{Hz} sinusoid with gradual rise/fall (ongoing phase cue, minimal transients)
1\ \text{ms} electrical click (single transient)

Broadband noise (random, headphone limited): 10\,\mu s
Noise 150\text{–}1700\ \text{Hz}: 9\,\mu s
Noise 2400\text{–}3400\ \text{Hz}: 44\,\mu s
Single click: 28\,\mu s
Click train (30 clicks/2 s): 11\,\mu s
Pure tones (selected):
- 90\ \text{Hz} → 75\,\mu s
- 500\ \text{Hz} → 17\,\mu s
- 1000\ \text{Hz} → 11\,\mu s
- 3200\ \text{Hz} → 10\,\mu s (note: high-frequency tones encode lateralization via envelope rather than phase)

Repetitive or broadband stimuli supply many temporal markers ➜ lower ITD thresholds.
A single click offers just one marker, hence coarser discrimination.

Short-duration tones (\
Dye & Hafter (1984): with click rates \le 200\ \text{clicks\,s}^{-1}, ITD thresholds remain as small as 10\,\mu s ⇒ binaural system maintains microsecond precision even at sparse rates.

Defined three ITD categories:
1. Start-time disparity
2. Ongoing disparity (phase alignment during sustained portion)
3. End-time disparity
Combined 1 & 3 ➜ Transient disparity; 2 ➜ ongoing.
Transient disparity loses efficacy once duration exceeds \approx 150\ \text{ms}.
ITD discrimination for noise bursts improves with duration up to \approx 700\ \text{ms}, asymptoting near 6\,\mu s.

Increasing IID pushes perceived image toward ear with greater intensity.
Complete lateralization typically occurs at 10\ \text{dB} IID (Békésy; Babkoff) though reports vary (Flanagan et al.: less; Guttman: more).
For pure ITD stimuli, low-frequency content (\<1200\text{–}1500\ \text{Hz}) drives lateralization.

ITD \approx 20\text{–}40\,\mu s → fused image slightly off midline toward leading ear.
ITD 500\,\mu s \text{–} 1\ \text{ms} → single image perceived at lead ear.
ITD 2\text{–}4\ \text{ms} → percept splits: strong lead-ear transient + faint lag-ear echo (onset of the precedence effect).

Click trains filtered (high-pass vs low-pass) and/or masked with complementary noise bands.
Task: discriminate centered (simultaneous) vs delayed-left click.

Removing low frequencies (>1500\ \text{Hz} high-pass) degraded performance.
Removing high frequencies (low-pass) caused little loss.
Adding low-pass noise (masking LF energy) impaired discrimination; high-pass noise had minimal effect.

Low-frequency content dominates ITD-based lateralization.
Some discrimination possible from high-frequency envelope cues but less precise.

Low-frequency pulses vibrate a larger portion of the basilar membrane (apex ➜ base) → engage more auditory nerve fibres → richer temporal code.
Transient LF waveforms have periods longer than neuronal refractory period, allowing successive cycles to trigger distinct neural spikes.
High-frequency transients localize vibration to basal end; successive peaks fall within neural refractory window → fewer effective spikes → poorer ITD coding.

Discrimination of arrival-time differences relies more on low-frequency (apical) information than on high-frequency (basal) information.
Confirmed through both spectral filtering and spectral masking paradigms.

Basal cochlear turn (high-frequency encoding) initially thought to dominate transient lateralization; however, converging evidence shows apical (low-frequency) contributions provide finer ITD acuity.
Temporal cue effectiveness is frequency-dependent; intensity cues interplay but cannot fully compensate when low-frequency ITD information is absent.

Design of binaural hearing aids & cochlear implants must preserve microsecond ITD cues, especially for low-frequency channels.
Virtual reality & spatial audio engines should embed accurate onset ITDs and IIDs ≲10\ \text{dB} for convincing internalized imagery.
Diagnostics: Elevated ITD thresholds for broadband transients may indicate pathologies affecting brainstem timing (e.g., lesions in medial superior olive).

Everyday transients: Splash, Roof! (door slam), Poof, Sna (sudden /s/ noise) – recall that sharp onsets carry strongest lateralization information.
Rule of thumb: "10\,\mu s / 10\ \text{dB}" → ~minimum reliable ITD & IID for full internal displacement.

Best ITD threshold for broadband noise: \le 10\,\mu s.
Transient disparity useful up to \approx 150\ \text{ms}; improvements continue until \approx 700\ \text{ms}.
IID for full lateralization: \sim 10\ \text{dB}.
Click-train optimum (\le200\ \text{clicks\,s}^{-1}): thresholds remain \sim10\,\mu s.

Klumpp & Eady (1956) – foundational ITD thresholds.
Tobias & Schubert (1959) – transient versus ongoing disparities.
Yost (1977); Yost, Wightman & Green (1971) – frequency dependence in click lateralization.
Babkoff series (1975–1982) – psychophysics & physiology of transient lateralization.
Dye & Hafter (1984) – intensity effects on high-frequency click ITDs.