Tipultech logo

Pitch Perception

Dawn Senathi-Raja

Beckwith (2003) has provided evidence concerning MR's pitch discrimination impairment. However, temporal and spectral aspects of complex pitches for fine-grained and higher-level temporal durations were not examined. Tasks assessing MR's ability to process these aspects of pitch were administered to provide a more complete picture.

Fine Temporal Processing of Pitch

Iterated rippled noise is a type of stimulus used to assess fine temporal processing of pitch presented at the level of tens of milliseconds. It is produced by taking noise with a flat spectrum and adding it to itself in a delay-and-add cycle (Patterson, Handel, Yost, & Datta, 1996). From this temporal change, a pitch emerges from the noise. MR's ability to process iterated rippled noise was tested using a well established, psychophysical task (Griffiths et al., 2001). For each of the 120 trials of this task and the remaining psychophysical tasks administered in this thesis, MR was presented with two sounds. One sound contained the target stimulus (in this task, a pitch) and the other sound excluded the target stimulus (in this task, white noise). From the two sounds, MR was required to choose which contained the target stimulus. A sigmoid (Weibull) function running from 50% correct (chance) up to 100% correct was fitted to the six levels of the task (20 trials for each level). This fit enabled a threshold to be determined, which was defined as the point at which a score of 75% correct was achieved. As shown in Table 2.9, MR's 75% threshold was significantly below the normal range (z = 3.68, p < 0.001), indicating impaired fine temporal processing of complex pitch (see Appendix L for MR's psychometric function).

Table 2.9

Seventy-five Percent Threshold for Griffiths et al.'s (2001) Iterated Rippled Noise Taskb

MR's threshold Controls threshold (SD)a
0.40*** -1.05 (0.18)

a Data based on 15 controls matched for age to MR from Griffith et al.'s (2001) norms.
b Data are expressed as log transformations of gain in octave steps. Gain varied between 0.01 and 0.32.

Temporal and Spectral Processing of Pitch

Timbral processing relies on the ability to perceive the spectral and temporal components of the complex pitch of individual sounds. Given MR's fine temporal processing deficit for pitch perception, Samson and Zatorre's (1994) timbre discrimination task was administered to confirm this deficit and to additionally test MR's spectral processing. The task was re-created using Protools (2002). Stimuli consisted of three complex tones with varied spectral information (i.e. the tones contained either the first three, four, or five harmonics) and three complex tones with varied temporal information (i.e. the tones contained onsets of 1, 50 or 190 milliseconds). Spectral-difference pairs varied in spectral information but had a constant temporal content of one millisecond, while Temporal-difference pairs varied in temporal information but had a constant spectral content of four harmonics. Identical pairs were constant in spectral and temporal content. The test was composed of 48 pairs, of which 24 were identical and 24 different, and the participants' task involved identifying them as such.

As shown in Table 2.10, MR's performance was below controls for the Spectral-difference condition, indicating a spectral impairment for complex pitches of short temporal duration. MR's poor performance for the brief stimuli of the Temporal-difference condition confirmed his impaired discrimination between tones of short temporal duration. MR's performance for identical pairs was inferior to the controls, except for pairs involving longer temporal durations. These results confirm MR's spectral and temporal processing deficit for brief stimuli, and indicates preserved spectral and temporal processing for tones of longer durations.

Table 2.10

Percentage of Correct Responses for Samson & Zatorre's (1994) Timbre Discrimination Task

Cue MR's score % Controls' score a Mean % (SD)

3 vs 4 harmonics 33.33%* 80.01% (29.89)

4 vs 5 harmonics 25.00%* 78.30% (26.15)

Total 29.16%* 80.88% (24.63)

1 vs 50 msec 58.33%** 93.33% (10.91)

50 vs 190 msec 41.66% 65.00% (14.99)

Total 50.00%** 79.21% (10.65)

3 harmonics, 1 msec 50.00%* 85.00% (13.76)

4 harmonics, 1 msec 33.33%*** 62.52% (6.30)

4 harmonics, 50 msec 25.00%*** 77.55% (10.53)

4 harmonics, 190 msec 87.50% 90.00% (10.51)

5 harmonics, 1 msec 37.50%* 75.06% (15.33)

Total 47.22%* 75.55% (13.41)

a MR's five matched controls completed this task.
*p < 0.05, **p < 0.01, ***p < 0.001

Higher-order Temporal Processing of Pitch Patterns

Given MR's impaired temporal and spectral processing for complex pitch perception of brief individual sounds, two tasks were used to assess his temporal processing for complex pitch patterns at the level of hundreds of milliseconds or seconds.

Gap Detection Task

Non-clinical laterality studies have used gap detection tasks to assess left hemisphere superiority for fine temporal processing (Nicholls, 1996). Recent clinical studies have found that gap detection stimuli can provide a measure of higher-order temporal pattern processing of segmented sounds relevant to speech and music (see Griffiths, Rees, & Green, 1999 for a review). Griffiths et al.'s (2001) gap detection task required MR to detect a temporal gap between a 10-ms leading and a 300-ms trailing marker. As shown in Table 2.11, MR was able to make correct responses on 75 percent of trials when the gap duration lasted for 200 milliseconds (log transformation value of 2.30) or more (see Appendix L for MR's psychometric function). MR's normal threshold indicates preserved segmented sound processing for stimuli presented over longer durations.

Low Modulation Tasks

Amplitude modulated (AM) or frequency modulated (FM) sounds are perceived as varying in loudness or pitch respectively. The higher temporal level of hundreds of milliseconds or seconds for sound patterns parallels the level of low modulation rates at tens of Hz or less (Kay, 1982; Moore & Sek, 1995). Thereby, a low modulation task, developed by Griffiths et al. (2001), assessed MR's temporal pattern processing for continuous sound by requiring him to detect a sinusoidally modulated tone.

As shown in Table 2.11, MR's 75 percent threshold was within the normal range for both FM and AM tasks (see Appendix L for MR's psychometric function), suggesting preserved higher-order temporal structure for pitch and loudness cues of continuous patterned sound occurring over a longer temporal window.

Table 2.11

Seventy-five Percent Thresholds for Griffiths et al.'s (2001) Higher-order Temporal Processing Tasks

Psychophysical task MR's threshold Controls threshold (SD)a
Gap detectionb
2.30 1.70 (0.35)
Low frequency modulationce

2 Hz FM 0.16 0.01 (0.18)

40 Hz FM -1.13 -1.22 (0.16)
Low amplitude modulationde

2 Hz AM -1.20 -1.02 (0.26)

40 Hz AM -1.39 -1.30 (0.14)

a Data based on 15 controls matched for age to MR from Griffith et al.'s (2001) norms.
b Data are expressed as log transformations of milliseconds
c Data are expressed as log transformations of the ratio of the frequency excursion to the modulation rate.
d Data are expressed as log transformations of the proportional depth of modulation.
e The carrier frequency and duration of each sound was 500 Hz and 1 second respectively.

Musical Illusions

In order to gain an understanding of the nature of MR's musical illusions, a detailed account of MR's phenomenology was obtained. As shown in Table 2.12, similarities were found between MR's case and cases of musical hallucinosis concerning their deafness levels, and the perceptual experience and quality of their illusory episodes. However, notable differences between MR and the controls were found concerning MR's stroke as the potential index event, his amusia, his less frequent and controllable episodes, his inability to recognise the illusory tunes, and the pleasant emotional experience accompanying his illusions. These disparities indicate that the underlying processes responsible for MR's illusions are not identical to those responsible for the controls' illusions.

Table 2.12

Phenomenology of Musical Illusions Experienced by MR and Musical Hallucinosis Patients

Phenomenology MR Controls, if similar to MRa Controls, if different to MRb
Index event Stroke/hearing loss N/A Hearing loss
Deafness (years) 3 5 M = 36.25 (SD = 16.10)
Measured deafnessc 54 (refer to Appendix C) 52 M = 87.87 (SD = 15.62)
Evidence of amusia Yes N/A No
Perceptual experience Like real music playing Same as MR N/A
Quality of illusory percept Akin to the quality of music prior to index event. Not distorted. Superior to current perception of actual music Same as MR N/A
Trigger of episodes Environmental sounds, e.g. Japanese garden chimes. TV program No trigger
Frequency of occurrence Daily occurrence initially but only occasionally now. N/A Continuous or near-continuous
Paroxysmal nature of illusions Initially spontaneous/ uncontrollable. Currently brought under control after spontaneous onset. Spontaneous onset Uncontrollable after spontaneous onset.
Recognisable music No N/A Yes
Emotional experience Pleasant feeling N/A Distressed/depressed

a,b Data based on 6 musical hallucinosis participants (Griffiths, 2000)
c Mean hearing level in dB I.S.O, measured at intervals of 1 octave from 250Hz to 8kHz.

<< >>