Tipultech logo

Auditory Cognition for Music

Dawn Senathi-Raja

Beckwith (2003) previously established MR's agnosia for musical sounds, finding that he was unable to recognise the lyrics and tunes of familiar music, or rhythmically process novel tunes. Some of Beckwith's tests were re-administered and new tests conducted to more fully assess MR's melodic processing for novel tunes.

Discrimination of Melodic Components

One such re-test involved Peretz's (2003) melodic tasks, which assessed the integrity of tonal, contour and intervallic processing by systematically manipulating a critical pitch in one melody of a pair (see Figure 2.2). The critical pitch manipulation for the Contour, Interval, and Tonality tasks created contour-violated, contour-preserved but interval-violated, and scale-violated alternate melodies respectively. Each task comprised 30 items requiring a "same" or "different" response.

Examples of melodies from the Montreal Battery of Amusia discrimination 
						tasks (taken from Liegois-Chauvel et al., 1998, p.1858). The melody shown 
						at (A) is the standard. The melody (B) has changed melody (A) at the * by 
						lowering the note producing a contour violation. Melody (C) changes the same 
						note making it out of key but preserving contour. Melody (D) changes this 
						note but preserves both contour and key. These are examples of items from the 
						Contour, Tonality and Interval tasks, respectively.

Consistent with Beckwith's (2003) findings, MR performed below the controls on the Interval and Tonality tasks, but performed within the normal range on the Contour task (see Table 2.2), suggesting preserved contour representation.

Table 2.2

Percentage of Correct Responses for Peretz's (2003) Melodic Tasks

Discrimination Task MR's previous percentage correct (Beckwith, 2003) MR's current percentage correct Controls percentage correcta Mean (SD)
Contour 73.33% 83.33% 91.33% (6.91)
Interval 43.33%*** 66.67%** 88.67% (6.91)
Tonality 53.33%*** 53.33%*** 88.87% (4.47)

a Data based on six controls, matched for age and sex to MR, from the norms for the Montreal Battery of Evaluation of Amusia (Peretz, 2003) **p < 0.01, ***p < 0.001

Discrimination of Melodic Contour

The right STG has been implicated in contour processing (Liegeois-Chauvel et al., 1998). MR's lesion encroaches on the STG, thus the finding suggesting preservation of MR's contour processing appears anomalous. Peretz's (2003) discrimination tasks, however, only assessed MR's ability to process contour in basic sequences. Thus, four discrimination tasks designed by Trehub, Bull and Thorpe (1984) were recreated and administered to further interrogate MR's contour processing. With the exception of items from the 'Identical' condition, melodic pairs were subject to the follow manipulations: (1) Transposition, in which the absolute pitch of individual tones was changed but the pitch ratios of successive tones remained intact; (2) Contour-violation, wherein ascending tones became descending tones or vice versa; (3) and Octave-dispersed/contour-preservation, where contour and chroma (note name) were retained but without regard to the octave from which the individual notes were drawn. Randomly chosen pairs from each of the conditions were combined to form a 24-item test. Participants were told to respond 'same' when the contour of the melodies was the same irrespective of transposition or octave change, or 'different' if otherwise. A time limit of seven seconds was imposed. An example of an original melody and all transformed comparisons is depicted in Figure 2.3.

Graph of original melody and transformed comparisons from Trehub et al.'s (1984) tasks.

Contrary to MR's normal performance on Peretz's (2003) Contour task, MR failed to discriminate same and different contours in Trehub et al.'s (1984) more subtle Contour- violation and Transposition tasks (z = 3.05, p < 0.01 and z = 4.92, p < 0.001 respectively). This suggests a failure for contour representation. Moreover, MR performed significantly below the controls on the Identical task (z = 4.10, p < 0.001). This result suggests a low-level perceptual deficit, which may have prevented the formation of melodic representations.

Octave dispersion facilitated MR's ability to recognise preserved contour in melodic pairs, as shown by his score within the normal range (see Table 2.3). Octave dispersion exaggerates contour cues, which may explain MR's improved performance.

Table 2.3

Percentage of Correct Responses for Trehub et al.'s (1984) Melodic Contour Discrimination Tasks

Condition MR's percentage correct Controls percentage correcta Mean (SD)
Identical 66.67%*** 96.67% (7.31)
Transposition 50.00%*** 86.67% (7.45)
Octave-dispersed/contour-preserved 83.33% 86.67% (13.94)
Contour-violated 66.67%** 94.00% (8.94)
Total 66.67%*** 89.29% (5.05)

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

To test the possible contribution of exaggerated contour cues on MR's normal contour-preserved performance, a Linear-transformation/contour-preserved task created by Massaro, Kallman, and Kelly (1980) was recreated. This task reduced the total pitch range of the original sequence by half, but preserved contour and the relative size of the original intervals. Randomly chosen melodic pairs from the Linear-transformation/contour-preserved condition and the Octave-dispersed/contour-preserved condition were combined to form a test of 24 items.

As shown in Table 2.4, MR's normal performance on the Octave-dispersed/contour- preserved task was replicated. MR performed significantly below controls on the Linear- transformed/contour-preserved condition (see Table 2.4), indicating that he can reliably represent the contours of novel melodies only when the contour cues are substantial.

Table 2.4

Percentage of Correct Responses for Massaro et al.'s (1980) Preserved Contour Tasks

Condition MR's percentage correct Controls percentage correcta Mean (SD)
Octave-dispersed 83.33% 88.88% (4.31)
Linearly-transformed 41.67%** 75.01% (12.52)
Total 62.50%*** 80.67% (4.62)

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

Recognition of Instrumental Sounds

In an instrumental sounds task comprising 26 items, MR previously performed within the normal range (Beckwith, 2003). This result is surprising, given his reported experience of confusion when identifying orchestral instruments. Table 2.5 shows the results of a re-test, which now suggests that MR was impaired in identifying individual instruments and the families to which they belong (refer to Appendix F). Although MR's performance has remained stable, the change in statistical outcomes is due to the poorer performance of the previous control group compared with the current group. The current group appears more closely matched for musical background, for example, the 'time spent listening to music' variable, so the more plausible inference is that MR exhibits agnosia for instrumental sounds.

Table 2.5

Percentage of Correct Responses for Beckwith's (2003) Instrumental Sound Identification Task

Task MR's previous percentage correct (Beckwith, 2003) MR's current percentage correct Controls percentage correcta Mean (SD)
Instrument family 87.50% 87.50%* 98.36% (2.33)
Instrument name 65.38% 65.38%* 80.81% (6.10)

a MR's five matched controls completed this task
*p < 0.05

<< >>