showed that increment detection accuracy was correlated with the temporal microstruc-
ture profiles of expert performances suggesting that temporal increments of an interval
which is usually lengthened are more difficult to detect than temporal increments of inter-
vals for which no microvariations were introduced. In the cognitive literature, two hypo-
thesis, which are not mutually exclusive, have been proposed to explain variations in the
accuracy detection. A top-down hypothesis suggests that listeners’expectations reflect
expressive performance of temporal microvariations^45 whereas a bottom-up hypothesis
indicates that some expectations may be due to psychoacoustical characteristics of the
stimulus.^44
Based on these observations, we designed an experiment involving the detection of
inter-onset interval increments introduced in musical sequences to test perception of these
temporal microvariations in patients with unilateral temporal lobe lesions. The musical
excerpts used were very familiar in order to generate high expectations and to prevent bias
due to each listeners’musical background.
In the experimental task, half of the trials consisted of the presentation of a score (played
exactly as it is written in the score) version of a familiar tune whereas the other half con-
sisted of a modified version of the tune in which one inter-onset interval was increased by
25 per cent, producing an increment of 10–140 ms (mean85 ms; standard devia-
tion62 ms) depending on the size (or duration) of the modified interval within the tune.
The subject’s task was to decide whether the presented trial corresponded to the mechani-
cal or to the modified version of the excerpt using a two-alternative forced choice para-
digm. Two types of inter-onset interval were introduced. One type corresponds to expected
increments and the other one corresponds to unexpected increments. The level of expecta-
tion was determined by results of previous studies detailed elsewhere.^46 Basically, the
recording of pianists’performances were compared to a score (or computerized) version of
each selected tune to allow the analysis of temporal microstructure. The results of this
experiment allowed the identification of one interval that was systematically lengthened in
pianists’recordings as compared to metrical version, and another interval that was systemat-
ically preserved. Inter-onset interval increments were thus introduced at these specific loca-
tions and it was hypothesized that temporal increments located at ‘lengthened intervals’
would be more difficult to detect than temporal increments located at ‘preserved intervals’
since listeners would be expecting temporal increments in the first but not in the second
condition. The ability to distinguish these two types of inter-onset increments was subse-
quently tested in a perceptual task confirming therefore the relevance of this temporal
manipulation in normal nonmusician listeners. Based on these results, we designed an
experimental task to evaluate the ability to detect subtle temporal changes by taking into
account implicit knowledge of the rhythmical structure underlying musical listening. By
manipulating temporal increments and cognitive expectations, especially in regard to the
temporal (or rhythmic) dimension, it was possible to assess bottom-up as well as top-down
processing involved in musical rhythm perception.
Twenty-two patients who had undergone a right (n11) or a left (n11) temporal
lobe resection for the relief of medically intractable epilepsy as well as 14 normal control
subjects were tested in this experiment. None of the patients presented language disorders
or suffered from extra temporal lesions. Language was lateralized on the left side in all
211