Perception, 2014, volume 43, pages 1083 – 1096
doi:10.1068/p7789
Improved beat asynchrony detection in early blind individuals
Elodie Lerens, Rodrigo Araneda, Laurent Renier, Anne G De Volder §
Université catholique de Louvain, Institute of Neuroscience (IoNS), Avenue Hippocrate 54, UCL B1.54.09, 1200 Brussels, Belgium; e‑mail:
[email protected] Received 12 May 2014, in revised form 29 August 2014, published online 14 October 2014 Abstract. Although early blind (EB) individuals are thought to have a better musical sense than sighted subjects, no study has investigated the musical rhythm and beat processing abilities in EB individuals. Using an adaptive ‘up and down’ procedure, we measured the beat asynchrony detection threshold and the duration discrimination threshold, in the auditory and vibrotactile modalities in both EB and sighted control (SC) subjects matched for age, gender, and musical experience. We observed that EB subjects were better than SC in the beat asynchrony detection task; that is, they showed lower thresholds than SC, both in the auditory and in the vibrotactile modalities. In addition, EB subjects had a lower threshold than SC for duration discrimination in the vibrotactile modality only. These improved beat asynchrony detection abilities may contribute to the known excellent musical abilities often observed in many blind subjects. Keywords: visual deprivation, early blindness, rhythm processing, temporal processing
1 Introduction Early blind (EB) individuals are thought to have a better musical sense than sighted individuals. There are more individuals with absolute pitch (Hamilton, Pascual-Leone, & Schlaug, 2004), sharper pitch (Gougoux et al., 2004; Voss & Zatorre, 2012), and better melody (Voss & Zatorre, 2012) discrimination abilities in the EB population even when controlling for musical experience (Wan, Wood, Reutens, & Wilson, 2010a). Rhythm and beat are two other essential components of music. Rhythm refers to the temporal organization of sound and silence in a musical sequence while beat is a particular element of the rhythm that marks equally spaced points in time and which produces a salient regular periodicity (Cooper & Meyer, 1960; Large & Palmer, 2002; Lerdahl & Jackendoff, 1983; Patel, Iversen, Chen, & Repp, 2005). Our perception of the beat is what allows us to accurately clap hands and dance to music in spite of the complex temporal structure of most musical sequences. In fact, when a beat is present, the occurrence of the following sound events can be predicted and thus the movements can be anticipated and prepared in order to synchronize them on the beat (Grahn & Rowe, 2013). In complex sequences of music the beat can be marked by a note of a higher intensity or of a different frequency in the auditory flow, but it can also be induced by the specific temporal grouping of sounds and silences that stresses regular periodic points in time (Drake, Jones, & Baruch, 2000; Essens & Povel, 1985; Grahn & Brett, 2007). In contrast to the relatively high number of studies on pitch perception in humans (Gougoux et al., 2004; Moore, 2008; Pring, Woolf, & Tadic, 2008; Tervaniemi, Just, Koelsch, Widmann, & Schröger, 2005; Voss & Zatorre, 2012; Zatorre, 1988; Zatorre, Evans, & Meyer, 1994), very few studies have focused on the temporal processing abilities in EB subjects and no study has focused on rhythm and beat perception in this population. Because usually the auditory system (and not the visual one) is the most precise sense in the temporal domain (Berger, Martelli, & Pelli, 2003; Burr, Banks, & Morrone, 2009; Holcombe, 2009; Shams, Kamitani, & Shimojo, 2000), subjects with congenital visual impairment should rely mainly on their spared audition and may therefore § Corresponding author.
1084
E Lerens, R Araneda, L Renier, A G De Volder
develop better abilities in temporal perception. Interestingly, improved temporal processing was reported in EB individuals in comparison with sighted controls (Röder, Krämer, & Lange, 2007; Röder, Rösler, & Spence, 2004; Stevens & Weaver, 2005; Van der Lubbe, Van Mierlo, & Postma, 2010). In particular, Van der Lubbe and colleagues (2010) reported superior auditory and vibrotactile duration discrimination abilities in EB individuals in comparison with sighted individuals. Finer temporal order judgment was also demonstrated in EB individuals in the auditory modality (Boas, Muniz, Neto, da Silva, & Gouveia, 2011; Stevens & Weaver, 2005) as well as in the tactile modality (Röder et al., 2004) during experiments in which participants had to determine which ear or hand was first stimulated. The improved temporal abilities observed in the EB individuals in comparison with the sighted ones led the authors to suggest that the EB could also possess better beat perception abilities. However, though beat and duration processing have long been considered to share some common mechanisms (Ivry & Spencer, 2004; McAuley & Jones, 2003), there are recent reports about a partial dissociation between both processes as well as a partially different neural network involved in the two (Grube, Cooper, Chinnery, & Griffiths, 2010a; Grube & Griffiths, 2009; Grube, Lee, Griffiths, Barker, & Woodruff, 2010b; Teki, Grube, Kumar, & Griffiths, 2011; Parncutt, 1994). A striato-cortical network is involved in beat processing whereas a cerebellar-cortical network is recruited by duration processing (Grahn & Brett, 2007; Grahn & Rowe, 2009; Grube et al., 2010a, 2010b; Ivry & Spencer, 2004; Teki et al., 2011). Consequently, beat processing of a regular sequence and duration discrimination of two single intervals could be differently affected by visual deprivation. In this study we aimed to test the effect of early visual deprivation on the accuracy of duration discrimination as well as on the accuracy of beat asynchrony detection. We used an adaptive procedure in order to measure the individual threshold for duration discrimination and for beat asynchrony detection. We chose a beat asynchrony detection task rather than sensorimotor synchronization with a beat (eg a tapping) because beat asynchrony detection had the advantage of minimizing the motor component. In addition, in contrast to most studies on asynchrony detection, we used complex beat sequences, in which the beat was produced by a regular temporal organization of the sequence rather than by simple isochronous sequences, in order to force participants to search for and extract the beat (Grahn & Rowe, 2009). This way, we assumed that beat asynchrony detection could not be achieved with a duration discrimination strategy. We administered both tasks in the auditory modality and in the vibrotactile modality. We used stimulation sequences with identical timing in the two sensory modalities, in order to allow the comparison between them. 2 Methods 2.1 Subjects The study involved fourteen EB individuals and fourteen sighted controls (SC), matched for gender (ten men), age (EB: 39.07 years ± 13 years; SC: 39.79 years ± 10 years), and the number of years of formal musical lessons (EB: 8.04 years ± 3.67 years; SC: 8.79 years ± 4.69 years, table 1 and appendix table A1). The musical experience was taken into account in order to control for its potential effect on beat and duration perception (Drake & Palmer, 2000; Repp, 2005; Strait, Kraus, Parbery-Clark, & Ashley, 2010). We paid special attention to identifying the subjects who reported percussion experience, given the particularly good beat perception and synchronization ability of percussionists (Ehrlé & Samson, 2005; Gérard & Rosenfeld, 1995). Seven EB participants and two SC had practised percussion instruments. The EB subjects were affected by complete blindness (no residual vision) as the result of bilateral ocular or optic nerve lesions from birth or before the age of 18 months. A summary of their medical history is provided in table 1. All blind participants were proficient Braille readers, with two hours’ practice daily. They all learned it around the age of 5 years, before entering
Beat perception in early blindness
1085
Table 1. Profile of the early blind (EB) subjects. a
Subject
Age
Sex
Handedness
EB 1 EB 2 EB 3
37 40 29
male female male
right ambidextrous right
EB 4 EB 5 EB 6
57 46 37
male male male
right right right
EB 7 EB 8 EB 9 EB 10 EB 11 EB 12 EB 13 EB 14
32 62 28 35 33 50 36 35
female male male female male male female male
right right left right right ambidextrous left ambidextrous
Onset of blindness
Diagnosis
Musical b experience
birth birth birth
Leber congenital amaurosis 7 retinopathy of prematurity 8* persistent hyperplastic primary 6 vitreous involving both eyes birth 13 severe retinal dystrophy c birth Leber congenital amaurosis 6 birth anterior chamber cleavage 14* syndrome (Peters syndrome) birth 1.5* severe corneal dystrophy c 800 ms: delay = 20% of the reference interval). Participants were required to judge whether the target interval had the same duration or was longer than the reference interval. Different target intervals were used, starting with a target interval of 880 ms, using an adaptive transformed up–down procedure based on subjects’ responses (see below).
(a)
Reference interval of 800 ms
Silent interval
Target interval of H 800 ms
Perceived beat
(b)
Reference IBI of 900 ms
= 80 ms stimulation
“Is this interval duration the same as the first one?” “Is this last simulation synchronized with the beat?”
Target IBI of H 900 ms = standard place of the final stimulation
= delay of the final stimulation
Figure 1. Examples of sequences. (a) Duration discrimination task: participants had to compare two duration intervals—a reference interval of 800 ms and a target interval that could last 800 ms (same trial) or more than 800 ms (different trial, eg 850 ms). (b) Asynchrony detection task: participants had to judge whether the last stimulation of a sequence was synchronized with the beat or not. The reference interbeat interval (IBI) was 900 ms (ie each beat occurred every 900 ms). The last stimulation could occur either 900 ms after the last beat (synchronized with the beat) or later (not synchronized, eg 980 ms).
2.2.2 Asynchrony detection. The sequences used in the asynchrony detection task were composed of stimulations provided with different inter-onset intervals (IOIs: 225 ms, 450 ms, 675 ms, or 900 ms) that were temporally grouped in order to make units of 900 ms (ie 450 + 450 ms or 225 + 675 ms, or 225 + 225 + 225 + 225 ms or 900 ms; see figure 1b), which was the length of the reference interbeat interval (IBI). This regular temporal structure led to the perception of a sequence of four ‘beats’, one at the beginning of each 900 ms unit with a similar methodology as in the pioneer paper of Povel and Essens in 1985 (see also Grahn & Brett, 2007; Grahn & Rowe, 2009). Each stimulation lasted 80 ms. A last stimulation of 80 ms was added at the end of the sequence, after the last interval of 900 ms, to provide a supplementary fifth beat that could be either synchronized with the previous ones or delayed (see below). The same sequence was played for all the trials in both sensory modalities, except that the final stimulation of the sequence could be delayed (eg occurring at 980 ms) in some trials to create trials with target IBI of the same duration (900 ms) and trials with target IBI longer than the reference IBI (> 900 ms: delay = 20% of the reference interval). Participants were required to judge whether the last stimulation of the sequence was or was not synchronized on the beat.
Beat perception in early blindness
1087
2.3 Experimental design Participants were tested individually in a quiet room in a sitting position. Sighted subjects saw the equipment and received the instructions before being blindfolded. A familiarization session was administered before both tasks. In particular, during the familiarization to the asynchrony detection task, participants listened to a sequence of 20 beats with the same temporal structure and tempo as the one to be used in the task until they perceived the beat. They were required to tap in synchrony with the perceived beat via a computer mouse while listening to the sequence. Tapping times on 20 beats (1 sequence) were recorded and a measure of accuracy was calculated for each sensory modality. This measure was adapted from the ‘global measure of accuracy’ used by Chen, Penhune, and Zatorre (2008) in order to assess whether the subjects were able to perform the task. Tapping time was considered as correctly synchronized when it occurred within half the reference IBI (450 ms) before or after each beat onset. If more than one tapping response fell within this time window, the first response was considered and the other ones were ignored. During duration discrimination and asynchrony detection experiments, participants were allowed to use their own strategies in both tasks (eg tapping, counting, and internal verbalization). Verbal subjects’ responses were noted by the experimenter, and a feedback was provided after each response to optimize subjects’ performance similarly as in previous studies using an adaptive procedure (Amitay, Irwin, Hawkey, Cowan, & Moore, 2006; Ehrlé & Samson, 2005). An adaptive transformed up–down procedure was used—that is, the individual response at one trial determined the delay of the last stimulation in the next trial (Levitt, 1971). We decided to use delayed shifts only, as in the study of Ehrlé and Samson (2005), because previous studies showed similar asynchrony detection thresholds for anticipated and delayed shifts in isochronous sequences with a similar tempo as in the present study (Halpern & Darwin, 1982; Hibi, 1983, cited in Friberg & Sundberg, 1995). For the first trial, the delay of the final stimulation was fixed and corresponded to 10% of the reference interval—that is, 80 ms for the reference interval of 800 ms in the duration discrimination task and 90 ms for the reference IBI of 900 ms in the asynchrony detection task. For the subsequent trials, two successive correct answers resulted in a delay decrease of 1% (–8 ms in the duration discrimination task and –9 ms in the asynchrony detection task), whereas the delay was increased by 1% in case of error (Levitt, 1971). Each shift value that produced a change in the shift direction (increased delay vs decreased delay) was recorded until six direction changes were obtained (Ehrlé & Samson, 2005). In each task the average of the four last shift values that triggered a direction change was considered as the individual threshold for the duration discrimination task and for the asynchrony detection task, respectively. The absolute values of individual thresholds were further expressed in relative values; that is, there were converted in percentages of the reference interval (Ehrlé & Samson, 2005), which allowed the comparison of performance between the two tasks. We added sequences without a delay (same trials) in order to have a validity index and to ensure that subjects actually seriously performed the task and did not respond randomly. Sequences without a delay (one third) and with a delay (two thirds) were randomly mixed. As a result of the adaptive procedure, the number of trials differed for each subject. One run of the adaptive procedure was administered for each task and modality, as in a previous study by Ehrlé and Samson (2005). An independent study on three additional subjects showed no difference between the threshold obtained using one run and the threshold obtained using four runs. The order of task and tested sensory modality were counterbalanced between the subjects. It should be noted that the reference interval in the duration discrimination task (800 ms) was different from the reference IBI in the asynchrony detection task (900 ms) to minimize task influence and to control for practice-related changes in performance. Breaks of 2 min were introduced between tasks as well as every 5 min to reduce the risk of any potential decrease of attention.
1088
E Lerens, R Araneda, L Renier, A G De Volder
2.4 Data analysis We first verified that our data were normally distributed in each group, modality, and task using the Shapiro–Wilk test. For the experimental conditions we performed two separate 2 (groups: EB vs SC) × 2 (sensory modality: audition vs tactile) analyses of variance (ANOVAs): on the mean individual relative thresholds for duration discrimination and for asynchrony detection. In addition, simple group and modality effects were tested on the thresholds using Student’s t-tests. We performed a one-tailed Pearson correlation analysis to test the relation between auditory and vibrotactile thresholds, for each task and in each group separately. Finally, we performed a one-tailed Pearson correlation analysis to test the relation between the musical experience and individual thresholds, for each task and in each group separately. 3 Results 3.1 Performance during the familiarization session. On average, the subjects correctly performed the tapping task in both modalities. The tapping times of one EB and two SC subjects were not recorded for technical reasons. Mean tapping accuracy was considered as 100% of correct taps because no participant tapped outside a time window of 450 ms (half of the IBI) before or after beat onset. The largest gap was 400 ms, and it was observed once only for one blind participant and three times for one sighted participant, both in the vibrotactile modality. 3.2 Performance during the experimental tasks On average, all the subjects performed correctly the tasks in all conditions: the average accuracy (percentage of trials with a longer target interval than the reference interval that had been correctly detected) was 75.2% [mean expressed in percentage ± standard error of the mean (M ± SEM): EB: 77.8% ± 2.1%; SC: 72.5% ± 2%, t26 = –2, p
