Uncovering Beat Deafness: Detecting Rhythm Disorders with Synchronized Finger Tapping and Perceptual Timing Tasks

Podsumowanie

Behavioral tasks that allow for the assessment of perceptual and sensorimotor timing abilities in the general population (i.e., non-musicians) are presented. Synchronization of finger tapping to the beat of an auditory stimuli and detecting rhythmic irregularities provides a means of uncovering rhythm disorders.

Streszczenie

A set of behavioral tasks for assessing perceptual and sensorimotor timing abilities in the general population (i.e., non-musicians) is presented here with the goal of uncovering rhythm disorders, such as beat deafness. Beat deafness is characterized by poor performance in perceiving durations in auditory rhythmic patterns or poor synchronization of movement with auditory rhythms (e.g., with musical beats). These tasks include the synchronization of finger tapping to the beat of simple and complex auditory stimuli and the detection of rhythmic irregularities (anisochrony detection task) embedded in the same stimuli. These tests, which are easy to administer, include an assessment of both perceptual and sensorimotor timing abilities under different conditions (e.g., beat rates and types of auditory material) and are based on the same auditory stimuli, ranging from a simple metronome to a complex musical excerpt. The analysis of synchronized tapping data is performed with circular statistics, which provide reliable measures of synchronization accuracy (e.g., the difference between the timing of the taps and the timing of the pacing stimuli) and consistency. Circular statistics on tapping data are particularly well-suited for detecting individual differences in the general population. Synchronized tapping and anisochrony detection are sensitive measures for identifying profiles of rhythm disorders and have been used with success to uncover cases of poor synchronization with spared perceptual timing. This systematic assessment of perceptual and sensorimotor timing can be extended to populations of patients with brain damage, neurodegenerative diseases (e.g., Parkinson’s disease), and developmental disorders (e.g., Attention Deficit Hyperactivity Disorder).

Wprowadzenie

Humans are particularly efficient at processing the duration of events occurring in their environment¹. In particular, the ability to perceive the beat of music or the regular ticking of a clock and the ability to move along with it (e.g., in dance or synchronized sports) is widespread in the general population (i.e., in individuals who have not received musical training)^2,3. These abilities are underpinned by a complex neuronal network involving cortical brain regions (e.g., the premotor cortex and the supplementary motor area) and subcortical structures, such as the basal ganglia and the cerebellum^4-7.

Disruption of this network and consequent poor temporal processing can result from brain damage^8-10 or neuronal degeneration, as observed in patients with Parkinson’s disease¹¹. However, poor perception of duration and poor synchronization to the beat of music can also manifest in healthy individuals in the absence of brain damage. In spite of the fact that the majority can perceive auditory rhythms and synchronize the movement to the beat (e.g., in music), there are notable exceptions. Some individuals have major difficulties in synchronizing their body movements or finger tapping to the beat of music and can exhibit poor beat perception, showing difficulties in discriminating melodies with notes of different duration. This condition has been referred to as “beat deafness” or “dysrhythmia” ^2,12-14. For example, beat deafness was described in a recent study¹³, in which the case of a patient named Mathieu was reported. Mathieu was particularly inaccurate at bouncing to the beat of rhythmical songs (e.g., a Merengue song). Synchronization was still possible, but only to the sounds of a simple isochronous sequence (e.g., a metronome). Poor synchronization was associated with poor beat perception, as revealed by the Montreal Battery of Evaluation of Amusia (MBEA)¹⁵. In an additional task, Mathieu was asked to match the movements of a dancer to the music; interestingly, Mathieu exhibited unimpaired pitch perception.

Poor rhythm perception and poor synchronization, in beat-deaf individuals with spared pitch perception, were observed in further studies^2,12,14, thus providing compelling evidence that rhythm disorders can occur in isolation. Beat deafness is therefore distinct from the typical description of congenital amusia (i.e., tone deafness), a neurodevelopmental disorder affecting pitch perception and production^16-19. Interestingly, poor rhythm perception and production can co-occur with poor pitch processing in congenital amusia^12,16,20. Nevertheless, poor rhythm perception in this case depends on the ability of an individual to perceive pitch variation. When pitch variations in melodies are removed, congenital amusics can successfully discriminate rhythm differences²¹.

Important individual differences have been observed in beat deafness; this fact deserves particular attention. In most cases, both rhythm perception and synchronization to the beat of music are deficient^2,12-14; however, poor synchronization can also occur when rhythm perception is spared². This dissociation between perception and action in the timing domain has been shown using synchronized tapping tasks with a variety of rhythmic auditory stimuli (e.g., a metronome and music) and using different rhythm perception tasks (e.g., the discrimination of melodies based on different note durations and the detection of deviations from isochrony in rhythmic sequences). This finding is particular relevant because it points to the possible separation of perception and action with regard to timing mechanisms, as previously observed in pitch processing^17,22-25. Further dissociations were highlighted depending on the stimulus complexity². Most poor synchronizers exhibited selective difficulties with complex stimuli (e.g., music or amplitude-modulated noise derived from music), while they still showed accurate and consistent synchronization with simple isochronous sequences; other poor synchronizers showed the opposite pattern. In summary, these results converge in indicating that there are a variety of phenotypes of timing disorders in the general population (as observed in other domains of musical processing such as pitch^25,26), which require a sensitive set of tasks to be detected. Characterizing the patterns of rhythm disorders is particularly relevant to shed light on the specific mechanisms that are malfunctioning in the timing system.

The goal of the method illustrated here is to provide a set of tasks that can be used to uncover cases of beat deafness in the general population and detect different subtypes of timing disorders (e.g., affecting perceptual vs. sensorimotor timing or a particular class of rhythmic stimuli). Sensorimotor timing abilities have mostly been examined using finger tapping tasks with auditory material. Participants are asked to tap their index finger in synchrony with auditory stimuli, such as to a sequence of tones equally spaced in time or to music (i.e., in a synchronized or paced tapping task^27-29). Another popular paradigm, which has been the source of considerable modeling efforts^29-32, is the synchronization-continuation paradigm, in which the participant continues tapping at the rate provided by a metronome after the sound has stopped. Rhythm perception is studied with a variety of tasks ranging from duration discrimination, estimation, bisection (i.e., comparing durations to 'short' and 'long' standards), and detection of anisochrony (i.e., determining whether there is a deviant interval within an isochronous sequence) to the beat alignment task (i.e., detecting whether a metronome superimposed onto music is aligned with the beat)^1,2,20,33,34. Most studies have focused on time perception, beat production or sensorimotor timing, which were tested in isolation. However, it is likely that such different tasks refer to somewhat different abilities (e.g., interval timing vs. beat-based timing, perceptual vs. sensorimotor timing) and do not reflect the functioning of the same timing mechanisms and the associated neuronal circuitry. This issue can be circumvented by using recently proposed batteries of tasks that assess both perceptual and sensorimotor timing abilities. These batteries allow researchers to obtain an exhaustive profile of an individual’s timing abilities. Examples of such batteries are the beat alignment test (BAT)³⁴, the Battery for the Assessment of Auditory Sensorimotor Timing Abilities (BAASTA)³⁵, and the Harvard Beat Assessment Test (H-BAT)³⁶. These batteries consist of tapping tasks with a variety of rhythmic auditory stimuli ranging from music to isochronous sequences as well as perceptual tasks (e.g., duration discrimination, detection of the alignment of a metronome to the beat of music, and anisochrony detection). In all cases, the same set of musical excerpts was used in perceptual and sensorimotor tasks.

In this paper, we illustrate a set of tasks that are particularly efficient at revealing patterns of rhythm disorders in beat-deaf individuals and poor synchronizers, as shown in previous studies². These tasks are part of a larger battery of tests, the BAASTA³⁵. Sensorimotor timing abilities are tested by asking participants to tap their finger to the beat of simple and complex auditory stimuli (e.g., isochronous sequences, music, and rhythmic noise derived from musical stimuli)^27,28. Perceptual timing is tested with an anisochrony detection task^2,20,33,37. A set of isochronous tones is presented. In some cases, one of the tones (e.g., the penultimate) is presented sooner or later than expected based on the isochronous structure of the auditory sequence. Participants are asked to detect deviations from isochrony. The advantage of these sensorimotor and rhythm perception tasks is that they both involve sequences of stimuli (instead of single durations) and stimuli of different complexity. Thus, based on previous evidence, these tasks provide the optimal conditions to uncover different phenotypes of beat deafness and poor synchronization. Particular attention is paid to the technique adopted in the analysis of synchronization data. This technique is based on circular statistics, an approach that is particularly well-suited for examining inaccurate and inconsistent synchronization to the beat.

Protokół

1. Synchronization Tasks

1. Preparation of Instruments:
   1. Connect a standard MIDI percussion instrument to the computer via a conventional MIDI interface.
      NOTE: Data acquisition is realized via a MIDI electronic percussion instrument. The device captures the exact timing of the finger taps during the motor synchronization tasks.
   2. Open the dedicated software for stimulus presentation and response recording.
      NOTE: The synchronization task is implemented using standard software for the presentation of audio material and recording of data from a digital MIDI musical instrument (with 1 msec precision).
2. Sound Material and Procedure:
   1. From the software interface, select the pacing stimulus to be used in the synchronization task from among three choices (isochronous sequence, music, and amplitude-modulated noise obtained from the wave envelope of the musical stimulus).
      NOTE: The isochronous sequence is composed of 96 isochronously presented tones (duration = 30 msec). The musical stimulus is a computer-generated piano version of a fragment of the Radetzky March (Opus 228) by Johann Strauss that includes 96 beats (beat = quarter note). Excerpts of the three stimuli are provided as Additional Materials to this manuscript.
   2. Select the appropriate tempo for the selected pacing stimulus (450, 600, or 750 msec Inter-onset-interval (IOI) / Inter-beat-interval (IBI)) as indicated in the software interface. Ensure that the stimuli are delivered at a comfortable volume level over the headphones.
   3. Ask the participant to sit in a quiet room in front of the computer monitor.
   4. Ask the participant to tap on the MIDI percussion instrument using the index finger of her or his dominant hand in synchrony with the tones of the isochronous sequence or with the musical beats for more complex stimuli (music or noise). Instruct the participant to tap as regularly as possible, without changing the tapping rate, while synchronizing with the pacing stimulus.
   5. Start the stimulus presentation and recording of taps.
   6. End recording of taps after presenting the last tone or musical beat.
3. Data Analysis:
   NOTE: Analyze the data from the synchronized tapping tasks using circular statistics^38,39. This method is particularly well-suited for the analysis of synchronization data^40,41; moreover, circular statistics are sensitive to individual differences in timing abilities and are therefore able to uncover cases of poor synchronization^2,40. The analysis procedure outlined below is implemented using Matlab software (using the CircStat toolbox³⁹).
   1. Transform the time of the taps relative to the pacing stimuli into angles on the unit circle (from 0-360°) following the procedure indicated by Berens³⁹. 0° (which is equal to 360°) corresponds to the time of the occurrence of the pacing stimulus (i.e., the sounds or musical beats). Use the following formula to obtain the angle for each tap time: [angle(radians) = 2 × π × (time of the tap / IOI)]. Convert radians into degrees with the circ_rad2ang function³⁹.
   2. Plot the angles obtained in the tapping trial as a distribution of dots on the unit circle. Do this using the circ_plot function³⁹. Provide angles in radians as the argument for the function to display the plot (see example in Figure 1).
   3. For each tapping trial, use the angles (dots on the circle) to compute the mean resultant vector R ^38,39,42 (see Figure 1). Use the circ_mean³⁹ and circ_r³⁹ functions, which allow for the computation of synchronization accuracy and consistency, respectively.
   4. Compute the synchronization accuracy (i.e., on average, how far from the pacing stimulus the participant taps in a synchronized tapping trial), which corresponds to the angle θ of vector R. Use the circ_mean function³⁹. Provide angles in radians as the argument for the function.
   5. Submit the tapping data to the Rayleigh test⁴³ to assess whether the distribution of the dots around the circle is random, using the circ_rtest function³⁹. Provide angles in radians as the argument for the function.
      NOTE: In the Rayleigh test, reject the null hypothesis (i.e., circular uniformity, randomly distributed dots around the circle) if the R vector length is large enough (e.g., greater than 0.4), indicating that participants tapped at a given phase relationship with respect to the pacing stimulus above chance. Only when Rayleigh test is significant (i.e., when the distribution of dots around the circle is not random) synchronization accuracy can be properly interpreted.
   6. Compute the synchronization consistency (i.e., the variability in the discrepancy between the time of the taps and the pacing stimuli), which corresponds to the length of the vector R (from 0 to 1). Use the circ_r function³⁹. Provide angles in radians as the argument for the function.
      NOTE: The consistency is 1 when all of the taps occur at exactly the same time interval before or after the pacing stimuli; the consistency is 0 when the taps are randomly distributed around the circle.
4. Evaluation of Individual Results:
   NOTE: Compare the performance of a participant to a normative group or to a control group to uncover cases of poor synchronization accuracy or poor consistency. To perform this comparison, run a corrected t-test⁴⁴ implemented in the singlims computer program (http://homepages.abdn.ac.uk/j.crawford/pages/dept/SingleCaseMethodsComputerPrograms.htm).
   1. Open the singlims computer program. Enter the mean and SD of the synchronization accuracy, and the sample size of the normative or control group. Provide the synchronization accuracy for the participant to be compared to the normative or control group. Click on the “Compute” button to obtain the results of the corrected t-test.
      NOTE: The participant performed significantly poorer than the normative or control group when the two-tailed probability of the corrected t-test is below 0.05.
   2. Enter the mean and SD of the synchronization consistency and sample size of the normative or control group. Provide the synchronization consistency for the participant who is to be compared to the normative or control group.

2. Rhythm Perception Tasks (Anisochrony Detection)

1. Preparation of Instruments:
   1. Open the computer program used for implementing the anisochrony detection tasks. Ensure that the keys of the computer keyboard are properly set to record participants’ answers.
      NOTE: The rhythm perception tasks are implemented using standard software for running behavioral experiments (i.e., stimulus presentation and recording of behavioral responses).
2. Sound Material and Procedure:
   1. Select the stimulus (either isochronous stimulus or music) as indicated by the software interface. Choose the appropriate tempo (450, 600, or 750 msec IOI/IBI) of the selected stimulus. Ensure that the stimuli are delivered over the headphones at a comfortable volume level.
      NOTE: Stimuli are based on the same auditory material used in the Synchronization tasks. Each stimulus includes only 8 isochronously presented tones or musical beats instead of 96. For each stimulus type, there is a “change” version (50% of the trials, n = 24) and a “no-change” version (50% of the trials, n = 24). In the change stimuli, the penultimate sound or musical beat occurs earlier or later than expected (by 8, 12, or 16% of the sequence IOI/IBI) based on the previous IOIs/IBIs. In the no-change stimulus, the IOIs/IBIs are completely isochronous.
   2. Instruct the participant to sit in a quiet room in front of the computer monitor, listen to the stimulus and then judge, after its presentation, whether a change in the interval between the stimuli or beats (i.e., anisochrony) is present or not. Encourage the participant to pay attention to the entire sequence.
   3. Start the stimulus presentation. Ask the participant to respond by pressing one of two keys on the computer keyboard (i.e., one key for “change” or the other key for “no-change” responses) after the presentation of the stimulus.
3. Data Analysis:
   NOTE: Analyze the data obtained from the rhythm perception task by calculating the discriminability index (d’) at each level of change (at 8, 12, 16% of the IOI/IBI) and for each IOI/IBI. The higher the d’ value, the greater the sensitivity to anisochronies.
   1. Consider the responses (n = 48) yielded by each participant for a given stimulus, recorded in the output file by the software used to run the behavioral experiment. Count the number of responses when the anisochrony present in the stimulus has been correctly detected. Compute the Hits rate (i.e., number of Hits / number of change stimuli).
   2. Count the number of responses when the participant reported a change in the interval between stimuli or beats when there was no change. Compute the False-Alarm (FA) rate (i.e., number of FAs/number of no-change stimuli).
   3. Calculate the z-score for the Hits rate and FA rate, using the norminv Matlab function (z-score = norminv(Hits rate or FA rate)). Subtract the z-score for the FA rate from the z-score for the Hits rate to obtain d’.
4. Evaluation of Individual Results:
   NOTE: Compare the performance of a participant to a normative or control group to uncover cases of poor rhythm perception. Regarding the results of the synchronization tasks, perform a corrected t-test using the singlims computer program.
   1. Open the singlims computer program. Enter the mean and SD of d’ and the sample size of the normative or control group. Provide the d’ value for the participant who is to be compared to the normative or control group.
      NOTE: The participant performed significantly poorer than the normative or control group when the two-tailed probability of the corrected t-test is below 0.05.

Wyniki

The tasks described above have been used with success to characterize the timing abilities of individuals without musical training^2,34-36. In a recent representative study on beat-deafness², a group of 99 non-musicians (university students) were screened using two simple synchronization tasks. Participants synchronized their finger tapping with an isochronous sequence and a musical excerpt at a comfortable tempo (with an IOI/IBI of 600 msec). Ten of the participants showed particularly poor synchron...

Dyskusje

The goal of the described method is to provide a set of tasks and analysis strategies to characterize the timing abilities of the majority of individuals and detect cases of beat deafness or poor synchronization. The critical steps of the protocol involve 1) the setup of the instruments used for stimulus presentation and collection of finger tapping data and subjects’ responses, 2) data collection using two sets of tasks (synchronization and rhythm perception), 3) analysis of synchronization data with circular stat...

Materiały

Name	Company	Catalog Number	Comments
Matlab	Mathworks		High-level language and interactive environment for numerical computation, visualization, and programming
MAX MSP	Cycling '74		Software for data acquisition from MIDI-controlled interfaces, and stimulation presentation
Presentation	Neurobehavioral Systems		Software for conducting experiments in experimental psychology. Allows precisely-times stimulus delivery and collection of behavioral responses.
Roland HPD- 10	Roland		Hand percussion pad (MIDI instrument)
EDIROL FA-66	Roland		MIDI interface to connect the MIDI instrument to the computer.