RPPW 98

Under delayed auditory feedback (DAF) speech control is disturbed, that is, speech errors of all types increase, voice pitch and level are enhanced, and extra time is needed to perform a spoken sequence. Speech control and response time are disrupted most by feedback delays of about 200 ms. One of the first attempts to explain disturbances under DAF was made by Lee (1950). He proposed that auditory feedback is essential for continous control of speech production. In contrary to such more traditional explanations of DAF effects, the Displaced-Rhythm Hypothesis" (Howell, Powell, & Khan, 1983) proposes that the disturbances arise from the disruptive effects caused by the rhythm of the delayed signal. This hypothesis implies that the semantics of the delayed speech is of no importance, that is, the content of the (delayed)feedback and the speech don't have to be necessarily identical. We tried to test this strong implication of the Displaced-Rhythm Hypothesis by introducing the following four conditions: Subjects had to perform sequences of 10 syllables each. In a first condition participants repeatedly pronounced "ta-ta" without any accentuation, in a second condition every second syllable had to be stressed. A third and a fourth condition changed the phonology of the syllables by speaking out "ta-ti" with and without stressing the "ta". Each condition consisted of one control trial with usual feedback and four experimental trials with delayed feedback from the fifth syllable onwards.

In a first analysis the number of speech errors occuring under each condition was calculated. Results show a significant influence of changes in phonology on the number of speech errors but no influence of accentuation. In that way, displaced rhythm cannot be the main factor being responsible for the well-known effects of delayed auditory feedback.

EXPERIMENTAL AND COMPUTATIONAL INVESTIGATIONS OF THE PERCEPTION OF RHYTHM IN FRENCH AND ENGLISH SPEECH: ARE CROSS-LINGUISTIC DIFFERENCES PRIMARILY METRICAL IN ORIGIN? Lee, C.S and Todd, N.P.McAngus.

Department of Psychology, University of Manchester

It has long been noted that the world's languages vary considerably in their rhythmic organisation, with different languages privileging different phonological units (mora, syllable, or stress foot) as their basic rhythmic unit. However, attempts to explain how such differences arise at the acoustic/auditory level have so far failed, since they are all predicated on the claim that units at the privileged level are particularly likely to display a tendency toward isochrony (or to be perceived as isochronous): a claim which a variety of studies (mostly of English and French) have failed to substantiate.

Failure to find evidence of isochrony might be taken as casting doubt on a crucial (though unspoken) assumption underlying such attempts: namely, that privileged units acquire their position by becoming metrically prominent. However, although it is true that metricality prototypically implies isochrony, there is ample evidence from musical styles such as Western classical music that listeners regularly infer metre even when there are quite large deviations from strict timing. Could it be, then , that notwithstanding apparently contrary evidence, listeners do extract rhythmic periodicities from speech signals and, more particularly, that privileged units yield the strongest periodicities? Indirect supporting evidence comes from an auditory model developed by Todd and his associates (see Todd and Lee, in press) which contains a mechanism for extracting periodicities >from signals and placing resultant "beats" on the appropriate events. We discuss the performance of the model with spoken sentences of French and English, and report experimental findings which bear on our theoretical claims.

TAPPING TO COMPOUNDS Peter Janker

Zentrum für Allgemeine Sprachwissenschaft,Typologie und Universalienforschung (ZAS)

The p-center, as defined by Marcus, is the psychological moment of occurrenceof a syllable. Its position within the syllable depends on the actual acousticmake-up of that syllable. The set of stimuli for this investigation consists of a click signal and eight disyllabic compounds. They have in pairs the same phonemic structure but the main stress is either on the first or on the second part of the compound:

/bl'u:t#Q"a6m/ (anaemic) /bl"u:t#Q'a6m/ (very poor)

/bl'aU#gr"y:n/ (blue-green) /bl"aU#gr'y:n/ (bluish-green)

/b'aIn#h"art/ (hard as ivory) /b"aIn#h'art/ (tough)

/St'aIn#r"aIC/ (lots of stones)

/St"aIn#r'aIC/ (very rich)

The participants had to perform a synchronisation task by tapping to sequences of binaurally presented disyllabic stimuli. A sequence consisted of 15 repetitions of the same stimulus with an inter stimulus interval of 1400 ms.

Due to the different rhythmic structure the stimuli also differ slightly in their acoustic make-up. The question now is whether or not these differences are sufficient to disambiguate between the two members of a pair.

EXPERIMENTAL AND COMPUTATIONAL INVESTIGATIONS OF BEAT INDUCTION IN A LARGE CORPUS OF SAMPLED MUSICAL PERFORMANCES, INCLUDING THE 48 FUGUE SUBJECTS OF THE WELL-TEMPERED KLAVIER BY J.S.BACH Todd, N.P.McAngus and Cohen, S

Department of Psychology, University of Manchester

In order to give the model described in Todd et al (these proceedings) a thorough testing, we presented it with a large number of samples of real performances of a variety of musical styles, such as jazz, latin, contemporary dance music and classical including the 48 fugue subjects of the Well-Tempered Klavier by J.S. Bach. In parallel we also presented the same samples to a small number of human subjects who were required to tap to the perceived beat. Both model and subjects exhibit a transient phase and a steady state phase. Once the steady state had been reached we extracted four measures of performance: (1) the period of the beat (2) the phase of the beat (3) the length of the transient phase and (4) stability of the steady state. Overall the model compared reasonably well with the human performance. We discuss these results and implications for model improvement.

References Todd, N.P.McAngus, O'Boyle, D.J and Lee, C.S. (these proceedings) Blueprint for a dancing robot: Further developments of a sensory-motor theory of rhythm, time perception and beat induction.

USING A NETWORK OF RELAXATION OSCILLATORS TO TRACK PERIODIC AND QUASI-PERIODIC PULSED INPUT Douglas Eck, Michael Gasser & Robert Port

Cognitive Science and Computer Science, Indiana University

Relaxation oscillators, used to model the dynamics of neural firing (Fitzhugh 1961; Nagumo et. al. 1962; Morris & Lecar 1981), synchronize more quickly than phase pulling oscillators when driven by constant input (Somers and Kopell 1995). I report on the behavior of a network of relaxation oscillators (Huber 1998) when driven by periodic and quasi-periodic pulsed input. I suggest a model of rhythmical pattern tracking which uses such a network to control a robotic arm having mass-spring oscillator characteristics.

EXPERIMENTAL AND COMPUTATIONAL INVESTIGATIONS OF BEAT INDUCTION IN AN ANAPAEST RHYTHM: PREFERRED TEMPO IS PREDICTED BY BIOMECHANICS. O'Boyle, D.J., Todd, N.P.McAngus and Brady, N.

Department of Psychology, University of Manchester

We report some exploratory experiments carried out in order to investigate a radical corollary of the sensory-motor theory (Todd et al, these proceedings). The principal assumption of the theory is that beat induction is mediated by an internal representation, or motor image, of the musculoskeletal system. This implies that the dynamical properties of an individual's motor image will be uniquely determined by the biomechanical properties of that individual's musculoskeletal system. A secondary assumption of the theory is that the particular motion induced in the motor image is locomotory.

According to the 'spinal engine' theory of locomotion, the central role of the spine in the locomotion of lower vertebrates has been retained in human bipedal locomotion. Although the power source for human locomotion is the hip extensors, this power is tranfered back to the spine, via gravity, in the form of a torque, in order to drive hip rotation. The system has two resonant frequencies, unique to an individual, depending upon whether the energy transfer is mediated by muscle (walking) or ligament (running). It is this individually unique resonant frequency, determined by the biomechanical properties of the trunk, which we claim is the origin of preferred tempo and which will determine the metrical level at which the tactus is placed. To put this another way, the way in which we individually 'hear' or 'see' a rhythm is determined by the individually unique physical properties of our upper bodies.

In order to test this claim, we carried out a series of experiments involving four independent sets of measurements: purely sensory measures, both auditory and visual, purely biomechanical measures, and motor measures of locomotion. The sensory measures were obtained by using the method of constant stimuli for auditory clicks and visual flashes whose inter-onset-interval patterns were arranged to be that of an anapaest rhythm. The rhythms were presented at a number of rates, from fast to slow. At the slow rate the anapaest rhythm could be identified with the tune 'Three Blind Mice', at an intermediate rate with the tune 'Jingle Bells' and, at a fast rate, with the rhythm of the 'William Tell Overture'. Subjects were required to make a forced choice of which tune they heard or saw. From the resultant psychometric functions, a value for each subject's perceptually preferred tempo could be reliably obtained. The results of these experiments showed a striking relationship between some biomechanical variables and perceptual measures. We discuss these results, their implications and some computational simulations.

CONCEPTIONS AND MISCONCEPTIONS OF DISCRIMINATING ISOCHRONOUS AND ANISOCHRONOUS SOUND SEQUENCES: A CRASH COURSE IN CLASSIC PSYCHOPHYSICS Gert ten Hoopen, Edwin van Buuringen, Stephanie van den Berg, & Jiska Memelink

University of Leiden

Unfortunately, the starting point of this paper should be a quick recap of my LA ICMPC talk (ten Hoopen, 1992), because only few of the estimated 60 attendants seem to have digested my elementary analysis (but at least Stephen Handel and Bruno Repp did). The message was that the threshold for temporal perturbation in isochronous sound sequences remains constant (approximately 10 ms) when the base interval duration varies from 100 to 250 ms (10 to 4 sounds per second). In other words, it is the absolute limen rather than the relative limen that remains invariant, defying Weber's law in a very systematic sense (see also lizuka & Nakajima, 1996). But from 250 ms on, this law appeared to hold: the limen increased at constant rate (approximately 3 ms per 100 ms). Whereas I based this mainly on the literature, Friberg & Sundberg (1995) independently and empirically established the same pattern of limens as a function of tempo (hurray for Zeitgeist). They found a breakpoint between two discriminatory mechanisms at 250 ms as well. However, their conclusion was not at all:....... that we have maximum sensitivity to time intervals of around 300 ms (Friberg and Sundberg, 1995)......... as Neil Todd wrote in his E-mail of May 23, 1997 to the "Multiple recipients of list Auditory." Anyway, the question (1) how base interval duration globally affects the sensitivity to anisochrony, seems to have been clearly answered. Four other questions form the core of recent "tempo-sensitivity" research: (2) Is the limen dependent on the position of the perturbation in the sequence? Recently, two simple studies on perturbing initial and final intervals were reported. Tanaka et al. (1 994) gave a confirmatory answer: the limen at the end is greater than at the beginning. Ten Hoopen et al. (1996) could not replicate this sensitivity-difference between start and end, but instead they found a significant increase of the PSE of the last interval. (3) Is the threshold for temporal deviation affected by the number of isochronous intervals? Drake & Botte (1993) found evidence for the multiplelook notion, but ten Hoopen et al. (1994) and ten Hoopen et al. (1995) did not. (4) Is the limen differentially affected by speeding up or slowing down the tempo of the isochronous sequence? McAuley & Kidd (1994) and Vos et al. (1997) suggest that such a difference exists. (5) Is the limen dependent on the relational pattern structure and musical skill? (Desain & Honing, 1994, Repp, 1992; Riess Jones & Yee, 1997). We put questions 3 & 4 to a series of psychophysical experiments (the method of constant stimuli, and not only the classic z-, but probit, and logit analyses were done too). We replicated the effect of base duration (invariant absolute DL for fast, invariant relative DL for slow sequences). We found partial support for the appealing multiple-look notion, but we could not find any support for the "direction of tempo-change" notion. Besides yet unpublished results, our story contains a moral: Some researchers tend to confound and sacrifice such simple psychophysical concepts as: POE, PSE, CE, LL, UL, DL, and JND in their craving for presenting a new model.

References Desain, P., & Honing, H. (1994) Does expressive timing in music performance scale proportionally with tempo? Psychological Research, 56, 285-292.

Drake, C., & Botte, M-C. (1993) Tempo sensitivity in auditory sequences: Evidence for a multiple-look model. Perception & Psychophysics, 54, 277-286.

Friberg, A., & Sundberg, J. (1 995) Time discrimination in a monotonic, isochronous sequence. Journal of the Acoustical Society ofamerica, 98, 2524-253 1.

Iizuka,T., & Nakajima, Y.(1996) On the perception of an empty duration preceded by an isochronous context. (in Japanese but with an English abstract ). Trans. Tech. Com. Psycho. Physio. Acoust., H-96-102, 1-7.

McAuley, J.D. & Kidd, G.(1994) Differential sensitivity to increases and decreases in tempo: Evidence for an entrainment model. Journal of the Acoustical Society ofamerica, 96, 3257.

Repp, B.K.(1992) Probing the cognitive representation of musical time: Structural constraints on the perception of timing perturbations. Cognition, 44, 241-281.

Riess Jones, M. & Yee, W.(1997) Sensitivity to time change: The role of context and skill. Journal of Experimental Psychology: Human Perception and Performance, 23, 693 -709.

Tanaka, M., Tsuzaki, M. & Kato, H.(1994) Discrimination of empty duration in the click sequence simulating a mora structure. The Journal of the Acoustical Society of Japan (E), 15, 191-192.

ten Hoopen, G. (1 992, February) Temporal processing of fast auditory patterns, Paper presented at the 2nd International Conference on Music Perception and Cognition, Los Angeles, CA.

ten Hoopen, G., Boclaarts, L., Gruisen, A., Apon, I., Donders, K., Mul, N. & Akerboom, S. (1994) The detection of anisochrony in monaural and interaural sound sequences, Perception & Psychophysics, 56, 1 10120.

ten Hoopen, G., Hartsuiker, R., Sasaki, T., Nakajima, Y., Tanaka, M. & Tsumura, T. (1995) Perception of auditory isochrony: time-shrinldng and temporal patterns. Perception, 24, 577-593.

ten Hoopen, G., Beumer, M. & Nakajima, Y. (1 996) What differs between the first and last interval of a click sequence simulating a mora structure, the DT or the PSE?: A replication of Tanaka, Tsuzaki, and Kato (1994). The Journal of the Acoustical Society ofjapan (E), 17, 155-158.

Vos, P. G., van Assen, M., & Fraiiek, M. (1 997) Perceived tempo change is dependent on base tempo and direction of change: Evidence for a generalized version of Schulze's (1978) internal beat model. Psychological Research, 59, 240-247

TIME-SHRINKING AND GESTALT PRINCIPLES Yoshitaka Nakajima, Daigoh Suetomi

Kyusu Institute of Design

When an auditorily presented empty duration shorter than about 0.5 s and another one whose physical duration is a little longer neighbor each other, the second empty duration is often underestimated. This illusory phenomenon, which we call 'time-shrinking', gives an insight into the fundamental mechanism of the perception of short time intervals. We employed patterns consisting of three, instead of two, neighboring empty durations. Previous studies had shown that time-shrinking could take place also in such cases. Our present aim was to compare the influences of the first and the second duration upon the perception of the last duration, and to examine whether and how the first and the second duration interact. The results indicated that the influence of the second time interval was greater than that of the first time interval. The effect of the first duration often seemed indirect. Gestalt principles could be invoked to describe the interaction between these time intervals.

TIME DISCRIMINATION: BEYOND WEBER'S LAW Devin McAuley

The Ohio State University

The effect of temporal context on time discrimination has received increasing attention in the recent rhythm perception literature, with most research contrasting two distinct theoretical approaches. Interval-based models account for context effects in time discrimination in terms of the statistics of the time intervals comprising a pattern, trial, or session, whereas entrainment models have emphasized the rhythmic structure of patterns within a trial, and those that comprise the experimental session. Entrainment models consider how stimulus events align with or deviate from the perceivers temporal expectations, and make predictions about time sensitivity based on the magnitude of temporal deviations relative to an established rhythm.

This talk reviews a range of factors that have been shown to influence tempo discrimination of isochronous tone sequences. The review is in three parts. First, the question of whether time perception obeys Weber's Law is revisited in relation to recent data demonstrating faster/slower asymmetries in discrimination. Second, the rhythmic structure of a trial is considered by examining within-trial manipulations that influence tempo sensitivity. Third, trial-to-trial uncertainty and carry-over effects are examined in relation to new data contrasting the predictions of interval and entrainment models. It is argued that entrainment models provide the most consistent explanation of time discrimination given current data.

SPATIAL ATTENTION DEFICITS AND THE PERCEPTION OF TIME Sophie Scott

MRC Applied Psychology Unit

Does a disorder of spatial attention lead to a distortion of temporal perception? We investigated whether patients with right hemisphere parietal lesions, who showed left sided extinction, made systematic deviations in their judgements of what is 'even' in sound sequences presented to alternate ears. This was compared to their monaural performance, and to that of control participants. The patient data indicate that a 100-200ms shift away from physical regularity is required for binaural presentations to be perceived as 'even', or 'regular' in their rhythm. Such distortions are not present in monaural presentations where each beat occurs within the same attended auditory space. The result demonstrates the temporal distortions which ensue from lateralised attentional impairment within the auditory domain. It provides further evidence for the importance of normal attention in the perception of events in time In addition, due to the relatively long intervals between events in the regularity judgement task, it describes a methodology whereby the effects of perceptual distortions may be disambiguated from competition for limited capacity between temporally simultaneous or adjacent events.

PERCEPTION OF TIMING PERTURBATIONS IN REAL TUNES: ROLE OF LEFT TEMPORAL LOBE STRUCTURES Ehrlé, N., Samson S. & Baulac M.

University Pitié-SalpÍtriËre Hospital, Paris

This study examines the contribution of temporal lobe structures to the detection of inter-onset interval increments in real monodic tunes. Using Repp's paradigm to explore the expressive timing microstructure, a set of familiar tunes whith systematic temporal extensions and respects was determined from pianistic performances. These results were then correlated with the expectations of normal subjects. The detection abilities of epileptic patients with either left (LTR) or right (RTR) temporal resection were compared with those of normal control (NC) subjects in a task involving the discrimination of temporal increments. Performances were measured for two increments (easy versus difficult to detect) using a forced-choice paradigm (changed versus unchanged melody). The results showed that patients with LTR but not RTR were significantly impaired as compared to NC subjects in the increment detection (p<0001), all groups showing significant differences between the two levels of difficulty. This finding suggests the contribution of left temporal lobe structures in perceiving local perturbations in real auditory sequences whereas they do not demonstrate any trouble in top-down processing.

HIERARCHICAL MODELS OF RHYTHMIC TIMING Ralf Krampe

University of Potsdam

In a series of experiments we assessed participants' timing performance on production tasks varying in complexity. The tasks used comprised simple (i.e., isochronous) repetitive tapping as well as rhythmic tasks with simple or complex ratios (including polyrhythms). From the interindividual difference perspective, we tested participants differing in their expertise level and also compared young and older individuals. Using a variant of the rhythm programming hypothesis proposed by Vorberg & Wing we propose a hierarchical model which includes the effects of rate fluctuation and error propagation when rhythms are performed at different tempos. For highly trained experts performaing polyrhythmic tasks at rapid tempos, we show that the assumption of integrated timing based on a single central clock can not be aligned with the data and that different, parallel modes of timing must be considered. With respect to aging, our data point to specific problems in serial order control which are way beyond potential age-related declines in the accuracy of internal clock mechanisms. These serial order control problems reveal an attentional component which becomes most crucial when rhythmic patterns must be switched within a timing task.

TEMPORAL CONTROL OF REPETITIVE MOVEMENTS IN THE SYNCHRONIZATION TASK: AN EMPIRICAL TEST OF THE WING-KRISTOFFERSSON MODEL Frank Miedreich, Gisa Aschersleben, Jˆrg Gehrke

Max-Planck-Institut für Psychologische Forschung

Wing and Kristofferson (1973) have introduced a model for the analysis of the tapping task in a continuation paradigm. In this paradigm subjects have the task to execute a simple motor task in a regular isochronic beat. Typically the task starts with a short synchronization phase.

The model postulates two independent processes, one for the generation of intervals, another for motor control. To use the model in the analyses of synchronization tasks it was extended by Vorberg and Wing (1994).

The application of the model was investigated in a series of experiments. The task always started with a synchronization phase followed by an equally long continuation phase. The interval length, feedback and effector was varied in three designs. Interval length varied between 300 and 1000 ms, feedback was either with ot without acoustic feedback, and effector was either the index finger or the foot.

The results are discussed in the context of existing models for synchronization.

Vorberg, D. & Wing, A. (1994) Modelle für Variabilit‰t und Abh‰ngigkeit bei der zeitlichen Steuerung. In H. Heuer & S.W. Keele (Eds.), Psychomotorik. Enzyklp‰die der Psychologie, 3, 181-262.

Wing, A. & Kristofferson, A.B. (1973) Response delays and the timing of discrete motor responses. Perception and Psychophysics, 14, 5-12.

SYNCHRONIZATION ERROR: AN ERROR IN TIME PERCEPTION ? Andreas Wohlschläger

Ludwig-Maximilians-Universitaet Muenchen

When synchronizing simple movements (taps) with regular auditory signals (clicks), the tap leads over the click (negative asynchrony). Based on the idea that the emptiness of the interval between two clicks (ISI) might distort time perception, we investigated whether structuring the ISI has an influence on synchronization performance. In Experiment 1 we added a rhythm to the pacing signal. In Experiment 2 five additional clicks were added per ISI at random points in time. In both experiments asynchrony was reduced. After turning off the pacing signal, tapping was slower in the random click condition than in the standard task. Hence, structuring the ISI reduced asynchrony, and this effect is obviously based on the improved perception of elapsed time.

Experiment 3 and 4 investigated the synchronization performance of different effectors and showed the smallest asynchrony for the finger, followed by elbow, knee, and toe. A Fourier analysis showed significant harmonic components of a 2nd, 3rd, 4th, and 5th order for the finger-movements, but not for the other effectors. In Experiment 4, subjects performed more complex movements (1 or 2 contact-free, inbetween taps). Asynchrony was reduced the more complex the movements were. In summary, results showed that structuring the ISI by adding auditory signals or limb movements improves time perception and leads to a reduction of the asynchrony.

COMPENSATION FOR SUBLIMINAL TIMING PERTURBATIONS IN PERCEPTUAL-MOTOR SYNCHRONIZATION Bruno Repp

Haskins Laboratories

It has been suggested in the literature that the temporal discrimination threshold (>3% of the baseline interval) and especially the temporal order threshold (>15 ms) limits the accuracy of error correction in perceptual-motor synchronization tasks (Michon, 1967; Mates, 1994; Vorberg & Wing, 1996). Experiment 1 uses "pulse" perturbations (phase shifts) in metronomic click sequences and, by averaging over a number of trials, demonstrates that reliable compensation in finger tapping can occur for perturbations as small as 1-2% of the baseline interval (500 ms). Experiment 2 shows that compensation for similarly small deviations is observed when participants tap in synchrony with music. However, unlike the detectability of (larger) temporal irregularities, which varies greatly with position in the music (Repp, 1992, in press-a), the error correction process seems to be insensitive to musical structure. And whereas the conscious detection of larger perturbations in metronomically timed music is inhibited by previous exposure to expressively timed music (Repp, in press-b), Experiment 3 shows that the error correction mechanism is immune to such pre-exposure. These results suggest that the subconscious processes involved in perceptual-motor synchronization have access to timing information that is considerably more accurate and less susceptible to contextual influences than the timing information available to conscious auditory perception. Nevertheless, musical structure does modulate the magnitude of the synchronization error (Exps 2 & 3; Repp, submitted), so that the unconscious detection of and corrective response to temporal perturbations in music takes place within a nonstationary response process.

FACTORS INFLUENCING PRECISENESS OF DOUBLING AND HALVING OF TEMPO Marek Franek, Jiri Mates

Czech Academy of Sciences

The paper is devoted to an investigation of musicians' skill to make tempo change in exactly defined temporal ratios. Results of the series of experiments concerning tempo doubling and halving are presented. The skill of tempo changes is studied in the context of a specific temporal domain. Michon (1964), Ellis (1991), and Drake et al., (1992) showed that the threshold for discrimination of a tempo change varied in dependency on a speed range of tempo. In the four experiments we studied the ability of amateur musicians t execute change of a tempo of their finger tapping in temporal ratio 2:1 (doubling) or 1:2 (halving) under tempi with duration of beat interval ranging from 400 to 1500 msec.

The results showed that doubling and halving of a tempo was not performed exactly. The accuracy of doubling/halving of a tempo was affected by the speed of the base tempo. The most accurate responses occurred in tempi with beat intervals between 500 - 600 ms. In faster tempi the difference from expected values corresponding to 2:1 (1:2) ratio decreased with the increase of a speed of the tempo, while in slower tempi the trend was the opposite.

Further experiments showed that not only a speed of a base tempo, but also the range of presentation of tempi in the experimental session might be the factor influencing preciseness of the performance. If the session started with the slowest tempo, the deviation from the 2:1 ratio was smaller in the course of whole experiment than in the case, when the session started with the fastest tempo. The phenomenon might be explained in the terms of specific "calibration" of a pacemaker of an internal clock. It might be that the pacemaker is "calibrated" according to the speed of the first beginning tempo.

TEMPORAL-TRACKING PERFORMANCE: EFFECTS OF MANIPULATIONS OF MODALITY OF METRONOME, OF MODALITY OF SENSORY FEEDBACK, AND OF ATTENTIONAL PROCESSES. O'Boyle, D.J., Al-Attar, Z. & Cody, F.W.J.

University of Manchester

The source of the small mean negative synchronization error during normal temporal tracking of a regular auditory metronome has recently been the subject of a good deal of interest. We are particularly interested in this issue as we have observed that patients with Parkinson's disease tap significantly further in advance of the beat than do neurologically-intact controls (O'Boyle et al., 1995; O'Boyle, 1997). In this paper, we describe the results from several experiments in which we have addressed the question of the source of the synchronization error in normal subjects by manipulating, i) the modality (auditory, visual) of the metronome and of the sensory feedback available to the subject, ii) the site of delivery on the body of an electrical cutaneous metronome, and iii) the degree to which subjects orient and attend, during tracking, to the tapping hand or to the contralateral hand to which a vibrotactile metronome is delivered.

O'Boyle, D.J. (1997). On the human neuropsychology of timing of simple, repetitive movements. In: Bradshaw, C.M. & Szabadi, E. (Eds.), Time and behaviour. Psychological and neuro-behavioural analyses (pp. 459-515). Amsterdam: Elsevier Science. O'Boyle, D.J., Freeman, J.S. & Cody, F.W.J. (1995). Sensorimotor synchronization during temporal tracking is impaired in subjects with Parkinson's disease (PD). Journal of Physiology, 485.P: 15P.

INTERNAL CLOCKS VERSUS NONLINEAR OSCILLATORS? TOWARDS AN INTEGRATED MODEL Lieke Peper, Peter Beek and Andreas Daffertshofer

Free University, Amsterdam

So far, polyrhythmic tapping has been studied from two theoretical perspectives, timekeeper models and nonlinear oscillator models. The former approach has concentrated on patterns of temporal covariance, and the latter approach has concentrated on pattern (in)stability and spontaneous transitions between patterns. Although timekeeper models with linear feedback loops have been developed that can account for stable tapping in synchrony with one or two metronomes, nonlinear feedback ('coupling') terms are needed to account for multistability, loss of stability followed by transitions between patterns, chaos and hysteresis during both paced and unpaced tapping. On the other hand, although models of nonlinearly coupled oscillators exhibit all of these properties, they can only reproduce the observed patterns of temporal covariance if two (hierarchically ordered) noise sources are incoporated in the model. Thus, as it stands, neither timekeeper models nor nonlinear oscillator models provide a comprehensive account for the full phenomenology of polyrhythmic tapping. To resolve this situation, a comprehensive model for polyrhythmic tapping is proposed that combines the strengths of the two types of models, that is, two hierarchically ordered noise sources (associated with timekeeping and the actual production of movement, respectively) and a nonlinear couplings between two postulated neural oscillators ('timekeepers').

NONLINEAR STOCHASTIC MODELING OF RHYTHMIC MOVEMENTS Ralf Engbert

University of Potsdam

Most studies on cognitive processes of the temporal control of simple movements can be divided into two different classes: The first class the so-called mental time-keeper models, dominates the literature in cognitive psychology. From the mathematical point of view, thes models are linear and stochastic. Corresponding research focuses, therefore, on stochastic variability in timing. In the dynamical framework, which is the second approach, cognitive control of movements is investigated in terms of dynamical systems theory which concentrates on the qualitative analysis of behaviour. In both cases, corresponding models can be used to make specific predictions on a variety of experimental paradigms. Because of the differing domains of the two approaches, i.e. stochasticity versus dynamics, they seem to be fundamentally different. In contrast to this we show that the two approaches do not exclude each other and even have to be combined in order to derive realistic models of movement control. In the case of rhythmic movements this leads us to a stochastic model of timing processes with nonlinear feedback control. As a result, the variability in the control of movements is a consequence of the interaction between stochastic timing components and dynamical processes.

R. Engbert, C. Scheffczyk, R.T. Krampe, M. Rosenblum, J. Kurths & R. Kliegl: Tempo-induced transitions in polyrhythmic hand movements. Physical Review E 56, (1997) 5823-33.

SYMBOLIC DYNAMICS OF BIMANUAL RHYTHMIC MOVEMENTS Giovanna Guasti

University of Potsdam

In our contribution we present a quantitative approach to the complexity in cognitive processes. In particular we analyse experiments of bimanual production of polyrhythmic and isochronous movements. The investigation of symbolic sequences is a general framework for the analysis of complexity, in particular for information processing in natural systems. These methods have been developed in the theory of complex dynamical systems and are well-adapted for the analysis of natural dynamical processes. We show that a coarse-graining of experimental data can be used to extract important structures generated by the underlying dynamical processes. In the case of short, noisy and nonstationary time series, typically obtained in psychological experiments, we develop measures for the characterization of the performance on the level of individual subjects. Furthermore, the technique of symbolic codings is useful for the comparison of experimental data with numerical simulations of corresponding models.

R. Engbert, C. Scheffczyk, R.T. Krampe, J. Kurths and R. Kliegl, "Symbolic dynamics of bimanual production of polyrhythms". In: H. Kantz, J. Kurths and G. Mayer-Kress (eds.) Nonlin ear analysis of physiological data. (Springer, Berlin-Heidelberg-New York, 1998).

P. Tass, J. Kurths, M. G. Rosenblum, G. Guasti, and H. Hefter, "Delay Induced Transition in Visually Guided Movements", Physical Review E, 54(3), pp.R2224-R2227, 1996

LEARNING DYNAMICS IN TAPPING MULTIFREQUENCY PATTERNS Chris Jansen, Lieke Peper & Peter Beek

Free University, Amsterdam

Dynamical systems theory has shifted the focus of research on motor learning from investigating task improvement to studying changes in the system dynamics. This study investigates the degree to which relative phase (RP) (deGuzman & Kelso, 1991) and frequency ratio (FR) (Haken et al. 1996) are crucial for the performance in multifrequency tapping. Two groups of four subjects learned to produce two multifrequency patterns. The first pattern for group A was a FR of 2:5 with RPs of 0� and 180�, for group B it was a FR of 2:3 with RPs of 120� and 300�. The second pattern was the same for both groups: a FR of 2:5 with RPs of 120� and 300�. If group A would outperform group B in learning the second task, FR would be more crucial than RP in producing multifrequency patterns. If the performance of group B would be better on the second task, RP would be more important than FR. The dynamics of the system during the learning process were monitored by scanning the attractor landscape of RP and FR. It was assumed that changes in the attractor landscape would indicate to what extent the system as a whole was affected by learning. However, no differential transfer effects between the two learning scenarios were found.

deGuzman, G.C., & Kelso. J.A.S. (1991). Multifrequency behavioral patterns and the phase attractive circle map. Biological Cybernetics, 64, 485-495.

Haken, H., Peper, C.E., Beek, P.J., & Daffertshofer, A. (1996). A model for phase transitions in human hand movements during multifrequency tapping. Physica D, 90, 179-196.

LEARNING AND TRANSFER OF MULTIFREQUENCY PATTERNS Jeffrey Summers

University of Southern Queensland

In studies of multifrequency coordination from a dynamical perspective, typically participants are required to maintain a steady beat with one hand while tracking with the other hand a frequency scaled pacing signal (e.g., Peper, et al., 1995). Thus, within each frequency scaled trial the participant 'passes through' a range of frequency ratios (e.g., 1:2, 2:3, 2:5,3:5 etc). Of particular interest in these studies has been the width of the resonance zones at various coupling strengths. In contrast, researchers >from a cognitive perspective have tended to examine the 'intentional' learning/production of a single polyrhythm at a time, That is, the same pattern is presented/produced repetitively within each trial and special instructions regarding the structure of the pattern are frequently given to participants (e.g., Summers, Ford & Todd, 1993). The present paper reports the results of work-in-progress examining the question of what is learned, in terms of changes to the underlying coordination dynamics and transfer performance, in 'incidental' and 'intentional' research paradigms.

References Peper, C. E., Beek, P.J., & van Wieringen, P.C.W. (1995). Multifrequency coordination in bimanual tapping: Asymmetrical coupling and signs of supercriticality. Journal of Experimental Psychology: Human Perception and Performance. 21, 1117-1138.

Summers, J.J., Ford, S.K., & Todd, J.A. (1993). Practice effects on the coordination of the two hands in a bimanual tapping task. Human Movement Science, 12, 111-133.

TIME JUDGMENTS IN RHYTHMIC CONTEXTS Ralph Barnes and Mari Riess Jones

The Ohio State University

We report some tests of an oscillator model of dynamic attending. The case for attending having a rhythmic aspect is more challenging than it is for motor gestures, in part because attention is not directly observable. We consider "Can an approach that posits internal (nonobservable) oscillators actually explain attention to temporal events such as speech and music?"

The answer comes in three steps. First we observe that a hallmark of attending is selectivity. We assume that attending can not only operate selectively in space, as many attention theories propose, but it can also operate selectively in time. Attending can be targeted to one temporal location and not another. Research that supports this view is cited. It illustrates that it is necessary to provide a temporal context in which oscillatory attending in somehow controlled in a predictable manner by the rhythmical structure of an ongoing pattern. In such contexts, certain temporal locations in the unfolding sequence are more likely than others to become future targets of attending.

Secondly, an oscillatory mechanism must be stipulated that can respond to the time structure, namely the tempo and rhythm, of an environmental event in a "meaningful way". By meaningful way, I mean that one (or more) oscillator(s) must be able to describe attentional selectivity in time, but not so rigidly that the system becomes unresponsive to naturally occuring changes so characteristic of acoustical patterns we deal with. In music, for instance, rubato, fine temporal nuances, playing ahead or behind the beat all convey additional meaning and are part of the overall expressive content of musical meaning. These properties defy a rigidly periodic account of an attentional process. In short, the oscillator must be able to adapt or follow modulations in rate or rhythm to a certain degree while at the same time realizing useful selectivity in the targeting of attending. That models incorporating nonlinear oscillators can meet this criterion is evident in several recent nonlinear oscillator models, including an elegant one developed by Edward Large. In these models, attending is viewed as an oscillatory process with an adaptive phase and period which can "lock into" and follow i.e., entrain to corresponding periodicities within an acoustical event. Thus, attending is heightened at some points in an oscillator's period and not others, and this period can itself adapt as a sequence speeds up or slows down.

Finally, the postulated activities of an attentional oscillator must be connected to measureable behaviors that reflect attending. We report the results of several experiments in which listeners are required to make comparative judgments about time intervals presented in different rhythmic contexts. These experiments show that judgments are best for time intervals that are rhythmically expected.

INFORMATION PROCESSING IN THE CENTRAL EXECUTIVE: EFFECTS OF CONCURRENT TEMPORAL PRODUCTION AND MEMORY UPDATING TASKS Scott W. Brown & C. Tigg Frieh

University of Southern Maine

The present research was designed to test the hypothesis that timing is a function controlled by the Central Executive component of working memory. Subjects performed both temporal and nontemporal tasks separately as well as concurrently. The temporal task required subjects to generate a continuous series of 5-sec temporal productions by pressing a key on a computer-linked mouse device. The nontemporal task was memory updating, a task previously associated with the Central Executive. In this task, a series of random digits was presented sequentially on a computer screen; the sequence was stopped at unpredictable times and subjects were required to report either the 3 most recently presented digits (easy condition) or the 5 most recently presented digits (difficult condition). A comparison of single-task versus dual-task conditions showed that the memory task interfered with timing by making production responses more variable. Likewise, the timing task disrupted memory performance in the easy (3 digit) condition by lowering recall scores. Memory performance was not affected by the concurrent timing task in the difficult (5 digit) condition, possibly because of floor effects. Mutual interference between the timing and memory updating tasks implies that both tasks may rely on the same set of attentional resources. The results are consistent with the idea that timing processes are handled by the Central Executive.

THE PERCEPTION OF ILLUSORY TEMPO CHANGES IN MUSICAL COMPOSITIONS Marilyn Boltz

Haverford College

This research describes two experiments that investigate a set of factors that may influence the perceived rate of an auditory event. In a paired-comparison task, subjects were presented with a set of musical patterns that differed in their relative number of contour changes and magnitude of pitch skips (Experiment 1) as well as the compatibility of rhythmic accent structure with the arrangement of pitch relations (Experiment 2). Results indicated that relative to their standard referents, comparison melodies were judged to unfold more slowly when they displayed more changes in pitch direction, greater pitch distances, and an incompatible rhythmic accent structure. These findings are suggested to stem from an "imputed velocity hypothesis" in which people overgeneralize certain invariant relations that typically occur between melodic and temporal accent structure within Western music.

RHYTHMIC ASPECTS OF VIBRATO Peter Desain, Henkjan Honing, Rinus Aarts & Renee Timmers

NICI, University of Nijmegen

Vibrato, the periodic fluctuations of pitch, amplitude and/or timbre, is an important expressive component of musical performances, especially for singers and string players. Next to measuring its characteristics in isolation we believe it is important to relate it to other temporal aspects of performance (global tempo, expressive timing, rhythmic structure), especially since we have discovered that some intriguing hypotheses about such relationships exist in the literature and in the intuitive knowledge of musicians.

In this paper we will focus mainly on pitch-vibrato. We will describe an experimental methodology (including ways to extract fundamental pitch from audio data) which was developed to check various hypotheses. Experiments were conducted using a fragment of the "Cygne" piece of Saint-Saens for five different instruments (violin, cello, tenor, theremin, oboe) in different tempi. The questions addressed are the following: Is vibrato rate adapted to tempo; is vibrato phase adapted to the timing of the end of the note; how is the vibrato contour adapted to note duration; are vibrato cycles somehow synchronized to the pulse at a small metrical level? Results will be presented that lead to answers to these questions.

GRACEFUL TIMING: ORNAMENTS, TEMPO AND MUSICAL STRUCTURE Luke Windsor, Peter Desain, Henkjan Honing, Rinus Aarts, Hank Heijink and Renee Timmers

NICI, University of Nijmegen

Earlier work on the timing of grace notes in music (Desain and Honing, 1994) has suggested that their durations are far from being proportionally related to global tempo. This work also showed that the relationship between their duration and global tempo might interact with the local rhythmic context and the differing functions of particular types of grace notes. Such results seem to go against the notion that expressive timing is relationally invariant across different global tempi for all types of musical events.

This earlier work, however, only analysed data at three tempi and across three repetitions at each tempo. The study reported here manipulates the overall tempo of a performance of Beethoven's variations on a Paisiello theme with a much finer grain and over a larger number of repetitions. The timing of the grace notes is analysed with respect to their different functions (e.g. appogiatura; acciaccatura), the local rhythmic context, and global performance tempo. In conclusion, some general principles are suggested for modeling the different relationships between grace note duration, musical structure, and global tempo.

References Desain, P. & Honing, H. (1994). Does expressive timing in music performance scale proportionally with tempo? Psychological Research, 56, 285-292.

DO LISTENERS USE TIMING PATTERNS TO DECODE THE EMOTIONAL EXPRESSIONS OF MUSIC PERFORMANCES? Guy Madison & Patrik N. Juslin

Uppsala University

A large number of studies have indicated that timing patterns tend to reflect the structure of the music (cf. Clarke, 1996). But are timing patterns also capable of mediating other communicative contents? Recent studies have shown that performers are able to communicate iconic representations of specific emotions to listeners, and that tempo, timbre, articulation, and loudness were used as expressive cues (Juslin, 1997). However, the timing patterns within a performance were not considered, and the question we pose is to what extent listeners use these to decode intended emotions. In one experiment we removed the timing patterns from performances made with different expressive intentions (e.g. "sad") to see whether this would impair the decoding accuracy of listeners. In a second experiment we removed all cues but the timing patterns, to see whether the listeners would be able to decode the intended emotion solely on the basis of the timing pattern. The results are discussed in terms of their implications for research on music performance and emotional communication.

Key words: music performance, timing, emotional communication, music listening, expressive timing

THE PERCEPTION OF GROUPING STRUCTURE INDUCED BY EXPRESSIVE TIMING Amandine Penel and Carolyn Drake

Laboratoire de Psychologie Expérimentale

Previous work (Penel & Drake, in press) suggested that a large part of expressive timing may be related to the hierarchical grouping structure of music. We have previously proposed a psychological segmentation model of expressive timing in which we claim that performers implement a hierarchical grouping organisation corresponding to the hierarchical grouping structure of the music they play. This organisation would be communicated to the listeners through expressive timing, by means of group-final lengthening, at each level in the hierarchy. In the excerpt chosen (the first 8 measures of the Traumerei and their repeat), four hierarchical levels could be highlighted.

We tested the perceptual implications of this model: either a global or local lengthening in music performance should indicate the end of a group. The question we investigated was whether this is perceived by listeners. We chose the lowest level in the hierarchy (notes grouped within basic units), because the ends of phrases are probably perceived by listeners, whatever the expressive timing. We selected 6 performances from the previous study, which differed in the number and position of lengthenings produced within the group of 13 eighth notes of the first phrase. These performances were equalized in articulation (everything was played legato) and intensity. Musician and non musician listeners were asked to indicate their perceptions of grouping within this group, for all these performances, and also for a perfectly mechanical performance. Results indicate that basic grouping is first influenced by the musical structure, which is very clear in our excerpt, but also by expressive timing.

References: Penel, A. and Drake, C. (in press). Sources of timing variations in music performance: a psychological segmentation model. Psychological Research.

COMPLEXITY MEASURES OF MUSICAL RHYTHMS Ilya Shmulevich & Dirk-Jan Povel

NICI, University of Nijmegen

The study of the factors that determine the complexity of rhythms has theoretical as well as practical relevance. On the theoretical side, a successful measure of complexity will shed light on the encoding and memory processes involved in rhythm perception. Hence, it will be able to indicate whether a given rhythm will be easy or difficult to memorize and reproduce. On the practical side, such a measure will complement current research in rhythm similarity and music pattern recognition in general.

In this paper, three measures of complexity of musical rhythms are compared. The first measure is based on the work of Tanguiane (1993), and uses the idea that a rhythmic pattern can be described in terms of (elaborations of) more simple patterns, simultaneously at different levels. The second measure is based on the complexity measure for finite sequences proposed by Lempel and Ziv (1976), which is related to the number of steps in a self-delimiting production process by which such a sequence is presumed to be generated. The third measure is rooted in the theoretical framework of rhythm perception discussed in Povel and Essens (1985).

The predictive strength of the three measures has been evaluated in an experiment in which listeners judged the complexity of a set of selected rhythms on a 5-point scale. Results of the experiment will be reported and the strengths and weaknesses of the complexity measures as well as their relationship to current work in rhythm similarity will be discussed.

TAPPING IN TIME WITH MECHANICAL AND MUSICAL SEQUENCES Carolyn Drake, Amandine Penel and Luca Stefan

Laboratoire de Psychologie Expérimentale, Paris

Tapping in time with music is a relatively easy task for most people. It involves the extracting of temporal regularities in the sequence in order to identify the underlying beat (beat extraction). In this paper we examine listeners' abilities to tap in time with musical excerpts depending on the temporal micro-regularity of the sequences. The musical sequences were played with three types of temporal micro-structure: 1) mechanical, that is by computer with all note durations exactly proportional as indicated in the score and all notes of identical intensity, 2) accented mechanical, the mechanical versions with intensity accents added to events falling on the first beat of each measure, and 3) musical, played by a concert pianist with all temporal, intensity and pedal variations. Synchronisation models predict that the best synchronisation performances should be observed for the mechanical and accented mechanical performances because they had the lowest level of temporal incertitude. Cognitive models of music performance however would predict better synchronisation for the musical performances because the various performance variations (timing, intensity, articulation) highlight important structural aspects and transmit a more structural representation to the listener, thus facilitating the process of beat extraction.

Two groups of subjects (musicians and non-musicians) first familiarised themselves with one of the three versions of each of six musical excerpts (Bach, Scarletti, Beethoven, Chopin, Bartok, Ravel). They were then asked to synchronise with it by tapping with a drum on a drum pad. First they synchronised at the most 'natural' rate, then faster and faster still, and then slower and slower still.

The results indicated that, for non-musicians, synchronisation was more accurate for the mechanical and accented mechanical versions than the musical versions. However, for musicians, synchronisation was more accurate and especially more elaborate (taps at more hierarchical levels) for the musical versions than for the mechanical and accented mechanical versions. This interaction suggests that surface temporal regularity facilitates the detection of the beat and measure level for subjects who have little practice with the type of sequence whereas performance variations do enable listeners who are familiar with the language (musicians) to create a more elaborate mental representation for the sequence, and therefore to better adapt their taps to the heard sequence. This more elaborate mental representation also allows more complex and varied synchronisations.

BLUEPRINT FOR A DANCING ROBOT: FURTHER DEVELOPMENTS OF A SENSORY-MOTOR THEORY OF RHYTHM, TIME PERCEPTION AND BEAT INDUCTION. Todd, N.P.McAngus, O'Boyle, D.J and Lee, C.S.

Department of Psychology, University of Manchester

In this paper, we develop further a sensory-motor theory of rhythm (Todd and Lee, 1998), time perception and beat induction. Using the theory, we attempt to account for these phenomena by the interaction of sensory systems, which represent temporal information in the cortex in terms of the spatio-temporal power spectrum of the sensory image (Todd, 1998), and the motor system, which has certain natural frequencies. The central assumption of the sensory-motor theory is that beat induction is not a passive process but, rather, a form of sensory-guided action involving all of the sensory and motor components that that entails, i.e. almost the entire nervous system, including primary and secondary sensory cortices, multi-modal sensory cortex, the cerebellum, basal ganglia and primary and secondary motor cortices, and the musculoskeletal system which it controls. Even if the musculoskeletal system ('the plant', to use the jargon of control theory) is not activated, i.e. there is no motor output, the higher supraspinal levels of the motor system ('constituting the controller') are. To be more specific, the principal agent mediating beat induction is the internal representation of the musculoskeletal system and its dynamic properties, which the 'controller' requires as a 'feedforward' model. This theory is implemented in the form of a computational model which takes sound signals as input and synchronises a locomotory motion to simulate beat induction.

References Todd, N.P.McAngus. (1998) A model of auditory image flow I: Architecture. Proc. 16th International Congress of Acoustics. Seattle, June, 1998.

Todd, N.P.McAngus, Lee, C.S. and O'Boyle, D.J. (1998) A sensory-motor theory of rhythm, time perception and beat induction. J. New Music Research.