Skilled human activity often involves precise temporal coordination at multiple levels. For example, in musical ensembles, performers coordinate their own body parts when manipulating instruments in order to produce sounds that are coordinated with the sounds of other performers. The intricate interlocking of instrumental parts in Balinese Gamelan music (especially in the kotekan style of playing) illustrates superbly how adept humans can become at such multilevel coordination. Here, and in more common activities such as dancing to music, there are two primary levels of coordination: (1) intra-personal
– between an individual’s own body parts (e.g., two hands or feet), and (2) inter-agent
– between one’s own actions and externally controlled events (e.g., another’s actions or the effects thereof).1
Keller and Repp (2004)
investigated simultaneous intra-personal and inter-agent coordination using a task that requires the left and right hands to move in alternation (intra-personal coordination) to produce finger taps at the midpoint between the tones of an auditory metronome (inter-agent coordination) (see ). Thus, antiphase coordination is required at both levels. This coordination mode, dubbed Alternating Bimanual Syncopation (ABS), proved to be relatively challenging. The variability of tap timing was far greater for ABS than for any of the other coordination modes that were examined, including unimanual syncopation where taps are made (with either the left or right hand) after every metronome tone, unimanual ‘skip’ syncopation where taps are made after every other tone, bimanual synchronization with alternating hands, unimanual synchronization (tapping with every tone), and unimanual skip synchronization (tapping with every other tone). The relative difficulty of ABS was more pronounced at moderately fast movement rates than at comfortable rates.
Schematic illustration of the intra-personal and inter-agent components of alternating bimanual syncopation.
Borrowing from the dynamical systems approach to movement coordination, Keller and Repp (2004)
concluded that the difficulty with ABS derives from competition between multiple attractors in the dynamical landscape wherein simultaneous intra-personal and inter-agent coupling takes place. According to this view, each of the two coupling collectives—inter-personal (i.e., hand-hand, or action-action) and intra-agent (metronome-hand, or perception-action)—harbours an antiphase attractor and an in-phase attractor. Note that in the case of the inter-agent collective, antiphase coordination is syncopation and in-phase coordination is synchronization. It is known from other work that antiphase attractors are weaker than in-phase attractors, and that perception–action coupling is weaker than action–action coupling (e.g., Byblow, Chua, & Goodman, 1995
; Chua & Weeks, 1997
; Haken, Kelso, & Bunz, 1985
; Schmidt, Bienvenu, Fitzpatrick, & Amazeen, 1998
). Thus, there are asymmetries both within and between the two collectives involved in ABS.
The asymmetries within collectives compromise the stability of ABS to the extent that antiphase movements are drawn towards the in-phase mode. Attraction to the in-phase mode is stronger at fast rates than at comfortable rates because perception-action and action-action coupling grow weaker with increasing movement frequency, and the effects of this weakening on stability are greater for antiphase than for in-phase coordination (see Kelso, 1995
; Schmidt et al., 1998
; Turvey, 1990
). At critically fast rates, transitions from the antiphase mode to the in-phase mode are often observed in intra-personal coordination and in inter-agent coordination (e.g., Kelso, 1984
; Fraisse & Ehrlich, 1955
; Kelso, DelColle, & Schöner, 1990
; Schmidt et al., 1998
). Such phase transitions are usually preceded by increased variability in movement timing (Kelso, Scholz, & Schöner, 1986
; Schmidt, Carello, & Turvey, 1990
). In ABS, the potential for transitions from antiphase to in-phase coordination may be relatively high because attraction takes place at the intra-personal and inter-agent levels simultaneously. Although phase transitions can be avoided by the intention to maintain double antiphase relations—at least by musically trained individuals tapping at the moderate rates tested by Keller and Repp (2004)
—it may be effortful to do so. The high movement timing variability for ABS observed by Keller and Repp (2004)
suggests that their musician participants were teetering on the edge of the boundary between antiphase and in-phase coordination.
The second type of asymmetry that characterises ABS—between intra-personal and inter-agent collectives—may compromise stability by influencing the allocation of attentional resources (Keller & Repp, 2004
). ABS can be viewed as a dual task wherein both components (intra-personal and inter-agent antiphase coordination) require attention (see Mayville, Jantzen, Fuchs, Steinberg, & Kelso, 2002
; Temprado & Laurent, 2004
; Temprado, Zanone, Monno, & Laurent, 1999
). The intra-personal (alternating hands) component of ABS may divert attentional resources from the relatively demanding task of maintaining antiphase coordination with the metronome. Inter-agent coupling may consequently weaken, making coordination accuracy at this level more variable from cycle to cycle. Such attentional constraints may apply in real-word instances of simultaneous intra-personal and inter-agent coupling when, for example, an instrumentalist falls out of sync with the rest of the ensemble due to excessive attention to fingering, and dancers lose the music (or their partners) due to excessive attention to their own footwork.
The current study investigated how multilevel coordination stability is influenced by the way in which the goals of the task are represented by the actor. The main aim of our experiment was to test whether coordination stability during ABS is affected by the degree to which the inter-agent and intra-personal components of ABS can be integrated into a unified task-goal representation. This was manipulated through the use of supplementary auditory feedback in the form of tones triggered by taps.
Previous work on intra-personal coordination has shown that bimanual coordination stability is influenced by the degree to which tones triggered by separate hands can be integrated into a single perceptual stream. For example, in a study of 2:3 polyrhythm production (i.e., repeatedly making two isochronous taps with one hand in the time that it takes to make three taps with the other hand), Summers, Todd, and Kim (1993)
found that timing was more accurate when the pitch separation between tones produced by each hand was narrow than when it was wide. Given that sequences of tones close in pitch are more likely to be perceived as a single stream than sequences of tones separated by large pitch differences (see Bregman, 1990
), the results of Summers et al. suggest that bimanual coordination is more stable when the two hands are controlled with reference to integrated rather than independent movement goals (also see Klapp, Hill, Tyler, Martin, Jagacinski, & Jones, 1985
; Jagacinski, Marshburn, Klapp, & Jones, 1988
; Summers, 2002
). This phenomenon generalizes across tasks and modalities, as benefits of integrated goal representations have also been observed in studies addressing the influence of transformed visual feedback on the stability of intra-personal coordination (e.g., Bogaerts, Buekers, Zaal, & Swinnen, 2003
; Mechsner, Kerzel, Knoblich, & Prinz, 2001
) and inter-agent coordination (e.g., Roerdink, Peper, & Beek, 2005
In the current study, it was hypothesized that the addition of supplementary auditory feedback during ABS would simplify the perceptual goals of the task if the feedback tones and pacing signal tones can be integrated into a single Gestalt. We assumed that such integration would encourage the actor to adopt a common (auditory) perceptual locus of control across intra-personal and inter-agent levels of coordination, and that this perceptual locus of control would stabilise performance by reducing the impact of coupling constraints that implicate the motor system, and/or by discouraging the disproportionate allocation of attention to the alternating hands. Thus, integrated goal representations may redress the asymmetries both within and between the intra-personal and inter-agent collectives involved in ABS.
A strong version of the integrated goal representation hypothesis predicts that multilevel coordination stability should be greatest when feedback tones and pacing tones are indistinguishable, because under such conditions a single perceptual stream is guaranteed. However, the benefits of having a simple perceptual goal may in this case be tempered by structural incongruence between the auditory goal representation (consisting of a series of identical elements) and the motor program that drives movement (consisting of a series of disparate elements: left-hand taps, right-hand taps, and, presumably, intervening ‘rests’ that coincide with pacing signal events2
). Research addressing the production of musical sequences has shown that timing accuracy can be affected by whether or not structural relations between movements and their auditory effects are compatible (e.g., Keller & Koch, 2006
; Pfordresher, 2003
). Such findings suggest that, during ABS, the benefits of integrated goal representations may enter into a trade-off with factors such as the degree of structural congruence between goal representations and their associated movement patterns.
To test the above hypotheses, coordination stability during ABS was compared across conditions that varied in terms of (1) whether or not finger taps triggered tones (i.e., presence of supplementary auditory feedback), and (2) the relationship between tone pitches produced by the two hands and the pacing sequence. It was assumed that integrated goal representations would be encouraged to the extent that feedback tones and pacing sequence tones are perceived as a single auditory stream, and that such perceptual organisation would arise when the tones are close in terms of pitch height. To gauge the effects of these feedback manipulations, coordination stability was indexed using an objective measure of the fastest at rate at which ABS could be performed accurately. This ‘rate threshold’ measure, determined using a simple adaptive staircase procedure, yields information about the critical rate at which participants’ intentions are no longer effective in preventing transitions from antiphase to in-phase coordination (or total loss of coordination).
If integrated goal representations facilitate coordination stability, then ABS thresholds should indicate faster critical rates when the pitch separation between feedback tones and pacing tones is small than when it is large. Whether such benefits extend to the case where feedback and pacing tones are identical was a question of interest. The perception of a single stream (and, by implication, a perfectly integrated goal) is unavoidable under such circumstances. However, this situation is characterised by structural incongruence between the auditory goal representation (a sequence of uniform elements) and the required movement pattern (a ‘left-rest-right-rest’ varying sequence).
We examined the role of structural congruence by varying the assignment of low and high pitches to the two hands in the conditions where feedback tones were distinctive from pacing tones. Feedback pitches could either be the same or different across hands. Structural congruence was assumed to be highest when a different pitch is associated with each hand, because in this case the auditory goal representation and the movement pattern both consist of a set of three distinct elements. Structural congruence was manipulated further in the conditions where the two hands were associated with different pitches by varying the assignment of low and high tones to the left and right hands. Congruence was assumed to be higher when the left hand produced low tones and the right hand produced high tones (which is compatible with the key-to-tone mapping on a piano keyboard) than with the reverse (incompatible) arrangement (see Lidji, Kolinsky, Lochy, Karnas, & Morais, 2007
; Repp & Knoblich, 2007a
; Rusconi, Kwan, Giordano, Umiltá, & Butterworth, 2006