Effects of divided attention and sensorimotor synchronization on detection of timing perturbations in auditory sequences Bruno H. Repp Haskins Laboratories, New Haven, CT Rutgers University, Newark, NJ Peter E. Keller Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany

Phase shifts vs. tempo changes Tempo changes are easier to detect than phase shifts (e.g., Friberg & Sundberg, 1995). Why?

Hypothesis 1 Detection is based on phase discrepancies between temporal predictions of an internal oscillator and sequence events (Large & Jones, 1999; McAuley & Jones, 2003).

Hypothesis 2 Detection is based on interval comparisons. A remembered (standard) interval duration is compared with the sequence inter-onset intervals (“multiple looks”; Drake & Botte, 1993).

Hypothesis 3 Phase-based and interval-based processes may be operating simultaneously. (Unparsimonious but perhaps realistic.)

Attentional requirements Attention may be required for any kind of conscious detection. Interval-based processes may require more attention than processes that are based on time points (i.e., relative phase). In sensorimotor synchronization, period correction is reduced when attention is diverted by a secondary task, whereas phase correction is unaffected (Repp & Keller, 2004). The same may hold for period and phase correction of an internal oscillator.

Question and predictions What will be the effect of adding a secondary task to a detection task (phase changes and tempo changes), so that attentional resources are diverted from detection? (1) Detection performance should be poorer overall. (2) If detection is based only on phase discrepancies, detection of tempo changes should improve relative to detection of phase shifts because period correction of the internal oscillator is inhibited, which leads to even larger phase discrepancies following tempo changes. Overall performance should suffer relatively little because phase-based processes require little attention.

Predictions (continued) (3) If detection is based only on interval comparisons, detection of tempo changes should deteriorate relative to detection of phase shifts because interval comparisons require attention. Overall performance also should suffer. (4) If detection is based on both phase and interval discrepancies, the two effects may cancel, so that there may be little change in the detection of tempo changes relative to that of phase shifts. Overall performance should suffer only moderately.

Materials ∆t = ±5, ±10, ±15, ±20, ±25, ±30 ms 10 blocks of 30 trials (10 tones) each: 12 phase shift, 12 tempo change, 6 no change trials

Conditions Sessions 1 & 3: Detection task only Sessions 2 & 4: Detection task + arithmetic task Arithmetic task: Mental addition of 8 digits (1, 2, 3) displayed in synchrony with tones 3-10 Within each session: Tapping along in synchrony with the tones in half the blocks (odd or even); sitting still in the other half

Purpose of tapping If tapping is controlled by the same internal oscillator as detection, then tapping is a way of externalizing the oscillator’s predictions and making them observable. Tapping provides external phase discrepancies (asynchronies) on which detection could be based. That may enhance detection performance (if internal and external phase discrepancies provide independent information) or it may not (if external phase discrepancies are redundant with internal ones).

Procedure Participants: 7 “master tappers” (musically trained) Phase shifts and tempo changes were explained. Participants gave a “speeded up”, “slowed down”, or “no change” judgment at the end of each trial. They were not asked to report which kind of change had occurred. After giving their judgment, they entered the result of their arithmetic calculation (in Sessions 2 and 4). They were asked to focus their attention on the digits but also to report any changes they heard in the sequences and not to neglect their tapping either.

Results: Arithmetic task Average score: 55% correct (range 29%–84%) Chance = ?? Most participants were probably well above chance.

Results: Detection performance (without arithmetic) Phase shift Tempo change blue = “speeded up”; green = “no change”; red = “slowed down”

Results: Effect of arithmetic on detection performance Phase shift Tempo change solid lines = without arithmetic; dotted lines = with arithmetic

Results: Effect of arithmetic on detection performance (p < .11) This result seems most compatible with the hypothesis that detection is based primarily on phase discrepancies, perhaps with occasional additional interval comparisons.

R & K (2004) vs. present results for tempo changes Repp & Keller (2004) (1,2) Present results (1,2,3) Smaller effect here, despite more complex arithmetic.

Results: Effect of tapping on detection performance (n.s.) Asynchronies between taps and tones did not serve as perceptual cues to phase shifts or tempo changes. (They were either redundant or were ignored.)

Results: Adaptation of tapping to phase shifts and tempo changes 1.0 = perfect adaptation

Summary of tapping results Adaptation to phase shifts and tempo changes was similar. Overcorrection suggests involvement of period correction in both cases. (2) The arithmetic task reduced adaptation in both cases (p < .05), presumably by reducing period correction. (3) Because asynchronies were similar for phase changes and tempo changes, it seems that the better detection of tempo changes derived either from internal phase discrepancies that differed from the observable asynchronies or from interval comparisons.

Conclusions Arithmetic interfered less with detection than it did in Repp & Keller (2004), even though it was more complex. This could indicate that inclusion of phase changes led to a greater reliance on internal phase discrepancies, whereas interval comparisons played a greater role in R&K (tempo changes only). (2) The small effect of arithmetic on detection is consistent with detection being based primarily on phase discrepancies. (3) However, asynchronies between taps and tones did not contribute to detection performance.(Recently replicated.) (4) This redundancy may suggest that tapping is controlled by the same internal oscillator that governs temporal expectancies. However, the external (= internal?) asynchronies are not sufficient to explain the superior detection of tempo changes. Perhaps that is still due to additional interval comparisons.