Abstract
Humans fluctuate naturally between two attentional states – characterized in some literature as good "on-task attention" and bad "mind wandering." However, these states can also be characterized in terms of attentional effort available (high vs. low). Here, we ask whether dancers have gained perceptual expertise that enables them to perform complex tasks even with divided attention or while in a "mind-wandering" state. In dance expertise, the effectively infinite movement possibilities dancers face when mimicking/learning dance sequences ensure that changes in required attentional effort are not attributable to proceduralization alone. Experiment 1 tested whether dance expertise transforms dance movement mimicry into a low-effort attention task. In a dual-task paradigm, we asked expert dancers (n=7) to watch and mimic 10s dance videos 1) with no secondary task, 2) while counting randomly presented auditory beeps, or 3) while performing mental math with the beeps. Subjects were recorded with an Xbox One Kinect sensor, and three expert dancers rated the movements' accuracy and timing. The dancers' performance was unaffected by either of the secondary attention tasks. Experiment 2 investigated whether, under naturally fluctuating attention, low-effort attention supports real-time reacting, but not remembering (due to added processing demands that require high-effort attention). Dancers (n=20) watched and mimicked dance videos, which stopped randomly every 40-80s. They indicated their attention state and tried to replicate their last few mimicked movements. Sensor recordings of the mimicked and replicated movements were analyzed using a Kinect machine learning gesture recognition software trained on the recordings from Experiment 1. As hypothesized, memory for the movements suffered significantly during low-effort attention (t=2.35, p=0.02), while real-time mimicry remained unimpaired (t=-0.45, p=0.67). These studies suggest that it may be adaptive (even optimal) to perform low-effort tasks in a mind-wandering state, and b) introduce a new method of studying complex actions with the Kinect.
Meeting abstract presented at VSS 2018