Grounding Early Time Perception in Motor Activity – GETPIMA

The goal of this project is to investigate the role of embodiment – specifically, infant motor activity – in the early development of time perception. Typically, adults from all cultures and backgrounds can accurately anticipate short-term events, can catch objects thrown to them, and can perform the enormous range of human activities that require timing. This universality may seem to imply that the sense of time does not rely on learning, a position implicitly taken by the well-known pacemaker-accumulator models and multiple-oscillator/coincidence-detection models of time perception. But there are problems with these models, among them, the problem of resetting the system for every event for which a time judgment might later be required. Recently, the reset problem has been avoided by suggesting that time perception depends on memory-trace decay. We will propose a novel memory-trace model of the early development time perception in which learning and coordinated motor activity in infancy play a key role. This hypothesis is explored through a combination of behavioral studies of infant time keeping and computational modeling.

Specifically, the aims of the projects are to:
• develop a novel memory-based computational model of time-perception
• explore the developmental trajectory and implications of this model for understanding emerging time perception
• verify empirically the key hypothesis of the model – i.e., that time-perception in infants is learned behavior driven by coordinated motor activity
• explore the relation between infant motor activity and infant timing accuracy.
• develop more advanced statistical methods to analyze infant EMG data, thus facilitating the use of EMGs in future developmental studies

Empirical work:
Each of four research streams will explore a facet of how motor-activity manipulations affect infants’ time perception. The infant experiments will use a “missed-beat” time-assessment methodology where a highly salient auditory-visual stimulus appears then disappears. This will be repeated at fixed intervals and will be skipped every nth interval. Infants’ responses to the missed event will be the measure of their anticipation of that event – and, therefore, of their encoding of the inter-stimulus interval.

Advanced EMG data analysis
We will measure anticipative postural adjustment EMG signals, which appear before the onset of visible arm movements, thus giving a better and more objective assessment of movement onset responses than with movement observation. However, infant EMGs tend to be noisy, which will require developing novel techniques to analyze the data. Discriminating the EMG time series for infant groups with regularly timed stimuli from those with randomly timed stimuli implies that the former group perceived the time-intervals.

Computational modeling:
Underlying our computational model is an unexpected observation: that the distributions of neurons in the cortex, of time-perception errors and of motor-activity errors are all scale-free (i.e., at all scales of observation, there is the same distribution of items, errors, etc.). We will design a learning- and memory-based computational model of infant time perception based on scale-free neuronal organization, use (simulated) motor-activity to calibrate it, and test it empirically. We believe it will reproduce a number of standard, empirically observed time-perception markers in infants.