Abstract
Purpose: To characterise the temporal tuning of the motion aftereffect (MAE) using temporally filtered dynamic random noise (DRN).
Methods: Adaptation to orthogonal bivectorial motion was used to produce the MAEs, whose direction was measured. Adapting speeds were either: 1.5 & 6, 1.5 & 24, or 6 & 24 deg/s. Each speed pair was tested with DRN filtered into one of 5 octave-width passbands with centre frequencies: 1.1, 2.1, 4.3, 8.5 & 17 Hz. MAE directions were then re-measured using DRN interleaved with a static noise pattern. Finally, MAEs were measured on a static-only noise pattern.
Results: Dynamic only: MAE direction for the 4-octave speed difference (1.5/24 deg/s) changed smoothly from opposite the slower vector to opposite the faster vector as test temporal frequency increased. The 2-octave speed pairs also varied smoothly with test frequency but over smaller directional ranges. Static only: MAE direction for static test patterns was directly opposite the slower vector for the 1.5/24 & 6/24 deg/s pairs, and opposite the vector sum for the 1.5/6 deg/s pair. Dynamic+static: Interleaving a static component with the temporally filtered DRN had little or no effect on MAE direction.
Conclusions The asymmetries in the dynamic-only conditions fit a two-channel model of temporal processing, comprising a low-pass and a high, bandpass channel. Dynamic MAEs seem to reflect the combined adaptation in both channels. The static MAE behaves quite differently and does not appear to interact with the dynamic MAE, being possibly mediated by mechanisms in the form pathway.