Storage of an oculomotor motion aftereffect

Abstract

Adaptation to motion produces a motion aftereffect (MAE), where illusory, oppositely-directed motion is perceived when viewing a stationary image. A common hypothesis for motion adaptation is that it reflects an imbalance of activity caused by neuronal fatigue. However, the perceptual MAE exhibits storage, in that the MAE appears even after a prolonged period of darkness is interposed between the adapting stimulus and the test, suggesting that fatigue cannot explain the perceptual MAE. We asked whether neural fatigue was a viable explanation for the oculomotor MAE (OMAE) by testing if the OMAE exhibits storage. Human observers were adapted with moving, random-dot cinematograms. Following adaptation, they generated an oculomotor MAE (OMAE), with both pursuit and saccadic components. The OMAE occurred in the presence of a visual test stimulus, but not in the dark. When the test stimulus was introduced after the dark period, the OMAE reappeared, analogous to perceptual MAE storage. The results suggest that fatigue cannot explain the OMAE, and that visual stimulation is necessary to elicit it. We propose a model in which adaptation recalibrates the motion-processing network by adjusting the weights of the inputs to neurons in the middle-temporal (MT) area.

Figure 1

Schematic representation of the adaptation protocol for experiments without and with a gap period. Each block of trials began with 60 sec adaptation (a static spot was provided for fixation). a) When there was no gap, a 1.5 sec RDC pursuit stimulus moving orthogonal to the adaptation stimulus was next presented. Each trial was immediately followed by 10 sec of top-up adaptation. b) In the gap condition, a blank, dark screen (1 sec) was presented immediately following the initial and top-up adaptation epochs, followed by a 1 sec pursuit stimulus. The pursuit stimulus in all conditions moved in one of five directions spaced every 10 deg from −20 to +20 deg, centered about rightward (0 deg). In control trials, the stimulus presentation was the same except the adapt stimuli were static.

Figure 2

Eye movement data for the no-gap condition. a) Mean vertical velocity difference plotted as a function of time for three observers. Smooth eye velocity showed a deflection in the downward direction (~200–800 msec), opposite the adapt direction. The downward velocity is suggestive of an oculomotor MAE. b) 2-dimensional position plot of pursuit and saccadic eye movements made to rightward-moving (0 deg) targets either after adaptation to a static RDC (control) or an RDC moving upward (adapt) for one observer (HY). For clarity, only the first eight trials of each condition are shown.

Figure 3

Summary of the OMAE eye movements. a) Median regression slope values for each target direction for both control (dashed lines) and adapt (solid lines) conditions in the no-gap experiment 200–800 msec after target onset (early period) for all three observers. Note that with few exceptions overall, for all target directions, adapt trials produced larger downward slopes for all observers. b) Median regression slope values for each test direction 800–1200 msec after target onset (late period). Note that the control and adapt trials now show similar direction eye movements. c) The difference in median slope (adapt - control) of eye position traces averaged over all target directions in the early and late periods. Note that all observers’ eye movements exhibited a downward bias indicative of an oculomotor MAE early but that the effect dissipates quickly, within about 1 sec of pursuing the visual target motion.

Figure 4

Eye movement data for the gap condition. a) Mean vertical eye velocity difference (adapt - control) plotted as a function of time for three observers. During the gap, smooth eye velocity initially showed an upward deflection that was in the same direction as the adapt stimulus. When the rightward-moving test stimulus was presented (post-gap), eye velocity abruptly reversed to downward, opposite the adapt direction, again suggestive of an OMAE. Panels b and c show position plots of the first eight control and adapt trials for one observer (HY). Note different axis ranges. b) A 2-dimensional position plot of pursuit and saccadic eye movements made during the gap (following upward adaptation) when the screen was blank. The eyes appear to move randomly, showing little difference between control and adapt trials. c) A 2-dimensional position plot of pursuit and saccadic eye movements for control and adapt trials made to rightward-moving (0 deg) targets presented after a 1.0 sec gap. Note that as in the no-gap condition, the eyes show a clear downward bias relative to controls after the test stimulus was presented.

Figure 5

Comparison of gap and post-gap eye movements. a) Median regression slope values for each test direction for both control (dashed lines) and adapt (solid lines) conditions for the gap condition for all three observers for the time period 200–800 msec after extinction of the adapt stimulus. Note that eye movement data in adapt and control trials have similar slopes. b) Median regression slope values for the post-gap period, 200–800 msec after visual target onset. Here, the eyes show a downward deflection consistent with an OMAE, demonstrating storage of the adaptation effect. c) The difference in median slope (adapt - control) of eye position traces averaged over all target directions during the gap and post-gap periods. All observers’ eye movements exhibited a downward bias indicative of an oculomotor MAE with the visual test stimulus.

Figure 6

Schematics of a model for the oculomotor MAE showing responses at various levels. a) The state of the model in the dark, following adaptation to upward motion. Four example motion detectors representing cardinal directions of motion are not activated without visual stimulation (small grey arrows). The synapse that the upward unit makes with MT has been changed by adaptation (smaller, light grey circle; larger medium grey circles, other synapses). Because the motion detectors are not active, the effect of the synaptic adjustment is not realized in the MT neurons. When their output is summed, the processed motion signal is zero and the eyes do not move. b) The model during rightward test stimulation. The rightward motion detector is activated, as well as the up and down ones to a lesser degree. Now, the upward MT unit is less activated than the downward one, resulting in a downward bias of the processed visual motion signal and the resultant eye movement.