Short-Term Monocular Deprivation Enhances Physiological Pupillary Oscillations

Abstract

Short-term monocular deprivation alters visual perception in adult humans, increasing the dominance of the deprived eye, for example, as measured with binocular rivalry. This form of plasticity may depend upon the inhibition/excitation balance in the visual cortex. Recent work suggests that cortical excitability is reliably tracked by dilations and constrictions of the pupils of the eyes. Here, we ask whether monocular deprivation produces a systematic change of pupil behavior, as measured at rest, that is independent of the change of visual perception. During periods of minimal sensory stimulation (in the dark) and task requirements (minimizing body and gaze movements), slow pupil oscillations, "hippus," spontaneously appear. We find that hippus amplitude increases after monocular deprivation, with larger hippus changes in participants showing larger ocular dominance changes (measured by binocular rivalry). This tight correlation suggests that a single latent variable explains both the change of ocular dominance and hippus. We speculate that the neurotransmitter norepinephrine may be implicated in this phenomenon, given its important role in both plasticity and pupil control. On the practical side, our results indicate that measuring the pupil hippus (a simple and short procedure) provides a sensitive index of the change of ocular dominance induced by short-term monocular deprivation, hence a proxy for plasticity.

Figure 1

Diagram of the experimental procedure. Baseline binocular rivalry dynamics and pupillary measurements were obtained before deprivation (2 × 180 sec binocular rivalry blocks, 2 minutes of pupil measurement in total darkness). Short-term monocular deprivation was achieved by having observers wear a translucent eye-patch over the dominant eye for 2 hours. Immediately after eye-patch removal, 4 × 180 sec binocular rivalry blocks were acquired within a temporal interval of 18 minutes. Binocular rivalry tests were followed by 2 minutes of pupillary measurement in the dark.

Figure 2

Preprocessing of pupillometry data. (a) Segment of pupil recording showing three artifacts: two typical blink artifacts and a short dilation artifact (eye-lashes disturbance). All three were removed by our preprocessing algorithm (black line). For comparison, the result of applying a more standard blink removal algorithm is shown by the red line. (b) Full preprocessed trace, same block from which the segment in (a) was extracted. (c) FFT spectrum of the pupil recording in (c), highlighting the frequency bands used in the main analysis. For this trace, the PUI value was 3.07.

Figure 3

Binocular rivalry results. (a) Scatter plot of the individual subjects' mean phase durations for the deprived and nondeprived eye obtained before (light grey symbols) and during the first 8 minutes after monocular deprivation (black symbols). (b) Deprivation Index (see (1) in Methods) summarizing the increase in deprived eye-predominance for the first 8 minutes after eye-patch removal and for the interval between 10 and 18 minutes after deprivation. The Deprivation Index value of 1 (designated by the dashed line) would indicate no change in ocular dominance after deprivation; values smaller than one indicate increased deprived eye-predominance. Error bars represent 1 ± s.e.m.; asterisks represent statistical significance (^∗∗∗p < 0.001,  ^∗p < 0.05).

Figure 4

Pupillometry results and correlation with binocular rivalry. (a) Scatter plot of the FFT energy in the 0–0.8 Hz range, measuring the amplitude of the pupil hippus, before deprivation versus 20 minutes after deprivation. The text inset shows the result of a paired t-test comparing values on the x- and y-axis. Similar values are obtained using the alternative preprocessing algorithm (blink removal only: t-test comparing before/after hippus amplitude: t(9) = 6.743,  p < 0.001). (b) Difference of FFT energy in the hippus range across deprivation, plotted against the binocular rivalry Deprivation Index (see (1)). The text inset shows Spearman's rank correlation coefficient Rho between values on the x- and y-axis (note that Spearman's Rho is insensitive to whether the axes are logarithmic or linear). The thick red line is the best-fit linear function through the data points. Each symbol is one subject, consistent across panels (a)-(b) and with Figure 3(a).