In Exp.1, we measured shifts in perceived phase occurring with brief exposure to specific amounts of vertical coma (Fig.4). We used AO to correct all optical aberrations and induce vertical coma–a HOA resulting in asymmetric blur that specifically affects horizontally-oriented spatial information. We computed the expected contrast reduction and phase shift in the presence of specific amounts of vertical coma from optical theory (see Methods). Specifically, contrast reduction and phase disruption are characterized by the Modulation Transfer Function (MTF) and the Phase Transfer Function (PTF), respectively. First, we predicted shifts in perceived phase (PSEs) for various amounts of vertical coma by simulating retinal images using Fourier propagation (Fig.4B). As expected, larger shifts in relative phase were predicted for larger magnitudes of vertical coma and for higher SFs, which were in opposite directions for positive and negative vertical coma (Fig.4B). We selected a range of vertical coma values for testing (±0.5µm, for f=3 cycles/deg.) that could be effectively evaluated using the compound appearance task (see Methods). We validated the predictions for these conditions directly from the AO visual display by capturing images of horizontal gratings under various amounts of AO-induced vertical coma. The phase shifts we obtained from the display matched the predictions from optical theory (Fig.4C).

Three participants with typical eyes judged compound grating stimuli (f: 3 cycles/deg) under various amounts of AO-induced vertical coma (Fig.4D-F). Under full AO correction, PSE estimates were centered on 0º (Fig.4D; -0.003º [-3.8º +3.4º]). In the presence of vertical coma, participants experienced shifts in perceived phase that matched optical theory predictions for mild-to-moderate amounts of vertical coma, both in terms of the magnitude and direction of the shift (Fig.4D,F). However, PSE shifts were attenuated compared to optical theory predictions and AO display validation for ±0.5µm of vertical coma (predicted: ±52.3º; perceived: -42.82º [-38.9º -46.4º] and +36.77º [+32.8º +39.9º], for +0.5 and -0.5 µm, respectively) (Fig.4D-F).

What could explain the mismatch between observed and expected phase shifts for larger amounts of vertical coma? We know that we can measure PSE shifts up to ±60º using high-contrast stimuli (Fig.S1). Moreover, we validated that there was no phase deviation in our AO display that could explain the deviations observed with perceived phase shifts (Fig.4C). We hypothesized that this mismatch might therefore reflect the impact of blur-induced contrast reduction on perceived phase. Although we maintained the contrast ratio between the f and 2f components in the presence of vertical coma (see Methods), reductions in contrast at ±0.5µm of vertical coma might be too severe for the 2f component and affect the appearance of compound gratings, thus attenuating the magnitude of PSE shifts.

We directly tested this hypothesis by measuring PSEs under AO correction using compound stimuli physically-adjusted both in terms of the amplitude and phase of the f and 2f components (i.e., using MTF and PTF values for ±0.5µm of vertical coma). Relative to the actual offset added to these MTF-adjusted stimuli (±52.3º), participants showed attenuated PSE shifts (Fig.4E; -42.02º [-36.5º -45.1º] and +41.3º [+38º +46.1º], for ±52.3º phase offset) that matched with the phase shifts observed under ±0.5µm of vertical coma (−42.82º [-38.9º -46.4º] and +36.77º [+32.8º +39.9º] for +0.5 and -0.5µm, respectively). In other words, contrast reductions can affect the magnitude of perceived phase shifts. The results were consistent across individual participants (see Fig.4F for individual and group average PSE estimates). Furthermore, using higher-SF stimuli (f: 4 cycles/deg; Fig.S2) led to a stronger attenuation of PSE shifts, which was observed for vertical coma values above 0.3µm. These findings show that brief exposure to vertical coma results in perceived phase shifts that match predictions from optical theory but can be attenuated by contrast reductions.

In Exp.2, we assessed whether the visual system can adaptively compensate for the impact of vertical coma on phase perception over relatively short periods of time (Fig.5). To do so, we measured PSEs before, during and following short-term (∼1h) exposure to -0.4µm of vertical coma. We selected this value based on Exp.1 to maximize our chance to observe adaptation effects while minimizing the impact of blur-induced contrast reductions. Each experimental session was divided into 3 periods during which participants continuously performed the phase appearance task (Fig.5A): baseline (∼8min), blur adaptation (∼70min), and post-adaptation (∼12min). Baseline and post-adaptation periods were similar: participants judged the appearance of compound gratings presented under AO correction (PSE_predicted: 0º). During adaptation, participants were exposed to a sustained amount of AO-induced vertical coma (−0.4µm; PSE_predicted: +29.5º). The first 3 adaptation blocks (∼3min) consisted of compound gratings only to accurately estimate the blur-induced PSE shift. Then, each compound grating was sandwiched between a series of high-contrast grayscale images consisting of complex natural images and checkerboard patterns (Fig.5B,C). These images appeared blurry owing to the presence of vertical coma, providing invariants the visual system can use to detect and adapt for the impact of blur (e.g., hard edges should have all SFs in sine phase). Adaptation stimuli encompassed a wide range of visual properties and varied rapidly to homogenize local light adaption and minimize afterimages (Elliott et al., 2011, 2007; Sawides et al., 2011a), ensuring adaptation to the presence of vertical coma rather than to specific image properties. To better assess PSE changes over time, we computed PSEs by binning trials from adjacent (±1) blocks, except if the trials belonged to different time periods (see Methods, and Fig.S3 for results using unbinned data).

Our results provide compelling evidence of neural adaptation to the phase spectra during short-term exposure to vertical coma (Fig.5D-F). Before blur exposure, PSEs were near 0º, with a slight positive bias on average during baseline (+3.93º±0.81º; Fig.5D). To treat 0º as reference point, we subtracted the mean of PSE estimates during baseline from all PSE estimates (Fig.5E). Inducing vertical coma resulted in a significant PSE shift relative to 0º baseline (+26.04º [+18º +33.8º]), which matched the prediction from optical theory (PSE_predicted: +29.5º). Critically, the magnitude of the blur-induced phase shift decreased during blur exposure, by -12.7º by the end of the adaptation period (last block: +13.34º [+6.2º +20.8º]). This attenuation over time was well captured by a power function (R²_adj =.77, p<.001), corresponding to ∼47% reduction by the end of adaptation. In addition, when we removed AO-induced vertical coma (post-adaptation), PSEs did not return to baseline levels immediately. Instead, an aftereffect was observed in the opposite direction (−14.72º [-21.8º -8.0º]), which represents a hallmark of visual adaptation. PSEs returned to baseline level after ∼3-4min (5^th post-adaptation block: -3.18º [-8.8º +2.7º]). Importantly, this pattern of results did not depend on whether PSE estimates were obtained from binned (Fig.5) or unbinned data (Fig.S3). All participants showed an attenuation of PSE shifts during adaptation (by -12.1º±6.4º on average; Fig.S3), with 3 out of 4 participants showing an aftereffect (−13.3±7.6º).

In a control condition, we ruled out confounding factors such as changes in observers’ response bias over time due to fatigue or serial dependency effects, as well as low-level effects of the adaptation stimuli presented during the adaptation period. To do so, we maintained AO correction during the entire control experiment, thus removing any blur-induced phase disruptions the visual system could adapt to. Instead, during the adaptation period, a phase offset of similar magnitude and direction than with -0.4µm of vertical coma (+29.5°) was added to compound stimuli. In the absence of vertical coma, adaptation stimuli should be perceived in-focus and should not induce blur adaptation, whereas the phase offset should result in a change in the observer’s response pattern. As expected, we initially observed a significant PSE shift relative to baseline that matched the phase offset added to the stimuli (+30.4º [+23.4º +37.7º]), matching the added phase offset (Fig.5G-I). However, the PSE shift remained stable during the entire exposure period (last block: 28.86º [+21.7º +35.3º]), returning immediately to baseline once the phase offset was removed (+1.36º [-5.82º +8.1º]). This pattern was clear across both participants (Fig.S4). Thus, the attenuation of perceived phase shifts during short-term blur exposure, and the subsequent aftereffect, likely reflect neural adaptation to phase spectra in the presence of vertical coma, rather than non-perceptual changes in response bias or low-level interactions during blur adaptation.

Finally, we assessed the effects of long-term exposure (months to years) to optical aberrations on phase perception. To do so, we measured perceived phase in patients with keratoconus (KC)(Fig.6A). KC patients are chronically exposed to large amounts of uncorrected HOAs in their daily life, with particularly large amounts of vertical coma. Long-term exposure to poor optical quality induces pronounced neural compensatory adjustments in KC patients. For instance, when tested under full AO correction, KC patients show poorer acuity and altered contrast sensitivity than healthy participants tested under similar, AO-corrected optical conditions (Barbot et al., 2021; Hastings et al., 2021; Ng et al., 2022; Rouger et al., 2010; Sabesan et al., 2017; Sabesan and Yoon, 2009). Here, we tested whether KC patients show evidence of altered phase perception under fully-corrected optical quality, which would suggest long-term neural compensatory adjustments to phase spectra.

To do so, we measured perceived phase under AO correction in participants with typical eyes (N=11) and patients with mild-to-severe KC (N=6)(Fig.S5). For each participant, we estimated the habitual optical quality as the total RMS error (tRMS) and higher-order aberration RMS error (hRMS) obtained from wavefront measurements (see Methods). Relative to participants with typical eyes (tRMS: 1.06±0.31µm; hRMS: 0.4±0.13µm), KC patients were exposed to poorer retinal image quality in their daily life (tRMS: 4.04±2.32µm; hRMS: 2.47±1.33µm). As shown in previous studies(Barbot et al., 2021; Hastings et al., 2021; Ng et al., 2022; Rouger et al., 2010; Sabesan et al., 2017; Sabesan and Yoon, 2009), AO correction allowed us to bypass optical factors and maintain total RMS around 0.05µm, with no difference between typical (0.051±0.01) and KC (0.058±0.01) eyes (Welch’s t-test: t=1.22, p=.260, Cohen d=0.66) (Fig.S6A). This allowed us to measure PSE estimates for each participant under similar, AO-corrected optical quality (Fig.6 and Fig.S5).

Under fully-corrected optical quality, participants with healthy eyes had PSEs distributed around 0º (Fig.6B; +1.17º±3.07º; Wilcoxon test: V=42, p=0.465, d=0.27). In contrast, KC patients exhibited clear shifts in perceived phase away from 0º under similar AO-corrected optical quality (Fig.6B), resulting in larger variance in the distribution of PSE estimates between the two groups (Levene’s test for equality of variance: F=39.17, p<0.001). Relative to the magnitude of absolute PSE shifts observed in participants with typical eyes (2.04º±2.52º), KC eyes exhibited significantly larger PSE shifts (14.86º±7.26º)(Welch’s t-test: t=4.19, p=.007, Cohen d=2.36). It is worth re-emphasizing that both control participants and KC patients were tested under similar aberration-free optical quality, for which PSEs are expected to remain around 0º. Moreover, we found no correlation between the magnitude of individual PSE shifts and the amount of residual aberrations under AO correction (Fig.S6A; r=.26, p=.31). Thus, the difference in PSE we observed likely reflects the impact of neural compensatory adjustments on phase processing following long-term exposure to optical aberrations. Consistent with this idea, the magnitude of individual PSE shifts correlated with the amount of habitual optical aberrations experienced by each participant in their daily life (Fig.6C–hRMS: r=.87, p<.001; Fig.S6B–tRMS: r=.82, p<.001). Finally, larger PSE shifts under AO correction were associated with stronger deficits in high-contrast letter acuity under full AO correction (Fig.S6C; second-order polynomial: F=23.3, p<.001, R²_adj = 0.74). These results provide evidence that long-term exposure to severe optical aberrations alters perceived phase in KC patients, consistent with the existence of neural compensation adjustments to phase spectra.

All participants were cyclopleged with tropicamide (1%) to dilate their pupil and paralyze accommodation during testing, after being screened by one of our ophthalmologists to ensure dilation was safe and provide standard information (e.g., corneal curvature, refractive error). An additional screening session was performed to ensure we could obtain good quality wavefront measurements and reach stable AO correction. Habitual optical quality was estimated for each participant from wavefront measurements collected using their everyday correction method, if any. The square root of the sum of Zernike coefficients for a 6-mm pupil was used to estimate total RMS error (tRMS: all coefficients, except piston and prisms) and HOAs RMS error (hRMS: 6–65th coefficients). Three participants (3 males; 22-51yo) with typical optical quality (tRMS: 1.04µm, range: 0.57-1.42µm; hRMS: 0.44µm, range: 0.35-0.54µm; 6-mm pupil) completed Exp.1 and S1 (Fig.4,S1). Exp.2 included an additional participant (25yo female) with typical optical quality (RMS: 1.59µm, hRMS: 0.56µm). Exp.3 included eleven participants with typical eyes (8 males, 32±11.8yo; tRMS: 1.06±0.31µm; hRMS: 0.4±0.13µm) and six KC patients (5 males, 35.7±15.5yo; tRMS: 4.04±2.32µm; hRMS: 2.47±1.33µm). All control observers were emmetropes or low myopes/hyperopes with well-corrected optical quality in both eyes. All patients had been diagnosed with KC at least 12 months before testing (reported onset around their twenties), and had been wearing the same habitual corrective method without substantial changes in their vision for at least 6 months before testing. The eye with the most severe optical aberrations was selected for monocular testing as it is generally associated with stronger adaptation effects (Barbot et al., 2021; Ng et al., 2022; Sabesan et al., 2017; Sabesan and Yoon, 2009), except for one patient due to the presence of scarring in the worst eye incompatible with AO correction. Two KC patients and one control observer did not pass the screening stage, and were not tested further along. The Research Subjects Review Board at the University of Rochester approved all experiments, which adhered to the tenets of the Declaration of Helsinki and were conducted after obtaining informed, written consent. Participants were compensated$12/hr.

Optical quality and visual performance were assessed under white light conditions using an Adaptive Optics Vision Simulator (AOVS). As previously detailed (Barbot et al., 2021; Ng et al., 2022, 2021; Sabesan et al., 2017; Sabesan and Yoon, 2010, 2009; Zheleznyak et al., 2016), our AOVS (Fig.3A) consists of a custom-built Shack–Hartmann wavefront sensor that measures observers’ wavefront aberrations, and a deformable mirror (ALPAO-DM97, Montbonnot, France) used to correct and/or induce wavefront aberrations online. An artificial pupil controls pupil size while visual stimuli are presented using a modified digital light processor (8bits, SharpXR-10X, Osaka, Japan). The visual display subtended 3.98×2.98deg (1024×768, 75Hz; ∼0.23 arc/min per pixel) and was gamma-corrected with a PR-650 SpectraScan Colorimeter (PhotoResearch, Chatsworth, CA). A superluminescent diode (840±20nm) was used for wavefront sensing. Head movements were stabilized with a headrest and a dental impression bite bar mounted to x-y-z translation stages. A camera imaging observers’ pupil was used to maintain pupil alignment. The AOVS operates in a continuous closed loop between wavefront sensing and deformable mirror correction, at ∼8Hz. Full correction of all coefficients was applied during testing, with the exception of defocus for which subjective values were used to correct axial chromatic aberrations and ensure the best focus in white light conditions. Subjective defocus was estimated by asking participants to adjust defocus using a motorized Badal prism while correcting all other aberrations in order to maximize image quality. We verified that subjective defocus values provided the highest acuity and perceived image focus under AO correction. Consistent with previous studies (Barbot et al., 2021; Ng et al., 2022, 2021; Sabesan et al., 2017; Sabesan and Yoon, 2010, 2009; Zheleznyak et al., 2016), the AOVS offered full control of wavefront aberrations over sustained periods (∼0.05µm residual RMS), in both ‘healthy’ and severely aberrated eyes.

Suprathreshold horizontal compound stimuli were created by adding two sinusoids with frequency f and 2f. The relative phase of the two components was shifted from a reference point at which the troughs of the two sinusoids are aligned and compound stimuli appear ‘square-wave-like’ (0º relative phase)(Fig.3B). Varying the f-2f relative phase resulted in ‘saw-tooth-like’ stimuli pointing either ‘downward’ (−90º) or ‘upward’ (+90º). The absolute phase of the fundamental f was randomly set on each trial to avoid low-level repetition effects related to local image contrast. A raised-cosine envelope (2.6deg diameter) was used to avoid sharp, high-SF borders. The amplitude of f and 2f was set to 66% and 33% contrast, respectively. To account for contrast reductions under AO-induced blur and maintain the f/2f contrast ratio (Exp.1,2), the amplitude of the two components was adjusted using corresponding MTF values. A fundamental frequency of 3 cycles/deg (2f: 6 cycles/deg) was used in all experiments (except Exp.S1).

On each trial, participants reported whether a compound grating stimulus presented foveally for 80ms appeared upwards or downwards (Fig.3B,C). No feedback was provided. All participants were first familiarized with the task using compound gratings presented until response. We computed the percent of upwards trials as a function of the f-2f relative phase. We used the Palamedes Toolbox (Prins and Kingdom, 2018) to fit data with a Cumulative Normal psychometric function using maximum likelihood estimation, and estimate the point-of-subjective equality (PSE)–i.e., the relative phase at which stimuli were perceived “aligned”. Non-parametric bootstrapping was used to compute 95%-CIs of PSE estimates and assess differences in PSE estimates. We randomly resampled individual trials with replacement to generate and fit resampled datasets. This procedure of resampling and refitting was repeated 10,000 times to generate bootstrap distributions of PSE estimates. Goodness of fit was verified using log likelihood with a global maximum successfully found for all fits.

Multiple visual acuity (VA) thresholds (10±6 on average) were measured for each participant under AO correction using a 4-AFC letter orientation task. On each trial, participants reported the orientation of a high-contrast, tumbling E acuity letter (Sloan) presented for 250ms. Letter size was adjusted from trial-to-trial using the QUEST staircase method (Watson and Pelli, 1983) to estimate 62.5%-correct thresholds (40 trials/staircase). Note that all included thresholds were obtained from successfully converged staircases, and were highly consistent within each observer.

The transfer function of the eye, like any optical system, can be characterized by the Modulation Transfer Function (MTF) and Phase Transfer Function (PTF), which specify the amplitude reduction and phase shift as a function of the signal SF, respectively. To obtain the MTF and PTF, we first compute the generalized pupil function containing the eye’s aberrations, as follows: where P(x,y) is the pupil function at x,y Cartesian coordinates, A(x,y) the amplitude function, and W(x,y) the wavefront aberrations. This function is then used to calculate the point spread function (PSF) and optical transfer function (OTF) using 2-dimensional Fourier Transform (fft2): The OTF can further be decomposed into the MTF and PTF by taking the modulus (magnitude) and phase of the complex OTF: Predicted shifts in relative phase between f and 2f components were computed using PTF values for various amounts of vertical coma (Fig.4B). These functions were used to generate simulated images for specific optical conditions, with or without compensating for the impact of MTF or PTF (Fig.1).

Predicted phase shifts under specific amounts of vertical coma selected for testing in Exp.1 were validated directly from the AOVS (Fig.4C). A foldable mirror was inserted into the path to redirect it onto a CCD sensor and take measurements of the display. A collimated single mode diode laser was used to provide a signal to the Shack-Hartmann sensor and induce aberrations without the eye’s optics. Gratings of varying f-2f frequencies were displayed under AO-induced vertical coma ranging from -0.5 µm to +0.5 µm. Ten images were collected for each condition and averaged together. Segments of f and 2f images were fit with cosine waves of variables amplitude, period, phase and vertical translation. The relative phase between the f and 2f was then computed from these fits to validate predictions from our theoretical simulations.

Observers judged the appearance of compound gratings (f: 3 cycles/deg; relative phase: -90º to +90º, 10º steps) under nine conditions of AO-induced vertical coma (0, ±0.2, ±0.3, ±0.4, ±0.5µm). We selected this range of values to ensure that the magnitude of observed phase shifts could be effectively evaluated using the compound grating appearance task (i.e., within ±90º in relative phase, for reliable PSE estimation). As described above, the amplitudes of f and 2f components were adjusted according to MTF values to account for blur-induced contrast reductions and maintain the f/2f contrast ratio. PSE estimates were compared to optical theory predictions (Fig.4D-F). Each condition was tested on separate blocks counterbalanced across sessions. In the MTF-adjusted control condition (Fig.4E), we tested the impact of blur-induced contrast reductions on perceived phase by testing participants under AO correction using compound stimuli whose contrast and relative phase were adjusted according to both MTF and PTF values for ±0.5µm vertical coma (i.e., phase offset: ±52.3º). 95%-CIs are provided to assess changes in PSEs across conditions. A total of 6097 trials was collected on average per observer (∼553 trials for each of the eleven tested conditions). Additional data was collected using higher-SF compound gratings (f: 4 cycles/deg) and positive vertical coma (Fig.S2).

We assessed the effects of sustained exposure to vertical coma on perceived phase (Fig.5). Participants judged compound gratings (f: 3 cycles/deg) with relative phase ranging from -90º to +90º (15º steps). Each experimental session (Fig.5A) consisted of 1599 trials divided into 41 successive blocks (39 trials each) across 3 periods: baseline, adaptation, and post-adaptation. The baseline (pre-adaptation) consisted of 8 blocks (312 trials; ∼8 min) collected under AO correction (PSE_predicted: 0º). During blur adaptation, observers were exposed to AO-induced vertical coma (−0.4µm; PSE_predicted: +29.5º) over 21 blocks (819 trials; ∼70 min). The first 3 blocks (117 trials; ∼3 min) only contained compound stimuli to estimate the initial blur-induced shift in PSE. The following 18 adaptation blocks (702 trials; ∼67 min) contained natural images (e.g., animals, buildings, walls, humans) and checkerboard textures presented before each compound stimulus, serving as perceptual cues the visual system could adapt to in the presence of blur (Fig.5B,C). Each trial started with 6-12 adaptation stimuli, each presented for 240ms and separated by 13ms. After a 253-ms delay, a compound stimulus was presented for 80ms. Note that adaptation stimuli changed rapidly to homogenize local light adaptation and minimize afterimages, as done in previous studies(Elliott et al., 2011, 2007; Sawides et al., 2011a). Each adaptation stimulus had an equal probability to be a natural image or a checkerboard stimulus. Natural images were randomly selected from ∼700 natural grayscale images and could not repeat within the same trial. Checkerboard stimuli had random square size (3,4,5 or 6 square width) and phase. Immediately following blur adaptation, AO-correction was switched back on and 12 post-adaptation blocks (468 trials; ∼12min) were collected with only compound gratings presented, as in the baseline. Participants were allowed to take short breaks (∼1-2 min each) away from the bite-bar at specific moments within each session (Fig.5A). To minimize the impact of these interruptions on the adaptation speed or depth, participants were asked to stay in the experimental room under the same lighting conditions. During blur adaptation, the first trial following a break started with top-up adaptation (60 adaptation images)(Fig.5B). The control condition was identical to the blur adaptation condition, except that participants were always tested under full AO-correction. Instead of inducing blur, a phase offset of +29.5º was added to compound stimuli, mimicking the phase shift under -0.4µm of vertical coma. Thus, participants’ responses were similarly biased during the adaptation period. However, there was no clear phase disruption the visual system could detect and adapt to as adaptation stimuli were presented under full AO correction.

Each participant performed at least two experimental sessions on separate days. To better characterize changes in perceived phase across time, PSE estimates at a specific timepoint t (i.e., block number) were computed by binning trials from adjacent blocks (±1; i.e., trials from t-1, t, and t+1). This binning procedure was adjusted to ensure not to bin trials across different experimental periods (baseline, adaptation and post-adaptation), for which trials from only two blocks were binned together. Importantly, our results did not depend on this binning procedure (Fig.S3). 95%-CIs were provided to assess changes in PSEs across time. In order to better visualize PSE changes relative to baseline, the mean of PSE estimates during baseline was subtracted from all PSE estimates, correcting for small individual biases and treating 0º as reference.

We assessed the effects of long-term (i.e., months to years) exposure to poor optical quality on phase perception. We measured perceived phase in patients with keratoconus (KC)–a corneal disease causing severe optical blur in neurotypically-developed adults (Fig.6A). KC results in large amounts of HOAs, particularly of vertical coma that primarily affects resolution along the vertical axis (i.e., horizontally-oriented signals). As previously detailed (Barbot et al., 2021; Ng et al., 2022; Sabesan et al., 2017; Sabesan and Yoon, 2010, 2009; Zheleznyak et al., 2016), our AOVS is able to correct all optical aberrations and estimate visual functions under similar, aberration-free conditions in both healthy and KC eyes. We measured PSE estimates under full AO correction in eleven control participants with healthy eyes and six KC patients (Fig.6B,S5). Relative to the optical quality of control participants, habitual optical quality in KC eyes was severely degraded despite their everyday optical correction (tRMS: 1.06±0.31µm in controls, 4.04±2.3µm in KCs), particularly when looking at the amount of uncorrected HOAs (hRMS: 0.38±0.13µm in controls, 2.47±1.33µm in KCs). Poorer habitual optical quality was associated with stronger alterations of visual functions (Fig.6,S6). Note that all participants were tested across ±90º range in relative phase, but the exact values used slightly varied mainly due to time constraints, with either 19 phase levels (10º equal steps), 13 phase levels (15º equal steps), or 9 phase levels (±90º,±45º,±15º,±7.5º, and 0º). A total of ∼900 trials was collected on average per participant.