Regaining Visual Acuity Does Not Restore Motion Extrapolation Deficits in Amblyopia

Abstract

Purpose

We previously reported that amblyopes exhibit deficits in motion extrapolation and in correcting for overextrapolation. In this study, we explored whether these motion deficits remain when normal visual acuity is restored after successful standard treatment in adults with former unilateral amblyopia.

Methods

Eleven clinically treated adult amblyopes and 11 control subjects participated in the study. We assessed visual motion processing functions by using two motion illusion paradigms: the flash-lag effect (FLE) and the flash-grab effect (FGE). We measured the monocular FLE and FGE magnitudes for all participants. Two contrast conditions (0.2 and 1) of FLE and two spatial frequencies (2 and 8 cycles) of FGE were tested.

Results

Compared to controls, treated amblyopes still exhibited a smaller FLE magnitude at the 0.2 contrast (F_1,20 = 5.69, P = 0.027) in both the former fellow eye (FFE) and the former amblyopic eye (FAE). Treated amblyopes had a larger FGE magnitude than controls (P ≤ 0.008) in both eyes, and the FGE magnitude of FAE was larger than that of FFE (F_1,20 = 15.9, P = 0.003). The FLE and FGE magnitudes were significantly correlated in most conditions in controls (P ≤ 0.035) but not in treated amblyopia (P ≥ 0.201).

Conclusions

We observed a smaller FLE and a larger FGE in treated amblyopes, suggesting that the motion extrapolation and the correction for extrapolation remain impaired in their brain. These may be due to processing delays and defective temporal integration from residual abnormal cortical connections in the amblyopic visual system.

Amblyopia is a neurodevelopmental visual disorder characterized by vision loss arising from uncorrected anisometropia or strabismus during early life in most cases. It affects 1% to 5% of the population.¹ Over the past 200 years, occlusion therapy has been the standard care to restore vision of the amblyopic eye.^2,3 The success of treatment is judged based on the visual acuity improvement of the amblyopic eye.^3,4 There is accumulating evidence to state that full recovery of visual acuity of the amblyopic eye does not mean fully restored normal spatial vision. The contrast sensitivity of the former amblyopic eye in treated amblyopia remains impaired, particularly in the mid-to-high spatial frequency range.^5–7 Follow-up studies revealed that some amblyopes still had impaired stereopsis following treatment.^8,9 Zhao et al.¹⁰ reported impaired binocular summation. Chen et al.¹¹ reported residual binocular imbalances over a wide range of spatial frequencies in treated amblyopia. These studies suggest that even if amblyopes regain normal visual acuity, they may still present residual spatial vision impairments.

Apart from spatial vision deficits, amblyopes also exhibit impaired temporal visual processing, such as a poorer temporal resolution¹² and a higher temporal synchronicity threshold.¹³ However, the effectiveness of monocular visual acuity (VA) restoration in improving the temporal deficits of amblyopes remains inconsistent. Chen et al.¹⁴ demonstrated that the temporal synchrony discrimination thresholds of former amblyopic and fellow eyes are similar to controls. Using visual evoked potential, Kelly et al.¹⁵ observed better phase alignment and shortened latency in the amblyopic eye after occlusion therapy. However, Watts et al.¹⁶ showed opposite results in treated strabismic amblyopic children. They found that both the amblyopic eye (AE) and fellow eye had markedly longer P100 latency than control subjects. In addition, motion perception deficits often persist despite improved visual acuity after patching.¹⁷

The transmission and processing of visual information in the human brain takes time,^18–20 meaning that our brain receives delayed sensory information. To accurately interact with the dynamic world, the brain may overcome these neural delays through motion extrapolation^21,22: using the motion trajectory information from the past to predict a moving object's current location. A body of visual motion-induced illusions has been used to study extrapolation mechanisms.^21,23–26 The most famous illusion is probably the flash-lag effect (FLE), in which a flashed stimulus is perceived as lagging behind a moving object when the two objects are physically aligned.²¹ On the other hand, another illusion, the flash-grab effect (FGE), has been used to study how the visual system is able to correct predictive errors when motion direction abruptly changes.^27,28

Applying these paradigms to the amblyopic population, we previously revealed that both the amblyopic eye and the fellow eye of amblyopes presented a smaller FLE^29,30 and a larger FGE magnitude,³¹ suggesting that motion extrapolation and correction for overextrapolation were both impaired in the amblyopic visual system. Additionally, we also observed an extended FGE specific to the amblyopic eye. Therefore, we believed that these two motion-processing deficits in amblyopia may be derived from two different neural mechanisms: a defect in V1’s horizontal connectivity for motion extrapolation²⁹ and a low-level delay in the information processing of the amblyopic visual system for motion correction.³¹ In this study, we ask whether restored visual acuity in the amblyopic eye following treatment also corrects one or both of these motion-processing deficits of presumably different origins.

We employed the FLE and FGE paradigms to evaluate motion extrapolation and correction of overextrapolation in treated amblyopes and controls. First, we investigated whether the difference in FLE and FGE between the two groups could be reduced after the visual acuity of the amblyopic eye is restored to normal levels. Second, we examined whether motion extrapolation and motion correction share common visual processing mechanisms.

Methods

Participants

Eleven treated unilateral amblyopes (age range 21–28 years, mean age 23.82 years) and 11 controls (age range 26–31 years, mean age 28.12 years) with normal or corrected-to-normal visual acuity participated in this study. The eye dominance of controls was defined by the Porta test. The visual acuity was measured by the Early Treatment Diabetic Retinopathy Study chart at a 4-m distance. Treated amblyopia was defined as a best-corrected visual acuity in the former amblyopic eye of ≤0.1 logMAR and a difference of visual acuity between the two eyes of no more than two lines. The clinical history details of subjects were recorded retrospectively, based on their memory. Treated amblyopes underwent a standard patching treatment plus bead-threading or surgery in some cases, detailed in Table 1. Subjects’ refractive error was fully corrected during the experiment. This research followed the tenets of the Declaration of Helsinki and was approved by the Ethics Committee of West China Hospital of Sichuan University. We obtained written informed consent from each participant before data collection.

Table 1.

Clinical Details of Treated Amblyopes

Subject	Age, Y/Sex	Type	Eye	Refraction	VA (LogMAR)	Squint (PD)	Stereoacuity Arc Seconds	History
A1	24/M	Aniso	OD	−1.25	0	∅	200	Detected at 8 y old, patched for 4 y
			OS	−2.50/−1.00	0
A2	23/F	Mixed	ODOS	−5.00/−1.75 × 180°−3.75/−3.50 × 180°	00	EX 2	100	Detected at 7 y old, patched for 1 y, received strabismus surgery at 10 y old
A3	21/F	Aniso	ODOD	−3.50−1.50/−0.50 × 165°	−0.2−0.2	∅	200	Detected at 7 y old, patched for 6 mo, bead-threading for 2 y
A4	27/M	Aniso	OD	+3.50/−1.00 × 110°	0	∅	160	Detected at 10 y old, patched for 10 mo
			OS	+1.25/−0.50 × 175°	−0.1
A5	28/F	Aniso	OD	+1.25/−1.50 × 175°	0	∅	40	Detected at 7 y old, patched for 3 y
			OS	−1.75 × 10°	0
A6	22/M	Aniso	ODOS	−2.00/−1.75 × 175°+3.00/−3.00 × 180°	−0.1−0.1	∅	100	Detected at 8 y old, patched for 2 y, bead-threading for 1 y
A7	21/F	Aniso	OD	−0.50	0	∅	200	Detected at 10 y old, patched for 2 y
			OS	−4.25/−0.50 × 10°	0
A8	22/F	Mixed	OD	pl	−0.1	ET 8	NA	Detected at 3 y old, patched for 5 y
			OS	+1.25/+0.25 × 85°	0
A9	24/M	Aniso	ODOS	pl−2.00/−1.50 × 170°	00	∅	NA	Detected at 4 y old, patched for 1 y, bead-threading for 1 y
A10	27/F	Aniso	OD	−1.75 × 170°	0	∅	NA	Detected at 5 y old, patched for 2 y
			OS	+3.75/−2.75 × 10°	0
A11	23/M	Aniso	OD	+5.00	−0.1	∅	NA	Detected at 4 y old, patched for 7 y
			OS	pl	−0.1

Aniso, anisometropia; ET, esotropia; EX, exotropia; mo, months; PD, prism diopters; pl, plano.

Apparatus

Experiments were run using MATLAB R2018b (MathWorks, Natick, MA, USA) with the PsychToolbox version 3.0.9 extension on a MacBook Pro laptop. Stimuli were displayed on a gamma-corrected CRT monitor (SUN GDM-5510, 21 in.; SONY, Tokyo, Japan) with a resolution of 1280 × 1024 pixels, refresh rate of 100 Hz, and max luminance of 72 cd/m². The participants sat 57 cm away from the screen and viewed the stimuli in a dimly lit room while wearing a black opaque patch over the untested eye.

Stimuli and Procedure

Flash-Lag Effect

The procedures of the FLE experiment were similar to our previous study.²⁹ The FLE stimulus consisted of a moving bar, moving horizontally in the left hemifield from left toward the vertical meridian, and a flashed bar, appearing at a fixed location, with a variable flash timing relative to the initiation of motion of the moving bar in each trial (Fig. 1A). The moving and flashed bars were the same size (5° × 1°). The vertical distance between the nearest edges of the two bars was 1°. An orange fixation dot was presented at the center of the screen throughout the experiment. The speed of the moving bar was 18°/s, and it would move for 1000 ms and then disappear. The flashed bar was briefly presented for 10 ms at a fixed position of 8° from the vertical meridian. The timing of the flash was varied within 12 time points (−150, −120, −100, −80, −60, −50, −40, −30, −20, −10, 20, and 40 ms) relative to the time when the moving bar reached the azimuth of the flashed bar. The negative values indicate that the flashed bar appeared ahead of the moving bar, whereas the positive values indicate that the flashed bar appeared behind the actual position of the moving bar. In each trial, the participant was asked to judge whether the flashed bar was to the left (behind) or to the right (ahead) of the moving bar. In each block, each time point was tested with 10 repetitions. Two configurations were tested: moving bar in the upper visual field and flashed bar in the lower visual field (as depicted in Fig. 1A) or the opposite. Two luminance contrasts of the flashed bar were tested: 0.2 and 1. Participants were tested with the left and right eye separately using a black opaque eye patch. In sum, there were a total of eight conditions (2 configurations × 2 contrasts × 2 eyes). One condition was tested per block and repeated two times. The order of the conditions was randomized.

Figure 1.

(A) Illustration of FLE. Left: the physical stimuli: one bar moved horizontally rightward toward the vertical meridian in the left hemifield. The flashed bar is presented at the time when the moving and the flashed bars are physically aligned. Right: the typical perception: the moving bar is perceived to be ahead of the flashed bar. (B) Psychometric function of one representative participant in the FLE experiment. Proportion of “left” response as a function of the timing of the appearance of the flashed bar. The estimated midpoint of the logistic function defines the PSE that characterizes the amplitude of the FLE.

Flash-Grab Effect

The procedures of the FGE experiment were similar to those of our previous study.³¹ The FGE stimuli consisted of an annulus with alternating black and white segments that had inner and outer radii of 2° and 3°, respectively (Fig. 2A). The annulus rotated either clockwise or counterclockwise at the angular velocity of 270°/s and reversed its direction every 670 ms with a 50-ms pause at the reversal. At each reversal, a pair of orange or green discs (0.5°) was flashed in a random order for 10 ms (one frame), superimposed on the annulus symmetrically from the vertical axis. The separation between the two pairs was varied within 10 physical offsets (0°, 10°, 15°, 20°, 25°, 30°, 35°, 40°, 50°, and 60°). Each physical offset was tested with 20 repetitions. In each trial, the participant was asked to judge whether the orange pair was clockwise or anticlockwise relative to the green pair. Two spatial frequencies (SFs) of the annulus were tested: low and high SF (two and eight cycles). Participants were tested monocularly using an opaque black eye patch. A total of four conditions (2 SFs × 2 eyes) were tested in random order.

Figure 2.

(A) Illustration of the FGE. Left: the physical stimuli: an annulus rotated either clockwise or counterclockwise. A pair of colored discs (orange or green) was briefly flashed on the annulus in a random order at the reversal. Right: the typical perception: the flashed discs’ positions are shifted toward the new motion direction. (B) Psychometric function of one representative participant in the FGE experiment. Proportion of correct responses as a function of the physical offsets between the orange and green discs. The estimated midpoint of the logistic function defines the PSE that characterizes the amplitude of the FGE.

Data Analysis

Data were analyzed with MATLAB R2018b (MathWorks). For both experiments, the psychometric functions were fitted individually with a logistic function forced between 0 and 1 to estimate the midpoint of the function, which defines the point of subjective equality (PSE). In the FLE experiment, the psychometric function describes the proportion of “left (behind)” response at each time point (Fig. 1B). The PSE would define the time point at which participants perceived an alignment between the moving and flashed bars, characterizing the FLE magnitude (Fig. 2B). For analysis, the FLE is reported in milliseconds (one frame = 10 ms).

In the FGE experiment, the psychometric function describes the proportion of correct responses (seeing the orange pair of discs clockwise or anticlockwise relative to the green pair) at each physical offset. At the PSE, the participants perceived the two pairs of colored discs aligned with each other. A significant PSE shift from zero would characterize the FGE magnitude. For analysis, the FGE magnitude was converted into milliseconds (disc separation/rotation speed). The PSE estimates that fell outside of the measured range were extrapolated as all fits were accurate (all coefficients of determination R² ≥ 0.722).

To further explore whether the impaired motion extrapolation of amblyopes could be altered after their visual acuity improved, we compared our data on treated amblyopia to the FLE and FGE we previously measured in untreated amblyopia: we extracted 11 amblyopes’ FLE data from Wang et al.²⁹ and 13 amblyopes’ FGE data from Wang et al.³¹ These previous experiments were run in a different setting but with the same paradigms on comparable equipment.

The statistical analyses were conducted by using IBM SPSS Statistics, version 26.0 (Armonk, NY, USA) and R software (open-source; R Foundation for Statistical Computing, Vienna, Austria). FLE and FGE magnitudes were separately compared between and within groups by using analysis of variance (ANOVA). Bonferroni correction was applied for multiple comparisons. Since our sample is relatively small, we chose the nonparametric test for post hoc analysis. The two-sided Mann–Whitney U test was used for between-group analysis, and the two-sided Wilcoxon signed rank test was used for within-group analysis. Spearman's ρ test was used for correlation analysis. The level of significance was established at P < 0.05.

Results

Flash-Lag Effect

We first validated our FLE estimation by checking the quality of our fits and the significance of the PSE shift. The mean coefficient of determination R² was high in all conditions in both control (0.952 ± 0.04) and treated amblyopic (0.951 ± 0.05) groups. The PSE shift from 0 was significant for both control (F_1,20 = 70.91, P < 0.001) and treated amblyopic (F_1,20 = 64.95, P < 0.001) groups, which indicates that a FLE was observable in all conditions.

In Figure 3, we plot the mean FLE magnitude and individual data for both controls and treated amblyopes. We can see that the controls have a larger FLE magnitude of approximately 20 to 30 ms than amblyopes. The FLE magnitude is also larger in the low-contrast condition than in the high-contrast condition in controls. To compare the difference in FLE magnitude between the two groups, we performed a repeated-measures ANOVA with eyes and contrast conditions. We found a significant main effect of group in the contrast 0.2 condition (F_1,20 = 5.69, P = 0.027) (Fig. 3A) but not in the contrast 1 condition (F_1,20 = 2.95, P = 0.101) (Fig. 3B). These results remained similar even when the two mixed amblyopes (A2 and A8) were removed. The interaction of groups and eyes was not significant in either contrast condition (P ≥ 0.579). The post hoc analysis showed the following: in the contrast 0.2 condition, the FLE magnitude of the former fellow eye (FFE) (56.1 ± 25.6 ms, mean ± SD) of treated amblyopes was significantly smaller than that of the dominant eye (DE) (92.1 ± 40.2 ms) of controls (P = 0.023), and the FLE of the former amblyopic eye (FAE) (61.5 ± 30.9 ms) of treated amblyopes was significantly smaller than that of the nondominant eye (NDE) (95.1 ± 40.6 ms) of controls (P = 0.033). We did not observe a significant difference between the two eyes in either control (F_1,10 = 0.18, P = 0.683) or treated amblyopia (F_1,10 = 0.068, P = 0.8) groups.

Figure 3.

The mean FLE magnitude of controls, treated amblyopes, and amblyopes (data from Wang et al.²⁹) in low-contrast (A) and high-contrast (B) conditions. Individual data points for each subject are represented by circles. Amblyopes’ data from Wang et al.²⁹ are reported in shaded areas. FE, fellow eye. Error bars represent the standard deviation. *0.01 < P < 0.05. *Only represent the significance of statistical tests between controls and treated amblyopes in the current data set. For readability, comparisons across data sets do not appear in the figure but are presented in the main text. The left y-axis indicates the FLE magnitude in frames. The right y-axis indicates the FLE magnitude in milliseconds.

We also plot the FLE data of amblyopes from Wang et al.²⁹ in the right-hand section of each panel (gray shaded area) in Figure 3 (see Methods). To check the validity of comparing the two data sets, we first compared the FLE magnitude of controls between the current and previous studies and confirmed no difference between these two data sets (F_1,20 = 0, P = 0.984). Therefore, we felt confident in comparing the FLE between the treated amblyopes from our current study to the amblyopes from our previous study. We did not find significant differences in FLE magnitude between the treated amblyopes and nontreated amblyopes in both contrast conditions (Fig. 2, P ≥ 0.159). Furthermore, we performed a correlation analysis between the VA of the AE and FLE magnitude in our previous data set of nontreated amblyopes and found no correlation in either the contrast 0.2 or 1 condition (P ≥ 0.487, Supplementary Fig. S1).

Flash-Grab Effect

To explore whether the impaired correction extrapolation could be improved after visual acuity restoration, we analyzed the FGE magnitude difference between treated amblyopes and controls. The average coefficients of determination R² were 0.971 ± 0.03 in controls and 0.920 ± 0.1 in treated amblyopes, indicating that the logistic function fits were accurate in both groups. The estimated PSEs were significantly different from 0 in both control (F_1,20 = 102.65, P < 0.001) and treated amblyopia (F_1,20 = 114.63, P < 0.001) groups. This is an indicator that participants in the current study experienced an FGE illusion in all testing conditions.

Figure 4 illustrates the mean FGE magnitude of the control and treated amblyopia groups across the different SF conditions. The treated amblyopia group had a larger FGE than the control group, and the FGE magnitude of the FAE was larger than the FFE, by a difference of approximately 20 ms. The statistics analysis showed a significant main effect of group in both low (F_1,20 = 9.14, P = 0.007) (Fig. 4A) and high (F_1,20 = 8.68, P = 0.008) (Fig. 4B) SF conditions. These effects remained significant even when the two mixed amblyopes (A2 and A8) were removed. The interaction of group and eye was not significant in the low SF (F_1,20 = 3.82, P = 0.065) but was significant in the high SF (F_1,20 = 6.72, P = 0.017). The post hoc analysis showed that (1) the FGE magnitude of the FAE in the treated amblyopes was significantly larger than that of the NDE in controls in both low and high SF conditions (P ≤ 0.006; Figs. 3A, 3B), and (2) the FGE magnitude of the FFE of the treated amblyopes was significantly larger than that of the DE of controls in both SF conditions (P ≤ 0.033; Figs. 4A, 4B). In the control group, we did not find a significant FGE difference between the two eyes (F_1,10 = 0.01, P = 0.926). However, the FGE magnitude of the FAE was significantly larger than that of the FFE in the treated amblyopia group (F_1,20 = 15.9, P = 0.003).

Figure 4.

The mean FGE magnitude of controls, treated amblyopes, and amblyopes (data from Wang et al.³¹) at low (A) and high (B) spatial frequencies. Individual data points for each subject are represented by circles. Amblyopes’ data from Wang et al.³¹ are reported in shaded areas. Error bars represent the standard deviation. *0.01 < P < 0.05, **0.001 < P ≤ 0.01. *Only represent statistical significance within each data set, respectively. For readability, comparisons across data sets do not appear in the figure but are presented in the text. The left y-axis indicates the FGE magnitude in space units (degrees). The right y-axis indicates the FGE magnitude in time units (milliseconds).

We also plotted the FGE data of amblyopes from our other previous study³¹ in the right-hand section of each panel in Figure 4. To check the validity of comparing the two data sets, we again compared the FGE magnitude of controls between the current and previous studies. We did not find any significant difference between the current control and previous control data (F_1,22 = 0.36, P = 0.553). Therefore, we then compared the FGE between the treated amblyopes and the amblyopes from our past study. There was no significant difference in FGE magnitude between amblyopes and treated amblyopes (Figs. 4A, 4B) (F_1,22 = 0.22, P = 0.644). In our previous study, we found no correlation between the FGE magnitude and the visual acuity of the amblyopic eye in all our six different stimuli conditions (P ≥ 0.24; see Supplementary Fig. S1³¹).

Correlation Analysis Between FLE and FGE

To explore whether motion extrapolation and motion correction share common mechanisms, we performed a correlation analysis between the FLE (2 contrasts × 2 eyes) and FGE (2 SFs × 2 eyes) in the control and treated amblyopia groups. The correlation coefficients are plotted in Table 2. We found moderate to high correlations between the FLE and FGE magnitude in the control group. The correlations were significant for almost all conditions (all P ≤ 0.035), except the correlation between contrast 0.2 of DE and SF 8 of NDE (P = 0.056). A subset of these correlations between FGE and FLE amplitudes in both eyes is presented in Figure 5. These results indicate that motion extrapolation and motion correction mechanisms are correlated in the control population. However, we did not find any significant correlation between the FLE and FGE in the treated amblyopia group (all P ≥ 0.201).

Table 2.

Correlation Coefficients of Correlations Between the Flash-Lag Effect and the Flash-Grab Effect in Control and Treated Amblyopia

	Control				Treated Amblyopia
Flash-Grab Effect	SF2 DE	SF2 NDE	SF8 DE	SF8 NDE	SF2 FFE	SF2 FAE	SF8 FFE	SF8 FAE
Flash-lag effect
C0.2 DE/FFE	0.782*	0.827*	0.818*	0.591	−0.164	−0.218	0.036	0.009
C0.2 NDE/FAE	0.809*	0.9*	0.809*	0.636*	−0.255	−0.245	−0.209	−0.136
C1 DE/FFE	0.7*	0.773*	0.7*	0.727*	−0.2	−0.2	−0.018	0.004
C1 NDE/FAE	0.718*	0.836*	0.791*	0.764*	−0.364	−0.418	−0.191	−0.173

C, contrast.

^*Significant correlations (P < 0.05).

Figure 5.

Examples of correlations between FLE and FGE in the control group. (A) The correlation between the FLE at contrast 0.2 and FGE of SF 2. (B) The correlation between the FLE at contrast 1 and FGE of SF 8.

Discussion

In the current study, we investigated whether motion perception deficits persist after the visual acuity of the amblyopic eye has recovered in treated amblyopes. We found that (1) treated amblyopes still exhibited a smaller FLE magnitude than controls. (2) Treated amblyopes had a larger FGE magnitude than controls, and the FGE magnitude of the FAE was larger than that of the FFE. However, these two differences were not as large as we previously observed in untreated amblyopes.^29,31 (3) Finally, we showed that the FLE and FGE magnitudes were significantly correlated in controls but not in treated amblyopia.

Visual impairments associated with amblyopia are usually characterized by spatial deficits, such as poor visual acuity and low contrast sensitivity. Other spatial deficits include poor positional acuity^32,33 and elevated visual crowding.^34,35 Amblyopic vision is therefore limited by abnormal spatial interactions.^36,37 These spatial interaction deficits may not be resolved even following effective treatment^38,39 and could be critical for spatiotemporal integration in motion perception.^40,41

Apart from these spatial visual deficits, the amblyopic visual system also exhibits unignorable impairments in the temporal domain. Accumulating evidence supports that the temporal deficits of the amblyopic brain are not derived from the spatial impairments. Using a psychophysical figure/ground segregation task, Spang and Fahle¹² found that patients with amblyopia showed a decrease in temporal resolution and confirmed in a control experiment that this was not due to the visual acuity loss. Huang et al.⁴² found no correlation between Snellen visual acuity and the synchrony threshold in the amblyopic eye. Hu et al.⁴³ also reported no correlation between the peak or width of the temporal window and the amblyopic eye's visual acuity. Gurman and Reynaud⁴⁴ reported that the interocular delay in amblyopia was positively correlated with the interocular visual acuity difference but did not observe a causal relationship. Finally, even the fellow eye with normal visual acuity can exhibit temporal visual deficits, such as impaired perception of a temporal order⁴⁵ or motion-defined form.⁴⁶ Physiologically, those deficits could be associated with reduced synchrony in the firing of neurons within the amblyopic visual system, such as observed in cats.⁴⁷

In the visual system, there is a constant interplay between motion extrapolation and correction.^48,49 Our results found that the FLE and FGE magnitudes are correlated in the control population. This would suggest that motion extrapolation and correction are underlaid by a common mechanism integrating cortical motion prediction mechanisms and the late sensory transient visual inputs.⁴⁹ However, this correlation was lost in treated amblyopia, suggesting that the cortical prediction mechanisms and the visual input remain not properly integrated even after visual acuity is recovered in amblyopia.

In fact, we observed that the FLE was equally reduced in the two eyes of amblyopes. We also found no correlation between the FLE magnitude and the VA of the amblyopic eye from our previous amblyopic data set (Supplementary Fig. S1). This suggests that motion extrapolation deficits in amblyopes are independent of their spatial impairments. The cortical structures and functions in amblyopia are hard to recover to normal after treatment,^50,51 which could continue to disturb the extent of propagating facilitatory activity waves across the cortex and impair motion prediction.^52,53 Additionally, we found that treated amblyopes still exhibited a larger FGE compared to controls. In our previous study, we also observed that the FGE magnitude was not correlated with the VA of the amblyopic eye. We proposed that amblyopes’ impairments in the correction for extrapolation could be explained by a global processing delay in their visual system, plus an additional specific delay in their amblyopic eye.³¹ Results from the current study suggest that the processing delay in amblyopia may not be reduced in treated amblyopia. Wang et al.⁵⁴ also reported a longer reaction time in a visual attention task for the recovered amblyopic group with normal VA.

Neurophysiological studies showed that visual evoked potentials remain attenuated and delayed after successful treatment of amblyopia.⁵⁵ The former AE showed a significantly longer P100 latency compared to the former fellow eye.¹⁶ Weiss and Kelly⁵⁶ observed an increase in P100 peak and a decreased latency of the AE, although this was not significant. They emphasized an abnormal late peak component that could reflect hierarchical processing and abnormal signal integration across the visual cortex,^57,58 particularly because those effects depend on spatial frequency.^16,56 The above evidence suggests that standard amblyopia treatment through patching, bead-threading, and surgery may not completely restore all cortical structures and not reduce the processing delays in the amblyopic visual system.

This study has a few limitations that need to be noted. One is that the medical history of our subjects was recorded based on their retrospective memory. Therefore, our findings cannot be directly applied to children with amblyopia. Furthermore, the data from the treated and untreated amblyopic groups were collected in different experiments, which may have affected the significance of our results.

In the current study, treated amblyopes had a smaller FLE and a larger FGE, suggesting that the motion extrapolation and correction for extrapolation remain impaired in their brain. These may be due to two interdependent processes: a vision processing delay and a more fundamental deficit in temporal integration from residual abnormal cortical connections in the amblyopic visual system. Thus, our results suggest that defining the recovery of visual acuity of the amblyopic eye as the hallmark of successful treatment of amblyopia is not sufficient; temporal vision remains defective even after the visual acuity of the amblyopic eye has recovered. These results indicate the need for continued development of therapeutic approaches for amblyopia that target both spatial and temporal vision.^59,60