One or both eyes open in archery: Hirota 2023 study

Fourteen archers from the Osaka University club, with a mean age of 20.7, shot 36 arrows at 30 metres on an 80 cm face with both eyes open and another 36 with the non-aiming eye covered by a patch, in crossed order between the participants. The aim was to see whether binocular vision, which supplies depth perception, changes anything in a movement that involves looking at a distant target and a near sight together.

The total score did not change to any statistically appreciable extent: 257.6 points with two eyes against 250.9 with one (p = 0.22). Nor did the centre of gravity of the group shift (4.67 cm against 5.33, p = 0.83). What did change was the distribution: with two eyes the arrows in the 9 ring averaged 11.86 against 9.36 with one eye, and this is the only difference crossing the threshold (p = 0.047). In the 4, 5, 6 and 8 rings the arrows were fewer in the binocular condition, without reaching significance; in the 10 and the 7 there was no difference. Horizontally the group was tighter in the binocular condition, and the peak of the distribution closer to the centre.

In the second part of the study, seven of the same archers shot wearing an eye tracker. With two eyes, five of seven moved their gaze back and forth between the target and the sight until release; with one eye two of seven did so, while the rest looked at the sight almost throughout (p = 0.051). Pupil diameter over the last second before release was smaller in the binocular condition, 1.42 against 1.64 mm (p = 0.018), which the authors read as a sign of the near reflex. Hence their interpretation: with two eyes the archers were gauging distance by alternating their gaze between the target and the sight.

All the participants were examined by an ophthalmologist: none had strabismus, amblyopia or other eye disease, and those who were myopic wore glasses or contact lenses. Ocular dominance was determined with a hole-in-card test; all aimed with the right eye, which was also the dominant one.

The monocular condition was created by covering the eye not looking at the sight completely, with a patch, rather than asking the archer to close it: the authors stress this because in the earlier study they compare themselves with, monocular vision was defined as viewing with one eye while both were open, which is a different condition.

In the first experiment the fourteen archers were split into two groups matched for age, score in their last three competitions and years of practice: one group shot the 36 binocular arrows first and the 36 monocular ones after, the other in the reverse order. Swapping the order was meant to keep fatigue in check. The holes were marked with adhesive seals, the face scanned at 1718 by 1628 pixels and the position of each seal computed with ImageJ, taking the centre of the face as the origin. For the archers with a dominant left eye the horizontal axis was reversed, because the direction in which the group shifts is the opposite. Each archer's centre of gravity is the mean of the 36 positions.

In the second experiment seven of the fourteen shot wearing an EMR-9 eye tracker, which derives eye position from the corneal reflex and the pupil centre at 240 Hz, with a stated error of 0.2 to 0.5 degrees at one metre, together with a scene camera at 29.97 frames per second. Calibration was done on a panel at 120 cm, chosen for an interpupillary distance of 60 mm so that movements between target and sight could be assessed equally, with nine fixation points; the centre of the scene camera was aligned with the centre of the target at 30 metres in the simulated full-draw position. Data were discarded when pupil diameter changed by more than 2 mm per frame, a sign of blinking, and missing values were linearly interpolated; the right eye was analysed. Each archer's horizontal focus point was estimated from the centre of the density map, taking the most negative value as the primary one where more than one appeared. The two conditions were compared with Fisher's exact test for the number of focus points, with the Wilcoxon signed-rank test for score, centre of gravity and pupil diameter, and with the paired t-test for the distribution of arrows across the scoring rings.

Fourteen university archers, seven in the eye-tracking part, and the authors give this as a limitation: they are not professional athletes, and they write that with top-level athletes the comparison between the two conditions might come out differently.

There is only one positive result: more arrows in the 9 ring under binocular vision, at p = 0.047. The total score does not change, the centre of gravity does not change, and the ring-by-ring comparison was made across eleven rings without correcting for the number of comparisons: with eleven tests, a p value around 0.05 should be read with caution. The differences in the 4, 5, 6 and 8 rings run the same way but stay above the threshold.

The discussion contains two statements at odds with the results. The authors write that the centre of gravity differed significantly between the two conditions, whereas in the results and the figure it did not (p = 0.83); and they write that the number of arrows in the 9 ring was reduced under binocular vision, immediately after saying the opposite, which is what the table and the figure show. This record follows the tables. One figure caption refers to three experiments, while the paper has two.

On the eye-tracking part, the comparison between conditions does not cross the threshold (p = 0.051, five archers against two out of seven), so the sentence about switching gaze between target and sight describes a tendency rather than an established difference. The authors flag a second limitation of their own: eye position is measured in relative rather than absolute terms, because the archers' heads shifted slightly with each arrow, their faces turning and heads tilting, which moved the origin of the measurements.

Finally the setting: 30 metres on an 80 cm face, not 70 metres. Everything the study says about depth perception, including the estimate that beyond 40 metres other cues are needed, comes from the literature it cites rather than from measurements made here.

For an archer the practical question is whether to shoot with the non-aiming eye open or closed, and this study does not settle it. The total score is the same in both conditions, and the difference that emerges is a fine one: with both eyes open the arrows clustered slightly more in the 9 zone and slightly less in the low rings, with a tighter group horizontally, but without the score moving.

It is worth noting exactly what was compared. The monocular condition here was one eye fully patched, not an eye closed or half-closed while the other aims, which is what archers who shoot that way actually do. The authors say so precisely in order to explain why their result does not match an earlier paper on traditional archery without a sight, where arrow convergence was better under monocular vision. Two different conditions, two different results, and neither corresponds exactly to habit on the shooting line.

The eye-tracking part is the most interesting and the least conclusive. With two eyes most of those seven archers moved their gaze back and forth between target and sight until release; with one eye they tended to stay on the sight. The smaller pupil in the binocular condition points the same way, since focusing on something near constricts the pupil. The authors conclude that the binocular archer was gauging distance, but the difference between conditions does not cross the statistical threshold: it is a hypothesis waiting for larger numbers.

Finally, the contribution the authors claim for themselves is not advice but a method: measuring the position of every single arrow, instead of the score alone, makes differences visible that the total hides. For fourteen university archers at 30 metres, with two eyes or one, the total did not move and the distribution did.