Clicker, brain and archery accuracy: Ertan 2021

Fifteen Turkish archers at national level, nine men and six women, shot 1,070 arrows in all at 18 metres wearing an EEG cap, with a switch mounted under the clicker marking the exact moment of the clicker's fall in the traces. The snap evokes the familiar long latency auditory response in the brain, the N1 and P2 waves.

Comparing arrows that landed close to the centre of the archer's own group with those that landed outside it, N1 amplitude was higher for the former. The difference emerges with a 95% confidence interval for the second and fourth of the ten area indexes used, and at 90% for the first and third; from the fifth index onwards it does not emerge at all. N1 latency, P2 amplitude and P2 latency separate the good arrows from the rest at none of the indexes.

The authors read the larger N1 as a sign of a higher state of attention and arousal, and note that this amplitude also grows when the exact moment of the stimulus is uncertain, as it is with a clicker, and when the task is demanding. They observe too that the response is larger than a sound as faint as a clicker ought to produce, and suggest it is not purely auditory: the clicker's fall is also felt as vibration in the bow hand, and seen.

The archers averaged 22.8 years of age, from 16 to 31, with 5.8 years of practice, from 2 to 14, and a mean personal best of 632 at 70 metres, from 602 to 661. They shot twelve practice arrows to settle into the measurement conditions, then between 64 and 72 arrows each, outdoors rather than in a laboratory. The EEG was recorded with electrodes in an elastic cap at twenty sites of the 10-20 system, sampled at 1,000 Hz, with impedances below 5 kilohms. A mechanical switch under the clicker brought the moment of its fall into the trace.

Each trace was cut from 200 milliseconds before to 800 milliseconds after the snap, filtered between 1 and 12 Hz, and the amplitudes and latencies of N1 and P2 were measured by hand in Matlab at the right mastoid electrode referenced to Cz. To pair every arrow with its own trace, the arrows were numbered and colour coded, and after each end the target face was photographed from the front and from both sides, so that the position and order of the holes could be reconstructed.

The point on which everything turns is how a good arrow is defined. For each archer the mean of the coordinates of his or her impacts is calculated, and a rectangle is drawn around that point: inside is the hit area, outside the miss area. Enlarging the rectangle produces ten progressively wider indexes; at the first and narrowest, 65 arrows out of 1,070 fell inside. The two areas are compared through confidence intervals on the difference between the means, at 95% and 90%: if the interval excludes zero, the difference is treated as significant.

The hit area is not the gold. It is a rectangle centred on the mean of that archer's own impacts, so an arrow counts as good if it lands near the centre of the group, not if it is worth ten points, and for an archer whose sight is off the rectangle moves with the group. The abstract instead describes a progressive widening from gold out to blue, and the two descriptions do not match. No score enters the analysis.

The pattern of results is not orderly: the difference is there at the second and fourth indexes at 95%, at the first and third only at 90%, and from the fifth onwards it is absent. With ten indexes and four measures there are forty comparisons, with no correction, and accepting 90% means working at a threshold of 10%. At the first index the good arrows number 65 in all, four or five per archer: the averaged traces for that area rest on very few repetitions, and an average of few repetitions is noisy precisely where the muscular noise of the shot is worst. Nothing in the text says how movement artefacts were handled.

Fifteen archers is few, and the data are old: the ethics approval carries a 2004 number, while the paper was received in 2020 and published in 2021. The study is cross-sectional and descriptive, so it does not say which way the relation runs: greater attention may raise N1 and bring the arrow near the centre, but a shot that is going as it should may equally come with a different state. The authors themselves list four possible explanations for the larger amplitude, from attention to uncertainty about the moment of the snap, without being able to choose between them.

The study makes one point worth pausing on: the clicker is not only a mechanical signal, it is a stimulus the brain registers, and it registers it more strongly on the arrows that land near the centre of the group. The difference lies in the amplitude of the wave, not in when it arrives: good arrows and the rest share the same latency, so it is not the speed at which the response forms that separates them.

Nothing to do on the range follows from that amplitude. It is not a quantity an archer can perceive, it takes an EEG cap and the average of many arrows, and with four or five centred arrows per archer the result is fragile. Above all the direction of the relation is unknown: the study records that the two appear together, not that one produces the other.

What does remain is the observation about timing uncertainty. The clicker goes when the archer brings it to go, but not at a moment the archer chooses: the point creeps through, the vibration in the arrow gives some warning, the snap itself does not. The authors use that uncertainty to explain why the response to so faint a sound should be so large, and in their account it is a property of the clicker rather than a fault in the archer.