Mental tension of an Olympic archer: Ohsuga 1992

Three engineers at the Mitsubishi Electric central research laboratory built a portable monitor of mental tension for one particular Japanese Olympic archer, and the paper describes how they arrived at the two numbers the monitor displays. From the electrocardiogram they derive the mean interval between beats and the size of its respiratory oscillation, the quantity they call RSA: the first responds to both the sympathetic and the parasympathetic branch, the second mainly to the parasympathetic one, which is why the two values do not always move together and are plotted on two axes instead of being reduced to a single index.

The starting data come from competitions in 1987: the national selection trials at Tsumagoi in June and the open practice for the US championships in August. On the first day of the trials the archer described himself as slightly tense and both values were low; on the second day, with first place already secured, jokes and laughter appeared in his comments and both values had risen; at the US championships RSA was particularly low, matching the very high tension he reported. Within a single competition, tension as estimated from the two indices rose after a miss, and when both were high came the remark that he had made a careless mistake through letting his guard down.

The resulting monitor weighs about 800 grams, measures 190 by 47 by 136 millimetres and reads the electrocardiogram alone. A button starts it, it measures for 30 seconds, discards the first 10 and analyses the remaining 20; meanwhile an earphone plays a slow three-beat tune, Greensleeves, to breathe in for three beats and out for three, which comes to 0.3 hertz. At the end it shows the point on a graph with the interval between beats across and RSA up, inside a target rectangle set provisionally between 600 and 800 milliseconds and between 10 and 20 milliseconds.

The competition recordings were made with a purpose-built unit: an electrocardiogram taken from the chest, chosen because it picks up fewer artefacts, and an abdominal belt with a tube for breathing, later moved inside the chest guard because it slipped. The archer spoke freely about how he felt, and the commentary was recorded alongside the signal on a portable cassette recorder.

Measurement ran throughout the competition, but the analysis uses only the last 20 seconds of the periods when the archer was sitting down between ends: the size of the respiratory oscillation also depends on posture and shrinks when standing. The competition format of the time had two archers alternating on the same target with three arrows each, twelve times per session, that is 36 arrows apiece, and four sessions at 90, 70, 50 and 30 metres made up one game.

The signal was sampled at one millisecond, the R wave picked out with a digital filter, and data lying more than 150 milliseconds from the window mean discarded as detection errors. The series of intervals was interpolated at a fixed 500 millisecond step, then filtered: a band-pass centred on 0.3 hertz and 0.2 wide for the respiratory component, a low-pass cutting above 0.15 hertz for the slow component, each assessed by its standard deviation. Breathing frequency, in the competition recordings, was read off the curve by eye.

The same archer also took part in an experiment with breathing paced by a metronome, four beats in and four beats out, at six rates from 0.2 to 0.45 hertz in steps of 0.05, one minute each, under four conditions: rising or falling rate, eyes open or closed. Breathing depth was only requested to be constant, not controlled. Three of the monitor's design choices come from there: 20 seconds of analysis suffice if breathing is steady, paced breathing has to be imposed because between 0.2 and 0.3 hertz RSA falls as the rate rises, and the slow component serves to reject contaminated readings. In the finished monitor breathing is not measured at all: its frequency is estimated from the zero crossings of the respiratory component of the heartbeat.

There is one subject: a Japanese Olympic archer, thanked by name in the acknowledgements. There are no statistics, no comparisons between people and no scores at all. The study does not relate the two indices to where the arrows went: the correspondence the authors show is between the indices and what the archer said he felt, read off the two-axis plots by eye, and his own account remains the reference.

The authors flag the weak points themselves. RSA is contaminated when a deep breath or a body movement inflates the slow component, which the band-pass filter cannot fully remove: in some recordings the archer reported high tension while RSA came out large, and that was the cause. The two indices are put on orthogonal axes for convenience although they are not independent, and the authors leave to a physiological model the job of estimating sympathetic and parasympathetic activity separately. The target rectangle between 600 and 800 milliseconds and between 10 and 20 milliseconds is declared provisional, derived from that same archer's competitions, and still to be refined during training. The monitor saves nothing: to keep a reading it has to be copied from the display by hand, which is why the authors say they cannot examine in detail how tension changes during a competition.

From the 1988 trials of the device what remains is a judgement, not a measurement: the archer reported that the values reflected his state well. Some of those recordings were made during visualisation, without paced breathing, and the breathing frequency they yield is not adequate. The long-wear adhesive electrodes held well, but on sensitive skin they could itch when the wearer sweated or leave the skin sore underneath. The paper is in Japanese, with the abstract and figure captions in English; the work was presented in 1990 and published in 1992, so it has to be read with the electronics of its day in mind.

This record matters for two things, and neither of them is a method to adopt. The first is the date: as early as 1987 an Olympic-level archer was being monitored during competition with an electrocardiogram over the twenty seconds he spent seated between ends, and the resulting monitor fitted in a bag and was left on the seat when he shot, with its connector unplugged.

The second is the idea that one number is not enough. The authors show cases where the interval between beats is long and the respiratory oscillation small, and cases the other way round, and that is why they refuse to flatten tension into a single index. Their original aim was to learn to come down from excessive tension in competition, but in use they found the opposite was needed too: in training the archer was too relaxed and had to come up to the level he had shown in competitions that went well.

What is not here is any link with points. No arrow was tied to a heart-rate value, no target band was derived from scores, and the band itself applies to that archer and to the way he breathed during the measurement. Read as an evidence record, it describes an instrument and a single case, not a criterion to go by.