It is often required in the analyses of the eye movement to divide the observed data into the saccadic and smooth components. Several computer algorithms have been proposed so far, which used the first- or second-order derivative of the movement with respect to the time for descriminating the two components. This paper describes a new method of using the third-order derivative. It separates the two components more clearly and is more insensitive to noises than the previous methods.
Experimental analysis is reported on signal confirmation errors by railway drivers on the occasion of a train's departure. Actual operation to start a train was simulated in the laboratory. Subjects were instructed to verify “clear” signal prior to departure time and reconfirm them before they begin the starting operation after a sign of departure is given with a pilot lamp or a tone. A secondary task such as memorizing figures was imposed on the subjects in order to distract their attention. Reaction time, error rates, and signal confirmation behavior recorded on VTR were analysed, comparing various types of departure signs. The results were as follows: (1) Subjects tended to miss the signal if they did not immediately notice the sign. (2) Preparatory confirmation was easily omitted owing to the distractor. (3) Probability of prior confirmation was improved by changing location or modality of the sign. (4) It was also effective to place a written warning on a pilot lamp. (5) The operation was sometimes triggered the sign even if the subjects noticed lights. (6) Automatic reaction without confirmation occurred when the operator was devoted to the secondary task. Based on experimental analysis, countermeasures were discussed.
A new method is proposed for monitoring the arousal level by detecting the grouping of blinks. An experiment of monotonous auditory task was performed where the subject was instructed to recognize the Morse-code for a period of 30 to 40 minutes. Blinking was recorded by using capacito-oculography (COG) developed by the authors together with EEG and task performance. In COG measurement the subject wears goggles having translucent metal electrodes for detecting a capacitance change due to blinking. Frequency modulated UHF signals are transmitted wirelessly from an oscillator mounted on the goggles. The blink grouping is defined as more than two successive blinks at the interval of less than 1 second. The blink grouping reached the maximum rate 10 to 25 minutes after the onset of the experimental session when the alpha index of EEG increased up to two to three times the initial level and the rate of missed signals began to increase rapidly. When the arousal level decreased further, the rate of blink grouping returned to a low level again. Therefore, the blink grouping may be used effectively as basic information to provide the operator with a warning signal to avoid hazards due to a lowered arousal level.