Position perception during linear movement was estimated quantitatively using a joystick device with and without visual information (VI). Eight healthy male subjects (19-24 years old) seated inside a capsule mounted on a linear accelerator were subjected to a constant g-load of 0.02 g along the X (antero-posterior) axis with varying displacement of 4 m, 10 m, 16 m, and to a constant displacement of 10 m with varying g-load of 0.02 g, 0.05 g and 0.08 g. The subjects were given the instruction to tilt a joystick in proportion to distance from starting position to present position during linear movement. The subject's position perception fitted Stevens' power law (R=kSn, R is output of joystick, k is a constant, S is displacement of linear movement, n is an exponent). The mean exponent was 0.49 for backward movement and 0.50 for forward movement without VI, and 0.80 for backward movement and 0.83 for forward movement with VI. Between backward and forward movements, there was no significant difference (p>0.1, paired t-test) of exponents and there was a correlation (r=0.90, p<0.01) of exponents under both visual conditions. Comparing absence of VI and VI, there was a significant difference (backward: p<0.002, forward: p<0.004, paired t-test) in exponents and no correlation of exponents (r=0.03, p>0.1). The position perception of 10 m displacement was not influenced by linear acceleration (0.02 g-0.08 g). We concluded that the method used in this study was useful for quantitative estimation of self-motion perception and could partly clarify properties of position perception for linear movement along the X axis.
