Correlation of Prism Adaptation with other Measures: A Cautionary Note

Abstract

The effects of deviated vision have been studied for some years and, given suitable conditions, adaptation has occurred with great reliability. However, the magnitude of adaptation has varied considerably among subjects. Some attempts have been made to link these differences with other measures of individual performance. Wilkinson ( 8 ) reported a slight connection between susceptibility to autokinesis and adaptation of internally registered eye-in-head position, and Melamed, Wallace, Cohen, and Oakes ( 2 ) found a correlation between field-independence and the 'correction effect.' Kahane and Auerbach (1) claimed that experienced dancers adapted less than non-dancers, but their measuring technique makes evaluation of their results difficult. Smothergill ( 4 ) found no correlation between visual capture and sensitivity to visual or proprioceptive position cues. N o significant correlations have been found between adaptarion to deviated vision and kinesthetic o r visual aftereffects (7), motor skill ( 5 ) , personality ( G ) , or adaptation to tilt ( 3 ) . The relative lack of correlations cannot be due to homogeneity of adaptarion between subjects; on the contrary, individual differences are very marked, which might reflect lack of consistency in the individual's adaptability as inspection of some previously collected experimental data suggested. Therefore, 16 subjects were tested in balanced order in two similar experiments on prism adaptarion of registered eye-in-head position. The product-moment correlation between experiments was 0.17. A further 16 subjects were tested in two more experiments involving substantially identical exposure conditions. Correlation between conditions was -0.19. T o attain a significance level as low as P = 0 .1 , the critical value of r with 14 d f has to reach 0.43. In every experiment, adaptation over all subjects was significant ( P < 0 .025 ) . It is strange that such a robust effect should show such a low test-retest reliability, but until the reason for this is found, it would seem prodigal of effort to pursue correlations between prism adaptation and other measures.