Predicting memory for pictures from rankings of interestingness, pleasingness, complexity, figure-ground and clarity

Pictures used in the test phase of a picture-recognition task were ranked by 10 university students for interestingness, pleasingness, complexity, figure-ground, and clarity. The scale values obtained from these rankings were then correlated with the errors made by other subjects in the test phase of the picture-recognition task. Figure-ground and clarity were found, for the most part, to be reliable predictors of the errors made to both the "old" and "new" subsets of pictures in two experiments. Complexity, while not as consistent a predictor of either figure-ground or clarity, was a reasonably good predictor. Interestingness and pleasingness were not reliable predictors of errors.