Alphabetic Letter Perceivability, Similarity, and Bias

The similarity of the letters in the Latin alphabet has been measured numerous times using a variety of methodologies and stimuli. We provide a comprehensive review of past attempts to construct similarity spaces of the Latin alphabet, and determine that accuracy has frequently been attributed to a subset of three common sources: perceivability, bias, and similarity. We present the results of two new experiments which, for the first time, allow for the simultaneous estimation of these factors. We found that perceivability was affected substantially by the visual similarity between target and mask, whereas the similarity space did not vary substantially with the mask, and the impact of bias was small. Finally, we demonstrate a method (related to choice theory) by which these three factors can be estimated simultaneously for our experiments.