Exploring Character Recognition through Modeling: Cognitive Reverse Engineering

What makes letters harder or easier to read? Moreover, what factors are most important to the brain for successful character recognition? These were the concepts investigated in this study. A “critical point theory ” was developed that suggested that the brain uses intersection points and points of inflection in letters for pattern recognition. An experiment was developed in which letters were shown to subjects that had certain factors controlled such as occlusion of the letter, skew, and frequency of the letter in English. Data was collected as the time it took for the subjects to name each letter. A multiple linear regression model was developed, analyzed, and refined based on this data. The model was intended to be used to recognize what factors were significantly important for character recognition, and to predict response time of hypothetical letters. Statistical evidence of the validity of the critical point theory was found. However, the study suffered from poor experimental design, and the resulting regression model was unsatisfactory for use in prediction. A follow-up study with a revised experiment is suggested, and our results are discussed. Most of the higher-level functionalities of the brain are still mysterious unknowns. W