Determining Visual Acuity Thresholds: A Simulation Study of Stimulus Presentation Strategies

摘要

An accurate and economical method for determining visual acuity thresholds was necessary for many tests in the Vision Test Battery (VTB). There were certain unanswered questions as to how the classical up and down method of presenting stimuli would perform in this regard with a four alternative forced choice task. To answer these questions computer simulation runs of the up and down method were carried out. A model of the psychometric function relating probability of correct detection to stimulus gap size was based on the cumulative normal distribution. The parameters of this model were chosen to characterize one of the acuity tests in the VTB. The intent of the simulation with this model was to generate a large sample of acuity thresholds, and, from this sample, form an estimate of the average acuity threshold and its variability for any given number of trials. An alternative method of presenting stimuli in a sequential manner was also studied via simulation. This method, called the bracket method, proved to be superior to the up and down method in calculating an average estimate of threshold acuity. The error attached to the threshold acuity estimate was also smaller in the bracket method. When the slope of the psychometric curve was varied over a large range, the bracket method retained these advantages over the up and down method. (Author)