Abstract
Adaptive Bayesian procedures are widely used in psychophysics to estimate detection thresholds. Simulations have revealed the importance of the parameters selected for the model likelihood function in obtaining threshold estimates with small bias and standard error (Alcalá-Quintana & GarcNúcleoa-Perez, 2004). One recommendation is that the spread (sigma) of the model function has to be bigger than the real sigma of the psychometric function of the subject. The selection of a smaller sigma can introduce a bias in the threshold estimation and increase the standard errors. Here we wanted to estimate the value of sigma and find whether it changes with participant's age using three different stereoacuity tasks. Studies were performed using a stereo display with polarized glasses. In all studies we used an adaptive weighted one-up one-down staircase procedure with 80-120 trials and a logistic psychometric function was fitted to the probabilities of correct detection in order to estimate stereoacuity thresholds and sigma. In the first study, we used a spatial two-alternative forced-choice (2AFC) task for detecting a 3D square created by introducing horizontal disparities in dynamic random dots. In the second study we used the same stimuli but in a spatial 4AFC task. In the third study we used 4AFC too but using a computerized version of the clinical Randot stereoacuity test. In the first study we tested a population of 70 people (aged 4.6-61 years, mean 20 years). Results showed that sigma (mean=1.23,SD=0.88) was independent of age (r=-0.01, p=0.92) although it correlates with stereoacuity (r=0.26, p=0.02; sigma increases by 1.23 for every doubling in threshold). The second and third studies were performed with younger populations (4.74-21.8 years, mean 10.3 years) but results were similar; again sigma did not correlate with age. Therefore, to estimate stereoacuity thresholds with adaptive Bayesian-staircases we recommend using sigma = 1.8 independently of age.
Meeting abstract presented at VSS 2016