Abstract
Adaptive testing methods serve to maximize the information gained regarding the values of the parameters of a psychometric function (PF). Such methods typically target only one or two (‘threshold’ and ‘slope’) of the PFs four parameters while assuming fixed values for the ‘nuisance’ parameters (‘guess rate’ and ‘lapse rate’). However, due to some redundancy among the parameters of the PF (e.g., Prins, J. Vis. 12(6):25, 2012) bias in parameter estimates results when the assumed values for the nuisance parameters do not match the generating values. This bias may be alleviated somewhat when observations are fitted in a subsequent procedure in which the nuisance parameters are allowed to vary. When this is done however, the uncertainty regarding the values of the nuisance parameters (not addressed by the adaptive method) contributes to the uncertainty in the values of the parameters of interest. Here I propose the Psi-marginal adaptive method which addresses this issue specifically. The method is based on Kontsevich & Tyler’s (Vision Res. 39(16):2729-37, 1999) Psi-method. However, in the proposed method a posterior distribution defined across all parameters of unknown value is maintained. Critically, selection of stimulus intensities is based on the expected information gain in the marginal posterior distribution defined across the parameters of interest only. The appeal of this method is that it will target nuisance parameters but only when doing so maximizes the expected information gain regarding the values of the parameters of interest. The method is extremely flexible: any combination of the PF’s four parameters may be included in the posterior and any combination of those included may be treated as nuisance parameters. Simulations indicate that use of the Psi-marginal method results in smaller bias and higher precision in threshold and slope estimates compared to the original Psi method.
Meeting abstract presented at VSS 2013