SDT: Gaussian Signal Detection Theory Model for Binary Responses

In EMC2: Bayesian Hierarchical Analysis of Cognitive Models of Choice

Gaussian Signal Detection Theory Model for Binary Responses

Description

Discrete binary choice based on continuous Gaussian latent, with no rt (rt must be set to NA in data).

Usage

SDT()

Details

Model parameters are: mean (unbounded) sd (log scale) and threshold (unbounded).

For identifiability in one condition two parameters must be fixed (conventionally mean=0 and sd = 1). When used with data that records only accuracy (so reponse bias cannot be evaluated) a single threshold must be assumed and fixed (e.g., threshold = 0).

At present this model is not fully implemented in C, but as its likelihood requires only pnorm evaluation it is quite fast.

Value

A model list with all the necessary functions to sample

Examples

dprobit <- design(Rlevels = c("left","right"),
           factors=list(subjects=1,S=c("left","right")),
           formula=list(mean ~ 0+S, sd ~ 1,threshold ~ 1),
           matchfun=function(d)d$S==d$lR,
           constants=c(sd=log(1),threshold=0),
           model=SDT)

p_vector <- sampled_pars(dprobit)

EMC2 documentation built on April 11, 2025, 5:50 p.m.