FitProbit: Fit probit psychometric function

In armanabraham/chb: Choice history biases

Description

Fits probit (cummmulative Gaussian) to proportion of rightward responses. The best fit after Weibull (with log transformed contrast values). Can fit with or without lapse rate.

Usage

FitProbit(dat, fitWithLapseRate = FALSE)

Arguments

dat	prorotion of rightward responses computed using the function ProportionRightwardResponses
fitWithLapseRate	if TRUE, fits probit that takes into account the lapse rate, which is computed as a proportion of incorrect responses to the highest stimulus intensity (this is safe to do for our experiment, cause we deliberately selected highest stimulus intensity to be 100 part of 'dat'.

Examples

oneRun <- droplevels(subset(glmData, SubjectID=='s025' & SessionID==3))
pRight <- ProportionRightwardResponses(oneRun)
probitModel <- FitProbit(pRight)
range <- c(-0.08,0.08)
predicted <- PredVals(probitModel, xvar="Contrast", yvar="pRightCorrect", type="response", xrange=range)
ggplot(pRight, aes(x=Contrast, y=pRightCorrect))+
 ylim(0,1) +
 scale_x_continuous(limits=range, breaks=seq(range[1],range[2], 0.02)) +
 geom_point() +
 geom_line(data=predicted, aes(x=Contrast, y=pRightCorrect))

armanabraham/chb documentation built on May 10, 2019, 1:39 p.m.