Decision Model based on Lak et al 2019:

Reinforcement biases subsequent perceptual decisions when confidence is low, a widespread behavioral phenomenon

https://elifesciences.org/articles/49834

WORKS:

decision model (figure 3a in Lak et al 2019)

psychometric(stimulus vs choice average), subset psychometric for each previous stimulus, difference matrix, hard vs easy choice updating % (figure 1d in Lak et al 2019)

psychometric for each previous choice (figure 1c in Lak et al 2019)

exploration of alpha and sigmas effects on updating matrix

plot reward prediction error in abhängigkeit von evidenz (stimulus)

SEMI-WORKS:

for some reason the model updating percentages are a lot lower than in the lak et al paper

in the subset psychometric and previous choice psychometric the expected shift is not really visible as much as it should be

TODO:

go back to equidistributed stimulus and fix the bin stimulus function

implement proper reward bias blocks

function for generating blocks

plot psychometric in a meaningful way to show that the model adapts to bias blocks -> maybe psychometric for each block

psychometric for previous lean/previous

implement a way to model time investment

timeinvestment(value, confidence) = avalue+exp(bconfidence)

look at time investment task design

look at previous attempts to implement time investment