Figure 6.

Comparison of approximate posterior distributions for different parameter priors. We show two-dimensional scatterplots of samples for each possible pair of parameters [(A) non-decision time T_nd vs. internal uncertainty $\widehat{\sigma }$, (B) T_nd vs. noise standard deviation σ, and (C) $\widehat{\sigma }$ vs. σ]. Green dots show samples from posterior with priors reported in Equation (45). Red dots show samples from posterior with double as wide priors. Lines visualize the corresponding means over samples. For comparison we also plot the parameter values found by EZ (blue star) and DMAT (black star). The fitted data were obtained from the “low evidence,” SHAM condition of the second half of the experiment of Philiastides et al. (2011). This specific condition showed the largest differences between posteriors (using the two different Gaussian priors), i.e., for all other conditions, we found smaller and often vanishing differences.