In this blog post, I explain what I do as a cognitive modeller - from decision making modelling, to neuroscience, fitting mathematical models to human data and more!
This blog post uses the pmwg package for a more generic example than has previously been shown (here for example).
PMwG The particle metropolis within Gibbs sampling package (pmwg) was created by a team at the University of Newcastle.
A blog post to help with writing log likelihood functions to be used in PMwG. The examples shown in the PMwG Sampler Doc are quite specific, and so this blog aims to outline the general process of constructing a likelihood function.
So you’ve made the PMwG sampler work and now you have a sampled object, what’s next? Well now is where the fun stuff happens - checking, inference, plots and posteriors.
This blog post presents a brief guide on how to create synthetic data, and perform parameter recovery, using the PMwG sampler. Parameter recovery is highly important in cognitive modelling procedures as it provides an indication of parameter identifiability (and in a way, model “power”).
There has been substantial research and debate over the non-decision time parameter of evidence accumulation models. For LBA and diffusion models, there is often a discrepancy, where diffusion models show longer t0 times, with the LBA known to poorly estimate this parameter.