chmathys
commented
9 months ago At the level of the generative model, the parameter rho is a constant drift parameter. This means that instead of assuming a Gaussian random walk without drift for rho=0 (the default), you will get a Gaussian random walk with drift when rho is non-zero.
In practical terms, a non-zero rho biases predictions in one or the other direction. This can be useful for instance when you want to estimate the extent (and possibly direction) of an individual's optimism bias.
On Tuesday, 13 February 2024 at 2:28 PM, a-yur @.***> wrote:
I have a question regarding one of the parameters. What is the meaning of the parameter "rho" and where was it introduced? It is present in the code, for example, it is define in tapas_hgf_binary_config.m as:
% Rhos % Format: row vector of length n_levels. % Undefined (therefore NaN) at the first level. % Fix this to zero to turn off drift. c.rhomu = [NaN, 0, 0]; c.rhosa = [NaN, 0, 0];
but I haven’t found it in the mathematical descriptions of the papers “A Bayesian foundation for individual learning under uncertainty”, Frontiers in Human Neuroscience, 2011 and “Uncertainty in perception and the Hierarchical Gaussian Filter”, Frontiers in Human Neuroscience, 2014.
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>
I have a question regarding one of the parameters. What is the meaning of the parameter "rho" and where was it introduced? It is present in the code, for example, it is define in
tapas_hgf_binary_config.mas:but I haven’t found it in the mathematical descriptions of the papers “A Bayesian foundation for individual learning under uncertainty”, Frontiers in Human Neuroscience, 2011 and “Uncertainty in perception and the Hierarchical Gaussian Filter”, Frontiers in Human Neuroscience, 2014.