Unless I am missing something it is using a multivariate Gaussian. What the probabilistic programming is doing is inferring the mean and variance from the data

Unless I am missing something it is using a multivariate Gaussian. What the probabilistic programming is doing is inferring the mean and variance from the data. cons[i] ~ normal(mu_cons, sigma_cons) T[0.0,]; latency[i] ~ normal(beta * cons[i], sigma_latency) T[0.0,]; // latency is linearly related to cons These are both normal distributions.