Predicting change: Approximate inference under explicit representation of temporal structure in changing environments
Fig 5
Example of a reversal schedule for simulated data.
In the condition with irregular reversals, the between reversal intervals were sampled from a geometric distribution (top right plot), whereas in the condition with (semi)regular reversals, the intervals were sampled from the negative binomial distribution with variance σ = 20 (bottom right plot). In both cases the mean interval duration is set to μ = 20.