Figure 6.

Recovery of parameters for duration distributions from analytic buildup functions. Recovery of percept duration distributions from a buildup function. We fitted parameters for the analytic buildup function by minimizing the squared error between the original and the fitted buildup function. This allows us to estimate the parameters of the underlying gamma density functions for percept durations. (A) Recovery of generating gamma density functions from buildup functions that were computed via Monte Carlo simulation (1000 trials). Usually the density functions corresponding to the fitted buildup function were a good match for those that had generated the sample buildup function. (B) Because the fitting process was very sensitive to noise, the density functions obtained from the fitted buildup function occasionally differed from the generating duration distributions. (C) From buildup functions that were computed analytically using our model, we always recovered the parameters of the gamma densities (i.e., α₀ = 1.47, μ₀ = 6.47; α₁ = 3.14, μ₁ = 21.45 for this buildup function) that had generated those buildup functions.