Figure 4.
The drift diffusion model.
As the time between updates decreases to zero (whilst keeping information content constant, independently sampled from a normal distribution, per unit of time), information moves according to a Brownian motion (Weiner process) with drift (positive drift if the item is good to approach).