Fig 6. Illustration of simplified detection model used to interpret the simulation results.

The continuous-time problem is approximated by a discretized process, obtained by “sampling” trajectories at times separated by the correlation time. In the spontaneous state, the probability that a time sample is not beyond the decision barrier θ is p₀. A: The peak decreases the local probability of not reaching θ, i.e. p₁ < p₀ and the effect size is ${\overline{Y}}_1$. B: The trough increases the local probability of not reaching θ, i.e. p₂ > p₀; the effect size is here ${\overline{Y}}_2$. C: in the case that both features are present and the changes in probability are small, the effect size is ${\overline{Y}}_{12}\approx {\overline{Y}}_1+{\overline{Y}}_2$, see text and Eq (22).