Reply to Vinken and Vogels.

In their comment, Vinken and Vogels [1] take issue with our claim [2] that "IT neurons encode long-term, latent probabilistic information about stimulus occurrence". They offer a biologically plausible model of our findings, which they argue is based on neuronal fatigue. However, like our account, their model includes latent variables that are modulated slowly with stimulus probability; models without such latent processes, such as those based on temporally local fatigue effects, cannot explain our findings. Although we share their desire for more clarity about the mechanisms underlying visual expectation, and appreciate their thoughtful critique, we argue here that their comment mostly restates our findings with a more complex model and alternative terminology.