Wiener filter estimation of transfer functions.

Kessel (2004) built a cleaner way to read messy data. He used a Wiener filter on repeated-trial records. The tool pulls the true response pattern out of the noise.

The paper is a how-to, not a new drug or reward system. It shows the math and gives code so you can copy it.

The filter cut the error in transfer-function plots. Lines looked smoother and matched theory better. No numbers were given, but the gain was visible.

Chou et al. (2010) also fight noise, but at the human end. They showed observer bias can twist scores as much as random jitter. Use both fixes: filter the data and rotate your staff.

Périkel et al. (1974) push Bonferroni t tests for the same kind of repeated-trial data. Robert adds a pre-clean step before any stats. Think of Wiener first, then correct for multiple tests.

Sosa et al. (2022) zoom out and call for a full feedback-control view of behavior. Robert gives the wrench you need before you can even speak that language.

If you run fluency charts, matching-law probes, or any repeated timing sessions, your raw counts bounce. One noisy probe can hide a trend. Run the Wiener script (it is short) and plot the smoothed line next to the raw one. Your eye will spot true changes faster and you will make program tweaks with more confidence.