Serial dependence in Oculomotor inhibition uncovers stimulus predictability: what, where, and when

Background: Making predictions about when and what comes next is a central brain function. We have previously found that the phenomena of Oculomotor Inhibition (OMI), in which microsaccades and eyeblinks are inhibited in response to perceptual events, is sensitive to violations of statistical regularities. Here we show that the OMI pattern reflects the buildup of perceptual predictions and their violation, and this applies to “what” as well as “when” in a similar manner. Method: In a set of experiments, observers passively viewed and silently counted sequences of 90 stimuli, briefly presented in around 1 Hz repetition rate at fixation. For testing the prediction of “what” we used separate experiments, each with a pair of stimuli presented in random order: faces-houses, color, auditory pitch, and contrast. For testing the prediction of “when” and the buildup of temporal anticipation, we used two randomly interleaved inter-stimulus intervals that differed by 100-500 ms in separate conditions. Results: We analyzed the results in terms of the serial dependency of the OMI on the history of preceding events. We found that repetition of the same stimulus (e.g., red in colors) shortened the OMI for the corresponding stimulus, while a change of stimulus increased it, with a magnitude that changed systematically with the number of recent (4-6) preceding events. We further found a similar OMI pattern for testing prediction for “when”. Conclusion: When perceiving a sequence of visual stimuli, the eyes freeze (OMI) for a shorter time for repeating stimuli (~10 ms per repeating item) and longer time for change, reflecting a serial dependence of the OMI. We interpret the results as reflecting a priming effect which implements an ongoing process of computing implicit predictions based on the recent past. This interpretation fits the data for predicting “what”, as well as “where” and “when”.