Irina Lavrova¹, Miranda Scolari²; ¹Texas Tech University, ²Texas Tech University

Implicit learning refers to the abstraction of structural regularities without conscious awareness, which widely studied using artificial grammar learning (AGL) paradigms. Eye movements can offer additional insights into the mechanisms of implicit learning, serving as involuntary indicators of acquired knowledge (Silva et al., 2017), and therefore revealing underlying learning processes. Across two studies, we tracked adults’ eye movements while they completed AGL tasks to test whether gaze patterns would reveal sensitivity to grammatical violations beyond what standard grammaticality judgments capture. In Study 1, participants (N = 65) were exposed to words generated by grammars of varying complexity while making same/different judgments that did not require explicit memorization. In a subsequent surprise grammaticality judgment test, participants showed increased dwell time (χ² = 4.57, p = .032) and more regressions (χ² = 24.89, p < .001) to letters that violated the grammar. They also fixated longer on simpler first-order violations compared to more complex second-order violations. Interestingly, participants’ accuracy judgements were not a significant predictor of eye movements. Study 2 tested whether this eye-movement signature would transfer to a new surface vocabulary. After exposure, participants (N = 44) judged strings following the same underlying grammar structure but composed of entirely novel letters. Participants again fixated longer on rule-violating elements (χ² = 143.48, p < .001). Furthermore, dwell time did not differ for violations created by repeated versus unique characters, indicating that participants did not rely solely on repetition patterns when processing a new symbol set. Instead, the data suggest that they had abstracted the underlying grammatical structure in a way that generalized beyond the specific letters or surface features of the stimuli. Taken together, these findings demonstrate that eye movements serve as indicators of implicitly learned patterns, their complexity, and internal structure, revealing learning that remains largely inaccessible to conscious awareness.