Invalid expectations induce perceptual bias yet enhance the accuracy of metacognitive judgments

Humans live in a world filled with uncertainty. To adapt to and process changes in their surroundings, the brain continuously generates predictions about the environment. Predictions can guide perception and action. However, given the ambiguous and highly unpredictable nature of the external world, individuals often lack sufficient information to form accurate predictions. This study investigates how non-informative prediction, specifically invalid prediction, shapes perceptual judgments and subsequent confidence estimates through five experiments. In a face/house judgment task, participants were required to determine whether a presented blurry image depicted a face or a house, and then rate their confidence in that perceptual judgment. Before the image was displayed, participants were asked to predict the category of the upcoming image (face or house). Experiment 1 (n = 38, perceptual accuracy ≈ 64%), Experiment 2 (n = 45, perceptual accuracy ≈ 75%) and Experiment 3 (n = 48, perceptual accuracy ≈ 84%) systematically manipulated task difficulty to examine how non-informative predictions modulate perceptual judgments and confidence ratings. Experiment 4 (n = 41) introduced a condition without predictions to provide a baseline for evaluating the predictive effects. In Experiment 5 (n = 40), the response keys for predictions and perceptual judgments were separated to eliminate potential action-related effects.The results revealed that: (1) non-informative predictions induced perceptual biases under high and moderate task difficulty conditions (except in Experiment 3 with low difficulty), systematically biasing individuals to align their perceptual judgments with prior predictions; (2) non-informative predictions affected subjective confidence, as participants reported higher confidence in trials where their perceptions aligned with their predictions compared to trials where they did not; and (3) non-informative predictions enhanced the accuracy of metacognitive judgments, with individuals exhibiting greater metacognitive efficiency when their perceptual judgments matched predictions compared to mismatched trials. In summary, while non-informative prediction induces perceptual biases, it enhances metacognitive judgments. These findings highlight the distinct effects of non-informative predictions on perceptual judgments and metacognitive functions. The study contributes to a deeper understanding of the interplay among expectations, perception, and metacognition. Finally, it offers practical implications for optimizing cognitive decision-making and improving metacognitive accuracy.