Perception is molded by sensory input, prior knowledge, and contextual cues, collectively giving rise to perceptual biases whose neural mechanisms remain poorly understood. This study aimed to elucidate these mechanisms by examining prefrontal cortex (PFC) neural recordings from monkeys performing a vibrotactile frequency discrimination task (Fig. 1a). Monkey’s behavior exhibited the contraction bias (Fig. 1b), a contraction of the perceived first frequency towards its mean. This bias increased with longer delay periods, suggesting a Bayesian explanation of this effect. Examination of PFC’s neuronal firing indicated that the activity reflected prior knowledge of the stimulus set, bolstering the Bayesian explanation. Additionally, PFC neurons encoded mutual information about the current stimulus but not about those in preceding trials, though information on stationary sensory history might still be present. Based on these findings, we proposed a novel Bayesian model that combines prior knowledge with current observations, obtaining a Bayesian estimate of the first stimulus. To explore the link between behavior and PFC activity, we analyzed the relative distances between neural trajectories in state-space for each first frequency value (Fig. 1c). Interestingly, from stimulus presentation through the delay period, these trajectories represented a sigmoidal transformation of stimulus frequency. Notably, this representation aligned with the predictions of its Bayesian estimator (Fig. 1d). This transformation of the first stimulus into its Bayesian estimator shows that PFC neural dynamics align with Bayesian concepts. This result enhances our comprehension of the neural basis of perceptual biases, underscoring the role of the PFC in shaping perceptual experiences.