Information diffusion and disease spreading in communication-contact layered network are typically asymmetrically coupled with each other, in which how an individual being aware of disease responds to the disease can significantly affect the disease spreading. Many recent studies have demonstrated that human behavioral adoption is a complex and non-Markovian process, where the probability of adopting one behavior is dependent on the cumulative times of the received information and the social reinforcement effect of these cumulative information. We study the impact of such a non-Markovian vaccination adoption behavior on the epidemic dynamics and the control effects. We find that this complex adoption behavior caused from the communication layer can significantly increase the epidemic threshold and reduce the final infection rate. By defining the social cost as the sum of the cost of vaccination and the cost of treatment, we show that there exists an optimal social reinforcement effect or optimal information transmission rate allowing the minimal social cost. We also develop a mean field based theory to verify the correctness of the simulation results.