Recently, the rapid development of artificial intelligence has led to the introduction of an array of different technologies. One such technology is fuzzy tilling activation. ReLU activation has been shown to work better for deep learning neural networks than other types of activation functions, but fuzzy tilling is a new type of activation function that has not been tried before. This work investigates the properties of FTA. Our work found that the use of tanh normalization is not dependent on bounds and can enhance the performance of FTA without any need for further tuning in the given environment. As a future direction, exploring the robustness of tanh-normalized DQN-FTA across multiple environments would be an intriguing prospect.