Deep reinforcement learning, as one of the powerful approaches in machine learning, has found extensive applications in controlling nonlinear and multivariable systems, including flying robots. However, conventional deep reinforcement learning algorithms still face challenges such as instability in convergence, high computational complexity, and sensitivity to network configuration. In this research, an emotional deep reinforcement learning controller based on the Inception network is introduced and compared with a conventional deep reinforcement learning controller for the path tracking problem in quadrotors. The proposed approach, inspired by the architecture of parallel networks, reflects the dual-path structure of the brain, including the logical and emotional pathways, enabling the formation of a bidirectional decision-making model. This mechanism enhances reaction speed, reduces cognitive load, and improves generalization in unfamiliar conditions. Additionally, in designing the reward function, both simplicity and intelligence are considered so that the agent can remain adaptable to diverse trainable movements. Simulation results indicate that the proposed controller offers significant improvements in flight stability, path tracking accuracy and learning efficiency compared to the conventional version.Furthermore, the integration of emotional components in decision-making processes allows the quadrotor to better respond to dynamic environmental changes, showcasing its potential in real-world applications.