During general anesthesia, anesthesiologists who provide anesthetic dosage traditionally play a fundamental role to regulate Bispectral Index (BIS). However, in this paper, an optimized type-2 Self-Organizing Fuzzy Logic Controller (SOFLC) is designed for Target Controlled Infusion (TCI) pump related to propofol dosing guided by BIS, to realize automatic control of general anesthesia. The type-2 SOFLC combines a type-2 fuzzy logic controller with a self-organizing (SO) mechanism to facilitate online training while able to contend with operational uncertainties. A novel data driven Surrogate Model (SM) and Genetic Programming (GP) based strategy is introduced for optimizing the type-2 SOFLC parameters offline to handle inter-patient variability. A pharmacological model is built for simulation in which different optimization strategies are tested and compared. Simulation results are presented to demonstrate the applicability of our approach and show that the proposed optimization strategy can achieve better control performance in terms of steady state error and robustness.