This paper presents a novel method for learning and tuning a fuzzy logic controller using reinforcements from a dynamic system. The Generalized Approximate Reasoning-based Intelligent Control (GARIC) architecture is introduced, which can learn and tune a fuzzy logic controller even with weak reinforcements, such as binary failure signals. The architecture includes a new conjunction operator for computing rule strengths and a localized mean of maximum (LMOM) method for combining the conclusions of multiple firing control rules. It also learns to produce real-valued control actions by integrating fuzzy inference into a feedforward network, which can adaptively improve performance using gradient descent methods. The AHC algorithm of Barto, Sutton, and Anderson is extended to include prior control knowledge of human operators. The GARIC architecture is applied to the cart-pole balancing problem, demonstrating significant improvements in learning speed and robustness to changes in system parameters compared to previous schemes.This paper presents a novel method for learning and tuning a fuzzy logic controller using reinforcements from a dynamic system. The Generalized Approximate Reasoning-based Intelligent Control (GARIC) architecture is introduced, which can learn and tune a fuzzy logic controller even with weak reinforcements, such as binary failure signals. The architecture includes a new conjunction operator for computing rule strengths and a localized mean of maximum (LMOM) method for combining the conclusions of multiple firing control rules. It also learns to produce real-valued control actions by integrating fuzzy inference into a feedforward network, which can adaptively improve performance using gradient descent methods. The AHC algorithm of Barto, Sutton, and Anderson is extended to include prior control knowledge of human operators. The GARIC architecture is applied to the cart-pole balancing problem, demonstrating significant improvements in learning speed and robustness to changes in system parameters compared to previous schemes.