This paper presents a binary feedback flow control scheme which uses the gradient of the system performance function to generate feedback from the network to the data sources. The setting of the single bit congestion indication communicated back to the users is based upon the sign of the gradient of the system performance function. The proposed scheme uses a simple gradient optimization method and a neural model of the system dynamics to determine the gradient of the performance function. Simulation results are presented to compare the performance of the proposed scheme to the conventional queue thresholding approach.