Humans are remarkably efficient at identifying motion of a biological origin. Computers have thus far been unable to emulate this ability. A gait recognition system which combines knowledge of the human visual architecture with computer vision techniques is proposed. The basic premise of the system is to model the human body as an articulated structure composed of straight line segments. The limbs of a walking individual are detected using the Hough transform, a technique which shares similarities with the primary visual cortex. The angles of the detected limbs are used as features to train and test a hidden Markov model (HMM) classifier. The system obtains a correct classification rate of 65% with a standard motion database. This performance is comparable to other leading systems.