The Neural and Computational Basis of Controlled Speed-Accuracy Tradeoff during Task Performance