Turing reduction is a method used in computability theory to determine if the solution to one problem can be efficiently transformed into a solution for another problem using a Turing machine. It provides a way to compare the relative complexity of decision problems by showing that if one problem is solvable, another can be solved using it as a subroutine.