Quadratic approximation is used to determine local maxima and minima by analyzing the behavior of a function near a point using its second derivatives. For multivariable functions, the Hessian matrix, which consists of second partial derivatives, is crucial in identifying the nature of critical points. A positive definite Hessian indicates a local minimum, while a negative definite Hessian suggests a local maximum. The concept of visualizing the function as a 3-D surface helps in understanding the location of bumps or dips that correspond to these extrema. Completing the square for quadratic functions can also aid in finding the vertex, which represents the maximum or minimum point.