Ann Hyland gives a brief overview of the development of the saddle in The Medieval Warhorse.

One of the first saddles we know of comes from the Pazyryk burials, dated to the 5th century BC. It was a pad type saddle, with two leather cushions attached to a rigid front and rear bow arch. These bow arches were quite small, certainly not cantles, but they did offer some security to the rider.

Here we lose the development of the saddle and pick it back up in the first century AD, where the Romans are using four tree saddles. These saddles have a similar saddle bow to the Pazyryk saddles, but a wooden post is attached to either side of the front and rear saddle bows. The trees provided curved in to extra support to the rider and allowed for more abrupt movements of the horse and provided extra stability when using a sword. In some cases, even this early, a metal plate might be attached to the trees, forming a cantled saddle.

By the 5th century, the use of a high cantled saddle appears to have been commonplace, and the design of the saddle doesn't seem to have changed much into the High Middle Ages, where my sources fail me again.

Interestingly, the use of a saddle - including cantled saddles - is not necessary to produce a strong strike with a lance, and it appears to allow for more consistent energies than some saddles. I suspect, therefore, that when these saddles were introduced, the biggest advantage was the stability they offered the rider during maneuvers and hand to hand combat. The stirrup further enhanced the stability of the rider, but did not noticeably alter the energy produced by a lance strike. It took much latter, much better (for jousting, at least) designed saddles to noticeably improve on bareback lance use.