Ok, cyclogenesis, the technical term for generation of a tropical cyclone (aka hurricane or typhoon) requires a few key things. One is a very large pool of warm water. (Warm water is key, as several other responders have pointed since convection driven by evaporation of water is the energy source. (This is because a tropical cyclone (TC) is a big engine that turns warm water into wind.)) The second thing is conditions where there is very little vertical shear, meaning the winds aloft aren't significant. (I'll come back to shear in a bit.) The third thing you need is a space big enough for the convection, or evaporation of water, to get organized, which is where lakes get affected.
In the formation of a tropical cyclone, first the convection has to set up into plumes, which we see as thunderheads. Once you get enough of these in the same general area a couple of things happen. One is I run away because those storms are big and scary looking. The second is that the rising air from all these big plumes in one general area leads to an airflow towards the convection source. If that happens on a big enough area, Coriolis (see below) comes into play and the moving air starts to curve. The curving motion of the air due to Coriolis is where the TC gets its characteristic spinning motion. (And since Coriolis goes the other way in the Southern Hemisphere, you will find Australian typhoons spin counter to hurricanes in the Atlantic. Anyway, with this spinning, the convection now organizes into bigger and bigger cells and once these cells close into a circle, voila, c'est tropical cyclone and I am heading inland.
Lakes have water that is plenty warm enough to get a TC going, but they lack size to get the convection organized and get Coriolis involved. Even the Great Lakes aren't big enough for that to happen. In contrast, not to dispute a previous responder, but the Gulf of Mexico is plenty big enough for hurricanes to spin up, Coriolis works just fine and only takes a few hundred kilometers or so. There are lots of cases where a tropical depression went to raging huge cyclonic storm in the Gulf.
Ok, I've rambled on a long time and you are starting to grumble: "What does the damned shear have to do with anything?." Well, see, if there is a lot of wind shear, the convective plumes gets blown away, sort of, before they have time to organize and get Coriolis involved.
Some of you may be wondering what the Sam Hill is Coriolis. That is a fictitious force that arises when you move things around on a spinning spherical body. Nobody wants me to try to explain Coriolis in detail, the last time I tried there were several fatalities from people chewing their brains off to get away. In a nutshell, Coriolis is caused by the fact that things rotating around on a sphere have different tangential velocities depending on their latitude on the sphere. Things towards the poles are moving slower than things on the equator. In the context of air, in the northern hemisphere, if air from the north is rushing south to replace air being convected aloft in a hurricane, the southward moving air isn't moving as fast to the east as the air already south so the southward moving air deflects westward. see here:
http://en.wikipedia.org/wiki/Coriolis_effect
Anyway, the bottom line is that lakes simply aren't big enough to have coriolis play a role in getting any convection organized. In addition, in general the surrounding land masses probably generate a lot of vertical shear due to onshore/offshore flows.
Free beer if you read this far.