Myth 1: Plates Are Rigid
Unlike dinner plates, lithospheric plates are not truly rigid, just stiff with a brittle crust on top. Rocks can and do deform, not only within the lower crust and upper mantle (that is, most of the lithosphere), but far from the active edges of plates. And of the world's plate boundaries, marked by crisp lines on the map, about 15 percent are actually soft and diffuse. The best example is the Tibetan Plateau.
Myth 2: Spreading Ridges Push
The thought (and footage) of red-hot lava rising at the deep midocean ridges plants the notion that rising magma is thrusting the plates apart. But spreading ridges are passive features. The main driving force of plate tectonics is gravity, specifically the downward fall of subducting slabs at the other end of the plate. There is a much lesser driving force called "ridge push," because the seafloor slopes downhill away from ridges—but at the ridge itself this too is a passive pull. It's the release of pressure where the ridge pulls apart that allows mantle rock to melt and rise by buoyancy, not the opposite.
Myth 3: Ridges Are Fixed
You always see pictures of Africa and South America splitting apart with the Mid-Atlantic Ridge sitting exactly between. Even though new oceanic crust usually moves away from ridges in both directions, the ridge itself moves sideways too. Consider Africa, almost surrounded by spreading ridges created as the Americas, Antarctica and India split away from it during the breakup of ancient Pangea. If you move those neighbors back toward Africa, the ridges move too. This is universal. As spreading ridges move, they crawl across the whole upper mantle releasing magmas from below. The geochemical record of those different magmas gets smeared in the process, homogenizing mid-ocean ridge basalts (MORB) and hiding much of the variation in the mantle beneath.
Myth 4: Subduction Is Forcible
Kids and writers love the image of plates crushing and grinding each other like bull hippos, until one plate is forced downward to its molten death in the hot mantle. But the process is more like two large men passing in a crowded streetcar. If there's any shoving it's brief—once a cold oceanic plate starts subducting, the dense slab sinks like a stone. The oldest, coldest slabs may sink straight downward. In fact, many slabs may sink faster than they move into the subduction zone. That makes their bending point move backwards in the process called slab rollback, sucking the other plate toward them. As the upper plate stretches into the gap left by the slab, it gives rise to back-arc extension, not compression.
Myth 5: The Mantle Convects
Rock comes up at spreading ridges, moves sideways on oceanic plates and goes down in trenches. This can give rise to a deceptive picture of an organized cycle driven from below, like a kettle of soup with a thick scum of continents. No wonder so many sites, even smart ones, close the cycle and draw nice rolling convection cells in the mantle. Many even repeat the early notion that convection cells are what drives plate tectonics. If you understand plates, though, the need for this kind of organized, cyclic convection vanishes. A more realistic word would be simply "mixing," but in fact that's all the word "convection" means. So the mantle indeed convects, but not in convection cells.
PS: If you dispense with these misconceptions, the case for hotspots and deep-mantle plumes weakens too. Here's more about that.