The simplest to understand -- if you know atoms and solids -- is that a quantum dot is intermediate between the cases.
Whereas an atom has few energy levels (and hence a sparse emission/absorption spectrum, since the energy levels are transitioned by pure energy -- photons, light), and whereas a solid has a quasi-continuous band of energy levels (meaning, practically speaking, any wavelength in the band will be absorbed; emissions however are a more complicated matter*), quantum dots have more energy levels than atoms, but still discrete, not so many as to have an effectively-solid block of levels.
*For example, LEDs emit light corresponding to the band gap (give or take a fairly small error spread, 5 or 10%). This is characteristic of direct bandgap semiconductors, while indirect bandgap semiconductors (like silicon) emit no light at all**.
**Actually you can make silicon emit light. You can add quantum dots (ha!) as a phosphor, or you can avalanche it (which causes the weak emission of green-yellow light). The latter is the mechanism through which Bob Widlar infamously posed the riddle: how can you generate a negative voltage from a 2N3055, simply by applying voltage or current to the remaining terminals? (You avalanche the E-B junction, and the B-C junction acts like a photodiode, generating a paltry -- but nonzero -- negative bias.)
The number of energy levels, and their spacing, is determined by the number of atoms interacting together. Equivalently, it depends on the geometry; so a qdot that's oddly shaped (rod shaped, say) will have different responses in different directions (which is perceptible, because photons can be polarized one way or another, and thus will only excite one axis or another).
The classic chemistry experiment is growing colloidal crystals of cadmium selenide (or related semiconductor), where the particle size is controlled by the concentration of reactants and temperature, and the fluorescence (light emission resulting from UV stimulation) color therefore as well.
The application to quantum computing is natural; anywhere there is a set of discrete states that are reasonably well separated in energy levels, and which can be potentially long-lived, is useful. The energy separation has to be much greater than the thermal energy at the operating temperature, so that random motion of the particles and surroundings does not upset the state***. Lifetime of a state depends on what forces cause it to relax to the ground state; in traditional semiconductors, an electron above the bandgap can transition below by getting trapped on a dopant or defect site, where it's much more likely to recombine. Qdots are nice because they are small crystals, which can potentially be completely free of defects. Alternately, a defect site can itself be used, as long as it's well separated from other defects.
***Or cause entanglements with the existing state; whatever the case, the effect is, time and heat mixes and scrambles the computer's state, and the information diffuses out into the surroundings, being diluted by meaningless ambient noise. Perhaps analogous to how your brain feels groggy, forgetful and incapable of concentration (coherent operation) when you're running a fever (high temperature!).
Maybe this, too, is not nearly basic enough; "state" and "bands" and "energy levels" can be hard to pin down as well. Let me know.
Tim