There are different ways you could approach it, but here is one.
Reformulate your dependent variable (DV) to be looking time, then make Object (the object they're looking at--floor, painting, etc) and Independent Variable. You could then test if they looked more at the painting than the floor in general, and if differences occur more in some lighting conditions (using an interaction term).
This approach has a few issues however. The DV might be negatively correlated in the different conditions just because the more time you look at one object, the less time you look at the other. In fact, if you have only two objects, it wouldn't work, because if you know how much time someone looked at one, by definition you know how much time they looked at the other. But if you're comparing looking times for two objects out of say, 8, they're less likely to be correlated.
The other thing is now it's a repeated measures study, so you need to treat it that way.