This is how I have always thought of it in simple terms. Take an example of a coloured object, such as a blue notebook. Now when the white light (from the sun or the ceiling light etc) hits the book the pigment molecules in the notebook absorb all of the wavelengths of light except the blue wavelengths. So what you are seeing is basically the "left over light". Of course this is not completely black-and-white (lol excuse the pun). So coloured objects can absorb over a range of wavelengths and your eyes will detect colours which are a mixture of wavelengths.
Now on the table you are talking about you will probably notice that an object that is percieved as blue will absorb mostly orange light (580-620nm or so). So if you were to start with a light bulb that only emitted orange light (or wavelengths 580-620nm) in theory you would not see any colour for the notebook.
So here is where the spectrophotometer works. Say for example instead of a blue notebook you now have a blue solution. You put this in the spectrophotometer and can select the wavelength you require (in this case the most appropriate wavelength will be the "orange" one so maybe something like 600nm). It is most appropriate to select this wavelngth simply because you want to use the wavelength that is absorbed the best or most, this leads to better results.
Now the organge light will go through the sample and some of it will be absorbed. The amount that is absorbed can be detected by the spectrophotometer (it know the initial amount of light and the final amount of light and tells you this in terms of absorbance). The beauty of this is that the more concentrated your solution is in the compound that is absorbing the more light of that wavelength it will absorb, this is a linear direct proportionality relationship known as Beer-Lambert Law.