Wait... they're not a blank slate?
Individuals build their knowledge by making connections to existing knowledge.1 This previous knowledge must be considered when trying to learn new content as it may not be correct, or students may not understand how to make the connections.
A few misconceptions seen in mechanics, which are easily revealed with the FCI:
The impact of previous knowledge can be seen when the normalized gain is plotted against prescore. Below is an example of this from an AJP paper.
This correlation is not always seen. In classes that are taught with traditional methods (lecture), the gains are small for all students. Also, if a population of students has all low, or high, prescores, the correlation will not be readily apparent. (This correlation does not rule out the strong possibility that both the prescore and normalized gain are dependent on a third variable.)