If the new knowledge is similar to the existing basis and structure of our knowledge, but requires some adjustment of that structure, it can be well-integrated by adjustments to part of the knowledge structure.
If the new knowledge simply extends the existing basis by extrapolating from it or interpolating within it, the energy this integration requires is quite low. This integration may be done metaphorically, by matching the new knowledge to a representation of the old knowledge. This metaphorical approach uses a powerful capability of the human brain. We assert this (something new) is like that (something we already know). This process allows us to “reuse” the existing framework. It works well when “this” truly is like “that” and we can recognize and allow for the differences. Problems arise when either the two things are actually different (and we are using a benchmark that is invalid) or how we perceive the similarities is flawed (our benchmark-matching process does not work).
Example: learning languages often uses this partial-integration technique. While people may learn the new words and their contexts, they tend to retain much of the structure of their native language. For instance, the Russian language does not use the definite article as does English, French, German and other languages. As a result of this, a native Russian speaker may tend to omit the definite article when speaking one of these languages.