If we really understand something, we are able to express it in words, at least for ourselves. Having the right words available may be essential. Language plays a basic role in thinking, If something new is found, if a new idea pops up the right word can crystallize it in language. By this, it can be used as a building block in further reasoning.
What holds for thinking also holds for observations. In reporting them characterizations may be needed and finding the right words can be essential. Classes of observations need names. They are used for memorizing and communication. Pattern recognition as a human ability is strongly related to the art of naming. Seeing a difference, being able to characterize it, finding or creating a concept should end in determining a name. Using the name brings back the concept. Being aware of the concept is needed to determine whether a new observation fits the related class.
Denying or emphasizing differences
Sometimes concepts seem to overlap strongly, or their difference is judged as irrelevant. It may be proposed to remove one of the names or to publicly state the equivalence. I encountered a striking example when I tried to distinguish two types of pattern recognition problems: problems in which the objects are human created, e.g. text, industrially produced items or art, and problems related to natural phenomena as in earth sciences or astronomy. I used the words cultural and natural. In one of the comments it was stated that this difference is non-existing as man-made objects are also natural as humans are a part of nature. So I tried to make a non-existing distinction.
Following this reasoning culture and cultural can be everywhere replaced by nature and natural. It simplifies the world by denying that it makes a difference whether human thinking or action is involved or not. However, discussions and thinking are impoverished when these concepts are not longer distinguished. Accepting that they are different on the other hand enriches our possibilities to express ourselves.
Pattern recognition and machine learning
The distinction between pattern recognition and machine learning fits in this story. Are these names referring to the same field or can they be distinguished? There are historical differences, but for many it is almost the same. A large part of the papers published under the flag of one of them could also have the other. So why distinguish them?
The answer is that it is an opportunity. It gives us the possibility to point shortly into a direction for which otherwise a longer explanation is needed. We may start to use the two names as follows. Machine learning emphasizes the mathematical description of the generalization tools. Pattern recognition uses these tools in recognition applications. It thereby emphasizes the study of representations. Properties of generalization tools are needed to construct a proper representation.
By using the two names like this, e.g. as key-words or for the titles of conference sessions, we make the distinction useful. Readers and participants will thereby better know what to expect. It should definitely be avoided to use machine learning and pattern recognition as referring to the same field. They should preferably be used as two directions of research: the mathematical description of the tools or their applicability.