The answer, which is complex in detail, can happily be summarized as follows: new things are understood when they are reshaped by old things to fit in with them.
The wording of your question may hinder the understanding of the answer. You have chosen as subject “brain” but you could have put “I”, “mind” or “consciousness” as well. What understand? Neurons? The prefrontal cortex? The conscious network? A metaphysical support of the “I”? The vagueness reigns.
The brain function becomes clear if you visualize it as mental representations organized among themselves, the most famous ones acting. These representations are physically supported by their dedicated neural networks. These are activated by the incoming information. Their successive stratification reflects the increasing complexity of the concepts and their interconnections. “Stratification” means, from one layer to the other, that first-level concepts are aggregated together to form second-level concepts. The second-level organization, if successful (the second-level concept is useful) stabilizes the first-level concepts in this arrangement. Feedback control.
A concept, basically, is an order, a regularity. The activity of interconnected neurons in an elementary network varies according to the regularities of the incoming signals. The pattern formed by the resulting signals contains the coding of this pattern. An example of a very elementary concept is a linear sequence of retinal signals juxtaposed and of the same intensity: a “line”. Constructed directly from sensory information, these basic concepts make it possible to represent objects. They individualize them. But also connect them, by common characteristics, sometimes between very disparate objects or abstractions. The circular shape can be that of an eye or a wheel. Different functions, but similar geometric properties for these two circles. The concept of the circle frees from objects. It is “abstract”. Neurologically, no difference between the networks ensuring the material and abstract representations. All form bridges and are organized into levels to build a pyramid, which rises from sensory-motor reflexes to consciousness, the seat of our most sophisticated retro-control.
If you have not given me up yet, I come to “new things”. There is nothing new at the foot of the pyramid. There are only rushes of data, processed by neural networks without moral concern, like all those who preceded them. Each level seeks the regularities that it is able to recognize, and codifies them. The information thus follows its path and eventually ends up bursting into consciousness if it corresponds to a concept recognized as “noticeable” by the unconscious structure. Notation that includes criteria as diverse as the duration of persistence of the signal, its associations with other remarkable representations. A small part of the data received by the base accesses consciousness. Most are only subconsciously treated. An unrecognized pattern, “new”, is a priori ignored since there is no network to identify it. However, if it repeats itself regularly, it causes the formation of its dedicated network. Because neurons are designed in such a way that they coordinate on patterns. Above all, it is the future utility of the new formed concept that decides its durability. The retro-control confirms or not the value of this brand new network according to its interactions with others of the same level.
This initiation of a network dedicated to a new regularity can occur at any level of the pyramid. On the stages close to consciousness and conscious, representations integrate abstract, logical and mimetic patterns specific to the pre-frontal cortex. Retro-control reshapes other representations to make them consistent with these patterns, like a laser beam sculpting the shape of the mind. There is nevertheless a large amount of new information that is not processed (we are far from being interested in everything), either by default networks able to do it (not recognition of novelty), or because the task does not expect significant reward, either because the conscious attention is not very vigorous at that moment (it can be defined as the focus of the retro-control).
Finally, the “new thing”, when it manages to go up to consciousness, is a particular organization of neural schemas like the others, but it has not stabilized its connections so assuredly. “Raw” representation, nevertheless consisting of known sub-concepts. Everything is not new in this thing. Thus the creative mind tries to apply to it proven filters, “size” the raw representation with already acquired schemas, to give it a little more known look, allowing to connect it to others.
It is the variety of our available patterns and their versatility that makes or breaks the success of the operation. Note that when the new thing is understood it is transformed by interaction with our higher patterns. To appropriate a new thing is to change what it was originally. Have we lost in passing its true meaning? It is not uncommon for us to misunderstand…