Thank you for posting this fascinating question. I can give some initial answers regarding how concepts are encoded in the brain but this is not my main area of research. I hope others can join the discussion and bring different perspectives.
Firstly, the answer might depend on what you mean by concept. It is generally agreed that when we recognize an object regardless of the representation that is being used (image, written or spoken words, etc.), we have conceptualized it. Some experiments show that certain visual areas are not sensitive to the attributes of an image of, let’s say the Eiffel Tower, and contribute to conceptualize it.
At a higher cognitive level, the evidence we have so far seem to show that each concept is coded in a small number of neurons (small compared to the 80 billion neurons of the brain!). It appears reasonable to assume that these neural networks include neurons in both unimodal (sensory areas) and multimodal cortices (like the hippocampus or frontal lobe), depending on what the concept is and how the concept is represented (visual stimulus, auditory stimulus, etc.).
However, it has been suggested that the hippocampus and the temporal medial cortex are key areas for forming long term memories, including concepts, and that these neural networks are distinct from the neural networks associated with perception. They could play an essential role in making sense of the representations but would be required to recognize, categorize, and manipulate representations.
There is obviously a lot of pending questions in this area and I hope that as neuroscientists will soon bring new evidence on neural correlates of higher order cognitive skills like conceptualization.
R. Quian Quiroga, Concept cells : the building blocks of declarative memory functions, Nature Reviews Neuroscience, vol. 13, pp. 587-597, 2012.
N. Suthana et I. Fried, Percepts to recollections : insights from single neuron recordings in the human brain, Cognitive Sciences, vol. 16(8), pp. 427-436, 2012
R. Quian Quiroga et al., Sparse but not “Grandmother-Cell” coding in the medial temporal lobe, Trends in Cognitive Sciences, vol. 12(3), pp. 87-91, 2008