The mapping is learned from experience! One neuron, or neuron assembly, learns to fire whenever you see a house, while other neurons, or assemblies, learn to respond to windows, doors, and roofs, for example. The collective pattern of activation of all of these neurons together corresponds to our experience of the external scene.

And in dreams and hallucinations, the same constellation of neural activations produces the same kind of visual experience as a perception of the corresponding real scene!