A critical function of central neural circuits is to integrate sensory and internal information to cause a behavioural output. Evolution modifies such circuits to generate adaptive change in sensory detection and behaviour, but it remains unclear how selection does so in the context of existing functional and developmental constraints. Here, we explore this question by analysing the evolutionary dynamics of insect mushroom body circuits. Mushroom bodies are constructed from a conserved wiring logic, mainly consisting of Kenyon cells, dopaminergic neurons and mushroom body output neurons. Kenyon cells carry sensory identity signals, which are modified in strength by dopaminergic neurons and carried forward into other brain areas by mushroom body output neurons. Despite the conserved makeup of this circuit, there is huge diversity in mushroom body size and shape across insects. However, an empirical framework of how evolution modifies the function and architecture of this circuit is largely lacking. To address this, we leverage the recent radiation of a Neotropical tribe of butterflies, the Heliconiini (Nymphalidae), which show extensive variation in mushroom body size over comparatively short phylogenetic timescales, linked to specific changes in foraging ecology, life history and cognition. To understand the mechanism by which such an extensive increase in size is accommodated through changes in lobe circuit architecture, we first combined immunostainings of structural markers, neurotransmitters and neural injections to generate, to our knowledge, the most detailed description of a Papilionoidea butterfly mushroom body lobe. We then provide a comparative, quantitative dataset which shows that some Kenyon cell populations expanded with a higher rate than others inHeliconius, providing an anatomical parallel to specific shifts in behaviour. Finally, we identified an increase in GABA-ergic feedback neurons essential for non-elemental learning and sparse coding, but conservation in dopaminergic neuron number. Taken together, our results demonstrate mosaic evolution of functionally related neural systems and cell types and identify that evolutionary malleability in an architecturally conserved parallel circuit guides adaptation in cognitive ability.