31The evolutionary diversification of animal behavior is often associated with changes in 32 the structure and function of nervous systems. Such evolutionary changes arise either through 33 alterations of individual neural components ("mosaically") or through scaling of the whole brain 34 ("concertedly"). Here we show that the evolution of a specific courtship behavior in Malawi 35 cichlid fish, the construction of mating nests known as bowers, is associated with rapid, 36 extensive, and specific diversification of orosensory, gustatory centers in the hindbrain. We find 37 that hindbrain volume varies significantly between species that build pit (depression) compared 38 to castle (mound) type bowers and that hindbrain features evolve rapidly and independently of 39 phylogeny among castle-building species. Using immediate early gene expression, we confirmed 40 a functional role for hindbrain structures during bower building. Comparisons of bower building 41 species in neighboring Lake Tanganyika show patterns of neural diversification parallel to those 42 in Lake Malawi. Our results suggest that mosaic brain evolution via alterations to individual 43 brain structures is more extensive and predictable than previously appreciated. 44 45 behaviors vary widely, as do their neural phenotypes [1]. Evolutionary 63 neuroscience identifies how the brain diversifies over time and space in response to selective 64 pressures [2]. A key goal of evolutionary neuroscience has been to identify whether brain 65 structures evolve independently ("mosaically") or in tandem with each other as they reflect key 66 life history traits, especially behavior [3][4][5][6]. While a number of studies have linked variation in 67 brain structure with other traits across evolutionary time [2, 7-9], it remains unclear whether or 68 not this variation is predictable. Specifically, when similar behavioral traits evolve among two or 69 more species, do their neural bases evolve correspondingly? If parallel brain evolution is 70 predictable then it may be possible to understand general principles of neural organization and 71 function across animals. This would expand our ability to manipulate brain function, but if this is 72 not true, new strategies will be needed to reveal the mechanisms of brain evolution. 73Fishes, as both the most speciose (50% of extant vertebrates) and most varied vertebrate 74 radiation [10] offer opportunities to answer these questions. Fish species live in diverse 75 ecological, sensory, and social environments and have evolved elaborate variations in neural 76 structure and function from a common basic ground plan [11] making rapid and variable 77 diversification of brain structures a broad and general feature of their evolution [10]. 78The cichlid fishes of Lake Malawi, Africa offer a particularly striking model of these 79 patterns of diversification. Although geologically young (less than 5 million years old), Lake 80Malawi contains at least 850 species of cichlids [12] that, based on molecular phylogenetic 81 analyses...