Manakin males (Aves: Pipridae) attract females with acrobatic displays varying in complexity across species. Previously, we found that brain mass (BrM) and body mass (BdyM) increase with display complexity. We suspect the BdyM relationship is due to muscle hypertrophy and dense bones associated with acrobatics. If so, body size measured by tarsus length cubed (Tar³) is unlikely to relate to complexity. We predict that BrM is driven by mosaic evolution of the the cerebellum (CB), which is essential for procedural learning, and the arcopallium (AP), a sensorimotor region that includes a motivation-linked limbic nucleus (taeniae, Tn). In 12 manakin species and a closely related flycatcher, we measured the volume (Vol) of Br, AP, Tn, CB, and a visual thalamic nucleus as a control (rotundus, Rt). Brain regions were adjusted for bird size and these adjusted values and the raw values for BrVol and Tar³ were corrected for phylogeny. Tar³ and Vols of Br, CB, and AP but not Tn or Rt were each positively associated with display complexity. To determine which variables best explain variation in complexity, we ran 3 mixed models with two random effects; species and a phylogeny covariance matrix. The best brain region model included CB, AP minus Tn, Tn, and Rt; CB had a positive association and Tn had a negative association with complexity. While BdyM is positively associated with complexity, in models with Tar3 or BrM, it is redundant. Tar³ is the best predictor of complexity, but BrM also predicts a unique portion of the variation in complexity. Phylogeny does not explain complexity. Thus, sexual selection for acrobatic complexity boosts the capacity for procedural learning via enlargement of CB Vol and increases Tar³ size, possibly in relation to the link of Tar³ with body condition.