Of three types of evidence available to evolution theorists – comparative, continuity, and computational – the first is largely productive rather than predictive. Although comparison between extant species or languages is possible and can be suggestive of evolutionary processes, leading to theory development, comparison with extinct species and languages seems necessary for validation. Continuity and computational evidence provide the best opportunities for supporting predictions.

van der Velde & de Kamps (dvV&dK) present a response to Jackendoff's four challenges in terms of a computational model. This commentary supports the position that neural assemblies mediated by recurrence and delay indeed have sufficient theoretical power to deal with all four challenges. However, we question the specifics of the model proposed, in terms of both neurophysiological plausibility and computational complexity.