A comparison of the archaeological evidence for symbol-based behaviours with the predictions of the social brain hypothesis has created ‘Dunbar's dilemma’. The dilemma lies in a disjuncture between evidence and theory, marked by a long chronological gap between the predicted cognitive potential for syntactic language (500 ka), communal religion (200 ka), and the earliest accepted archaeological evidence for these symbol-based behaviours (135 ka). One possibility is that the predictions of the social brain hypothesis may simply be wrong and these cognitively demanding behaviours developed with later populations of Homo sapiens. Alternatively, the social brain hypothesis may indeed be correct and the dilemma arises from the limited conceptual vision of how to interpret the material record. The resultant narrowing of the evidential gaps that comprise Dunbar's dilemma highlights the potential analytical insight that can be gained from an integrated approach involving evolutionary psychology, neurobiology and archaeology.