Gestures for controlling audio algorithms may be selected arbitrarily or based on post-hoc analysis of preferences of users. In this study, a wearable device which uses head gestures to control a blind-source-separation-based assistive hearing device was developed. Gestures were drawn from a corpus of simple primitive movements and the most appropriate gestures to use to generate control signals were identified by analysing preference rankings using a principal-component-analysis-based approach. Subjects were asked to rank the appropriateness of head gestures for specific functions of the source separation algorithm. Subjects were also asked to rank social acceptability of gestures and the usefulness of individual functions of the device. Although naive analysis of ranks of mean responses indicate some preferences within the population, principal component analysis suggests that within the set of responses independent dimensions of preference and consensus could be identified. By weighting preference scores by consensus, functions of the assistive device can be assigned to control gestures with more confidence that they will be acceptable to users and allow for intuitive control.