Research has shown that face recognition accuracy can be improved by prior global processing and impaired by prior local processing (Macrae & Lewis, 2002). The aim of this study was to test the processing bias account of face recognition, using the composite face task (Young, Hellawell, & Hay, 1987), a test of featural recognition. Undergraduate volunteers (N = 75) participated in a between-subjects design that tested their ability to recognize face halves within a composite, following either global or local Navon processing or a control task. Results showed that, as compared with the control task, local processing speeded ability to recognize face halves. These results provide support for the processing bias account of face recognition.