A version of the Hebb repetition task was used with faces to explore the generality of the effect in a non-verbal domain. In the baseline condition, a series of upright faces was presented and participants asked to reconstruct the original order. Performance in this condition was compared to another in which the same stimuli were accompanied by concurrent verbal rehearsal to examine if Hebb learning is dependent on verbal processing. Baseline performance was also compared to a condition in which the same faces were presented inverted. This comparison was used to determine the importance in Hebb learning of being able to visually distinguish between the list items. The results produced classic serial position curves that were equivalent over conditions with Hebb repetition effects being in evidence only for upright faces and verbal suppression as having no effect. These findings are interpreted a posing a challenge to current models derived from verbal domain data.