Selective binding of facial features reveals dynamic expression fragments

The temporal correspondence between two arbitrarily chosen pairs of alternating features can generally be reported for rates up to 3–4 Hz. This limit is however surpassed for specialised visual mechanisms that encode conjunctions of features. Here we show that this 3–4 Hz limit is exceeded for eye gaze and eyebrow pairing, but not for eye gaze and mouth pairing, suggesting combined eye and eyebrow motion constitutes a dynamic expression fragment; a building block of superordinate facial actions.