Different neural mechanisms underlie repetition suppression to facial identity for same-size and different-size faces in the occipitotemporal lobe

摘要

Repetition of the same stimulus leads to a reduction in neural activity known as repetition suppression (RS) (or fMRI-adaptation). RS is frequently used to probe the response properties of face-selective regions in the occipitotemporal lobe. For example, a region showing RS to the same face across changes in stimulus size is inferred to hold a size-invariant representation of facial identity. This inference is based on the assumption that RS reflects locally based a??within-regiona?? changes such as neural fatigue. An alternative hypothesis characterises RS as a consequence of a??top-downa??, between-region modulation. Differentiating between these accounts is central to the correct interpretation of RS data and to understanding the functional role of different regions in the face-processing network. Using fMRI, we measured RS to familiar and unfamiliar faces within two face-selective regions of the occipitotemporal lobe - the occipital face area (OFA) and the fusiform face area (FFA). A univariate analysis revealed that both regions showed RS to images of the same face compared to different faces that persisted across changes in image size. Using Dynamic Causal Modelling (DCM) we determined how RS changed effective connectivity between these two regions. DCM demonstrated that repetition of identical face images produced changes in forward connectivity (OFA-to-FFA) only. By contrast, RS across image-size produced changes in backward connectivity only (FFA-to-OFA). A similar pattern was found for both familiar and unfamiliar faces. These results suggest that different mechanisms underlie different forms of RS in the occipitotemporal lobe. RS to the same face image is driven by a??bottom upa?? changes in connectivity, consistent with neural fatigue, whereas RS across size-changes is dependent upon a??top-downa?? modulation. Our findings challenge previous interpretations made using fMRI RS paradigms regarding the underlying nature of neural representations in the face processing network.