Abstract
The nervous system must resolve a trade-off between stability of existing neural connections and plasticity to encode new information. Selective attention may control learning so that only behaviorally-relevant, attended stimuli induce changes in neural circuitry. We tested this hypothesis using fMRI measurements of scene learning in the parahippocampal place area (PPA, Epstein & Kanwisher, 1998) and face learning in the fusiform face area (FFA, Kanwisher, McDermott, & Chun, 1997). As an index of learning, we measured adaptation, namely, reduced activation to repeated vs. novel images (Grill-Spector et al., 1999). To manipulate attention, we used overlapping scene and face images (O'Craven et al., 1999), and asked subjects to attend to either scenes or faces in a 1-back matching task. When scenes were attended, significant adaptation was observed for the attended repeated scenes in the PPA, while no adaptation was observed for unattended repeated faces in the FFA. Conversely, when faces were attended, no adaptation was observed for the unattended repeated scenes in the PPA, while significant adaptation was observed for the attended repeated faces in the FFA. A control experiment ruled out alternative explanations based on signal level reductions due to inattention. These results suggest that attention governs when neuronal representations will exhibit plasticity to repeated perceptual input.