Abstract
Real objects are more memorable than two-dimensional (2-D) images of the same items, a phenomenon known as the “Real Object Memory Advantage”, or ROMA (Snow et al., 2014). Although emerging evidence indicates that real objects are processed differently to images because they afford physical interaction (Gomez, Skiba and Snow, 2018), little is known about the underlying mechanism for the ROMA. Here, we used fMRI to identify brain areas that decode, at the time of recollection, the format in which an object was displayed during encoding. Participants first completed a behavioral learning task in which they were asked to remember a large set of everyday household objects. Half of the stimuli were presented as real-world objects; the other half were 2-D images of objects presented on a computer monitor. The images were matched closely to their real-world counterparts for size, apparent distance, viewpoint, background, and illumination, and all stimuli were presented within reach. Participants later completed a recognition task in the MRI scanner. During each scan, participants viewed text descriptors (e.g., ‘hammer’) and were asked to decide whether each item was viewed as a real object, a 2-D image, or was not viewed at all, during the study phase. Overall, most observers showed superior memory performance for items previously viewed as real objects versus 2-D images, consistent with earlier findings (Snow et al., 2014). Critically, searchlight multivariate pattern analysis (MVPA) of the fMRI data revealed that memory network areas (Posterior Cingulate Cortex and Retrosplenial) and dorsal somato-motor areas (Anterior Intraparietal Sulcus, Superior Parieto-occipital Cortex, Dorsal Premotor Cortex, and Ventral Premotor Cortex) were able to decode stimulus format, even though participants did not interact manually with any of the stimuli during the study phase. These results suggest that the ROMA is due to re-activation of dorsal somato-motor networks at the time of retrieval.