Grant T. Fairchild¹ (), Samantha C. Lee¹, Michael T. Compton¹, Sara Fabbri¹, Christian A.P. Nemeth¹, Lars Strother¹, Jacqueline C. Snow¹; ¹University of Nevada, Reno

Studies of human visual memory typically rely on artificial stimuli presented in the form of two-dimensional (2-D) images. Yet there is mounting evidence that 2-D images are processed differently in the brain compared to real-world solid objects, and behavioral studies point to a “real object advantage” in memory compared to 2-D images. Here, we used fMRI to compare brain responses to objects presented in real versus 2-D formats. During an initial learning phase in the laboratory, participants were shown a large set of everyday items presented either as real objects or as high-resolution colored 2-D images. During a subsequent recognition phase in the MRI scanner, participants viewed words that corresponded to items that had been presented during the learning phase, or had not been presented at all (i.e., “foils”). The participants’ task was to judge whether each item had been presented during the learning phase (or if it was a foil), and if it had, which format it was presented in (real object or image). A univariate analysis found that cortical networks commonly implicated in recognition memory were activated more strongly for real objects than for images; no regions showed the opposite pattern. Next, a multivariate searchlight classifier analysis revealed successful decoding of recognition memory for both real objects and images versus foils, but this decoding was considerably more widespread for real objects. Moreover, an additional multivariate analysis revealed that several of these regions, including the hippocampus and parahippocampal cortex, represented the format in which the stimulus was presented during the study phase. Together, our results show that brain networks implicated in recognition memory are activated more strongly, and more extensively, by real objects than by image displays, and that areas within this network represent the format in which a previously viewed item was seen.

Acknowledgements: This work was supported by a grant to J.C.S. (NEI-NIH R01EY026701), and by the National Institute of General Medical Sciences of the National Institutes of Health P30 GM145646.