Massive visual long-term memory is largely dependent on meaning

Previous research demonstrated a massive capacity of visual long-term memory (VLTM) for meaningful images. However, the capacity and limits of a "pure" VLTM that is independent of conceptual information still need to be determined. In the encoding phase of three experiments, participants viewed hundreds of images depicting real-world objects, along with visually similar images that were stripped of their semantic meaning. VLTM was evaluated using a four-alternative-forced-choice test including old and new images and their counterpart mirror transformations. The results revealed superior memory for meaningful than for meaningless stimuli and importantly, there was no hint of a massive VLTM for the meaningless items. Furthermore, when examining memory recognition of visual properties per-se (i.e., original/mirror state), memory was overall poor, and practically negligible for the meaningless items. Taken together, our findings suggest that meaning is critical for massive VLTM and for the ability to store visual properties.