Abstract
Studies of pictorial long-term memory have shown that humans have an astonishing ability to remember with high fidelity previously viewed scenes (Konkle et al., 2010) with robust memory for visual detail (Castelhano & Henderson, 2005). How to explain this massive pictorial long-term memory? We tested the robustness of scene memory under intentional and incidental encoding with 100 exemplars from each of four categories (cityscape, indoor scenes, waterscapes & landscapes) and examined if aesthetic valence affects memorability of scenes. We randomly assigned observers to two conditions, either intentional or incidental memory group. Both groups followed the same experimental procedure except that only the intentional group was aware their memory was going to be tested. In the first phase observers saw 200 scenes, one at a time, each for 3 seconds and were asked to rate the pleasantness of the scene on a 4-point scale. In the phase that followed, they were shown another 200 images, half of which were seen in the study phase and randomly intermixed with new images not previously viewed, and asked to respond old or new to each image. During the third phase observers rated again the pleasantness of the last 200 scenes of which half were seen during the first study phase. Observers unaware of the memory test performed at the same level as observers intentionally memorizing the scenes (d'=1.85 for both groups) with a significant correlation in memorability of images across the two conditions (.27 at p<0.01). Perceived pleasantness of scenes had no affect on memory performance in either of the conditions even though pleasantness rating increased with familiarity (p<0.02). High fidelity encoding into pictorial long-term memory does not seem to require any intent and the memorability of image is intrinsic to the image, reproducible under any encoding condition and independent from it's aesthetic valence.
Meeting abstract presented at VSS 2014