Predicting episodic memory formation for movie events

Hanlin Tang, Jed Singer, Matias Ison, Gnel Pivazyan, Melissa Romaine, Rosa Frias, Elizabeth Meller, Adrianna Boulin, James Carroll, Victoria Perron, Sarah Dowcett, Marlise Arellano, Gabriel Kreiman

Figure W1: Experimental design and performance consistency (Original version of Figure 1). Due to copyright issues, we had to replace the movie frames in the figure by artistic renderings. Here we show the original version of the figure and the actual frames that the subjects saw. A. Single frame (frame 50946) from the movie showing eye fixations from 25 subjects (each colored circle denotes a separate subject). B. During each recognition memory trial, subjects were presented with a single shot (here from frame 55576 to frame 55601, duration = 0.833 seconds). Subjects indicated whether or not they had seen the events in the shot during the movie. C. Raster plot showing the performance of each of the 41 subjects (one subject per row) for multiple shots from frame 55,500 to frame 56,800. Each vertical mark indicates the subject’s response (blue=correct, red=incorrect). Bottom: for each shot, if most subjects were correct, the height of the blue line indicates the percentage of subjects that were correct; if most subjects were incorrect, the height of the red line indicates the percentage of subjects that were incorrect (see Fig. S3 for a raster over the whole experiment).

figure w1