Multi-Stable Perception Task

Participants completed two 13.5-min scanning runs of a multi-stable perception task, adapted from classical binocular rivalry tasks requiring continuous report (e.g., Carmel et al., 2010).

In total, we presented 14 different bistable images (see Supplementary Information for details, some examples are displayed in Figure ​Figure1C),1C), separated by 30-s baseline period displaying either a fixation cross or a phase-scrambled bistable image. Participants viewed each image for a total of 90 s (Figure ​(Figure1A).1A). Above the images (on the top right and left corners of the screen), the words “FACE” and “LAND” appeared in gray. Participants used their right hand to hold down a key in the button box corresponding to either percept (face or landscape, respectively), and held down both keys if they saw the two percepts simultaneously. One or both of the words (“FACE” and/or “LAND”) were highlighted while participants held down the corresponding keys to provide feedback for their responses. We asked participants to avoid focusing on only one of the percepts, and instead to balance the total amount of time across all three percepts. Importantly, we also instructed participants to report as faithfully as possible the percept that they really perceived in that very moment. To increase the rivalrous aspect of the images, we split the 90-s presentation time into two 45-s periods. During the first 2 s of each of these periods, we presented two attentional cues. These cues could not be taken as emphasizing either the face or the landscape interpretation of the image. Instead, as the example in Figure ​Figure1B1B illustrates, a cue could be pointing to either be an eye or a mountain according to the specific percept. With this instruction, we aimed to prevent participants from simply focussing on different parts of an image (e.g., the sky for landscape viewing and the eye region for face viewing). We asked participants to focus on these regions, and try to interpret them as any one of three options: exclusively part of the face, exclusively part of the landscape, or as part of both simultaneously. Importantly, the attentional instruction cue was only displayed for 2 s and disappeared afterwards to avoid obstructing the images. Participants were asked to continue focussing on the cued feature of the image even after the attentional cue had disappeared.

We defined three conditions based on subjective report, namely face, landscape and simultaneous. Thus, the total duration of each trial was determined by each participant’s behavior.