Fig 1.
(A) The Necker cube in the middle with the two interpretations, where the view from above is typically preferred. (B) A point light walker, here illustrated by an ambiguous stick-figure, can be perceived as facing or facing away from the viewer, but a facing-the-viewer bias is typically preferred. (C) A SFM stimulus displaying a spinning half-cylinder is depth asymmetric in frontal views (convex or concave), but depth symmetric in side views and should then reverse more easily (possibly to uphold a convex percept). (D) An obliquely viewed wobbling SFM-cylinder (oscillating around the vertical and main axis) is observed with either its left or right side facing the viewer. In movie sequences of C and D only dots randomly spread on the surface of the objects are visible.
Fig 2.
Schematic illustrations of the ambiguities.
The two interpretations of (A) the spinning half-cylinder bulging toward or away from the viewer, (B) the cylinder slanted in depth, with its left or right side facing the viewer, and (C) the human figure obliquely facing away or toward the viewer. Figures (A-C) are displayed in a view from above with the frontal projection plane indicated by the dotted lines. (D) Schematic illustration of a stick-figure shown at 45° steps on a circular plot. The figure may be perceived as spinning clockwise (CW) or counter clockwise (CCW), with reversals in direction accompanied by depth-order reversals. Similar convex–concave depth reversals occur with the spinning half-cylinder (not shown). Perceived leftward spinning at 45° angle indicate that the figure is facing the viewer (FTV, blue shaded area within the circles, and black stick figures) whereas a perceived rightward spinning at the same angle indicate that the figure is perceived facing away (light grey stick-figures). Facing direction is not specified at side views (90° and 270°).
Fig 3.
Distribution of button presses indicating reversals.
Circular plots of the distributions of the reported reversals as a function of spinning angle (0-360°) during 168 sec across all participants where each response is indicated by a dot. From top to bottom: a spinning point-light walker, a rigid human shape in a walking posture, a spinning half-cylinder, and a wobbling cylinder. On the leftmost plots are distributions of all reversals. The responses are sorted as clockwise responses in the middle column (CW or left) and counter clockwise responses in the rightmost column (CCW or right). The CW responses indicate reversals from spinning rightwards to leftwards, and the CCW responses indicate reversals from spinning leftwards to rightwards. The blue shaded ranges indicate facing the viewer (FTV) responses, or bulging outward (convexity) responses for the half-cylinder, and the unmarked range indicate facing away, or bulging inward, responses. The mean direction where responses were made is displayed by the arrows, and resultant lengths are normalized to the unit circle.
Fig 4.
Raincloud plots of reversals per minute in the four stimulus conditions, and on the right are the corresponding box-whisker plots showing the medium, the interquartile extent, and ranges excluding data beyond 1.5 times the interquartile ranges from the upper and lower bound of the boxes.
Fig 5.
Reversal rate heatmap and stable durations.
(A) The heatmap show the Kendall Tau-b correlations, and Bayes factors, between reversal rates for the different stimuli (* p < .05, ** p < .01, one-tailed for positive correlations). (B) Shown are stable durations of perceived left vs. right spinning directions. For the wobbling cylinder left vs. right refer to the cylinder perceived as facing the viewer with its left or right side.
Fig 6.
Distribution of button presses sorted as facing the viewer and facing away responses.
Responses are sorted as facing the viewer responses in the left column (FTV+) and facing away responses in the right column (FTV-), across all participants where each response is indicated by a dot. For the spinning half-cylinder, FTV+ and FTV- corresponds to bulging toward (convex) and bulging away (concave) from the viewer.
Fig 7.
The FTV bias is scored as FTV- subtracted from the FTV+ responses for each participant, presented as a raincloud plot and the corresponding box-whisker plots showing the medium, the interquartile extent, and ranges excluding data beyond 1.5 times the interquartile ranges from the upper and lower bound of the boxes.
Fig 8.
The two interpretations of the half-cylinder.
(A) When bulging leftwards (90°) a counter clockwise (CCW) spinning is preferred. (B) When bulging rightwards a clockwise (CW) spinning is preferred (rightmost illustrations). During its phases in the spinning motion when an edge of the spinning half-cylinder begin to be self-occluded by its perceived near surface, it reverses in depth so that the edge is instead perceived moving in front. For illustrative purposes the half-cylinder is displayed with an oblique viewpoint slightly from above.