How small changes to one eye's retinal image can transform the perceived shape of a very familiar object

Abstract

Vision can provide useful cues about the geometric properties of an object, like its size, distance, pose, and shape. But how the brain merges these properties into a complete sensory representation of a three-dimensional object is poorly understood. To address this gap, we investigated a visual illusion in which humans misperceive the shape of an object due to a small change in one eye's retinal image. We first show that this illusion affects percepts of a highly familiar object under completely natural viewing conditions. Specifically, people perceived their own rectangular mobile phone to have a trapezoidal shape. We then investigate the perceptual underpinnings of this illusion by asking people to report both the perceived shape and pose of controlled stimuli. Our results suggest that the shape illusion results from distorted cues to object pose. In addition to yielding insights into object perception, this work informs our understanding of how the brain combines information from multiple visual cues in natural settings. The shape illusion can occur when people wear everyday prescription spectacles; thus, these findings also provide insight into the cue combination challenges that some spectacle wearers experience on a regular basis.

Keywords: binocular vision; cue combination; object perception; visual illusion.

Fig. 1.

Spectacles with a monocular horizontal magnifier cause real objects to appear distorted under natural viewing conditions. (A) An image of a mobile phone held frontoparallel to the camera was warped to match the average shape ratio that participants drew when wearing the control spectacles (plano lenses). (B) The same image was warped to match the average shape ratio that participants drew when wearing the experimental spectacles with a monocular horizontal magnifier. The increase in length on the right side is equally split between the bottom and top right corners; however, participants varied in whether they saw equal or unequal stretching of top right and bottom right corners of their phones. (C) Bar heights indicate the average shape ratio: the ratio of the length of the right side of participants’ drawings to the left side. This includes drawings of their own phone (Left) and an unfamiliar square object (Right). The black dots in the figure represent each participant’s shape ratio. Ratios greater than 1 indicate that the right side was drawn taller than the left side, and ratios less than 1 indicate that the left side was drawn taller than the right side. Error bars represent the 95% CI and horizontal lines represent significant differences. If we assume that perceived shape is determined based on the binocular disparities created by the spectacles, and assume a typical viewing distance of 35 cm, we expect participants to see a shape ratio of 1.05 with the experimental spectacles on.

Fig. 2.

How shape and slant cues may be combined to create a perception of a three-dimensional object. (A) A frontoparallel square (i.e., not slanted in depth) casts roughly square-shaped retinal images, and its geometric properties (pose and shape) are likely to be perceived accurately. Small deviations from rectilinearity in the retinal images arise because each eye views the square from a slightly different angle. (B) A square that is slanted to face the right eye casts trapezoidal retinal images due to perspective projection. The trapezoidal retinal image in the right eye is also wider than the retinal image in the left eye, resulting in horizontal binocular retinal disparities that provide useful cues to slant angle. (C) If an observer views a frontoparallel square and one eye’s image is magnified slightly in the horizontal direction, this simulates the binocular disparities associated with a slanted object but does not change perspective cues. Prior researchers have shown that observers perceive the object to be slanted, but at the same time, the perceived shape becomes distorted into a trapezoid. (D) A trapezoid that is slanted to face the right eye can create a roughly square image on one retina and an elongated rectangular image on the other, similar to the pattern observed in (C).

Fig. 3.

When viewing their phone and an unfamiliar square object in the experimental and control spectacles, some participants perceived a slant while others did not. (A) The percent of participants who perceived the objects to be slanted when placed flat against a wall and viewed through the experimental spectacles (orange) and the control spectacles (blue). Horizontal lines represent significant differences. (B) The percent of participants in the experimental spectacles who perceived the phone or the square to be slanted with the left side closer or right side closer. Error bars represent the binomial 95% CI.

Fig. 4.

Cross-fusible stereoscopic stimuli for the slant and shape adjustment tasks. (A) Random dot cloud presented during the slant task. (B) The stimulus presented during the shape task. In each panel, the right eye’s image is horizontally magnified 6%. The dot density, dot size, and luminance in all panels have been adjusted for visual clarity. During stimulus presentation, there was a larger space (about 34 deg) between the edge of the stimulus and the edge of the screen.

Fig. 5.

Results from the slant and shape adjustment tasks with horizontal, vertical, and uniform magnification. In each plot, circular markers indicate averages and error bars indicate the 95% CI across participants. (A) The slant that was perceived frontoparallel while experiencing monocular horizontal (blue) and vertical (yellow) magnification in the left eye or the right eye. The symbols beside the y axis indicate the direction of a positive and negative slant from a top–down view. (B) The shape that was perceived to be a square (specifically, the ratio between the height of the right and left side of the stimulus) while experiencing monocular horizontal (blue) and vertical (yellow) magnification simulated for the left or the right eyes. The icons next to the y axis represent shapes that would produce a shape ratio above and below 1 (the ratios are exaggerated for visibility). The lines indicate the predicted shape estimates based on the average responses from the slant judgment (fit of a third-order polynomial). (C) The slant percepts produced by monocular uniform magnification. (D) The shape percepts produced by a monocular uniform magnifier plotted in the same way as in B. The solid line again indicates the predicted shape estimates based on the average responses from the slant judgment.