Zhang, K., Huang, J., & van Buren, B. (2023). The automaticity of seeing in: Pictorial depth cues influence judgments of surrounding spatial relationships even when task-irrelevant. Poster presented at the annual meeting of the Visual Science of Art Conference, 8/26/23, Cyprus, Greece, and at the annual meeting of the Vision Sciences Society, 5/20/23, St. Pete Beach, FL. PDF (currently writing for publication)

Background Introduction

Many representational paintings give a strong impression of depth, despite being flat surfaces which lack any depth—a phenomenon often referred to as ‘seeing-in’. 

Cats "see-in" flat carpet

Driver "sees-in" flat wall painting

Motivation

It has been claimed that, by attending appropriately to pictorial artworks, we can sometimes perceive them as flat.  In contrast, here we report two experiments which show that observers cannot fully “turn off” their perception of depth in a picture.  Rather, when viewing a pictorial artwork, impressions of depth arise automatically, and irresistibly interfere with judgments of surrounding spatial relationships, even when the painting is entirely task-irrelevant. 

Experiment One

In Experiment 1, observers viewed a virtual gallery with two freestanding walls, and reported as quickly as possible which wall was closer to them.  Each wall displayed a painting—an abstract artwork featuring a luminance gradient, which, depending on its orientation, looked either convex or concave. On Congruent trials, the near wall displayed a convex-looking painting, and the far wall displayed a concave-looking painting. On Incongruent trials, this was flipped. 

Experiment One Stimuli Demo

Although observers were told to ignore the paintings, and to focus only on the wall placements, they were unable to ignore the paintings’ pictorial depth cues, and responded slower in the Incongruent condition. Thus, seeing-into a picture occurs automatically, and interferes with an orthogonal task even when we are trying to ignore the painting. 

Does this also occur for more complex paintings? 

Experiment Two

In Experiment 2, one of the walls displayed a Renaissance or traditional East-Asian painting, and the other a phase-scrambled version of the painting (abolishing depth cues). 

Eastern V.S. Western Paintings

Experiment Two Stimuli Demo

Observers responded faster when the original painting (with pictorial depth) hung on the far wall,compared to when it hung on the near wall. We conclude that seeing-in is automatic: even when trying our best to resist seeing depth in a picture, it is not possible to perceive a pictorial artwork as flat.

Does this also occur in the real in-person setting? 

Experiment Three

Participants will complete an experiment consisting of one block of practice trials and four blocks of actual trials. Each block will consist of 36 trials each featuring two images displayed on two monitors in a three-mionitor “triptych”.  There will be two spatial arrangements of the monitors: even participants will see an arrangement of close, far, close, and odd participants will see an arrangement of far, close, far.

On each trial, two of the monitors will each display an image.  One image will depict a shaded circle oriented to appear convex, and one image will depict a shaded circle oriented to appear concave.  Participants will be told to ignore the images themselves, and to simply report as quickly and accurately as possible which of the two display monitors was closer — by pressing the left and right arrow keys.

We are interested in whether task-irrelevant depth cues in the images irresistibly influence judgments about the spatial relationship of their host monitors.

We predict higher accuracy and faster responses in the Congruent condition.

Experiment Four

Participants will be invited to a room with a box containing a turntable. Participants will observe the contents of the box through two viewing holes. This experiment consists of one block of practice trials and four blocks of actual trials. Each block will consist of 10 trials featuring two images displayed on two freestanding walls on the left and right, one of which will be closer than the other.

On each wall, there will be a painting of a shaded circle, which depending on its orientation, will appear either concave or convex. In the Congruent condition, the closer wall will display the convex version of the painting, and the farther wall will display the concave version. In the Incongruent condition, the closer wall will display the concave version of the painting, and the farther wall will display the convex version.

The key dependent variable in this study is the response time of the participant. On each trial, the observer will be instructed to rapidly report which of the two walls is closer. We predict faster responses in the Congruent condition.

Data collection in progress…

Experiment Five

Testing in Virtual Reality and Mixed Reality…