A case study in smartphone vision science
Here are some demonstrations of the various conditions discussed in the following paper:
Liverence, B. M., & Scholl, B. J. (2015). Object persistence enhances spatial navigation: A case study in smartphone vision science. Psychological Science, 26(7), 955-963.These demonstrations are provided as Quicktime movies, which can be downloaded or viewed directly in most web-browsers. These movies are a bit large and choppy, but they should be sufficient to illustrate the basic conditions. As compressed versions of the original stimuli, these movies may not preserve the precise spatial and temporal characteristics of the originals.
Violations of spatiotemporal continuity disrupt performance in many attention and working memory tasks, but such experiments have been limited to the study of moment-by-moment online perception, typically assessed by passive monitoring tasks. Here we ask whether persisting object representations also serve as underlying units of longer-term memory and active spatial navigation, using a novel paradigm inspired by the visual interfaces common to many smartphones. Participants used keypresses to navigate through simple visual environments constituted by grids of "icons" (depicting real-world objects), only one of which was visible at a time through a static virtual "window". Participants found target icons faster when navigation involved persistence cues (via "sliding" animations) compared to when persistence was disrupted (e.g. via matched "fading" animations), with all transitions modeled on existing smartphone interfaces. Moreover, this difference occurred even after explicit memorization, demonstrating that object persistence enhances spatial navigation in an automatic and irresistible fashion.
Expt #1: Sliding vs. Fading (1 icon per page) (808 KB)
Replication: Sliding vs. Fading (4 icons per page) (1.3 MB)
In this experiment, participants navigated through a 4x4 grid of icon-pages -- with 1 icon visible per page -- using arrow keys on the keyboard. In the actual experiment, there was only 1 window (in the center of the screen), beneath which were displayed 4 "target" icons (not depicted here) that the participant had to navigate to (and click on), in order, as quickly as possible. This demo depicts two matched windows, corresponding to the views an observer would have in the two conditions of the experiment: Slide (left window) and Fade (right window). In this demo, the participant takes a clockwise route around the perimeter of the grid, starting and ending in the top-left corner, and viewing 12 of the 16 icons along the way (with an additional 4 unviewed icons in the center of the grid). The keypresses that the participant would have made while taking this route are highlighted in red on the "virtual keys" shown at the bottom of the display (not depicted in the actual experiment). We also include a similar demo of a replication (mentioned in footnote 3 in the manuscript) with 4 icons per page.
Expt #2: Sliding vs. Wiping (760 KB)
During Slide animations, icons are always drawn as fully saturated, but their shapes are only revealed gradually as they appear and disappear. During Fade animations, however, the icons' shapes are visible throughout the animation sequence, though they aren't fully saturated during the actual fading. Could such low-level differences in moment-to-moment visibility (rather than differences in object persistence) explain the robust difference between Slide and Fade RTs? To test this, we used a new Wipe animation that perfectly equated frame-by-frame visibility to the Slide animation, while disrupting object persistence. In the Wipe animation (shown in the right window in this demo), virtual occluding surfaces move across and thereby cover up the static icons at the same speed as the icons themselves move in the Slide condition (the left window, as before), thereby perfectly equating, frame-by-frame, the amount of the icons' surface area that is visible in each condition.
Expt #3: Controlling for Low-Level Motion (996 KB)
Whereas Slide animations involved motion in the direction opposite to each keypress, Fade animations involved no motion whatsoever, and Wipe animations involved a more ambiguous motion signal (i.e., of the occluder). Could such low-level differences in motion signals (rather than differences in object persistence) explain these effects? To test this, we created two new, duration-matched conditions that each involved both a sliding phase and a fading phase. During each condition, the initial icon disappeared twice -- first being replaced with itself, and then with the incoming icon. In the Fade-then-Slide condition (left window), the icon was replaced by itself as in standard Fade animations, and then was replaced by the incoming icon as in Slide animations from Experiment 1. In the Slide-then-Fade condition (right window), this order was reversed. The functional consequence was that both conditions were perfectly equated for all low-level visual properties (including motion), but while persistence was preserved in the Fade-then-Slide condition, it was robustly disrupted in the Slide-then-Fade condition.