Change Blindness and Attention

The article explores the phenomenon of change blindness, where observers fail to notice significant changes in a scene when there are brief interruptions (blank fields) between successive views of the scene.
It challenges the assumption that we perceive our surroundings in complete detail.
The study investigates the role of attention in perceiving changes and how attention is guided.

Brief-Display Paradigm:
- Initial display shown for 100-500 ms.
- Followed by a brief interstimulus interval (ISI).
- Second display with an item removed or replaced.
- Observers struggled to detect changes with ISI > 60-70 ms (Pashler, 1988; Phillips, 1974).
Eye Movement Studies (Saccade-Contingent Change):
- Examined change detection during saccades.
- Observers were poor at detecting changes, except in the saccade target (Bridgeman, Hendry, & Stark, 1975; Grimes, 1996; McConkie & Zola, 1979; Currie, McConkie, Carlson-Radvansky, & Irwin, 1995).
- Blurring during saccades masks motion signals, potentially leading to change blindness.
Change blindness in brief displays might be due to attention limits (Pashler, 1987; Pylyshyn & Storm, 1988; Wolfe, Cave, & Franzel, 1989).

Developed to investigate change blindness under more normal viewing conditions.
Original image (A) alternates with a modified image (A').
Brief blank fields are inserted between images (See Figure 1).
Observers freely view the display and report when they perceive a change, including the type and location.
Combines ISI manipulations with free-viewing conditions.
Provides ample viewing time to build a representation of the scene.

Flicker sequences: A, A, A', A', … with gray blank fields (Figure 1).
Each image displayed 240 ms, each blank for 80 ms.
Images presented twice before switching to create temporal uncertainty.
48 color images of real-world scenes, 27° wide and 18° high.
Single change in color, location, or presence/absence.
Changes divided by interest level: Central Interests (CIs) vs. Marginal Interests (MIs).
- CIs: Objects/areas mentioned by 3+ observers in independent descriptions.
- MIs: Objects/areas mentioned by none.
- MI changes were slightly larger (average = 22 sq. deg) than CI changes (average = 18 sq. deg).
Changes were large and easy to see once noticed (e.g., object appearance/disappearance, color switch, position shift).
Ten naive observers per experiment.
Instructed to press a key when they saw the change and describe it.
Dependent variable: Average number of alternations to see the change.
Averages taken only from correct responses.
Identification error rates were low (average 1.2%).

Basic flicker paradigm to induce change blindness.
Images displayed for 240 ms, blanks for 80 ms.
Hypotheses:
- Insufficient viewing time (brief-display experiments): changes seen within a few seconds.
- Saccade-specific mechanisms (saccade experiments): changes easy to see by keeping eyes still.
- Attentional mechanism: changes take a long time to see.
Results (Fig. 3a):
- Changes in MIs were very difficult to see (average 17.1 alternations, 10.9 s).
- Changes in CIs were noticed much faster (average 7.3 alternations, 4.7 s).
- Perception of MI changes took significantly longer than CI changes (p < .001 for presence vs. absence; p < .05 for color; p < .001 for location).
Without flicker (blanks removed), identification required only 1.4 alternations (0.9 s), with no significant differences between MIs and CIs.

Addresses the explanation that old and new scene descriptions could not be compared due to time limitations.
Blanks between images