These are some demos I have put together using some simple JAVA code. Feel free to appropriate them for your own presentations/research/etc, but please credit the original source.
The current demonstrations all use the same images, so if you want to sample each type of presentation paradigm you should only look at a few examples on each of the following pages.
In this very popular technique two versions of an image are shown repeatedly and the subject is asked to identify the changing element. By separating the display of each image with a 80+ ms blank screen, it is possible to defeat the perception of motion caused by the change. In such cases people often have great difficulty in detecting differences, and typically require several chances to inspect the two images before they can accurately determine the difference. There are many variations, such as how long to display each image, and how long to make the intervening blank screen.
Traditional: 240 ms per image, 80 ms gray ISI. (based on Rensink, 1997). This corresponds to the A, A` experiment in that paper. I personally find this duration of ISI to be so short that I can perceive the motion caused by the changes.
Extended ISI: 240 ms per image, 320 ms gray ISI. (based on Rensink, 2000). This increases the duration of the gray blank screen between images. Amazingly, it didn't decrease performance that much in Rensink's experiments as compared to the traditional 80ms ISI.
This condition allows the subject to inspect the image for an extended period of time before showing the changed version. If an accurate representation is built of the visual scene over time one would expect reasonable performance at detecting changes in this condition. Results of such experiments are mixed - subjects still miss many changes, but if given cues as to what objects to inspect for changes, their performance is increased significantly.
One-Shot: (12 seconds per image, 500 ms gray ISI.
This example is similar to our current work, except that this demo does not include crosshairs to indicate what subset of objects in the scene might change.
Practice.zip - 16 images. These images were constructed for use in a practice block. As such they include a lot of relatively easy to detect changes.
JAVA source code to the demos on this website can be downloaded here. The program is controlled by modifying the parameters to the applet tag inside the .HTML file. JAVA timing is not accurate enough for research purposes, however, and our actual experiments are written in Matlab.
Rensink R, O'Regan J, and Clark J (1997) To See or Not to See: The Need for Attention to Perceive Changes in Scenes. Psychological Science, 8, p. 368-373. http://www.psych.ubc.ca/~rensink/publications/download/PsychSci97-RR.pdf
Rensink R, O'Regan J, and Clark J (2000) On the Failure to Detect Changes in Scenes Across Brief Interruptions. Visual Cognition,7:127-145. http://www.cs.ubc.ca/~rensink/publications/download/VisCog_00.08.pdf
CSC Lab Webpage
Return to Alan's vision page
(c) 2003 Alan Robinson (robinsoncogsci.ucsd.edu)