User`s guide

E-Prime User’s Guide
Chapter 3: Critical Timing
Page 118
a few milliseconds, but enough to occasionally miss the 33ms vertical blanking signal
15
. This
results in a delay extended to the next refresh cycle (50ms spikes on images 527 and 584). At
image 602 the computer tried to present an image every 17ms. At this duration setting, there was
insufficient time to prepare the stimulus, and the onset delay increased to 27ms with occasional
spikes to 44ms (skipping to 1 or 2 later refresh cycles). The actual display time duration
remained at 33ms with some spikes to 50ms. Figure 24 shows a similar pattern of rate effects for
266MHz and 120MHz computers. With slower computers, the maximum rate of presentation of
images slows down, and the onset delays (black lines) increase.
The E-Prime variable OnsetDelay provides a good indication of whether the program can track
the desired presentation rate. Whenever the onset was near the expected value of 10ms, the
display times were accurate to the nearest refresh of the desired display time. However, if the
onset delay increased, display times typically increased by a full refresh duration, and that
duration was represented in the inter-stimulus interval between stimuli
16
. Program accuracy can
be checked quickly by looking at the OnsetDelay variable. If it is near the expected onset time,
the display times were accurate. If not, too much is being demanded from the computer. The
solution is to use a faster computer, or modify the experiment and/or the nature of the stimuli to
reduce preparation time.
High Rate Image Presentation
(266 MHz Processor, 70.1 Hz Refresh)
0.00
16.67
33.33
50.00
66.67
83.33
100.00
116.67
1 121 241 361 481 601 721
Sequential Image
Delay Between
Stimuli (ms)
High Rate Image Presentation
(120 MHz Processor 70.4 Hz Refresh)
0.00
16.67
33.33
50.00
66.67
83.33
100.00
116.67
1 121 241 361 481 601 721
Sequential Image
Delay Between
Stimuli (ms)
Figure 24. Presentation rates for progressively slower computers. Note, the expected lines are based on a 60Hz refresh rate and
the observed are shown at the nearest refresh for the given monitor.
Image caching can be used to reduce stimulus preparation time and present images at
faster rates. If a limited number of images must be presented at a fast rate, the images may be
pre-loaded into memory to reduce the preparation time during the time-critical portion of the
procedure. With fast display hardware, very fast rates of display presentation (e.g., 10ms) can
often be obtained. Before the experiment begins, the critical images need to be read in and
saved in an image cache that is stored in memory. Reading from the cache greatly reduces the
preparation time, allowing faster stimulus presentation rates.
15
Tests in this section were run with the Vertical Blank Simulation feature in E-Prime temporarily disabled to
purposely cause missed refresh time spikes. By default, E-Prime keeps this feature enabled as it allows the
system to use the real-time clock to detect a missed refresh and continue the presentation as if the refresh
occurred with only a short 1-3ms delay. To disable this feature you must use Inline script to insert the
following line at the beginning of the experiment: Display.VerticalBlankSimulationEnabled = False
16
This test was run with the Refresh Simulation feature turned off, causing missed refresh time spikes.
These can be reduced by enabling the Refresh Simulation feature (default mode), by manually setting
VerticalBlankSimulation to false via script.