An Interesting Observation Refresh Rate
Today I did a test to see if the refresh rate (RR) affects the trigger timing in the receiving site of computer (i.e., ActiViews recording). Quite interestingly, I changed the refresh rate to 100 Hz in the subject and experimenter monitors. The jitter of timing observed yesterday seemed to decrease. The original setting is 85 Hz in the experimenters monitor, but in the other end, for some unresolved issue, the subject monitor cannot be detected by the stimulus presenting computer, so presumably 60 Hz (which was reported by the software and is also supported by the recorded values in a E-prime experiment). The NVidia software just showed CRT-0 (or CRT-1, if I reversed the order of connection).
Because I did two main improvements, the real causes of the change cannot be completely certain. The first improvement was to connect both monitor directly to the stimulus presenting computer via one graphic card, which has two input connector (one VGA and the other DVI). The second improvement was to increase the refresh rate to 100 Hz and resolution to 1152 x 864. Moreover, the two monitors I use are the same brand (SONY), rather than like previous setting each monitor is from different manufacturers.
The data in the first section is:
| Theoretical values | Real values | |||||
|---|---|---|---|---|---|---|
| Millisecond | 500 | 499.51171875 | 500 | 500.48828125 | 500.9765625 | |
| Count | 3 | 43 | 7 | 31 | ||
| Millisecond | 250 | 250.48828125 | 250.9765625 | 251.46484375 | 251.953125 | 252.44140625 |
| Count | 8 | 11 | 14 | 40 | 11 |
The theoretical values are the delay I set in the attention task, so the extra time should be attributed to the time computer used to present, clear images and other procedures. Usually, I assume in regular behavioural task, these procedures take only minimal time that is far less then 1 ms.
It looks in the first case (500 ms), more data points match exactly the delay period set in the programme, although there were still considerable data showing 500.9766 ms. Not sure if this was the time when the graphic card and monitor drew the fixation image.
In the second case (250 ms), probably because more lines of command were executed here, the timing is more variable. However, comparing to yesterdays recording (bi-modal shape), more data fell in the category of 251.9531 ms and less went to 250.9766 ms.
 |
| The figures show the possible time jitters, calculated from 6720 observations in one participant run. |
alternative link download