DirectRT always takes a system time stamp (in milliseconds) at the start of a session. It uses that stamp as a relative marker for the values in the "started@" column in the log file, i.e., telling you how many milliseconds have passed at the time of any given trial's onset. You can then take the difference between "started@" values for any two sequential trials and see how many milliseconds, in total, each trial took.

The value of that first number is consequently meaningless--simply an indicator of the current state of your computer when the session began. In more recent releases of v2006, we actually subtract this value from every "started@" value. As a result, you'll see the first value is always "0". Doesn't change the data in any meaningful way, but we thought it might be less alarming and less confusing to have a starting point of 0 rather than, say 621556!

A note for the curious:

If you're interested in some additional validation of your timing data, you can compare the difference in sequential "started@" values with the sum of the individual times within each trial. Times recorded within the trials are calculated in real time, as opposed to being relative to the start of the session.

Example:

Say you request that two pictures each be shown for 20ms and you have a 1000ms trial interval. You might expect the difference between this and the next trials started@ values to be 1040ms. Not necessarily. If your system can only handle 17ms display intervals, you may see in the log file that the pictures were actually shown for 17ms each and that the trial interval was 1003ms. Adding these up you find that the time elapsed between trials should be 1037ms.

Display Times vs Reaction Times:

Final note, remember that response times do not equal display times. Say a picture is displayed until the subject responds. The response occurs after 704ms. The picture must be displayed in multiples of the display's refresh rate. So if the refresh rate is 100hz, the display rate is 10ms. Consequently, in this case, the image would be on the screen for 710ms. That is why, in the log file, you would see 704ms as the RT and and 710ms as the display time for the image. It is the latter value you would use in calculating the total elapsed time for your trial.

Hope that helps,
Blair