Sorry, my explanations of the timeline weren’t clear, so I will walk through them in more detail.
The first lines, the ones with “0ms” on the left, correspond to the execution of the lines from your script starting with newImage("fixationpoint", row.Fixation), which is executed immediately after the end of your Header, up to newTimer(700).start().wait(): all of them are executed so quickly that they virtually take no time, which is why I associated them all to the 0ms time point.
After the two wait commands on your first (unnamed) Timer elements have finished running, the script reaches the line getImage("fixationpoint"): that’s where the timeline reaches the time point “1850ms.” All the lines from that one up to newTimer("exposure", 300).start().wait() are executed in the blink of an eye, including the callback command of your Key element. Crucially, that command has an effect lasting for most of the rest of the timeline: from that time point on, whenever the participant presses the left or the right arrow, both the “exposure” and the “timeout” Timer elements will be stopped (if they were running during the keypress, that is).
After that time point, the timeline is no longer purely deterministic, because the course of events will depend on whether the participant does or doesn’t press a key during the execution of the Timer elements. Starting at time point 1850ms, the “exposure” Timer element starts and will run for 300ms unless the participant presses the left-arrow or right-arrow key. If the participant does not press a key, then the “exposure” Timer element elapses entirely and you are 2150ms in the trial. If the participant does press one the two arrows before that, the “exposure” Timer element is stopped prematurely and you are anywhere between 1850ms and (strictly before) 2150ms.
In either case, the “exposure” Timer element has stopped (either because it elapsed fully, or because it was ended prematurely) and the script reaches line getKey("pressOnArrow").test.pressed() and the corresponding failure command. That failure command will take effect only if the participant has not pressed one of the two arrows until then. Now remember, at this point, there are two possibilities: either the timer was ended prematurely by a keypress on one of the two arrows and the time point is not 2150ms yet, or the “exposure” Timer element has elapsed fully because no key was pressed and it has been 2150ms. Only in the latter case will the failure command take effect, and will the “timeout” Timer element start running and logging. If the participant did press a key (and it hasn’t even been 2150ms), then the “timeout” Timer element will not run and will not report any line to the results file.
This means that if your results file does report lines for the “timeout” Timer element for an item, your participant has not pressed one of the two arrow keys during the 300ms “exposure” Timer element: any relevant keypress therefore happened more than 2150ms after the end of the Header. However, if you see no such lines for an item, it means that the participant did press one of the two arrow keys before the 300ms of the “exposure” Timer element elapsed entirely, and therefore the keypress happened before 2150ms-after-the-end-of-Header. Item 60 corresponds to the first case: subtracting the time event for the keypress from the time event for the end of the Header gives 3956ms, which is well past the 2150ms signaling the end of the (uninterrupted) “exposure” Timer element. Item 68, on the other hand, corresponds to the second case: the same subtraction gives 2044ms, which is before the “exposure” Timer element has had time to fully elapse, and so the failure command did not take effect and the “timeout” Timer element was not run at all for item 68.
I hope the timecourse of events of your trials is clearer now. As I said, if you’re planning to re-run this, you could add a log command to your “exposure” Timer element so you get lines added to your results file reporting when that Timer element started (it should always be around 1850ms) and when it ended (anywhere between 1850ms and 2150ms).
Let me know if you have any questions