Experiment 1: Trial-by-trial variability in latency

In this first stage we checked the latency of the photodiode comparing to the trigger.
For this purpose we ran an experiment in a dark room where in each of 100 trial the PC screen changed from black to white via a psychopy experiment. We measured the “time to peak” that is, the time latency from trigger onset to photodiode’s maximal voltage value.
We started by comparing the latency with and without the KVM controller and found similar results (see figure 1A-B).
However, we were surprised to find a substantial trial-by-trial variability in time to peak (7 samples=14ms; see Table 1)
Thus, we ran another experiment where recording of data is via the EEGO software and not the LabRecorder and again found similar results (see figure 1C). Next, we changed the triggering script from the Lab streaming layer (lsl) to the parallel port (ttl) connection. This led to a ~40ms (20 samples) drop in latency. However, it increased the variability of latencies from 1.8 to 2.4 (see Figure 1D).
Another try was examining the impact of lengthened and varied inter-trial-intervals (ITIs), having the change from black to white occur every 3-5 seconds - however this only led to greater latency (~14ms) with a similar variability (see Figure 1E). Finally, we tried to replicate the former results with a different BIP input in the sandbox and found results similar to condition D, thus leading to the conclusion that the results from experiment E may have been coincidental (see Figure 1F).

Figure 1 - Trial-by-trial variability in photodiode latency from trigger under different conditions

Table1

Comparing the summary results of latencies under different conditions
Name Mean Std Max Min Range
With KVM(lsl) 289.51 1.845 293 286 7
Without KVM(lsl) 289.58 1.913 294 286 8
EEGO(lsl) 288.32 1.979 292 284 8
EEGO(parallel port) 267.14 2.370 271 263 8
EEGO(parallel port) long trials 276.48 2.218 281 272 9
EEGO(parallel port) long trials different input 267.51 2.410 272 262 10

Experiment 2: Latency from peak to peak and trigger to trigger in high change frequency experiment.

In this second experiment, we changed the design such that the screen changed color repetitively. Specifically, we had a color shift every 100ms therefore aiming for 200ms latency between each peak of the photodiode. Furthermore, we introduced triggers every 10 color shifts and wanted to measure the latency between each two triggers. We find that the photodiode peak-to-peak latency could be divided to two categories, one where timing was exact (200ms) and one where a constant delay occurred of 16ms (total of 216 ms; see Figure 2A). Moreover, we found that this delay (16ms) is schematic and appears every 3 or 4 color shifts (see Figure 2B).

Figure 2 - Photodiode peak to peak difference lags in 16ms every 3-4 color shifts

Next, we measured the latency between one trigger to another and found a 40-50ms delay which can be attributed to an aggregation of delay from the 10 loop iterations happening between each two triggers. This finding aligns with former result showing a 3-5ms delay in triggers in each trial.

Figure 3 - Trigger to trigger timing lags in 40-50ms due to ten within-triggers color shifts