| Table of Contents |
|---|
Great signals and easy interpretability
The following example shows the co-occurrence of spikes in the respiration and the EDA signal. Channel 8 reflects the onset and offset of the signal to take a deep breath. We also see a deeper breath (possibly speech) before the onset of the sigh task and smaller SCRs after the end of the sigh task.
...
Unmatched SCRs
In this example, we see some “unmatched” SCRs: while SCRs occur with each sigh, there are also some SCRs that occur shortly before the second sigh and a few seconds after the third SCR. Unmatched means that it is unlikely that these occurred due to the event (sigh signal in channel 8, reflected by the light bulbs in the global channel).
...
Unmatched but somewhat related SCRs
Below we see an SCR right before the onset of the 4th sigh. You can see from the respiration signal that the participant took a longer breath around the same time. Therefore, this might mean that the SCRs are related to deeper breaths, but deeper breath can also occasionally happen without a signal to do so. This depends on the participant. For example, the person might yawn, sneeze or cough. We should not count these unrelated SCRs but they are useful in understanding that the SCRs might be related to breathing events and not solely to the stimulus on the screen.
...
Similar examples of more unrelated SCRs
This person shows a lot of unrelated SCRs.
...
Declining responses
This dataset shows an extremely strong response at the onset of the task without any breathing related events (no sharp peaks or troughs in the respiration channel). In addition to several unrelated SCRs, the first sigh elicits an SCR followed by more unrelated SCRs and another one that may be too close to the sigh signal onset to really be related to anything. You see that after that, the SCRs decline and remain absent until the end.
...
Absence of SCRs
...
Absence of SCRs during events but unrelated SCR in the middle
If we run the SCR analysis on this subject’s data, we get one big massive SCR and otherwise nothing (until after the task ends). Remember to always cross-check other channels if strange events occur. In this case, we are also looking at the EMG signal (channel 11; muscle activity from the right forearm). You can see that there was muscle activity at the onset of the big SCR. We don’t know whether that was a big body movement, a sneeze, a cough, or a yawn or if the person got startled (e.g. unexpected noise, someone knocked on the door or similar → can always check for notes for this subject’s session on confluence). In any case, we need to exclude this subject from the analysis.
...
Many SCRs, several unrelated to task events
In this example, the respiration signal is so poor, it can’t be used to determine when a deep breath was taken. The EDA channel shows SCRs each time the protocol channel shows a sigh, but also 2 SCRs before the first sigh signal and several SCRs between sighs 1 and 2, 2 and 3, and 3 and 4. SCRs can occur within a few seconds after an event, but not for too much longer after an event onset. Therefore, we have to assume that ,many of these SCRs are unrelated to the sighs. We here checked whether the EMG from the right forearm can help us understand whether the SCRs were evoked by movement. While we cannot exclude this possibility entirely (i.e. the EMG is recorded from the right forearm, but other parts of the body might be able to move without causing arm muscle contraction), we do not know where the SCRs come from. There are 2 options of things we can do:
Exclude this subject from analysis
Define a period of time during which we investigate the SCRs and ignore everything that’s outside that range.
CURRENTLY DO NOT HAVE AN ANSWER
FIND OUT WHAT HAPPENED AT THE REST5 SECTION
...
How do I know the data have sufficient quality to analyze?
In this example, the range of the signal is small (ca. 1.75-2.25 microsiemens for the visible signal range - compare to subject above peaking above 9 micro siemens). Remember, in the EDA preferences, you selected a baseline, threshold and a rejection percentage of amplitude height that will all determine whether a peak qualifies as an SCR. Therefore, we leave it up to the software to decide whether the fluctuations are large enough to report as an SCR. It is important to apply the same threshold to all subjects and data.
...
Artifacts that can be removed
Artifacts can be brief signal spikes on the EDA signal that are clearly not part of the signal pattern. To remove the artifact, select the area around it (as tightly as possible), then select the EDA channel you want the signal to be removed from, then click Transform → Math functions → Connect endpoints.
This means that everything in the selection area will be flattened to connect the beginning point of the selection with the end point of the selection. This will not happen in your original EDA channel. Instead, a new EDA channel will be created that is now lacking that particular artifact.
...
Partly usable with effort
In the following example, the signal is good until the last section of the task. In the focus area labeled “Rest5”, there are several signal artifacts. If you run the SCR analysis, markers for SCRs will show up on the artifacts, THESE NEED TO BE REMOVED. The reason this is ok in this particular case is that SCRs occur within a few seconds after an event, but not minutes after. Since these SCRs are not related to the (in this case) sigh, it is ok to delete them.
...
Real signal disturbance, do not use
In the following example, there are clear technical artifacts. This signal needs to be discarded, there is no way to rescue this.
...
This is the same file, just zoomed in around “Sigh2”
...