Analyzing startle data
Starting with ANSLAB 2.4, there are two different ways of analyzing startle data:- The raw data file contains a marker channel that indicates when the startle probe was given ("channel").
- The timepoints of the startle probes are given in a timing file ("timingfile").
You have to specify either "channel" or "timingfile" in the startle
options, using the marker type dropdown box. For the case of a marker channel, the marker needs to go
up when the noise burst starts and down when the noise burst ends. After loading
the data file containing the EMG- and the marker-signal, you are prompted for this threshold (see image to
the right).Every timepoint, for which the marker channel rises above this threshold, is considered to be the onset of a startle tone. For the case of timing files, startle onsets are taken from an associated timingfile. All entries in the timing files are considered as startle trials independent of the segment value.
Next, the first startle probe (=trial) is displayed:
Here is how to read the startle data display window:
- The first graph shows the trigger signal.
- The second graph shows
- the raw EMG signal.
- an indicator for the detected response onset.
- The third graph shows
- the rectified and lowpass-filtered EMG signal used for onset detection.
- an indicator for the detected response onset.
- the startle response onset threshold (red line).
- The fourth graph shows
- the filtered and rectified EMG signal for startle amplitude measurement (this is what you should focus on).
- the probe onset (the gray vertical line) and two horizontal gray lines from the probe onset to the detected respone onset and the detected response maximum.
- the baseline window (the two green vertical lines). By default, the baseline window ends at the probe onset time point.
- a marker for the detected response maximum and response latency (red dot)
- a marker for the response onset derived from third graph (blue dot)
- two horizontal green lines, where the lower line shows the startle response baseline value (average of the baseline window) and the upper line is an indicator for five standard deviations of the signal within baseline window above the average baseline.
- two vertical red lines showing the startle scoring area (ANSLAB only looks for a response in the window between 20 and 120 ms after the startle probe as this is physiologically the time window a startle response is expected to occur. Sometimes there will be a second, partially voluntary response, after this time window which we do not want to include).
Adjusting the detection parameters
In the example below, the baseline interval and the response window are not adequate for the data: the baseline window covers part of the response and the response window misses the response's maximum.
Note: if either the baseline window or the response window are adjusted, ANSLAB automatically recomputes the response onset and the response maximum based on the new window positions. ANSLAB also stores the window positions once the editing is confirmed (by clicking accept, invalid, zero, or inv. onset in the 'dynamic'-section). If you just switch to another trial (e.g. by using the buttons < or >) the window positions for the current trial are reset to the default positions. Hitting detect also resets the window positions to the default values.
The response onset latency is identified using the third graph, that shows a lowpass filtered rectified EMG signal. An onset is detected if the curve rises above a the mean value during the baseline interval plus a customizable number of standard deviations of the baseline signal. This threshold is illustrated by the horizontal red line in the third graph. You can adjust the threshold by using the x stddev of baseline field in the startle-control-section of the command window.
In the example below this threshold was clearly chosen too large, resulting in an overestimated onset time:
Once you have adjusted the windows and the detection parameters to the right values, you may save the values as default values to be used for all other trials and other files by clicking set (the values will then be saved to your 'anslabdef.m'-file in the study folder). In case of the response window and the baseline window the respective positions are then just set for the remaining, non-edited trials. To reset the window positions for an already edited trial to these default values, select the respective trial and hit detect.
Scoring startle trials
Use the buttons accept, invalid, inv. onset, and zero to jump from trial to trial and decide wether trials are valid or not. The different buttons have the following meanings:| Button | Meaning |
| accept | saves the current trial |
| invalid | will set the parameters for the current trial to NaN (not a number) |
| inv. onset | will set the parameters regarding the onset of the current trial to NaN (not a number) |
| zero | will set the parameters for the current trial to zero |
The extracted variables are displayed in the command window. This output will look something like this:
last edited: trial 1 : at 0.009321s : amplitude 21.8695 : baseline 1.9437 : onset latency 57ms : response latency 103ms : condition 196 pending approval : trial 2 : at 19.478s : amplitude 12.1778 : baseline 1.7035 : onset latency 79ms : response latency 112ms : condition 161 -------------------------------------------------------------------------- last edited: trial 2 : at 0.019477s : amplitude 12.1778 : baseline 1.7035 : onset latency 79ms : response latency 112ms : condition 161 pending approval : trial 3 : at 30.533s : amplitude 11.9867 : baseline 2.5102 : onset latency 82ms : response latency 109ms : condition 196 -------------------------------------------------------------------------- last edited: trial 3 : at 0.030532s : amplitude 11.9867 : baseline 2.5102 : onset latency 82ms : response latency 109ms : condition 196 pending approval : trial 4 : at 42.19s : amplitude 1.3704 : baseline 2.2463 : onset latency 49ms : response latency 53ms : condition 162Most important of these values are startle response magnitude (amplitude = peak minus baseline value), startle response latency (from tone onset to peak response as evident in the EMG average upper window), and startle onset latency (from tone onset to onset of EMG response as evident in the upper raw EMG window).
These variables are illustrated by the red and blue dot in the startle response graph. If the automatically detected positions of these dots do not seem appropriate to you, you can drag and drop the dots to a position along the response line - latency and magnitude values will be updated according to the new positions.
Note: similar to the detection parameters, changed startle onset and maximum positions for a trial will only be saved once the trial settings have been confirmed (e.g. by hitting accept). Clicking detect will reset the position of the startle onset and maximum to automatically detected values.
Setting the startle response maximum
If there are two equally plausible peaks, pick the one closer to the normal response time for that subject and above the 5 SD line (see example 2 below: the first response is closer to this subject's normal response latency).If there is no clear response, move the dot to a clearly identifiable peak that is approximately at the normal response time for that subject in order to not distort response latency. Do this only if the standard deviation is low (see example 3 below: this response might justifiably be set a little later on the second discernible peak if it reflects the subject's normal response latency; or alternatively, set to 0 response).
If the standard deviation is higher (i.e., much higher than any of the responses in that trial), consider excluding this response by hitting the button invalid, as this response measurement is probably not reliable due to an unstable baseline.
Analysis results display
After you have gone through one file trial-by-trial, startle magnitude and latencies are displayed for each response:
