Blink Inhibition

Blink Inhibition

Determine when a group of viewers is engaged

Prepare Data | Inputs and Settings | Outputs | Run Analysis | Troubleshooting

blink


Prepare Data  (Back to the top)

To run this analysis, you need:
  • A set of participants who viewed the same stimulus
  • Data indicating when each participant blinked
Blink data must be saved in a csv file. There are two acceptable formats:
  1. One subject per column
    1. Each column contains data from one participant
    2. Each row is a sample
      1. Data should be collected at a constant sample rate
      2. Samples should be aligned in time across participants
    3. There are three acceptable values:
      1. 1 indicates that a blink was occurring at a sampled time point
      2. 0 indicates no blink
      3. NaN indicates lost data (i.e. unknown whether the participant was blinking)
  2. Three column format
    1. Each blink (from all subjects) appears in a separate row
    2. Three columns (in the following order, no header row):
      1. Participant identifier – unique to each participant (numeric value)
      2. Blink start – sample when blink began (integer)
      3. Blink end – sample when blink ended (integer)
    3. If this format is selected, you need to specify the length of the data collected, in number of samples


Inputs and Settings  (Back to the top)

  1. Select "Raw Blinks" to load data
    After a csv file has been selected, dialog boxes will appear:
    1. Select file format (One set per column or Three column format)
    2. Enter sample length (only if file is in Three column format)
    3. Enter sample rate (in samples/sec)

      Instantaneous blink rate will be plotted in the application:

      rate
    4. Use the right and left arrows by the x axis to scroll through the data
    5. To view the plot in a separate window, right click and select "Pop out figure"
  2. Specify the number of permutations for statistical testing (max: 10,000)
  3. Advanced settings (optional)
    1. Specify bandwidths to consider in optimization of the Gaussian smoothing kernel. User can specify a single value or a range (min:[step]:max)
    2. Set significance thresholds:
      1. Percentiles of the permutations used as low and high significance cutoffs
      2. Minimum number of consecutive frames necessary to accept significantly decreased or increased blinking


Outputs  (Back to the top)

Results will be plotted in a separate window. Use checkboxes to specify which outputs should be saved.

Summary csv (BLINK_MODsummary.csv)
Comprehensive summary document, containing the following sections:
  • INPUTS
    • Sample rate
    • Number of individuals
    • Number of permutations
    • Number of consecutive frames (significance threshold)
  • SMOOTHING
    • Gaussian kernel bandwidth - standard deviation of the Gaussian kernel used to smooth data
  • SIGNIFICANT FRAMES
    • Decreased blinking - sample(s) in which group blink rate is significantly lower, assessed via permutation testing
    • Increased blinking - sample(s) in which group blink rate is significantly higher, assessed via permutation testing
  • ALL FRAMES (one sample per row)
    • Smoothed blink rate – smoothed group blink rate in each sample (blinks/min)
    • Low percentile – low percentile of the permutation test for each sample
    • High percentile – high percentile of the permutation test for each sample

Figure (BLINK_MOD.jpg)
One plot showing smoothed group blink rate, low and high percentiles of the permutation test, and times when the smoothed blink rate was significantly higher or lower (on the x axis).
Use drop down menu to select figure format (.jpg, .pdf, .eps, .fig, .png, or .tif).
rate1

.mat file (BLINK_MOD.mat) - a MATLAB file containing a single struct ("results"), with the following fields:
  • smoothedBR – smoothed group blink rate in each sample (blinks/min)
  • decreasedBlinking – sample(s) in which group blink rate is significantly lower
  • increasedBlinking – sample(s) in which group blink rate is significantly lower
  • lowPrctileLevel – low significance threshold
  • highPrctileLevel – high significance threshold
  • lowPrctile – low percentile of the permutation test for each sample
  • highPrctile – high percentile of the permutation test for each sample
  • optW – standard deviation of the Gaussian smoothing kernel
  • inputs – a struct with fields:
    • numIndividuals – number of individuals in the input
    • dataLen – number of samples in the input
    • numFrames – number of samples in the input
    • numPerms – number of permutations
    • sampleRate – sample rate (Hz)
    • smoothType – method by which bandwidth of smoothing kernel was selected (sskernel)


Run Analysis  (Back to the top)

After specifying inputs and outputs, press "Run Analysis".
Progress bars will display the status of the analysis.
A figure with the results of the analysis will appear in a separate window.

Troubleshooting  (Back to the top)

[Link to error list]