• Naghmeh

Naghmeh's Projects

1. Characteristics of saccade transients
   

2. Contrast sensitivity project
   

3. Peri-Microsaccadic Vision

Characteristics of Saccade Transients

MR 1/16/14: Resumed working on this. We had left the work on April 18, 2014, see corresponding report. We went over the materials, and agreed on preparing for next week a succinct report with the main figures for JV, so that we can involve him in the loop. Issues discussed:

• Model-based vs. direct method: We use the model-based method because it allows higher spatial resolution than the direct method. There was an attempt to further increase the resolution in the direct method, but that implied going outside of the image boundary and making assumptions on the stimulus (e.g., mirroring the image), so effectively introducing independence. The final data on the direct estimation with high-resolution images (3 subjects, spectra run on the cluster) show a reasonable correspondence with the model-based method. It is not perfect, but sufficient to proceed for now. We also discussed the possibility of trying other methods to test independence, like was done for the VR manuscript.
  Range of whitening: Note that there is an interesting limitation. The natural images we used deviate from 1/k² at high spatial frequencies, perhaps as a consequence of the optics of the lens. This has implications for the range of whitening: whitening would have been broader for pure 1/k² images. This is not a problem with the model-based estimation method.
  Comparison drift-saccades: I continue having doubts about this analysis. We want to capture the transients of saccades, but we are effectively capturing the entire trace, including the stationary periods before and after the saccade. This is because we use a method for stationary signals, the power spectrum, with a non-stationary one. The real solution to this is to finally start using spectrogram. Maybe an alternative, simpler way is to work with the derivative of the signal, rather than eye movements themselves. In this way we don't add power for saccades.
  

We did not go over the model. As I remember, there may still be issues there with differences in power that were not present in Boi's data. Maybe this is caused by using real images vs. natural noise (i.e., same problem of Range of whitening above)?

Contrast Sensitivity project

Study Objectives

• • Examine the impact of microsaccades (3-30 arcmin saccades) and small saccades (30-60 arcmin saccades) on the spatiotemporal characteristics of the stimulus on the retina.
  • Link the changes in visual stimulation with contrast sensitivity at low (0.8 cpd) and high (10 cpd) spatial frequencies in a discrimination experiment.
  

Time Frequency Analysis of the Retinal Stimulus Synchronized to the Occurrence of Microsaccades, Small Saccades, and Drift

Method

1. Eye movements of 5 human observers were recorded while performing a discrimination task. The task was to discriminate the tilt of a low (0.8 cpd) or high (10 cpd) spatial frequency stimulus presented on an Iyama HM204DT CRT monitor at 800 * 600 pixel resolution and 100Hz refresh rate for 1s. Eye movements were recorded with a Dual Purkinje Image eye tracker and sampled at 1 KHz.
   

2. Recorded traces were segmented into periods of drift, microsaccade and saccade based on eye velocity. Movements with speed smaller than 3 deg/sec were classified as drift. Movements with speed greater than 3 deg/sec were classified as saccade. Microsaccades were identified as saccades with amplitude smaller than 30 arcmin.
   

3. In order to analyze the frequency content of the retinal input under the influence of microsaccades, the eye movement traces were reconstructed composing of one microsaccade. The microsaccade was centered in a 512 ms time window filled with the same eye position before and after the microsaccade. Similar traces were reconstructed for small saccades (0.5-1 deg saccades).
   

4. In order to provide a comparison between the effect of microsaccades and drifts during fixation, drift eye movements longer than 512 ms were also selected from the recorded traces.
   

5. The spatio-temporal power spectrum (energy spectrum) of the movie was estimated using the QRad_fft² method
   

Data
All clean databases can be found in \\opus.cvs.rochester.edu\aplab\PublishedProjects\MostofiRucci15\Data

Detailed description of each database in available in the "read me" files within each folder.

Notes
10/10/2014. MR notes: Gone over reply strategy with NM. For Rev. 1, main issue is independence. How do they affect the input? We should also look at where exactly they go. Are they orthogonal to gratings? Do they have specific directions? Are small saccades corrective and re-centering the stimulus (see last comment from Rev. 1). Need to perform analysis on actual transients caused by microsaccades/saccades: changes, starting points, and landing points.
For Rev. 2, main point is simulation with drift. It should be explained very clearly what current analysis is meant to estimate, the transient change caused by saccades at fixation onset. Plan is to include one extra line in figure and write extensive reply. There is one extra issue on errorbars in Fig.4, in which one subject was eliminated. Simulate one additional subject to answer Reviewer's query. Reviewer mention a possible saccade improvement at high spatial frequency, which needs to be investigated better.
We also need the clean database to be freely shared in the lab. We have already discussed some refinements.

10/15/2014. NM notes: Reviewed the analysis NM did on the input luminance during microsaccades. To do list: 1- take an area with eg. 20*20 receptors instead of only one point corresponding to preferred retinal locus. 2- look at the change in total energy. 3- compare the input luminance change histograms with microsaccades occurring during stimulus presentation using the observed images with histogram of the input luminance with microsaccades occurring during fixation period assuming that they were made during the stimulus was on. If the 2 distributions turn out similar it shows that eye movements are independent of the observed images.

10/17/2014. NM notes: reviewed the analysis NM did on the input luminance during microsaccades. For each microsaccade occurring during stimulus presentation period, a movie was created by moving a 10*10 frame of the input grating around the preferred retinal locus along the microsaccade trace. The difference between the energy of the frame at the microsaccade starting point and ending point was calculated for each microsaccade. The distribution of energy changes were plotted for the low and high frequency trials seperately (due to the limited number of microsaccades, the data for all subjects was merged. All together we have 819 microsaccades for the low and 849 microsaccades for the high frequency conditions.) In order to investigate the independence between the eye movements and images, similar analysis was done, this time by swapping the eye movements and images, i.e. microsaccades made during high frequency condition were used on a sample low frequency grating and vice versa. Distributions were similar to the ones deriving from real retinal input.
MR was not convinced that this analysis really addresses the question.
To do list: 1- Do the analysis but this time using the microsaccades made during fixation period (when no stimulus was present).

10/20/2014. NM notes: Reviewed the distributions for the fixation microsaccades vs. stimulus presentation microsaccades. They are very similar and seem to be fine. For the fixation microsaccades we have all together 1891 microsaccades in the low and 2299 microsaccades in the high frequency condition. NM needs to plot the curve corresponding to the fixation on top of the distribution for stimulus presentation microsaccades. Final plan is to add 1-2 sentences to the manuscript explaining the fact that the spectral analysis looks at how microsaccades or small saccades change the spectral distribution of the input to the retina on average and independent from the specifics of the observed images. However in for this experiment, the distribution of the input energy to the retinal cells for different eye movements and stimuli show similar patterns confirming the validity of the assumption of independence between the eye movements and observed gratings.

Peri-Microsaccadic Vision

Progress report and meeting notes

NM 1/23/15

• Reviewed the experiment paradigm and the data collected and brain stormed about the possible ways of exploring the data. For now, I will look at contrast sensitivity functions for different time windows proceeding and preceding saccades. At this point we will average the positions to simplify the analysis. It is important that the sensitivity functions are not saturated and we can reliably estimate 75% contrast thresholds for each subject. I need to collect a good number of trials from myself to be able to do the analysis. Also it is important that in each time window, we make sure that no other saccade happen in close temporal relation to the one under study to remove interference.
  

  Finally I need to collect data from another subject to see if the same exploratory behavior is present in other observers as well.
  

Saccadic suppression literature review: LitReview.pdf