Datasets:
File size: 4,061 Bytes
eb72cfa | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 | # PolyCAT Acquisition Protocol
## Equipment
| Component | Specification |
|-----------|--------------|
| Eye tracker | EyeLink 1000+ (SR Research) |
| Mounting | Head-mounted (desktop mount) |
| Recording mode | Binocular |
| Sampling rate | 500 Hz per eye (1000 Hz hardware, split binocular) |
| Display | 27" 4K LCD monitor |
| Resolution | 3840 x 2160 pixels |
| Refresh rate | 144 Hz |
| Viewing distance | 70 cm |
| Background luminance | Gray (0.3 normalized) |
## Display Geometry
The pixels-per-degree (PPD) value is computed from the physical display properties:
```
Screen diagonal = 27" = 68.58 cm
Aspect ratio = 16:9
Screen width = 68.58 * (16/sqrt(16^2 + 9^2)) = 59.77 cm
Screen height = 68.58 * (9/sqrt(16^2 + 9^2)) = 33.62 cm
Pixel pitch = 59.77 / 3840 = 0.01557 cm/px
PPD = viewing_distance * tan(1 deg) / pixel_pitch
= 70 * tan(1 deg) / 0.01557
= 70 * 0.017455 / 0.01557
≈ 78.5 px/deg
```
Note: This gives the conversion for uniform pixel density. Since pixels are square on modern displays, horizontal and vertical PPD are equal.
## Calibration
- **Calibration type:** HV13 (13-point horizontal-vertical)
- **Validation:** Performed before each block
- **Drift check:** Before each trial (fixation on cue position)
- **Drift retry limit:** 1 attempt, max 10 seconds
- **Re-calibration:** Triggered if drift check fails
## Saccade Detection Parameters
The EyeLink built-in saccade detection was configured with:
| Parameter | Value |
|-----------|-------|
| Velocity threshold | 30 deg/s |
| Acceleration threshold | 8000 deg/s^2 |
Fixations are defined as periods between saccades (EyeLink cognitive configuration).
## Experiment Design
### Structure
Each participant completed 2 parts (A and B), typically on separate days.
Each part consists of:
- 9 mini-blocks of 39 trials each = 351 trials
- 36 image trials + 3 empty trials per block
- 1 memory probe after each block
### Trial Sequence
1. **Drift check** — Participant fixates on a cue at one of 9 grid positions (3x3 grid at 960/1920/2880 x 540/1080/1620 px). Max 10 s, 1 retry.
2. **Stimulus presentation** — Image displayed through polygon aperture for 4.0 seconds.
3. **Inter-trial interval** — Blank screen for 0.5 seconds.
### Stimuli
- **Images:** 600 images from the CAT2000 dataset across 6 categories:
- Fractal, Object, OutdoorNatural, Random, Satelite, Sketch
- 100 images per category (102 available, 100 used per part assignment)
- **Polygon apertures:** 27 shapes with systematically varied geometric properties
- See `data/manifests/polygon_geometry.csv` for center coordinates
- See `data/stimuli/polygons/*.json` for vertex definitions
- **Aperture scale factor:** 1987 (applied uniformly)
### Memory Task
After each block, a memory probe image was shown for 3 seconds:
- Participants judged whether the image was "old" (seen in that block) or "new"
- Response time and accuracy recorded
- Purpose: ensure attentive viewing
### Fixation Cue Positions
A 3x3 grid of possible fixation cue locations:
| Grid ID | X (px) | Y (px) |
|---------|--------|--------|
| grid_11 | 960 | 540 |
| grid_12 | 1920 | 540 |
| grid_13 | 2880 | 540 |
| grid_21 | 960 | 1080 |
| grid_22 | 1920 | 1080 |
| grid_23 | 2880 | 1080 |
| grid_31 | 960 | 1620 |
| grid_32 | 1920 | 1620 |
| grid_33 | 2880 | 1620 |
## Data Collection Timeline
Data was collected between January 15, 2026 and February 25, 2026. Most participants completed Part A and Part B on separate days.
## Split Sessions
Some participants required multiple session folders to complete a single part (due to experiment restarts or technical issues). In these cases:
- Trials are concatenated across sessions using `trial_uid` as the unique key
- For overlapping trials (recorded in multiple sessions), the later session's data takes precedence
- The `sessions.csv` metadata file labels these as `split_primary`
|