Table 1.
Observed scanning strategies in the eye-tracking studies of multiple object tracking (MOT).
| Study | Stimulus Size (°) | Set-Size | Speed (°/s) | Analysis Method | Scanning Strategy |
|---|---|---|---|---|---|
| Fehd & Seiffert [8] | 2.1 | 1 | 15 | Shortest distance rule | Target: 96% |
| Follow-up | 3 | Centroid: 66%; Target: 9% | |||
| 3 | Centroid: 42%; Target: 11% | ||||
| 4 | Centroid: 42%; Target 9% | ||||
| 5 | Centroid: 42%; Target 8% | ||||
| Fehd & Seiffert [9] | AoI: 5° in diameter | ||||
| Experiment 1 | 2.1 | 4 | 3, 6, 12, 18, 24 | Centroid:~25%; Target: ~10% | |
| Experiment 2 | 0.06–0.3 | 4 | 3, 6, 8, 12, 24 | Centroid: 34%; Target: 8% | |
| Experiment 3 | 1.8 | 3 | 12 | Centroid: 43%; Target 13% | |
| Zelinsky & Neider [10] | 0.5–1.1 | 1 | 1.1 | Shortest distance rule | Target: 94% |
| 2 | Centroid: 47%; Target: 37% | ||||
| 3 | Centroid: 39%; Target: 42% | ||||
| 4 | Centroid: 24%; Target: 52% | ||||
| Zelinsky & Todor [11] | 0.5–1.1 | 2–4 | Not reported | “Rescue saccade”: landing 1° from a target | Anticipatory rescue saccades were initiated to occluded targets |
| Huff et al. [12] | 1.3–2.2 | 3 | AoI: same size as the objects | ||
| Experiment 1 | 2 | Centroid:~7%; Target: ~10% | |||
| 4 | Centroid:~7.5%; Target:~8.5% | ||||
| 6 | Centroid:~10%; Target:~8% | ||||
| Experiment 2 | 4 | Centroid:~11%; Target:~8% | |||
| 10 | Centroid:~12.5%; Target:~5.5% | ||||
| Vater et al. [13] | 1 | 4 | 6, 9, 12 | AoI: 5° in diameter | Centroid: 30%; Target: 11% |
| Vater et al. [14] | 1 | 4 | 6 | Gaze-vector distances | Gaze was closer to centroid than to target regardless of target changes |
| Vater et al. [15] | 1 | 4 | 6 | Relative gaze distance to the targets | Gaze was closer to crowded than uncrowded targets. Anticipatory saccades were initiated to targets colliding with border. |
| Lukavský [16] | 1 | 4 | 5 | AoI: 1° for centroid, 2° for objects; Normalized scanpath saliency | Centroid: 7.7%; Target: 12.6%; Anticrowding point: 12.2%; Target eccentricity minimizing point: 9% |
| Dechterenko & Lukavsky [17] | 1 | 4 | Adaptive | Normalized scanpath saliency | The model accounting for the crowding effect yielded the best performance. |
| Lukavský & Děchtěrenko [18] | 1 | 4 | 5 | Local maximum of similarity | Gaze position lagged by approximately 110 ms behind the scene content. |
| Děchtěrenko, Lukavský, & Holmqvist [19] | 1 | 4 | 5 | Correlation distance | Scan patterns in flipped trials differed only slightly from those of the original trials. |
| Oksama & Hyönä [20] | |||||
| Experiment 1 | 2.1 | 2–5 | 2.6, 6.3, 10.3 | AoI: 3.4° | Blank area: 48%; Target: 21%; Centroid: 7%; Distracter: 4% |
| Experiment 3 | 2.1 | 2–5 | 2.6, 6.3, 10.3 | AoI: 3.4° | Blank area: 48%; Target: 24%; Centroid: 7%; Distracter: 5% |