RESEARCH


Our research is directed to understanding the processes that control the planning and programming of eye movements made while people perform natural visual and cognitive tasks. We are focusing on the role of planning, attention, memory, and sensory and perceptual cues, studied in tasks that are designed both to mimic the demands of natural scanning, and to allow us to understand major underlying mechanisms.

We record eye movements with a sensitivity of about 1 minute of arc using a Dual Purkinje Eyetracker, as shown in Figure 1. An example of typical eye movements during reading is shown in Figure 2.

Figure 1
Recording eye movements with the Dual Purkinje Image Eyetracker (Fourward Technologies).
      Figure 2
Example of eye movements during reading.



Some examples of recent and ongoing projects:

Visual search: How do people decide where to look in a scene when they are trying to find something? Recently, we found that choices are not always optimal. Instead of looking at locations where targets were likely to appear, the eye often visited unlikely locations simply because they were closer and took less time to access. Although such strategies seem to be irrational, they might be useful in the long run because they reduce the time needed to plan saccades. (Araujo, Kowler & Pavel, Vision Research, vol. 41, 3613-3625, 2001). We are continuing with these studies using search tasks of various levels of complexity (Figures 3 and 4).

Figure 3
Eye movements made while searching clusters of characters for the letter T. The probability of the T appearing in a red cluster is .8.
Figure 4
Another set of eye movements while searching for the T except the mixing of colors in a cluster makes t he probability level harder to determine.



Saccadic eye movements and attention: We found previously that it is difficult to attend to one place and saccade to another (Kowler, Anderson, Dosher & Blaser, Vision Research, vol. 35, 1897-1916, 1995). This shows that accurate programming of saccades makes some demands on attention. We are continuing to study links between saccades and attention using different visual tasks to understand how attention is distributed during active scanning of scenes.

Visual memory: After scanning a scene like the one in Figure 5, people can recall only about 3 or 4 of the objects. Memory improves when the same scene is presented again several minutes later, showing that visual memory is not fragile but can steadily accumulate over time (Melcher, D. Persistence of visual memory for scenes. Nature, vol. 412, 401, 2001; Melcher & Kowler, Visual scene memory and the guidance of saccadic eye movements, Vision Research, vol. 41, 3597-3611, 2001). Melcher is continuing work on many aspects of visual memory and eye movements.

Figure 5
Eye movements while scanning a visual scene and trying to remember the contents. (Melcher, D. Persistence of visual memory for scenes. Nature, vol. 412, 401, 2001; Melcher & Kowler, Visual scene memory and the guidance of saccadic eye movements, Vision Research, vol. 41, 3597-3611, 2001).



Using saccades to look at shapes: When you look at an object, it is not necessary to select the exact location you want the eye to land. Eye movements directed to an eccentric target shape, such as the L shown in Figure 6, land near the center of gravity with a high degree of precision. It is thought that this ability rests on visual pooling or averaging of information across attended regions. (Vishwanath & Kowler, Localization of concave shapes: Eye movements and perception compared, Vision Research).

Figure 6
Illusory shifts in visual direction accompany adaptation of saccadic eye movements (Bahcall & Kowler, Nature, vol. 400, 864-866, 1999). Panel a is a baseline trials in which a target stepped abruptly to the right and the eye followed with a saccade (dotted line is the target; solid line is the eye). In panel b the target jumped during the saccade. At first, saccades missed the target but after a few trials of adaptation (panel c) the saccades became accurate. Panel d shows one of the interleaved perceptual probe trials. Even though the probe was located about one-half degree to the right of the target, the observer saw the probe as aligned with the target.



Saccadic adaptation: If targets are moved while a saccade is in progress the saccade will of course miss the target, as is shown in Figure 7. But after a few trials of following such motion, saccades adapt and are accurate once again. We found that this adaptation produces a shift in visual direction such that targets flashed briefly right after an adapted saccade appear to be in the wrong place. This illusion means that the signal that the perceptual system uses to keep track of the position of the eye is generated at an early stage of eye movement programming, before the site of adaptation. (Bahcall & Kowler, Nature, vol. 400, 864-866, 1999).

Figure 7
Eye movements directed to an eccentric target shape, such as this L, land near the center of gravity with a high degree of precision. This illustrates the spatial pooling that is required to accurately localize selected objects quickly. (Vishwanath & Kowler, Localization of concave shapes: Eye movements and perception compared, Vision Research).


Last update: January, 2003.


:: Psychology   :: RuCCS   ::Rutgers home