Eye movements are visually controlled motor responses; however, they are not just this, since their operation also controls the sampling of the visual input. Their involvement with vision thus takes the form of an interactive loop of saccades being visually guided and at the same time controlling/modulating the sampling of the visual scene. The ultimate goal of my research is therefore to track down this visu-oculomotor loop at both the psychological and neural levels and to understand how does visual perception affect eye movements and in turn how does a saccadic decision determines perception.
In order to answer these questions and more, we conduct experiments involving eye tracking device and fMRI. This research is conducted together with my co-advisor Dr Galia Avidan from the Department of Psychology, Ben-Gurion University of the Negev, Beer-Sheva, Israel
I am involved in several studies:
Spatial patterns of fMRI activation in the human eye fields
Various brain regions, such as the frontal eye fields (FEF), supplementary eye fields (SEF), and the intraparietal sulcus (IPS), are known to be involved in eye movement planning and execution in both humans and primates. But it is still unclear whether the direction of a saccade, its origin, final destination point (relative to the head, body or world) or combinations of these elements are mapped in these regions. Recently, new multivariate classification methods have been used to assess the reliability of distinctive spatial patterns of the fMRI response (SPfR) in the visual cortex. We apply similar methods here to study the spatial nature of representation of memory-based saccades in cortical regions. Human subjects performed a memory guided saccade task in which equal-amplitude saccades were executed from several starting points in various directions. We find that the SPfR in the FEF, SEF and IPS is positively correlated when the same saccades are repeatedly executed. The patterns become uncorrelated when saccades of the same magnitude and direction are executed from different starting points. These findings suggest that all these regions contain a neural representation sensitive to both the direction and starting point of the saccade. The representation of memory guided saccades in the human cortex is therefore not entirely oculocentric, but rather, it also contains information in (at least) head-based coordinates.
Implicit guidance of gaze position during free viewing
In this study we analyzed the scanning pattern of subjects while examining the effects of visual primes. We show that, even when no explicit target is defined, the scanning pattern of subjects is robustly influenced by primes. Interestingly, subliminal primes which are shown in fixation attract the gaze to their corresponding location in the inspected image. This demonstrates that implicit processes are guiding attention within complex scenes, even without explicit awareness. Interestingly, only image primes but not word primes seem to influence attention. Implications on search models are discussed.
Accumulation and stability of visual information across multiple fixations.
When looking at a complex visual scene, humans often redirect their gaze to the same objects within the scene, even without being consciously aware of it. In this project, we sat out to examine the functional role of this behaviour and specifically to investigate the nature of the visual information that can be accumulated over recurrent fixations on the same objects. We designed a study in which subjects freely viewed scenes comprising of several objects and were subsequently asked to perform memory tasks related to a randomly chosen target object. In the first experiment, the scene was presented for a fixed duration (6 s) and subjects were then asked to first indicate whether a target object appeared in the stimulus set, then they had to pick the correct viewpoint in which the object was presented and finally they were asked to actively point to the location of the object within the scene by moving the mouse. Memory performance, as assessed by these tasks, was enhanced when multiple fixations were directed to the same object (target fixations). In contrast, performance decreased monotonically, as more fixations directed to other objects within the scene were executed after the last fixation on the target object (intervening fixations). Our results suggest that some information about objects, such as their presence and exact location, fades when fixations are directed elsewhere, and the gaze is redirected to target objects in order to refresh and accumulate more information regarding these objects.