Endogenous saccades are preceded by shifts of visual attention: evidence from cross-saccadic priming effects
Introduction
Over the past 20 years, a considerable amount of research has been conducted to determine the relationship between the visual attention and oculomotor systems. Specifically, researchers have attempted to determine whether a shift of attention precedes endogenous (voluntary) saccadic eye movements. The most common method to examine this issue has been the dual-task paradigm (e.g. Deubel and Schneider, 1996, Hoffman and Subramaniam, 1995, Kowler et al., 1995, Shepherd et al., 1986). The primary task of this paradigm is an eye movement task, in which the participant is required to execute a saccade in response to an imperative signal. The signal is usually endogenous, such as a central arrow pointing toward a peripheral saccade target. The secondary task is a manual response task that is sensitive to attentional allocation. Typically, the participant is either asked to detect the presence of a probe stimulus (detection task) or to identify a property of the probe stimulus (discrimination task). The rationale in the dual-task procedure is that, if attention does precede the saccade, identification and detection of a probe stimulus should be better when it is presented at the saccade destination than at other locations.
These dual-task studies have generally provided support for the notion that attention does precede the saccade. For example, Kowler et al. (1995) conducted a series of experiments in which participants were required to make a saccade while performing a letter identification task. Participants viewed displays containing eight pre-masks on a circular array surrounding a central fixation cross. A saccade was executed to one of the peripheral objects as indicated by a central arrow. Simultaneous with the onset of the arrow, the pre-masks were replaced by letters, which were masked 200 ms later. In the random report condition the letter `Q' appeared randomly at one of the display locations at the end of the trial, and participants had to report the letter that had appeared there. In the fixed report condition participants were always required to report the letter at a certain fixed location. On the one hand, in the random report condition identification accuracy was best when the saccade goal and letter target were at the same location, providing evidence for the notion that attention shifted to the location at which a saccade was being programmed. On the other hand, in the fixed report condition letter identification was always very good and did not depend on the location of the saccade target. However, this excellent performance was achieved at the cost of an increase in saccade reaction time (SRT) of 50–75 ms. According to Kowler et al. participants were concentrating on the perceptual task before programming the saccade. This illustrates one of the problems of the dual-task paradigm: even if attention necessarily precedes a saccade, it is possible to find no effect of saccade target location simply because participants delay their saccade in order to first concentrate on the perceptual task. Furthermore, in the random report condition participants needed to choose which letters to process first. Given the importance of the saccade destination for the primary (eye movement) task, it is possible that participants were biased to process the letter at this location first. Both tasks required the selection of locations, and in the random report condition participants might have used the selection of the saccade destination as a starting point for the letter identification task. Therefore, it is unclear whether the better perceptual performance at the saccade destination was the result of a shift of attention required for saccade programming, or whether it was just a consequence of the requirement to perform an attention-demanding task simultaneous with an eye movement task.
There have also been a few dual-task studies that have been unable to find a relationship between attention and eye movements (e.g. Remington, 1980, Stelmach et al., 1997). For example, Stelmach et al. (1997) used a temporal order judgment (TOJ) task to assess attentional allocation during saccade programming. In a TOJ task two stimuli are presented in the periphery, and participants are required to judge which of the two appeared first. Previous research has shown (e.g. Stelmach & Herdman, 1991) that this judgment is sensitive to attentional allocation. The critical condition is the condition in which the stimuli are presented simultaneously. If attention is directed to one of the stimulus locations participants will judge this stimulus to have appeared first with a greater than chance probability even though both stimuli appeared simultaneously. In Stelmach et al. (1997) participants executed a saccade to a location at which one of the stimuli would appear. It was hypothesized that more responses in favor of the stimulus that appeared at the saccade destination would be support for the view that attention preceded the saccade. However, there was no effect of saccade target location on the temporal order judgment, which seemed to suggest that attention did not precede saccades. According to Stelmach et al. it is possible that no relationship between attention and saccades was found because the attentional requirements of the temporal order judgment task may be different from discrimination tasks. Presumably, the TOJ task requires preparatory attention while the discrimination task requires selective attention. Stelmach et al. further proposed that selective attention may be relatively fast (tens of milliseconds) while preparatory attention may be relatively slow (hundreds of milliseconds). Therefore, it is possible that preparatory attention is simply too slow to be effective prior to the execution of a saccade. This reasoning points to another problem of the dual-task paradigm: whether a study finds a relationship between attention and saccades could depend on the type of perceptual task used to examine attentional allocation.
Taken together, previous dual-task studies have generally provided evidence for a tight relationship between attention and saccades. However, the dual-task paradigm is not without its problems. First of all, the simultaneous execution of an attention-demanding task and an eye movement task may result in strategic biases of participants to process stimuli at the saccade destination first. Second, participants may in fact delay the programming of the saccade in order to first concentrate on the perceptual task. Third, different perceptual tasks provide qualitatively different results, and the conclusions seem to depend on the type of perceptual task used. Therefore, it is important to examine what other evidence there is supporting the view that attention precedes saccades.
A number of other studies examining the relationship between attention and saccades have used the moving window technique (McConkie & Rayner, 1975), in which the display is changed in real time contingent on the gaze point of participants. In this way the degree of extrafoveal preview of objects to be identified (e.g. Henderson, Pollatsek, & Rayner, 1989) or words to be read (e.g. Henderson and Ferreira, 1990, Rayner et al., 1982) can be manipulated. These studies have typically found that when extrafoveal preview was available at the saccade destination, attention preceded the eyes to that location.
However, the question remains as to why attention preceded the eyes in these studies. One possibility is that attention preceded the eyes because attention was needed for the programming of the saccade. This is how the results of dual-task studies are generally interpreted (e.g. Deubel & Schneider, 1996). There is, however, an alternative cause for the presaccadic shift of attention. When a saccade is programmed to a certain object, this is done in order to facilitate the processing of the object. Since shifts of attention are faster than saccades (e.g. Henderson et al., 1989) shifts of attention prior to the saccade may simply be a way of getting a head start on the processing of the object. In this view attention precedes the saccade simply because attention and saccades have a common goal, namely facilitation of stimulus processing. Thus, the relationship between attention and saccades found in these studies might not have been due to a tight coupling between the attention and oculomotor systems, but might instead have been due to the fact that attention could start processing a stimulus before the eyes had reached it.
The present study is conducted in order to examine whether attention precedes the eyes using a task which avoids the problems of the dual-task paradigm and the moving window paradigm. The present task differs from these previous tasks in the sense that in the present task the only reason participants might have for attending to the saccade destination is directly related to the eye movement. This is done by using a new task in which attentional allocation is assessed by examining priming effects. The priming effect refers to the fact that identification of a stimulus (the target) is facilitated if it matches a stimulus (the prime) previously seen in the same context (Kahneman, Treisman, & Gibbs, 1992). Thus, reaction times (RTs) to identify the target should be lower when the prime is compatible to the target (i.e. the prime matches the target) than when it is incompatible to the target (i.e. the prime does not match the target). Importantly, it is assumed that if the prime is attended to, the processing of the prime is enhanced, which increases its effect on the RT to identify the target. There are two main advantages of using priming effects as an indicator of attentional allocation. First of all, it does not require a separate task to be performed simultaneous with the eye movement task as in the dual-task paradigm. Therefore, participants can fully concentrate on the eye movement task and do not need to allocate their attention in order to identify objects presented during saccade programming. Second, participants receive no preview of the target prior to the execution of the saccade as in the moving window paradigm. Moreover, if a prime, which is presented during saccade programming, is completely non-predictive of the target identity, participants would have no reason whatsoever to attend to it. Therefore, if the priming effect is greater when a prime is presented at the saccade destination than when it is presented at a different location it must be the case that attention is required at the saccade destination in order to program an eye movement.
On each trial three stimulus displays were successively shown: a fixation display, a prime display, and a target display. The fixation display consisted of a central fixation point and two square frames in the periphery, one left and one right of the central fixation point. Once the fixation had been achieved, the prime display appeared, in which an arrow (the imperative signal) replaced the central fixation point and two triangles replaced the peripheral squares. The arrow pointed to one of the triangles, and this indicated the saccade destination. One of the triangles in the prime display was the prime triangle, which was compatible or incompatible relative to the target triangle. The other was a distractor triangle, which was always neutral relative to the target triangle. Before a saccade could be completed the prime display was replaced by the target display, in which a target triangle and a distractor object were presented. Participants were required to identify the orientation of the target triangle as quickly and accurately as possible. The target was so small that it could only be identified when it was fixated.
According to the view that endogenous saccades are preceded by shifts of visual attention it was expected that the priming effect would be larger when the prime was presented at the saccade destination than when it was presented elsewhere. Furthermore, since no task-relevant information could be processed at the saccade destination prior to execution of the saccade (that is, the prime was completely uninformative of the target), this finding would indicate that attention was required at the saccade destination in order to program the saccade and not in order to identify a stimulus presented at that location.
The present study consists of three experiments. Experiment 1 examined whether priming effects could be found across saccades in the present design and whether the priming effects depended on the predictive validity of the prime. Specifically, would priming effects only occur when the prime was predictive of the target orientation, or would priming effects also occur when the prime was non-predictive of the target orientation and was therefore completely task irrelevant? Experiment 2 examined whether attention preceded a saccade by varying the time at which the prime was presented. The prime was presented at the saccade destination either prior to the central arrow or simultaneous with the central arrow. If a movement of attention preceded a saccade it was expected that the priming effect would be larger when the prime was presented simultaneous with the central arrow than when it was presented prior to the central arrow. Experiment 3 compared the priming effect of a prime at the saccade destination with the priming effect of a prime at a no-saccade location. According to the view that attention precedes saccades it was expected that the priming effect would be larger when the prime was presented at the saccade destination than when it was presented at a no-saccade location.
Section snippets
Experiment 1
The main purpose of Experiment 1 was to examine whether a priming effect could be found with the present design when a prime was presented at the target location during the programming of a saccade to that location. A subsidiary purpose of Experiment 1 was to examine whether the predictive validity of the prime was critical for the priming effect.
Experiment 2
In Experiment 2 the prime was presented either at the saccade destination or at a no-saccade location. Furthermore, the prime was presented prior to the onset of the central arrow or simultaneous with the onset of the central arrow. If no attention was required for the priming effect in the present paradigm, a priming effect (i.e. type priming) would still be found when the prime was presented at a location different from the target. This was the case when the prime was presented at the
Experiment 3
The main purpose of Experiment 3 was to examine whether attention precedes an endogenous saccade by comparing priming effects of primes presented at the saccade destination with priming effects of primes presented at the no-saccade location. If attention precedes a saccade the priming effect was expected to be larger when the prime was presented at the saccade destination than when it was presented at the no-saccade location. In order to compare priming effects of primes at the saccade
General discussion
The three experiments contained in the present study were designed to examine whether an endogenous shift of attention precedes saccades. A new task was used in which presaccadic allocation of attention was examined by cross-saccadic priming effects. Experiment 1 showed that priming effects occurred when a prime was presented at the saccade destination even when the prime was completely non-predictive of target identity. Furthermore, priming effects were larger when the prime was presented at
Acknowledgements
This research was supported by Natural Sciences and Engineering Council of Canada grants (0194537 and 0196162) to Jay Pratt and a grant from the Sigma Xi research society to Richard Godijn. The authors would like to thank Paul Cassar for collecting data for Experiments 1 and 2.
References (20)
- et al.
Saccade target selection and object recognition: Evidence for a common attentional mechanism
Vision Research
(1996) - et al.
The reviewing of object files: Object-specific integration of information
Cognitive Psychology
(1992) - et al.
The role of attention in the programming of saccades
Vision Research
(1995) - et al.
Reorienting attention across the horizontal and vertical meridians: Evidence in favor of a premotor theory of attention
Neuropsychologica
(1987) Selection for action: Some behavioral and neurophysiological considerations for attention and action
- et al.
Roles of object-file review and type priming in visual identification within and across eye fixations
Journal of Experimental Psychology: Human Perception and Performance
(1994) - et al.
Effects of foveal processing difficulty on the perceptual span in reading: Implications for attention and eye movement control
Journal of Experimental Psychology: Learning, Memory, and Cognition
(1990) - et al.
Covert visual attention and extrafoveal information use during object identification
Perception and Psychophysics
(1989) - et al.
The role of visual attention in saccadic eye movements
Perception and Psychophysics
(1995) - et al.
Changing views of attention and automaticity
Cited by (41)
An ERP study of preparatory and inhibitory mechanisms in a cued saccade task
2006, Brain ResearchCitation Excerpt :Interestingly, preparing and inhibiting an eye movement were not reflected in P1/N1 modulations of the imperative ERPs. However, the close coupling between attention and eye movements (Rizzolatti et al., 1987, 1994) and the evidence that attention precedes the eye movement towards the goal (Deubel and Schneider, 1996; Godijn and Pratt, 2002; Hoffman and Subramaniam, 1995; Irwin and Gordon, 1998; Van der Stigchel and Theeuwes, 2005a) may suggest that attention was allocated at the target location just before saccade execution. Our finding of the different cue related components also suggests that the cue was used and that spatial attention was allocated at the target location.
Eye movement trajectories and what they tell us
2006, Neuroscience and Biobehavioral ReviewsOrientation of Tactile Attention on the Surface of the Tongue
2022, Journal of Experimental Psychology: Human Perception and PerformanceSpelling ability influences early letter encoding during reading: Evidence from return-sweep eye movements
2021, Quarterly Journal of Experimental Psychology