Journal of Experimental Psychology: Human Perception and Performance 2004, Vol. 30, No. 1, 194 –211
Copyright 2004 by the American Psychological Association, Inc. 0096-1523/04/$12.00 DOI: 10.1037/0096-1523.30.1.194
Saccadic Inhibition in Reading Eyal M. Reingold and Dave M. Stampe University of Toronto In 5 experiments, participants read text that was briefly replaced by a transient image for 33 ms at random intervals. A decrease in saccadic frequency, referred to as saccadic inhibition, occurred as early as 60 –70 ms following the onset of abrupt changes in visual input. It was demonstrated that the saccadic inhibition was influenced by the saliency of the visual event (Experiment 3) and was not produced in response to abrupt but irrelevant auditory stimuli (Experiment 1). Display changes restricted to an area either inside or outside the perceptual span required for normal reading produced strong saccadic inhibition (Experiment 2). Finally, Experiments 4 and 5 demonstrated higher level cognitive or attentional modulation of the saccadic inhibition effect.
recovery from inhibition. Finally, following the peak, the proportion of saccades returned to initial levels. Although visual inspection of saccadic frequency histograms is an acceptable method for the purpose of demonstrating the general phenomenon, a more thorough exploration of how stimulus and task parameters affect saccadic inhibition requires the development and validation of quantitative strength and latency measures. The present experiments were designed, in part, to devise and validate such measures (see Figure 1D). In addition, in the present study we attempted to explore the potential implications of saccadic inhibition for reading research. Reingold and Stampe (2000) argued that saccadic inhibition is a low-level oculomotor effect. An important question is whether saccadic inhibition is sensitive to higher level cognitive or attentional factors. That is, can such factors enhance or attenuate the saccadic inhibition effect? If it can be shown that any component of the latency or strength of saccadic inhibition is sensitive to higher level attentional or cognitive influences, then it may be possible to use this effect to study processes that underlie task performance. This issue falls under the general topic of the relation between visual attention and saccadic control. At present, this is a controversial issue. For example, Rizzolatti and his colleagues (Rizzolatti, Riggio, Dascola, & Umilta, 1987; Rizzolatti, Riggio, & Sheliga, 1994) suggested that the same neural circuits subserve visuospatial attention and eye movements and that a shift of attention corresponds to the programming of an eye movement. They explained covert shifts of attention in the absence of concomitant eye movements (e.g., Downing, 1988; Posner, 1980; Posner, Nissen, & Ogden, 1978) as cases in which an eye movement is programmed but not executed. This position has been widely contested (e.g., Hodgson & Mu¨ller, 1995; Klein, 1980; Klein, Kingstone, & Pontefract, 1992; Klein & Pontefract, 1994; Posner, 1980). Instead, a growing body of research has suggested that whereas attention can be shifted covertly in the absence of eye movements, eye movements are preceded by an attentional shift to the saccadic target (e.g., Deubel & Schneider, 1996; Henderson, 1993; Hoffman, 1998; Hoffman & Subramaniam, 1995; Kowler, Anderson, Dosher, & Blaser, 1995; Rafal, Calabresi, Brennan, & Sciolto, 1989; Rayner, McConkie, & Ehrlich, 1978; Remington, 1980; Schneider & Deubel, 1995; Shepherd, Findlay, & Hockey,
During the performance of complex visual tasks such as reading and visual search, observers produce high-velocity eye movements referred to as saccades. The periods between saccades during which the eye is relatively still are referred to as fixations. Saccades are required to align the high-acuity foveal region of the visual system (the central 2° of vision) with objects of interest in the visual field (see Rayner, 1998, for a recent review). In previous studies, Reingold and Stampe (1997, 2000) devised a paradigm that explored the effects of task-irrelevant, sudden-onset visual events on scanning saccades produced during reading and visual search. For example, while participants were reading for comprehension, a text screen was replaced for 33 ms by a black screen at intervals that varied randomly between 300 and 400 ms, resulting in the subjective experience of a flicker. These studies documented a decrease in saccadic frequency time-locked to the visual event and occurring as early as 60 –70 ms following the onset of the flicker. It was argued that the short latency of this effect, which approaches the limits imposed by delays in the visual and saccadic systems, strongly suggests a low-level, reflexlike, oculomotor effect, which was referred to as saccadic inhibition. A histogram of saccadic frequency following flicker onset is shown in the flicker condition in Figure 1C, which illustrates the typical saccadic inhibition profile. Specifically, for the first 50 ms following the display change, the proportion of saccades remained constant. Approximately 60 –70 ms following the onset of the flicker, the proportion of saccades decreased below this initial level, forming a dip that reflects saccadic inhibition. Following the dip, an increase above the initial level of saccadic frequency occurred, forming a peak, which likely reflects the
Eyal M. Reingold and Dave M. Stampe, Department of Psychology, University of Toronto, Toronto, Ontario, Canada. Preparation of this article was supported by a grant to Eyal M. Reingold from the Natural Science and Engineering Research Council of Canada. We thank Elizabeth Bosman and Colleen Ray for their helpful comments on an earlier version of this article. Correspondence concerning this article should be addressed to Eyal M. Reingold, Department of Psychology, University of Toronto, 100 St. George Street, Toronto, Ontario M5S 3G3, Canada. E-mail:
[email protected] 194
SACCADIC INHIBITION
195
Figure 1. Results of Experiment 1. Histograms of fixation durations (A), saccadic amplitudes (B), and saccadic frequency (C) following the display change for the control, beep, and flicker conditions. D: Normalized saccadic frequency in the flicker condition, and saccadic inhibitions measures used in all of the experiments. L50% ⫽ latency to 50% of maximum inhibition; LMAX ⫽ latency to maximum saccadic inhibition.
1986; but see Stelmach, Campsall, & Herdman, 1997, for evidence against this preallocation hypothesis). Furthermore, the link between the oculomotor and attentional systems is supported by neurophysiological data (e.g., Goldberg & Wurtz, 1972; Kustov & Robinson, 1996; Mohler & Wurtz, 1976; Wurtz & Mohler, 1976) and work with neuropsychological populations such as neglect patients (e.g., Johnston & Diller, 1986; Walker & Young, 1996). Not surprising, the issue of the relationship between attention and saccadic control in the context of reading research is also controversial. Models of saccadic control in reading differ dramatically with respect to the hypothesized role of visuospatial attention in the programming and execution of eye movements (see Rayner, 1998, for a review). The attentional guidance model postulates tight coupling between attention and saccadic control in reading. An early version of this model, proposed by Morrison (1984; see also Just & Carpenter, 1980), argued that an attention shift in the direction of the next saccade occurs prior to its
execution. Specifically, this model assumes that attention is initially centered on the foveated word (word n) during fixation. Following lexical encoding of the fixated word, attention covertly shifts in the direction of reading, and a saccade aimed at fixating the newly attended word (word n ⫹ 1) is programmed. If parafoveal lexical encoding of word n ⫹ 1 is completed prior to the execution of the next saccade, attention further shifts to word n ⫹ 2, and a new saccade aimed at fixating word n ⫹ 2 is programmed. In such instances, either word n ⫹ 1 is skipped (i.e., the n to n ⫹ 1 saccade is canceled), or if the point of no return is exceeded, word n ⫹ 1 is only briefly fixated on route to fixating word n ⫹ 2 (see Rayner & Pollatsek, 1989, for a review and discussion). Morrison’s (1984) model was very influential, and several modified versions aimed at extending it were proposed (e.g., Henderson, 1992; Henderson & Ferreira, 1990; Kennison & Clifton, 1995; Pollatsek & Rayner, 1990; Rayner & Pollatsek, 1989; Reichle, Pollatsek, Fisher, & Rayner, 1998). In addition, alternative models
196
REINGOLD AND STAMPE
were formulated, which assume that nonlexical, low-level information determines saccadic control in reading (e.g., Kowler & Anton, 1987; McConkie, Kerr, Reddix, & Zola, 1988; McConkie, Kerr, Reddix, Zola, & Jacobs, 1989; O’Regan, 1990, 1992). Such models argue that the influence of higher level cognitive or attentional processes on saccadic control in reading is very limited. Currently, the issue of the relationship between attention and saccadic control in reading remains controversial (e.g., Brysbaert & Vitu, 1998; Deubel, O’Regan, & Radach, 2000; Kliegl & Engbert, 2003; McConkie & Yang, 2003; Pollatsek, Reichle, & Rayner, 2003; Rayner, 1998; Reilly & Radach, 2003; Vitu & O’Regan, 1995; Vitu, O’Regan, Inhoff, & Topolski, 1995; for a comprehensive review and discussion, see Reichle, Rayner, & Pollatsek, in press, and its related commentaries). One clear prediction that can be derived from the attentional guidance model is the occurrence of perceptual enhancement in the direction of the next saccade, due to preallocation of attention to the target of the saccade. The magnitude of this perceptual enhancement should be influenced by high-level cognitive or attentional task demands. It is in this context that the saccadic inhibition effect may provide a useful contribution to this debate, and Experiment 4 was designed to examine this possibility. In Experiment 4, a flicker was presented either in the direction of the next saccade (congruent condition) or in the opposite direction (incongruent condition). If perceptual enhancement in the direction of the next saccade occurs, then saccadic inhibition would be expected to be stronger in the congruent condition than in the incongruent condition. In addition, the size of such a congruency effect was compared across a reading condition (in which participants read for comprehension) and a “mindless reading” condition (in which all the letters in the text were replaced by the letter Z and participants pretended to read these Z-strings). The characteristics of saccades have been found to be quite similar between the reading condition and the mindless reading condition (Vitu et al., 1995; but see Rayner & Fischer, 1996), and consequently, a difference in the congruency effect across these conditions would provide strong support for a higher level, task-specific influence. Such a pattern was demonstrated in Experiment 4. The saccadic inhibition phenomenon may be relevant to reading research in another important way. Reingold and Stampe (2000) pointed out that several studies reported dips in fixation duration distributions in both reading (e.g., Blanchard, McConkie, Zola, & Wolverton, 1984; McConkie, Underwood, Zola, & Wolverton, 1985) and visual search (e.g., van Diepen, de Graef, & d’Ydewalle, 1995). These studies used gaze-contingent paradigms (see Rayner, 1998, for a review of gaze-contingent paradigms) in which a mask was presented, centered about or displaced from the point of gaze, at a fixed delay from the onset of fixations. For example, Blanchard et al. (1984) briefly masked the text at delays of 50, 80, or 120 ms after the fixation began. The mask was presented for 30 ms and was followed by the reappearance of the text. Although not reported in the original article, McConkie, Reddix, and Zola (1992) reported that in this study, the stimulus change produced a large dip in the distribution of fixation durations that began about 80 –90 ms following the visual change. This effect was attributed to a disruption of massively parallel and automatic registration or encoding processes that operate to provide utilization processes with the information required for reading comprehension (henceforth, the processing disruption hypothesis).
This model assumes that following the onset of fixation, a timelocked, fixed sequence of registration processes extract visual information from the text. At the termination of this sequence, visual information becomes available for use by higher level utilization processes, but the time at which this information is actually used is variable (i.e., registration but not utilization processes are time-locked to the beginning of fixation). A mask appearing during the period in the fixation when registration processes are active will result in a constant delay of the next saccade, rendering a certain range of fixation durations less probable and resulting in the dip in the fixation distribution. However, as pointed out by Reingold and Stampe (2000), it is possible that the dips observed in the fixation distributions may result from a reflexive, low-level saccadic inhibition effect that is time-locked to the flicker associated with the presentation of the mask (henceforth, the saccadic inhibition hypothesis). A crucial prediction of the processing disruption hypothesis is that the disruption caused by the mask should vary as a function of fixation onset to mask delay. This is the case because the mask may or may not affect registration processes, depending on the length of the delay. If the mask appears after registration processes are complete, fixation durations may still be impacted because of disruption to utilization processes. However, this disruption should not result in fixation distribution dips because utilization processes are not time-locked to the beginning of fixation, and consequently, the nature and the magnitude of the disruption should vary across fixations. Experiment 1 was designed, in part, to test this prediction. Inconsistent with the processing disruption hypothesis, Experiment 1 demonstrated that regardless of the point in the fixation at which a display change occurred, a stereotyped inhibition pattern time-locked to the onset of the display change followed. Furthermore, in clear opposition to the processing disruption hypothesis, Experiment 2 demonstrated that display changes produced strong saccadic inhibition regardless of whether the text being processed was altered or masked. In summary, in the present research we attempted to further explore the saccadic inhibition effect in the context of reading as the saccade-generating task. First, the saccadic inhibition paradigm was extended to provide quantitative strength and latency measures of this phenomenon. Devising and validating such measures is an important prerequisite for exploring the effect of experimental manipulations on the saccadic inhibition effect. Second, the potential implications of saccadic inhibition for reading research were examined. In the present experiments, participants read black text on a light background. This text was replaced by a transient image (e.g., a black screen) for 33 ms, at intervals that varied randomly between 300 and 400 ms. The display change (e.g., flicker) associated with the presentation of the transient image was not time-locked (i.e., synchronized) to the beginning of fixation and, consequently, could occur at any point during a fixation. In some experiments the transient image covered the entire screen, whereas in other experiments it was displayed only within the boundaries of a square window that was centered on or offset from the point of gaze (i.e., a gaze-contingent window). Experiment 1 examined the effects on saccadic frequency of a flicker (the screen briefly turned to black) and an auditory beep and compared these conditions to a control condition in which no stimulus change occurred. Only the flicker condition produced saccadic inhibition. Experiment 1 was also
SACCADIC INHIBITION
used to validate the quantitative measures of saccadic inhibition. Experiment 2 demonstrated that display changes restricted to an area either inside or outside the perceptual span required for normal reading produced strong saccadic inhibition. Experiment 3 examined the effects of the saliency of the display change on the saccadic inhibition measures. Experiment 4 compared the saccadic inhibition effect produced by display changes that were congruent or incongruent with the direction of the next saccade, in both a reading for comprehension task and a mindless reading task. The results of Experiment 4 demonstrated that higher level cognitive or attentional influences might modulate the strength of the saccadic inhibition effect. Finally, Experiment 5 established the generality of the findings obtained in the reading task in Experiment 4 by demonstrating comparable effects in both forward and regressive saccades and across a wide range of saccadic amplitudes.
Experiment 1 A valid estimate of baseline saccadic frequency (the rate of saccade production in the absence of a display change) is a prerequisite for the development of measures of saccadic inhibition. In the present paradigm, saccadic frequency during the first 48 ms following the onset of the display change (e.g., the flicker) was used to compute the baseline saccadic frequency. On the basis of available knowledge of neural transmission delays in the visual and saccadic systems (for reviews, see Reingold & Stampe, 2000, 2002), it was expected that during this period, saccadic frequency would not be affected by the display change. Therefore, saccadic frequency during this baseline period was used to calculate quantitative estimates of the strength and latency of saccadic inhibition. These measures are illustrated in Figure 1D. The magnitude of saccadic inhibition was defined as the proportion of saccades inhibited when inhibition was at its maximum (i.e., at the center of the dip). Two latency measures were computed: L50% and LMAX, which represent the latencies from the display change at which inhibition achieved 50% and 100% of its magnitude, respectively. To quantify how sustained the inhibition effect was, we defined a duration measure as the temporal interval during which inhibition was greater than or equal to 50% of its magnitude. Finally, to estimate the proportion of saccades produced during the recovery from inhibition, we defined the peak area measure as the proportion of saccades that were in excess of the baseline following the dip. Given the importance of the baseline estimate in the computation of all these measures, a major goal of Experiment 1 was establishing the validity of this estimate. The display change used in the experiment was a simple flicker, produced by replacing the black text on a light background with a black screen for 33 ms. The flicker condition was contrasted with the control (invisible display change) condition. The validity of the baseline estimate was tested by comparing the proportion of saccades during the baseline period of the flicker condition with the control condition. In a third condition, a beep replaced the flicker to evaluate whether inhibition is produced in response to a nonvisual transient stimulus. A group of 10 participants was tested to contrast the flicker, beep, and control conditions. An additional goal of Experiment 1 was a comparison of the quantitative measures of saccadic inhibition that were computed separately, on the basis of each individual participant’s histogram, and then averaged (i.e., in the flicker condition described above),
197
with measures that were computed on the basis of a group histogram (i.e., a histogram aggregating saccades across all participants). For the purpose of obtaining stable group estimates, an additional group of 50 participants was tested in the flicker condition only, and the saccades produced by all participants were pooled to create a single metaparticipant (henceforth, the aggregate flicker condition). The aggregate flicker condition was then compared with the flicker condition. Finally, the aggregate flicker condition was also used to derive a stable estimate of the onset of saccadic inhibition.
Method Participants. Two groups of participants who were all native English speakers were tested. Participants ranged in age from 19 to 23 years. The first group of 10 participants was tested in the flicker, beep, and control conditions. The second group of 50 participants was tested in the aggregate flicker condition. All participants had normal or corrected-to-normal vision and were paid CAN $10.00 (about U.S.$7.50) per hour. Apparatus. The SR Research Ltd. EyeLink eye tracking system used in this research has high spatial resolution (0.005°) and a sampling rate of 250 Hz (4-ms temporal resolution). The three cameras on the EyeLink headband allow simultaneous tracking of both eyes and of head position, computing true gaze position with unrestrained head motion. By default, only the participant’s dominant eye (assessed by the sighting test: Miles, 1929) was tracked in our studies. The EyeLink system uses an Ethernet link between the eye tracker and display computers, which supplies real-time gaze position data. The on-line saccade detector of the eye tracker was set to detect saccades with an amplitude of 0.5° or greater, using an acceleration threshold of 9500°/s2 and a velocity threshold of 30°/s. Participants viewed a 17-in. ViewSonic 17PS monitor from a distance of 60 cm, which subtended a visual angle of 30° horizontally and 22.5° vertically. The display was generated with an S3 VGA card, and the frame rate was 120 Hz. Materials and randomization. Participants read a short story for comprehension. The text was presented in black on a white background. Antialiased, proportionally spaced text was used, with an average of 2.2 characters per degree of visual angle and an average of 10 lines of text per page. Each trial displayed one page of text in the story, with pages presented in the same order to all participants. Participants in the aggregate flicker group read 22 pages of text. The other group of participants read a total of 72 pages of text, with 24 pages randomly assigned to each condition (flicker, beep, or control). The pairing of pages to conditions was determined randomly for each participant, constrained to allow no more than 3 contiguous trials of the same condition. In the flicker condition, a black (4.0 cd/m2) transient image briefly replaced the normal text image. The control condition used a transient image identical to the normal image, resulting in an invisible display change. In the beep condition, a 2000-Hz square-wave tone was produced during the period when the transient image would have been displayed. Both display changes and beeps lasted for 33 ms and occurred at random intervals between 300 and 400 ms. The average brightness of the normal image was 40 cd/m2. Procedure. A 9-point calibration was performed at the start of the experiment, followed by a 9-point calibration accuracy test. Calibration was repeated if the error at any point was more than 1° or if the average error for all points was greater than 0.5°. Before each screen of text, a black fixation target was presented at the top left corner of the display. The participant fixated this target, and the gaze position measured during this fixation was used to correct any postcalibration drift errors. Throughout the reading of each screen, the experimenter was able to view on a separate monitor the text the participant was reading, overlaid with a cursor corresponding to real-time gaze position. If the experimenter judged that gazetracking accuracy had declined, the experimenter initiated a full calibration before the next screen. This occurred very infrequently. Participants were
198
REINGOLD AND STAMPE
instructed to ignore the flicker and to read the text for comprehension. They were told that they would be asked questions about the content of the story when they finished reading. On average, participants answered 94.3% of these questions accurately, indicating that they complied with the instructions and were not simply scanning the text. After reading each page, participants pressed a button to end the trial and proceed to the next page of the story. Data analysis. The time of the display of the transient image was recorded, along with eye movement data for later analysis. Eye tracker data files were processed to produce histograms of saccade frequency as a function of latency from the display change. A separate histogram was compiled for each participant and condition and analyzed to produce measures of the evoked saccadic inhibition. For the aggregate flicker condition, a composite histogram was generated from the saccades of all 50 participants. Histogram generation. The times of display changes, and of the onset of all saccades with amplitudes of 0.5° or greater, were extracted from the eye tracker data files. Saccades occurring during display changes were discarded. Of the remaining saccades, only forward saccades were used (i.e., left to right saccades) to compile the histograms. The number of saccades for each participant in each condition exceeded 800 saccades, allowing for the generation of reliable saccadic frequency histograms. To maximize the temporal resolution of all histograms, we used a 4-ms bin width (i.e., the maximum temporal resolution of the eye tracker). These narrow bins resulted in noisy individual participant histograms (there were typically only 10 –30 saccades in each bin). To reduce this noise, we applied a 7-bin running average filter to each histogram, which replaced each bin with the average of itself, the three previous bins, and the three following bins. The figures in this article were produced from histograms aggregated across all participants and were not filtered because noise was reduced by the greater number of saccades per bin. Baseline. As documented by Reingold and Stampe (2000), a typical histogram of saccadic frequency by latency from display change can be divided into three distinct parts: the baseline, the dip, and the peak (see Figure 1D for an example). The frequency of saccades remains unaffected by the display change for at least 48 ms following the display change (the baseline period). At longer latencies, a reduction in frequency caused by saccadic inhibition is followed by a peak associated with higher saccadic frequency during recovery from inhibition, after which saccadic frequency returns to the baseline rate. In the histogram for each condition, the values corresponding to the number of saccades in the first 12 bins (48 ms) following the onset of the display change were averaged to estimate the baseline or expected saccadic frequency. Each bin of the histogram was then divided by the baseline saccadic frequency. This converts the histogram units into the ratio of measured to expected saccadic frequency. The strength of saccadic inhibition, which is the proportion of expected saccades eliminated by inhibition, can be computed for any time after the display change by subtracting the normalized saccadic frequency from 1.0. Measures of inhibition. The first step in analysis of the histogram was to locate minimum saccadic frequency. This was achieved by finding the lowest three-bin average for bins 13– 44 (49 –176 ms) following the onset of the display change. Ten percent of the difference between the baseline and minimum saccadic frequency was then added to the minimum saccadic frequency. The bottom of the dip was then defined as the period during which saccadic frequency was equal to or less than this value. The latency to maximum saccadic inhibition (LMAX) was then computed as the time interval from the onset of display change to the center of the bottom of the dip. Because the histogram had been normalized, the magnitude of inhibition (henceforth, magnitude) was simply computed as 1.0 minus the average normalized saccadic frequency corresponding to the center of the bottom of the dip. A magnitude of 1.0 means that no saccades were observed in the bin corresponding to the center of the bottom of the dip.
Latency to 50% of maximum inhibition (L50%) was then defined as the latency from the display change at which inhibition first reached 50% of its maximum strength. To minimize the effects of noise, we actually computed this measure by averaging the latency of all bins prior to the center of the dip in which inhibition was between 33% and 67% of its maximum strength. Similarly, the time at which inhibition returned to 50% of its maximum strength following the center of the dip was computed, and the difference between L50% and this value was defined as the duration of inhibition (henceforth, duration). That is, duration corresponds to the period in which inhibition remained above 50% of its maximum strength. Finally, we defined the peak area measure as the proportion of saccades that were in excess of baseline levels following the crossover point between the dip and the peak. It was not possible to derive fine-grained measures of latency and magnitude for the recovery peak in the same manner as that for the dip. This is the case because although the dip patterns were very stereotyped and similar across participants, the peak patterns were quite variable across participants.
Results and Discussion Before considering the patterns of saccadic inhibition, it was important to examine any differences in eye movement behavior across the flicker condition, the beep condition, and the control condition. Figure 1 plots the fixation duration distribution (A) and saccadic amplitude distribution (B) across these conditions. The means and standard errors for these measures across conditions are shown in Table 1. As can be seen by inspection of Figure 1 and Table 1, the flicker condition produced longer fixation durations compared with the other two conditions: flicker versus control, t(9) ⫽ 7.03, p ⬍ .001; flicker versus beep, t(9) ⫽ 6.13, p ⬍ .001. There was no difference in fixation duration across the beep and control conditions (t ⬍ 1). As for saccadic amplitude, there was a small but significant effect, reflecting an increase in the length of saccades for the flicker condition compared with the other two conditions: flicker versus control, t(9) ⫽ 3.94, p ⬍ .01; flicker versus beep, t(9) ⫽ 2.86, p ⬍ .05. There was no difference in saccadic amplitude across the beep and control conditions (t ⬍ 1). The effect of the flicker on fixation duration is ambiguous as it may reflect either saccadic inhibition or the disruption of reading comprehension processes. Similarly, the effect on amplitude in this paradigm is ambiguous because unlike studies of stimulus elicited saccades, in which the saccadic target is clearly specified, in reading, multiple destinations for the next saccade exist, and there is a substantial range of saccadic amplitudes (see Figure 1B). Furthermore, it is possible that the small effect on saccadic amplitude resulted from longer fixation durations that may have allowed for additional parafoveal processing to occur, which occasionally may have permitted the skipping of short words (i.e., longer saccades on average). Given these limitations, the measures of fixation duration and saccadic amplitude are included in Table 1 for descriptive purposes but are not analyzed in subsequent experiments. Instead, the emphasis is on analyzing the saccadic inhibition measures. Figure 1C plots saccadic frequency as a function of latency from display change for the control, beep, and flicker conditions. As can be seen by an inspection of Figure 1C, the flicker condition produced saccadic inhibition, whereas the beep condition did not differ from the control condition. Furthermore, saccadic frequency in the first 48 ms following the display change did not differ significantly across conditions (as measured by the proportion of
SACCADIC INHIBITION
199
Table 1 Means (and Standard Errors) of Saccadic Inhibition Measures, Saccadic Amplitude, and Fixation Duration Across Conditions in Experiments 1–5 Experiment and condition Experiment 1 Control Beep Flicker Aggregate flicker Experiment 2 18° central 18° peripheral Experiment 3 0.1° displacement 0.3° displacement 0.6° displacement Flicker Experiment 4 Congruent text Incongruent text Congruent Z-string Incongruent Z-string Experiment 5 Congruent ⬍ 2.3° Incongruent ⬍ 2.3° Congruent ⬎ 5° Incongruent ⬎ 5° Congruent forward Incongruent forward Congruent regressive Incongruent regressive
L50% (ms)
LMAX (ms)
Duration (ms)
Magnitude (prop.)
Peak area (prop.)
Saccade amplitude (deg.)
Fixation duration (ms) 240 (9.2) 241 (9.5) 252 (9.4) 251
69.8 (1.4) 70.0
95.0 (2.2) 94.0
52.2 (1.2) 48.0
.872 (.024) .791
.113 (.014) .089
3.61 (0.22) 3.62 (0.66) 3.74 (0.22) 3.50
73.0 (1.7) 72.0 (2.5)
98.4 (2.6) 94.4 (3.2)
51.5 (2.4) 42.4 (2.6)
.650 (.051) .684 (.028)
.101 (.013) .093 (.011)
3.65 (0.30) 3.78 (0.32)
248 (8.9) 243 (9.1)
90.6 (3.1) 85.2 (1.9) 79.8 (2.6) 68.6 (1.5)
109.0 (2.0) 105.2 (1.8) 105.8 (1.5) 92.8 (1.6)
31.1 (2.1) 35.6 (1.6) 43.6 (2.9) 50.0 (3.7)
.426 (.034) .668 (.044) .772 (.049) .789 (.042)
.034 (.006) .062 (.013) .080 (.012) .095 (.013)
3.66 (0.22) 3.66 (0.22) 3.59 (0.23) 3.64 (0.22)
233 (8.6) 236 (7.9) 239 (7.9) 239 (8.6)
74.9 (1.6) 75.0 (1.4) 76.2 (1.4) 76.2 (1.2)
99.9 (1.6) 96.0 (1.2) 102.1 (1.2) 99.8 (1.4)
47.1 (1.7) 38.4 (1.8) 49.8 (1.5) 44.6 (1.5)
.812 (.024) .661 (.030) .847 (.018) .795 (.025)
.127 (.11) .084 (.007) .204 (.019) .148 (.014)
3.91 (0.27) 3.89 (0.27) 4.44 (0.31) 4.47 (0.33)
235 (6.4) 236 (6.2) 284 (10.4) 289 (11.3)
82.9 (1.9) 79.0 (1.6) 68.8 (2.6) 68.0 (2.1) 76.4 (1.6) 75.2 (1.2) 78.4 (2.1) 79.1 (1.6)
107.4 (2.5) 103.0 (2.9) 97.6 (2.4) 93.2 (3.0) 102.4 (2.4) 98.4 (3.2) 106.8 (2.1) 100.6 (1.8)
47.7 (2.3) 46.6 (3.1) 49.6 (2.2) 46.8 (3.8) 46.8 (1.6) 43.4 (2.3) 49.0 (1.4) 41.4 (2.7)
.875 (.034) .748 (.040) .823 (.024) .644 (.040) .823 (.026) .687 (.033) .775 (.028) .580 (.057)
.122 (.013) .074 (.010) .132 (.016) .086 (.012) .122 (.010) .071 (.005) .083 (.014) .034 (.010)
1.72 (0.03) 1.72 (0.03) 6.16 (0.10) 6.20 (0.11) 3.83 (0.22) 3.83 (0.22) 2.44 (0.16) 2.53 (0.17)
256 (11.0) 260 (12.1) 245 (10.3) 245 (9.2) 248 (11.1) 251 (11.1) 220 (10.6) 215 (10.6)
Note. L50% ⫽ latency to 50% of maximum inhibition; LMAX ⫽ latency to maximum saccadic inhibition; prop. ⫽ proportion; deg. ⫽ degrees.
saccades in the first 48 ms of the histogram: flicker ⫽ .145, control ⫽ .140, beep ⫽ .136; all ts ⬍ 1), thus confirming the validity of its use as the baseline measure. Figure 1D plots the normalized saccadic frequency (i.e., divided by baseline saccadic frequency) in the flicker condition as a function of latency from display change. As well, saccadic inhibition measures are illustrated in Figure 1D. Table 1 provides the means and standard errors of these measures for the flicker condition. In addition, the values for the same measures based on the composite histogram for the aggregate flicker group are provided. A comparison of the saccadic inhibition estimates across the flicker and aggregate flicker conditions suggests that latency estimates are remarkably similar regardless of whether they are based on a composite histogram or individual participant histograms, with L50% equaling about 70 ms, LMAX equaling about 95 ms, and duration equaling about 50 ms. Note that LMAX equals L50% plus half of the duration. This is not surprising because, as can be seen in Figure 1D, the duration measure is defined as the distance between L50% and its corresponding point on the other side of the dip, and LMAX (i.e., the time corresponding to the center of the dip) is roughly midway between these two points. Thus, LMAX is redundant with L50% and duration and consequently is included in Table 1 but is not analyzed in subsequent experiments. The aggregate flicker estimates for the strength measures (i.e., magnitude and peak area) appear to underestimate the strength of inhibition by about 10% and recovery by about 20%, based on individual participant histograms. This underestimation is a result of inter-
participant variability in the shape of the histogram, which causes erosion of the peak and dip in the composite histogram and therefore results in lower magnitude and peak area estimates. The aggregate flicker condition was further analyzed to provide an estimate of the latency to the onset of inhibition. Such an estimate cannot be derived from the noisy individual participant histograms. To achieve this, we computed the average proportion of saccades and the standard deviation for the first 12 bins (equivalent to the first 48 ms following display change) in the aggregate histogram. The point at which the proportion of saccades decreased to three standard deviations below this average was determined as the onset of inhibition in the aggregate flicker group. Following this procedure, latency to the onset of inhibition was found to be 64 ms, which is only about 6 ms shorter than L50% for this group. Given that L50% can be reliably determined from histograms for each participant and condition, we use it in the remainder of the experiments to allow comparisons of inhibition onset across conditions. The remarkably short latency of the inhibition caused by the flicker replicates the 60- to 70-ms onset values estimated by Reingold and Stampe (2000) and strongly suggests that saccadic inhibition is a low-level, reflexive, oculomotor effect that approximates the limits of neural delays in the visual and saccadic systems. However, before we can conclude that the reduction in saccadic frequency time-locked to the flicker represents saccadic inhibition, an alternative explanation should be considered. Specifically, it is possible that the flicker disrupts encoding or lexical processes
200
REINGOLD AND STAMPE
associated with reading, resulting in a delay of the subsequent saccade. This interpretation was referred to earlier as the processing disruption hypothesis (e.g., McConkie et al., 1992). Upon further reflection, such an explanation is inconsistent with our findings. This is the case because the flicker can occur at any point during the fixation, and the processes operating are assumed to vary throughout the fixation duration (for such an assumption, see Blanchard et al., 1984; McConkie et al., 1992). Consequently, the flicker may disrupt different processes at different points during the fixation, rendering it extremely unlikely that such disruptions will be translated into a reduction in saccadic frequency at a constant latency time-locked to the flicker. To examine this issue, we separately analyzed data from the aggregate flicker condition by the timing of the flicker within
fixations. Fixations with flicker delays from fixation onset ranging from 10 to 210 ms were divided into 10 bins of 20 ms in width. For each fixation, the average flicker delay in the bin was subtracted from the fixation duration to align the fixation distributions by the flicker delay and allow visual comparison of the effect of the flicker across bins. The proportions of fixations were then plotted as a function of fixation duration minus flicker delay and are presented in Figure 2, both separately (A) and overlaid (B). Given that these histograms were shifted to account for differences in flicker delay, it is possible to visually and numerically compare the effects of the flicker across histograms. An inspection of Figure 2A reveals that all bins with average flicker delay equal to or greater than 60 ms show a clear dip. Furthermore, the center of the bottom of the dip for all these bins
Figure 2. Analysis of the aggregate flicker condition in Experiment 1 by timing of the flicker from the onset of fixations (i.e., flicker delay). Fixations were assigned to 10 bins of 20 ms in width, with average flicker delays of 20, 40, 60, 80, 100, 120, 140, 160, 180, and 200 ms. Histograms of the fixation distributions in each bin were aligned on the time of flicker onset and are plotted separately (A) and overlaid (B) to show the striking similarity of the pattern and timing of saccadic inhibition across different flicker delays. B also includes a histogram of saccadic frequency following flicker onset (shown in bold). C: Fixation duration distributions corresponding to the no-flicker control condition (shown in bold) and the 20-, 40-, and 60-ms flicker delay bins, which illustrate that inhibition is also present for fixation distributions that do not show a clear dip.
SACCADIC INHIBITION
is virtually identical (90 or 94 ms; note that temporal resolution is 4 ms). In addition, as the flicker delay increases, the proportion of fixations prior to the center of the dip increases, and the proportion following the center of the dip decreases. This is the case because as flicker delay increased, there were more fixations that occurred prior to the onset of inhibition. That is, in gaze-contingent, fixeddelay flicker, despite the fact that the timing of inhibition is constant, the pattern observed is dependent on the precise range of affected fixation durations. Figure 2C helps clarify that despite the fact that no dips were observed for the 20- and 40-ms bins, saccadic inhibition still occurred with similar timing. This is shown by comparing the fixation duration histograms for these bins with the fixation duration distribution for the control condition, as well as with the fixation duration distribution for the 60-ms bin. As flicker delay increases, the distribution of fixation durations is progressively shifted forward in time (i.e., the average duration increases). For the 20-ms delay, and more so for the 40-ms delay, short duration fixations appear to be inhibited. However, for these bins, no short duration fixations appear to escape inhibition. In contrast, for the 60-ms bin, a clear dip is observed as some of the very short duration fixations were spared. In Figure 2B, the histograms for all bins, as well as the regular saccadic inhibition histogram (which includes all saccades regardless of whether they terminate a fixation that included a flicker), are overlaid. Figure 2B represents a powerful visual demonstration that saccadic inhibition is identical across all flicker delays, thus rejecting the processing disruption hypothesis (see also Reingold & Stampe, 2000). Taken together, the short inhibition onset latencies and the finding that inhibition is time-locked to the flicker onset and not to the beginning of fixation strongly support a low-level, reflexive, oculomotor effect rather than a higher level processing disruption effect.
Experiment 2 As discussed earlier, the processing disruption hypothesis attributes dips in the fixation duration distributions in gazecontingent masking experiments to the disruptive effect of masking on encoding processes that are time-locked to the beginning of fixation (e.g., McConkie et al., 1992). The results of Experiment 1 were used to argue against this processing disruption hypothesis. Experiment 2 examined whether saccadic inhibition is obtained even under conditions in which the text being processed is unaffected by the flicker. Two flicker conditions were used. In the central condition, the flicker occurred inside an 18° square centered on the point of gaze. In the peripheral condition the flicker covered the entire display area outside an 18° square centered on the point of gaze. These two conditions were equated in terms of the total screen area that flickered and, consequently, in the global luminance change produced. The conditions were also equated in terms of the contours created by the flicker (i.e., the edges of the square). Most important, the flicker in the peripheral condition occurred at a distance of approximately 20 character spaces to the left and right of the point of gaze and, consequently, was outside the perceptual span required to maintain normal reading speed and comprehension (see Rayner, 1998, for a review). Furthermore, the contours created by the flicker were far enough from the text being processed that low-level masking, such as metacontrast masking, would not be a factor (for a review, see Breitmeyer, 1984; Di
201
Lollo, Enns, & Rensink, 2000). Thus, in this condition, at any point in time, the text being processed was not affected by the flicker. If processing disruption due to masking was responsible for the decrease in saccadic frequency, then it would be expected that the peripheral condition would produce little or no saccadic inhibition.
Method A group of 10 participants who had not taken part in Experiment 1 was tested. All participants had normal or corrected-to-normal vision and were paid $10.00 CDN per hour for their participation. A black transient image (4.0 cd/m2) was used in Experiment 2. During the flicker, a gaze-contingent window was used to limit the transient image to a region either inside or outside an 18° square centered on the participant’s point of gaze, constituting the central and peripheral conditions, respectively, as shown in Figure 3A. As participants moved their eyes, their gaze position on the display was computed and used to set the region in which the transient image would be displayed during the next flicker. The average delay between an eye movement and the update of the gazecontingent window was 14 ms. A total of 50 pages of text were presented, with 25 pages randomly assigned to each condition. In Experiment 2 participants answered correctly 92.8% of the reading comprehension questions. All other aspects of the design, procedure, and data analysis were identical to those in Experiment 1.
Results and Discussion Figure 3B plots the normalized saccadic frequency histograms for both conditions. The corresponding values of the inhibition measures are shown in Table 1. As an inspection of Figure 3B reveals and data analysis confirms, the inhibition pattern was extremely similar across conditions. The estimates for the magnitude, peak area, and L50% were virtually identical (all ts ⬍ 1). Looking at Figure 3B, it appears that the central condition has a longer inhibition duration when compared with the peripheral condition. However, this difference was not statistically reliable, t(9) ⫽ 1.45, p ⫽ .182. Taken together, the results of Experiments 1 and 2 cannot be reconciled with the processing disruption hypothesis and provide strong support for the saccadic inhibition hypothesis.
Experiment 3 Several measures of the strength and latency of saccadic inhibition were introduced in Experiment 1 and were used to compare central and peripheral flicker in Experiment 2. In Experiment 3, we used these measures to test the effects of the saliency of the display change (henceforth, the visual event) on the saccadic inhibition it evokes. We predicted that the magnitude of the saccadic inhibition and the corresponding measure of recovery (i.e., the peak area) would be correlated with the saliency of the visual event. The visual event used in this study was a brief (33-ms) vertical displacement of the text, with its saliency set by the size of displacement. The flicker condition used in Experiment 1 was also included for the purpose of comparison.
Method A group of 10 participants who had not taken part in previous experiments was tested. All participants had normal or corrected-to-normal vision and were paid $10.00 CDN per hour for their participation.
REINGOLD AND STAMPE
202
Figure 3. Results of Experiment 2. A: Areas of the display that flickered (turned black for 33 ms) in the central flicker (top) and the peripheral flicker (bottom) conditions; the flickered region moved to stay centered on the point of gaze (shown as a cross for illustration purposes; not seen by participants). B: Normalized saccadic frequency following the flicker onset in the central flicker and the peripheral flicker conditions. Note the very similar dips across conditions.
The normal image for all four conditions consisted of pages of text as in Experiment 1 (average brightness ⫽ 40 cd/m2). In the displacement conditions, the transient image was created by a downward vertical shift of the normal image and was displayed for 33 ms, which resulted in an intermittent jitter of the entire display. Transient images were shifted down by 0.1°, 0.3°, or 0.6°, with the largest shift displacing the text by the height of a lowercase character a. The flicker condition used a black transient image, identical to that used in Experiment 1. Each condition was randomly assigned to 20 pages of text, for a total of 80 trials. In Experiment 3 participants answered correctly 89.1% of the reading comprehension questions. All other aspects of the design, procedure, and data analysis were identical to those in Experiment 1.
visual events, these measures may reflect different aspects of the saccadic inhibition. Specifically, the difference in inhibition onset across the flicker and 0.6° displacement conditions likely reflects differences in neural transmission delays in the visual system, with the luminance change in the flicker condition being processed faster than the display change in the displacement conditions, which did not involve a change in luminance (see Stampe & Reingold, 2002). However, when comparing visual events of low and high saliency, as in the 0.1° versus the 0.3° and the 0.6° displacement conditions, both slower onset and smaller magnitude may result when saliency decreases (i.e., the measures are associated).
Results and Discussion Figure 4 plots the normalized saccadic frequency histograms for all four conditions. The corresponding values of the inhibition measures are shown in Table 1. As predicted, the magnitude of inhibition increased for larger displacements (0.1° ⬍ 0.3° ⬍ 0.6°; all ts ⬎ 3.30, all ps ⬍ .01), with a corresponding increase in peak area (0.1° ⬍ 0.3° ⬍ 0.6°; all ts ⬎ 2.67, all ps ⬍ .05). The flicker and 0.6° displacement conditions did not significantly differ for either the magnitude of inhibition (t ⬍ 1) or the peak area, t(9) ⫽ 1.61, p ⫽ .14. The flicker condition produced larger magnitude and peak area than the 0.1° and 0.3° displacement conditions (all ts ⬎ 3.45, all ps ⬍ .01). L50% was analyzed to compare conditions in terms of inhibition onset. The flicker condition produced faster inhibition than any of the displacement conditions (all ts ⬎ 4.68, all ps ⬍ .001). Inhibition onset for the 0.1° displacement was slower than that for the 0.3° displacement, t(9) ⫽ 2.59, p ⬍ .05, and the 0.6° displacement, t(9) ⫽ 3.41, p ⬍ .01. The difference between the latter two conditions was marginal, t(9) ⫽ 1.99, p ⫽ .078. The faster inhibition in the flicker condition, compared with the 0.6° displacement condition, is interesting given that the inhibition produced in these conditions did not differ in magnitude. This dissociation indicates that at least when comparing inhibition caused by salient
Figure 4. Results of Experiment 3: Normalized saccadic frequencies following the display change in the 0.1°, 0.3°, and 0.6° displacement conditions and the flicker condition. The strength and timing of the dip caused by inhibition is clearly different in each condition, becoming weaker and more delayed as the saliency of the display change is decreased.
SACCADIC INHIBITION
Finally, the pattern for the duration measure was similar to that seen for the L50% measure. Specifically, the flicker condition produced longer inhibition duration as compared with all displacement conditions (all ts ⬎ 6.43, all ps ⬍ .001). Inhibition duration for the 0.1° displacement was shorter than that for the 0.3° displacement, t(9) ⫽ 2.89, p ⬍ .05, and the 0.6° displacement, t(9) ⫽ 2.49, p ⬍ .05. The difference between the latter two conditions was not significant (ts ⬍ 1). Again, it appears that the magnitude and duration measures may be dissociable when salient visual events are compared, but a visual event of low saliency may result in smaller magnitude, and slower, less sustained inhibition. In summary, the results of Experiment 3 clearly demonstrate an increase in inhibition and recovery strength as a function of an increase in the saliency of the visual event. The temporal characteristics of saccadic inhibition (i.e., onset and duration) may be determined by multiple factors, such as neural transmissions delays (see Stampe & Reingold, 2002), and also appear to be affected
203
when the visual event is of low saliency. Of course, it is possible that the neural transmission is slower for visual events of low saliency and, consequently, that the effect of saliency on the time course of saccadic inhibition is indirect.
Experiment 4 The goal of Experiment 4 was to investigate whether a perceptual enhancement in the direction of the next saccade occurs and whether it is influenced by high-level cognitive or attentional task demands. Toward this end, we used two flicker conditions. In both conditions, the flicker was displayed within the boundaries of a gaze-contingent, 10° square window that was displaced such that its nearest edge was 1° to the left or right of the point of gaze, as illustrated in Figure 5C. For forward saccades (i.e., left to right saccades in English) the flicker to the left is incongruent and the flicker to the right is congruent with the direction of the saccade.
Figure 5. Results of Experiment 4. Histograms of fixation durations (A) and saccadic amplitudes (B) are shown for the text (normal reading) and Z-string (mindless reading) conditions; each histogram includes data from both the congruent and incongruent conditions. C: Areas of the display that flickered (turned black for 33 ms) in the congruent flicker (top) and the incongruent flicker (bottom) conditions; the regions moved with the point of gaze (shown as a cross for illustration purposes; not seen by participants). D: Histogram of normalized saccadic frequency following the flicker onset in the congruent text, incongruent text, congruent Z-string, and incongruent Z-string conditions. Note the clear difference in the inhibition evoked in the congruent and incongruent text conditions, as compared with the much smaller difference between the congruent and incongruent Z-string conditions.
204
REINGOLD AND STAMPE
We predicted that a congruent flicker would be perceptually enhanced, whereas an incongruent flicker would be attenuated, leading to stronger inhibition in the congruent relative to the incongruent condition. When investigating whether higher cognitive or attentional processes modulate saccadic inhibition, the reading task must be compared with an oculomotor control condition that produces scanning saccades similar to those made during reading and involves text displays (i.e., similar low-level visual characteristics). Vitu et al. (1995) introduced an ideal task for this purpose. In their study, all letters in a text were replaced with Zs (henceforth, Z-strings), while punctuation and spacing were preserved. Participants were instructed to scan the text as if they were reading. The results indicated that the global characteristics of saccades were quite similar for the “mindless reading” of Z-string text and normal reading (however, for differences in saccade control between these tasks, see M. H. Fischer, 1999; Rayner & Fischer, 1996). Mindless reading represents an ideal control in the present context because although the properties of the saccades generated are similar, mindless reading shares none of the higher level lexical, semantic, or syntactic processes involved in reading (see M. H. Fischer, 1999, for a related rationale). Accordingly, in Experiment 4, congruent and incongruent flicker conditions were used with both a reading and a mindless reading task. A stronger congruency effect in the former, relative to the latter, task would support the conclusion that saccadic inhibition is modulated, at least in part, by higher level cognitive or attentional processes. It would also be consistent with the conclusion that a perceptual enhancement in the direction of the next saccade occurs and that it is greater in reading relative to mindless reading.
Method A group of 20 participants were tested. None of the participants had taken part in the previous experiments. All participants had normal or corrected-to-normal vision and were paid $10.00 CDN per hour for their participation. Twenty-five pages were used for each of the four conditions (text congruent, text incongruent, Z-string congruent, Z-string incongruent). To avoid undue disruption to reading, we made the presentation of the flicker consistently congruent or incongruent within a screen. Text screens and Z-string screens were alternated, and the Z-string pages were generated from a story different from that used to generate the text pages, to ensure that no pages were matched in layout. The assignment of screens to congruent and incongruent conditions was random for each participant. Participants were instructed to pretend that they were reading the Z-string screens. As in previous experiments, participants were instructed to ignore the flicker and to read the text for comprehension. The gaze-contingent window constrained the flicker to a 10° square area, which was displaced such that its nearest edge was 1° to the left or right of the point of gaze, as illustrated in Figure 5C. Note that for both left-sided and right-sided flickers, the region in which the flicker was displayed could extend beyond the area of text. The implementation of the gaze-contingent flicker display was the same as that in Experiment 2. In Experiment 4 participants answered correctly 96.6% of the reading comprehension questions. All other aspects of the design, procedure and analysis were identical to those in Experiment 1.
Results and Discussion Table 1 presents the means and standard errors for saccadic amplitude and fixation duration for all four experimental condi-
tions. For each of these dependent variables, a 2 ⫻ 2 withinparticipants analysis of variance (ANOVA), which crossed flicker congruency (congruent vs. incongruent) and text type (Z-string vs. normal), was performed. For both fixation duration and saccadic amplitude, the main effect of congruency and the interaction with text type were not significant (all Fs ⬍ 1). Figure 5 plots histograms of fixation durations (A) and saccadic amplitudes (B) for the text and Z-string conditions, collapsed across the congruent and incongruent flicker conditions. Consistent with previous findings (M. H. Fischer, 1999; Rayner & Fischer, 1996; Vitu et al., 1995), mindless reading produced longer saccades, F(1, 19) ⫽ 6.18, p ⬍ .05, and longer fixation durations, F(1, 19) ⫽ 45.40, p ⬍ .0001, relative to normal reading. Figure 5D presents the normalized saccadic frequency histograms for all four experimental conditions, and the derived saccadic inhibition measures are shown in Table 1. As in previous experiments, the magnitude, duration, and onset (L50%) of inhibition, as well as the peak area, were analyzed. For each of these dependent variables, a 2 ⫻ 2 within-participants ANOVA, which crossed flicker congruency (congruent vs. incongruent) and text type (Z-string vs. normal), was performed. The interaction between congruency and text type was significant for both the magnitude, F(1, 19) ⫽ 16.72, p ⬍ .001, and duration, F(1, 19) ⫽ 6.40, p ⬍ .05, measures. As can be seen in Figure 5, these interactions reflect a larger difference between congruent versus incongruent flicker in the text versus Z-string conditions. Thus, when the direction of the saccade was congruent with the flicker, a stronger and more sustained inhibition resulted, and this effect was differentially greater in reading than in mindless reading. It is important to note that the congruency effect on magnitude and duration was significant for both the text (both ts ⬎ 4.77, ps ⬍ .001) and Z-string (both ts ⬎ 2.76, ps ⬍ .05) conditions. None of the main effects or interactions were significant for the L50% measure (all Fs ⬍ 1), indicating that saccadic inhibition onset latency did not differ across conditions. This is in agreement with the results of Experiment 3, demonstrating a dissociation between the onset and magnitude measures and further establishing that they tap independent aspects of saccadic inhibition. The fact that inhibition onset did not differ across conditions may indicate that low-level characteristics of the visual display and flicker were closely matched across conditions, resulting in comparable neural transmission delays. Finally, larger peaks were obtained for the Z-string condition than for the text condition, F(1, 19) ⫽ 41.72, p ⬍ .001, and for the congruent condition than for the incongruent flicker condition, F(1, 19) ⫽ 27.16, p ⬍ .001. The Congruency ⫻ Text Type interaction was not significant (F ⬍ 1). In previous experiments, we interpreted the peak area as reflecting delayed or rescheduled saccades that were made during recovery from inhibition, as the effects on the peak appeared to mirror magnitude effects. In Experiment 4, the interaction obtained for the magnitude measure was not reproduced for the peak area measure. The difference between Experiment 4 and the previous experiments may be because whereas in Experiments 1–3 the flicker was symmetric about the point of gaze, in Experiment 4 the flicker was lateralized. This difference should be evaluated in future research.
SACCADIC INHIBITION
Experiment 5 The results of Experiment 4 demonstrated a stronger congruency effect in a reading task relative to mindless reading task. Experiment 5 was designed as a replication of the reading task in Experiment 4 with two notable modifications. First, in Experiment 4 the flicker was consistently congruent or incongruent for any given text screen (i.e., each screen involved either flickers to the left or flickers to the right of the current gaze position). This within-screen blocking procedure may have permitted participants to use a strategy or an attentional set to minimize the impact of the flicker. Such strategies were rendered ineffective in Experiment 5, in that it used a random sequence of left and right flickers within each screen. Second, Experiment 5 used over six times the number of text screens used in Experiment 4. This enhanced power permitted more fine-grained analyses aimed at determining the effects of saccadic amplitude and saccadic direction on the congruency effect demonstrated in Experiment 4. As pointed out by Rayner and Fischer (1996; see also M. H. Fischer, 1999), mindless reading produces longer saccades on average as compared with normal reading (see also Figure 5B). Consequently, by contrasting the congruency effect for long versus short saccades, Experiment 5 was designed in part to rule out the possibility that such differences in saccadic amplitude mediate the Congruency ⫻ Task interactions demonstrated in Experiment 4. The increase in power in Experiment 5 also allowed for comparing the congruency effect for forward versus regressive saccades (in English, the latter are right to left saccades that are not return sweeps; see Rayner, 1998, for a review). Experiment 5 demonstrated that comparable congruency effects were produced regardless of saccadic amplitude or direction.
Method A group of 10 participants were tested. None of the participants had taken part in previous experiments. All participants had normal or
205
corrected-to-normal vision and were paid $10.00 CDN per hour for their participation. A total of 320 pages of text were used in the experiment. Unlike in Experiment 4, within each screen a random sequence of left and right flickers was used. In Experiment 5, histograms were generated for congruent and incongruent regressive and forward saccades. In addition, separate histograms were also generated for congruent and incongruent forward saccades with amplitudes smaller than 2.3° or with amplitudes greater than 5.0°. In Experiment 5 participants answered correctly 91.1% of the reading comprehension questions. All other aspects of the design, procedure, and analysis were identical to those in Experiment 4.
Results and Discussion In general, as can be seen by an inspection of Figure 6 and Table 1, the congruency effect obtained in Experiment 4 was replicated for every type of saccade examined in Experiment 5. We first discuss the saccadic amplitude analysis and follow with an examination of the saccadic direction analysis. Saccadic amplitude. Forward saccades were divided into two bins of short (size ⬍ 2.3°) and long (size ⬎ 5.0°) amplitude saccades. As shown in Figure 6A, for each of these amplitude bins normalized saccadic frequency histograms were plotted for congruent and incongruent flickers. The derived saccadic inhibition measures are shown in Table 1. As in previous experiments the magnitude, duration, and onset (L50%) of inhibition, as well as the peak area, were analyzed. For each of these dependent variables a 2 ⫻ 2 within-participants ANOVA that crossed flicker congruency (congruent vs. incongruent) with saccadic amplitude (short vs. long) was performed. As in Experiment 4 a very strong congruency effect for the magnitude of inhibition regardless of saccadic amplitude was observed, F(1, 9) ⫽ 55.34, p ⬍ .0001. There was also a main effect of saccadic amplitude, reflecting stronger magnitude of inhibition for short versus long saccades, F(1, 9) ⫽ 13.71, p ⬍ .01. Most important, there was no interaction between congruency and sac-
Figure 6. Results of Experiment 5. A: Histogram of normalized saccadic frequency following the flicker onset, by congruency and saccadic amplitude. B: Histogram of normalized saccadic frequency following the flicker onset, by congruency and saccadic direction. Note that clear congruency effects were produced regardless of saccadic amplitude or direction.
206
REINGOLD AND STAMPE
cadic amplitude (F ⬍ 1), indicating that the congruency effect on the magnitude of inhibition was comparable across saccadic amplitudes. As in Experiment 4 the peak area measure did not perfectly mirror the pattern of results for the magnitude measure. Specifically, whereas the congruency effect on peak area was highly significant, F(1, 9) ⫽ 38.98, p ⬍ .0001, the main effect of saccadic amplitude was not significant (F ⬍ 1). In addition, the Congruency ⫻ Amplitude interaction was not significant (F ⬍ 1). For the duration measure the main effects of congruency and saccadic amplitude and their interaction were not significant (all Fs ⬍ 1). Thus, in this analysis the congruency effect obtained in Experiment 4 on duration was not replicated, although numerically there was a small difference as a function of congruency (see Table 1). This may reflect the fact that the loss of power associated with contrasting extreme amplitudes reduced the reliability of this measure. In fact, as is discussed later, when all forward saccades are collapsed the congruency effect on duration is significant. Finally, the analysis of L50% produced no effect of congruency (Fs ⬍ 1), also replicating Experiment 4. As well, the interaction between congruency and saccadic amplitude was not significant (Fs ⬍ 1). However, a very strong effect of saccadic amplitude was observed, F(1, 9) ⫽ 53.25, p ⬍ .0001, demonstrating that the onset of inhibition was approximately 12 ms faster for long versus short saccades. In summary, for all the saccadic inhibition measures congruency did not interact with saccadic amplitude. In replication of the results of Experiment 4 the congruency effects on inhibition magnitude and peak area were significant, and there was no difference in inhibition onset as a function of congruency. The congruency effect on duration obtained in Experiment 4 was not replicated, due to the loss of power associated with the exclusion of midamplitude saccades. Saccadic amplitude had a strong effect on inhibition onset and magnitude but no effect on duration. To examine whether the effects of saccadic amplitude on the magnitude and onset of saccadic inhibition represent a reliable oculomotor effect, we divided all saccades less than 7° in amplitude for the metaparticipant from Experiment 1 (i.e., the aggregate flicker condition) into seven bins of 1° in width. For each bin a normalized saccadic frequency histogram as a function of the latency from display change was generated, and magnitude and
L50% were derived. Figure 7 plots the number of saccades (A), magnitude (B), and L50% (C) as a function of saccadic amplitude. In replication of the pattern obtained in Experiment 5, an inspection of Figure 7 clearly reveals that when saccadic amplitude increases, both inhibition onset and magnitude decrease. The decrease in inhibition onset appears to be almost linear, whereas most of the decrease in magnitude occurs for saccades less than 5°. (Of course, the majority of forward saccades in reading are less than 5°; for an example, see Figure 1B.) Saccadic direction. In Figure 6B, separate normalized saccadic frequency histograms are plotted for forward and regressive saccades in each of the congruent and incongruent flicker conditions. The derived saccadic inhibition measures are shown in Table 1. As in previous experiments the magnitude, duration, and onset (L50%) of inhibition, as well as the peak area, were analyzed. For each of these dependent variables a 2 ⫻ 2 within-participants ANOVA that crossed flicker congruency (congruent vs. incongruent) and saccadic direction (forward vs. regressive) was performed. For the magnitude of inhibition the main effect of congruency was highly significant, F(1, 9) ⫽ 80.76, p ⬍ .0001, replicating Experiment 4. The interaction between congruency and saccadic direction was not significant (F ⬍ 1), indicating that this congruency effect was comparable across saccadic direction (forward and regressive). The main effect of saccadic direction on inhibition magnitude was significant, F(1, 9) ⫽ 7.24, p ⬍ .05, reflecting stronger magnitude of inhibition for forward saccades as compared with regressive saccades. This is an interesting finding that requires further study. It is important to note that on the basis of the saccadic amplitude analysis, as saccadic amplitude decreases, magnitude increases. Consequently, solely on the basis of saccadic amplitude, one would predict a pattern of results opposite to the one obtained (i.e., would predict stronger magnitude of inhibition for the smaller amplitude regressive saccades as compared with the larger amplitude forward saccades). The results of the analysis for peak area replicated the pattern for the magnitude measure. A very strong main effect of congruency was observed, F(1, 9) ⫽ 48.86, p ⬍ .0001. The main effect for saccadic direction was also significant, F(1, 9) ⫽ 9.33, p ⬍ .05. The Congruency ⫻ Saccadic Direction interaction was not significant (F ⬍ 1).
Figure 7. The number of saccades (A), inhibition magnitude (B), and inhibition onset (C) by saccadic amplitude in the aggregate flicker condition in Experiment 1. Note that as saccadic amplitude increases, inhibition magnitude and inhibition onset decrease. L(50%) ⫽ L50% ⫽ latency to 50% of maximum inhibition.
SACCADIC INHIBITION
For the duration measure replicating Experiment 4, a main effect of congruency was observed, F(1, 9) ⫽ 6.35, p ⬍ .05. The main effect for saccadic direction and the Congruency ⫻ Saccadic Direction interaction were not significant (Fs ⬍ 1). For the L50% measure, neither the main effect of congruency nor the Congruency ⫻ Saccadic Direction interaction were significant (both Fs ⬍ 1). The main effect of saccadic direction was significant for the L50% measure, F(1, 9) ⫽ 5.97, p ⬍ .05, reflecting faster inhibition onset for forward as compared with regressive saccades, which is likely due to the smaller amplitude of the latter relative to the former. (See the saccadic amplitude analysis for a similar result.)
General Discussion The present research provided important new insights into the nature of the saccadic inhibition effect. Taken together, the present experiments and the results reported by Reingold and Stampe (2000) conclusively rule out the processing disruption hypothesis as a viable explanation of this phenomenon. First, although perceptual and cognitive processing are assumed to vary throughout the fixation duration (see Blanchard et al., 1984; McConkie et al., 1992), the effect of the flicker was identical regardless of the delay of the flicker from fixation onset (Experiment 1). Second, a flicker presented outside the perceptual span that is required for normal reading produced a robust saccadic inhibition effect that was very similar to the effect produced by a flicker that completely obliterated the text being processed (Experiment 2). In addition, the very short onset latency of the saccadic inhibition effect (estimated at 64 ms in the flicker condition in Experiment 1) and the remarkable consistency of the onset of this effect across participants support the conclusion that saccadic inhibition is a fast reflex of the oculomotor system that acts in response to sudden changes in visual input to inhibit or delay the production of saccades (see also Reingold & Stampe, 2002). It was also found that the saccadic inhibition effect varied as a function of the saliency of the visual event (Experiment 3) and was not produced in response to abrupt but irrelevant auditory stimuli (i.e., beep vs. flicker in Experiment 1). Most important, saccadic inhibition was shown to be sensitive to higher level cognitive or attentional factors. Specifically, in the reading condition in Experiment 4, the magnitude of saccadic inhibition was 22.8% stronger when the flicker location was congruent with the direction of the saccade relative to when the flicker was incongruent with the direction of the saccade (i.e., the congruency effect). Experiment 5 further demonstrated the generality and robustness of this congruency effect. In contrast, the congruency effect in the mindless reading condition in Experiment 4 was only 6.5%. The fact that the congruency effect was significant across both of these conditions is consistent with the view that eye movements are preceded by an attentional shift to the saccadic target (e.g., Deubel & Schneider, 1996; Henderson, 1993; Hoffman, 1998; Hoffman & Subramaniam, 1995; Kowler et al., 1995; Rafal et al., 1989; Rayner et al., 1978; Remington, 1980; Schneider & Deubel, 1995; Shepherd et al., 1986; but see Stelmach et al., 1997). Furthermore, given that the characteristics of saccades are quite similar between the reading condition and the mindless reading condition (M. H. Fischer, 1999; Rayner & Fischer, 1996; Vitu et
207
al., 1995), the difference in the size of the congruency effect across these conditions provides strong support for a higher level, taskspecific influence, which is consistent with the prediction of the attentional guidance model of saccadic control in reading (e.g., Henderson, 1992; Henderson & Ferreira, 1990; Kennison & Clifton, 1995; Morrison, 1984; Pollatsek & Rayner, 1990; Rayner & Pollatsek, 1989; Reichle et al., 1998). That is, reading is expected to involve greater preallocation of attention in the direction of the next saccade when compared with mindless reading, potentially explaining the stronger congruency effect found in the former, relative to the latter, condition. Thus, despite the fact that the saccadic inhibition effect constitutes a reflexive oculomotor effect, its sensitivity to higher level cognitive or attentional factors demonstrates that this effect may provide a useful tool for the investigation of processes that underlie task performance. It is important to contrast the results of Experiments 4 and 5 with the findings reported in a recent study by M. H. Fischer (1999) that used a similar rationale and methodology to test predictions from the attentional guidance model. In this study, a probe was presented either congruent or incongruent with the direction of the next saccade. Participants had to detect this probe and respond to it as quickly as possible. It was predicted that if the probe was presented late in the fixation, shorter reaction times should be obtained for the congruent condition. If, however, the probe is presented early in the fixation, before attention is preallocated, no such difference should be seen. This pattern was not obtained for either a reading task or a mindless reading task but was successfully demonstrated for visual search with readinglike displays of horizontal letter strings. There are important differences between Fischer’s study and the present experiments that may help explain the divergence of the findings across studies. In Fischer’s study participants were required to respond to the probe (i.e., a dual task condition), whereas in the present experiments the flicker was irrelevant, and participants were instructed to ignore it. Also, the size of the probe was very small (an asterisk appearing above a letter), whereas the flicker covered a 10° square. Finally, in Fischer’s study the appearance of the probe was time-locked to the beginning of fixation (i.e., a gaze-contingent fixed delay), whereas in the present experiment the flicker occurred at a random delay from the beginning of fixation. Recently, Reingold and Stampe (2003) provided preliminary evidence suggesting that the critical difference may be the size of the transient stimulus. Specifically, Reingold and Stampe used the saccadic inhibition paradigm and varied the size of congruent and incongruent flickers (10° square vs. 1° square), replicating both the present results and the pattern obtained by Fischer. It is important to examine the present findings in the larger context of the behavioral and neurophysiological studies of the saccadic system. To date, the vast majority of such studies have focused on reflexive saccades triggered by the sudden appearance of the saccadic target (i.e., stimulus-elicited saccades). However, as demonstrated in the present study, the investigation of saccades produced during the performance of complex oculomotor scanning may be particularly informative for the exploration of high-level cognitive control pathways within the saccadic system (see Findlay & Walker, 1999; Yang & McConkie, 2001). In contrast to the reflexive nature of stimulus-elicited saccades, scanning saccades are considered voluntary but may be produced by highly efficient
208
REINGOLD AND STAMPE
and automated procedures (e.g., in the context of tasks such as reading and visual search). The present investigation builds on a growing emphasis on the importance of the study of voluntary saccades for the development of a general model of the saccadic system. As part of this trend, in addition to investigations in behavioral studies (e.g., Abrams, Oonk, & Pratt, 1998; Craig, Stelmach, & Tam, 1999; B. Fischer & Weber, 1992, 1997; Forbes & Klein, 1996; Goldring & Fischer, 1997; Hallett, 1978; Hallett & Adams, 1980; Reuter-Lorenz, Hughes, & Fendrich, 1991; Reuter-Lorenz, Oonk, Barnes, & Hughes, 1995; Smit, van Gisbergen, & Cools, 1987), voluntary saccades have also been investigated in studies of patients with frontal lobe lesions (e.g., Fukushima et al., 1988; Guitton, Buchtel, & Douglas, 1985; Lasker, Zee, Hain, & Folstein, 1987; PierrotDeseilligny, Rivaud, Gaymard, & Agid, 1991; see Everling & Fischer, 1998, for a review), in studies that recorded neural activity in awake, behaving monkeys (e.g., Amador, Schlag-Rey, & Schlag, 1996; Everling, Dorris, Klein, & Munoz, 1999; Everling & Munoz, 2000; Schlag-Rey, Amador, Sanchez, & Schlag, 1997), in positron emission tomography studies (O’Driscoll et al., 1995; Sweeney et al., 1996), and in studies using the recording of event-related potentials (Everling, Spantekow, Krappmann, & Flohr, 1998). Framed within this larger context, the present study demonstrated that a transient, task-irrelevant visual event (e.g., a flicker) inhibited or delayed the production of voluntary saccades and that the strength of this saccadic inhibition effect was modulated by higher level cognitive or attentional factors. Reingold and Stampe (2000, 2002) suggested that the neurophysiological locus of the saccadic inhibition effect might be related to inhibitory processes in the superior colliculus that were documented by Munoz and colleagues (e.g., Dorris & Munoz, 1998; Dorris, Pare´, & Munoz, 1997; Munoz & Istvan, 1998; Munoz & Wurtz, 1992, 1993a, 1993b, 1995a, 1995b; see Munoz, Dorris, Pare´, & Everling, 2000, for a review). The fine temporal resolution of the present paradigm has provided data that may be used to further validate this hypothesis. On the basis of the available neurophysiological evidence, the superior colliculus is the primary candidate for mediating the short saccadic inhibition onset latencies (60 –70 ms) documented in the present study and by Reingold and Stampe (2000, 2002). The superior colliculus, located in the midbrain, receives direct visual input from the retina in as little as 35 ms (Rizzolatti, Buchtel, Camarda, & Scandolara, 1980), and collicular output directly activates the saccade generator in the brain stem (for reviews, see Moschovakis, Scudder, & Highstein, 1996; Sparks & Hartwich-Young, 1989; Wurtz & Goldberg, 1989). The present study demonstrated that the magnitude of saccadic inhibition can be modulated by higher level cognitive or attentional factors and that this modulation is significant at latencies of less than 100 ms following flicker onset (based on LMAX in Experiments 4 and 5). This finding is consistent with the results reported by Goldberg and Wurtz (1972), that the activity of many neurons in the monkey’s superior colliculus is modulated by an instruction to saccade to or to ignore the receptive-field stimulus (for a related finding, see also Everling et al., 1999). The present findings are also consistent with the results of a study by Kustov and Robinson (1996), which demonstrated that covert attentional shifts are associated with eye-movement preparation in the superior colliculus in monkeys.
Finally, it is probably the case that saccadic inhibition reflects a more general perceptual bias in favor of immediate processing of sudden visual onsets, which is integral to the architecture of the visual and saccadic systems (e.g., Breitmeyer & Ganz, 1976; Jonides, 1981; Mu¨ller & Rabbitt, 1989; Nakayama & Mackeben, 1989; Remington, Johnston, & Yantis, 1992; Theeuwes, 1991, 1994; Yantis & Hillstrom, 1994; Yantis & Jonides, 1984). The visual system is unable to process new input during saccades resulting in a momentary disruption. Consequently, saccadic inhibition may serve to give the perceptual system time to process the arrival of changes in visual input by delaying the execution of saccades.
References Abrams, R. A., Oonk, H. M., & Pratt, J. (1998). Fixation point offsets facilitate endogenous saccades. Perception & Psychophysics, 60, 201– 208. Amador, N., Schlag-Rey, M., & Schlag, J. (1996). Supplementary eye field (SEF) neuronal activity in monkeys can predict the successful performance of antisaccade tasks. Society for Neuroscience Abstracts, 21, 1195. Blanchard, H. E., McConkie, G. W., Zola, D., & Wolverton, G. S. (1984). Time course of visual information utilization during fixations in reading. Journal of Experimental Psychology: Human Perception and Performance, 10, 75– 89. Breitmeyer, B. G. (1984). Visual masking: An integrative approach. New York: Oxford University Press. Breitmeyer, B. G., & Ganz, L. (1976). Implications of sustained and transient channels for theories of visual pattern masking, saccadic suppression, and information processing. Psychological Review, 83, 1–36. Brysbaert, M., & Vitu, F. (1998). Word skipping: Implications for theories of eye movement control in reading. In G. Underwood (Ed.), Eye guidance in reading and scene preparation (pp. 125–148). Oxford, England: Elsevier. Craig, G. L., Stelmach, L. B., & Tam, W. J. (1999). Control of reflexive and voluntary saccades in the gap effect. Perception & Psychophysics, 61, 935–942. Deubel, H., O’Regan, J. K., & Radach, R. (2000). Commentary on Section 2. Attention, information processing and eye movement control. In A. Kennedy, R. Radach, D. Heller, & J. Pynte (Eds.), Reading as a perceptual process (pp. 355–374). Amsterdam: Elsevier. Deubel, H., & Schneider, W. X. (1996). Saccade target selection and object recognition: Evidence for a common attentional mechanism. Vision Research, 6, 1827–1837. Di Lollo, V., Enns, J. T., & Rensink, R. A. (2000). Competition for consciousness among visual events: The psychophysics of reentrant visual processes. Journal of Experimental Psychology: General, 129, 481–507. Dorris, M. C., & Munoz, D. P. (1998). Saccadic probability influences motor preparation signals and time to saccadic initiation. Journal of Neuroscience, 18, 7015–7026. Dorris, M. C., Pare´, M., & Munoz, D. P. (1997). Neuronal activity in monkey superior colliculus related to the initiation of saccadic eye movements. Journal of Neuroscience, 17, 8566 – 8579. Downing, C. J. (1988). Expectancy and visual–spatial attention: Effects on perceptual quality. Journal of Experimental Psychology: Human Perception and Performance, 14, 188 –202. Everling, S., Dorris, M. C., Klein, R. M., & Munoz, D. P. (1999). Role of primate superior colliculus in preparation and execution of anti-saccades and pro-saccades. Journal of Neuroscience, 19, 2740 –2754. Everling, S., & Fischer, B. (1998). The antisaccade: A review of basic research and clinical studies. Neuropsychologia, 36, 885– 899.
SACCADIC INHIBITION Everling, S., & Munoz, D. P. (2000). Neuronal correlates for preparatory set associated with pro-saccades and anti-saccades in the primate frontal eye field. Journal of Neuroscience, 20, 387– 400. Everling, S., Spantekow, A., Krappmann, P., & Flohr, H. (1998). Eventrelated potentials associated with correct and incorrect responses in a cued antisaccade task. Experimental Brain Research, 118, 27–34. Findlay, J. M., & Walker, R. (1999). A model of saccade generation based on parallel processing and competitive inhibition. Behavioral and Brain Sciences, 22, 661– 675. Fischer, B., & Weber, H. (1992). Characteristics of “anti” saccades in man. Experimental Brain Research, 89, 415– 424. Fischer, B., & Weber, H. (1997). Effects of stimulus conditions on the performance of antisaccades in man. Experimental Brain Research, 116, 191–200. Fischer, M. H. (1999). An investigation of attention allocation during sequential eye movement tasks. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 52(A), 649 – 677. Forbes, K., & Klein, R. (1996). The magnitude of the fixation offset effect with endogenously and exogenously controlled saccades. Journal of Cognitive Neuroscience, 8, 344 –352. Fukushima, J., Fukushima, K., Chiba, T., Tanaka, S., Yamashita, I., & Kato, M. (1988). Disturbances of voluntary control of saccadic eye movements. Biological Psychiatry, 23, 670 – 677. Goldberg, M. E., & Wurtz, R. H. (1972). Activity of superior colliculus in behaving monkey: I. Visual receptive fields of single neurons. Journal of Neurophysiology, 35, 542–559. Goldring, J., & Fischer, B. (1997). Reaction times of vertical prosaccades and antisaccades in gap and overlap tasks. Experimental Brain Research, 113, 88 –103. Guitton, D., Buchtel, H. A., & Douglas, R. M. (1985). Frontal lobe lesions in man cause difficulties in suppressing reflexive glances and in generating goal-directed saccades. Experimental Brain Research, 58, 455– 472. Hallett, P. E. (1978). Primary and secondary saccades to goals defined by instructions. Vision Research, 18, 1279 –1296. Hallett, P. E., & Adams, W. D. (1980). The predictability of saccadic latency in a novel oculomotor task. Vision Research, 20, 329 –339. Henderson, J. M. (1992). Visual attention and eye movement control during reading and picture viewing. In K. Rayner (Ed.), Eye movements and visual cognition: Scene perception and reading (pp. 260 –283). New York: Springer-Verlag. Henderson, J. M. (1993). Visual attention and saccadic eye movements. In G. d’Ydewalle & J. Van Rensbergen (Eds.), Perception and cognition: Advances in eye movement research (pp. 37–50). Amsterdam: Elsevier. Henderson, J. M., & Ferreira, F. (1990). Effects of foveal processing difficulty on the perceptual span in reading: Implications for attention and eye movement control. Journal of Experimental Psychology: Human Perception and Performance, 16, 417– 429. Hodgson, T. L., & Mu¨ller, H. J. (1995). Evidence relating to premotor theories of visuospatial attention. In J. M. Findlay, R. Walker, & R. W. Kentridge (Eds.), Eye movement research: Mechanisms, processes and applications (pp. 305–316). Amsterdam: Elsevier. Hoffman, J. E. (1998). Visual attention and eye movements. In H. Pasher (Ed.), Attention (pp. 119 –153). East Sussex, England: Psychology Press. Hoffman, J. E., & Subramaniam, B. (1995). The role of visual attention in saccadic eye movements. Perception & Psychophysics, 57, 787–795. Johnston, C. W., & Diller, L. (1986). Exploratory eye movements and visual hemi-neglect. Journal of Clinical and Experimental Neuropsychology, 8, 93–101. Jonides, J. (1981). Voluntary vs. automatic control over the mind’s eye’s movement. In J. B. Long & A. D. Baddeley (Eds.), Attention and performance IX (pp. 187–203). Hillsdale, NJ: Erlbaum. Just, M. A., & Carpenter, P. A. (1980). A theory of reading: From eye fixations to comprehension. Psychological Review, 87, 329 –354.
209
Kennison, S. M., & Clifton, C. (1995). Determinants of parafoveal preview benefit in high and low working memory capacity readers: Implications for eye movement control. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 68 – 81. Klein, R. M. (1980). Does oculomotor readiness mediate cognitive control of visual attention? In R. Nickerson (Ed.), Attention and performance VIII (pp. 259 –276). Hillsdale, NJ: Erlbaum. Klein, R. M., Kingstone, A., & Pontefract, A. (1992). Orienting of visual attention. In K. Rayner (Ed.), Eye movements and visual cognition: Scene perception and reading (pp. 46 – 65). New York: Springer-Verlag. Klein, R. M., & Pontefract, A. (1994). Does oculomotor readiness mediate cognitive control of visual attention? Revisited! In C. Umilta & M. Moscovitch (Eds.), Attention and performance XV: Conscious and nonconscious information processing (pp. 333–350). Cambridge, MA: MIT Press. Kliegl, R., & Engbert, R. (2003). SWIFT explorations. In J. Hyo¨na¨, R. Radach, & H. Deubel (Eds.), The mind’s eye: Cognitive and applied aspects of eye movements (pp. 391– 411). Amsterdam: Elsevier Science. Kowler, E., Anderson, E., Dosher, B., & Blaser, E. (1995). The role of attention in the programming of saccades. Vision Research, 35, 1897– 1916. Kowler, E., & Anton, S. (1987). Reading twisted text: Implications for the role of saccades. Vision Research, 27, 45– 60. Kustov, A. A., & Robinson, D. L. (1996, November 7). Shared neural control of attentional shifts and eye movements. Nature, 384, 74 –77. Lasker, A. G., Zee, D. S., Hain, T. C., & Folstein, S. E. (1987). Saccades in Huntington’s disease: Initiation defects and distractibility. Neurology, 44, 2285–2289. McConkie, G. W., Kerr, P. W., Reddix, M. D., & Zola, D. (1988). Eye movement control during reading: I. The location of initial eye fixation in words. Vision Research, 28, 1107–1118. McConkie, G. W., Kerr, P. W., Reddix, M. D., Zola, D., & Jacobs, A. M. (1989). Eye movement control during reading: II. Frequency of refixating a word. Perception & Psychophysics, 46, 245–253. McConkie, G. W., Reddix, M. D., & Zola, D. (1992). Perception and cognition in reading: Where is the meeting point? In K. Rayner (Ed.), Eye movements and visual cognition: Scene perception and reading (pp. 293–303). New York: Springer-Verlag. McConkie, G. W., Underwood, N. R., Zola, D., & Wolverton, G. S. (1985). Some temporal characteristics of processing during reading. Journal of Experimental Psychology: Human Perception and Performance, 11, 168 –186. McConkie, G. W., & Yang, S.-N. (2003). How cognition affects eye movements during reading. In J. Hyo¨na¨, R. Radach, & H. Deubel (Eds.), The mind’s eye: Cognitive and applied aspects of eye movements (pp. 413– 427). Amsterdam: Elsevier Science. Miles, W. R. (1929). Ocular dominance demonstrated by unconscious sighting. Journal of Experimental Psychology, 12, 113–126. Mohler, C. W., & Wurtz, R. (1976). Organization of monkey superior colliculus: Intermediate layer cells discharging before eye movements. Journal of Neurophysiology, 39, 722–744. Morrison, R. E. (1984). Manipulation of stimulus onset delay in reading: Evidence for parallel programming of saccades. Journal of Experimental Psychology: Human Perception and Performance, 10, 667– 682. Moschovakis, A. K., Scudder, C. A., & Highstein, S. M. (1996). The microscopic anatomy and physiology of the mammalian saccadic system. Progress in Neurobiology, 50, 133–254. Mu¨ller, H. J., & Rabbitt, P. M. A. (1989). Reflexive and voluntary orienting of visual attention: Time course of activation and resistance to interruption. Journal of Experimental Psychology: Human Perception and Performance, 15, 315–330. Munoz, D. P., Dorris, M. C., Pare´, M., & Everling, S. (2000). On your mark, get set: Brainstem circuitry underlying saccadic initiation. Canadian Journal of Physiology and Pharmacology, 78, 934 –944.
210
REINGOLD AND STAMPE
Munoz, D. P., & Istvan, P. J. (1998). Lateral inhibitory interactions in the intermediate layers of the monkey superior colliculus. Journal of Neurophysiology, 79, 1193–1209. Munoz, D. P., & Wurtz, R. H. (1992). Role of the rostral superior colliculus in active visual fixation and execution of express saccades. Journal of Neurophysiology, 67, 1000 –1002. Munoz, D. P., & Wurtz, R. H. (1993a). Fixation cells in monkey superior colliculus. I. Characteristics of cell discharge. Journal of Neurophysiology, 70, 559 –575. Munoz, D. P., & Wurtz, R. H. (1993b). Fixation cells in monkey superior colliculus. II. Reversible activation and deactivation. Journal of Neurophysiology, 70, 576 –589. Munoz, D. P., & Wurtz, R. H. (1995a). Saccade-related activity in monkey superior colliculus. I. Characteristics of burst and buildup cells. Journal of Neurophysiology, 73, 2313–2333. Munoz, D. P., & Wurtz, R. H. (1995b). Saccade-related activity in monkey superior colliculus. II. Spread of activity during saccades. Journal of Neurophysiology, 73, 2334 –2348. Nakayama, K., & Mackeben, M. (1989). Sustained and transient components of focal visual attention. Vision Research, 29, 1631–1647. O’Driscoll, G. A., Alpert, N. M., Matthysse, S. W., Levy, D. L., Rauch, S. L., & Holzman, P. S. (1995). Functional neuroanatomy of antisaccade eye movements investigated with positron emission tomography. Proceedings of the National Academy of Sciences, USA, 92, 925–929. O’Regan, J. K. (1990). Eye movements and reading. In E. Kowler (Ed.), Eye movements and their role in visual and cognitive processes (pp. 395– 453). Amsterdam: Elsevier. O’Regan, J. K. (1992). Optimal viewing position in words and the strategytactics theory of eye movements in reading. In K. Rayner (Ed.), Eye movements and visual cognition: Scene perception and reading (pp. 333–354). New York: Springer-Verlag. Pierrot-Deseilligny, C. P., Rivaud, S., Gaymard, B., & Agid, Y. (1991). Cortical control of reflexive visually-guided saccades. Brain, 114, 1472– 1485. Pollatsek, A., & Rayner, D. (1990). Eye movements and lexical access in reading. In D. A. Balota, G. B. Flores d’Arcais, & K. Rayner (Eds.), Comprehension processes in reading (pp. 143–164). Hillsdale, NJ: Erlbaum. Pollatsek, A., Reichle, E. D., & Rayner, K. (2003). Modeling eye movements in reading: Extensions of the E-Z reader model. In J. Hyo¨na¨, R. Radach, & H. Deubel (Eds.), The mind’s eye: Cognitive and applied aspects of eye movements (pp. 361–390). Amsterdam: Elsevier Science. Posner, M. I. (1980). Orienting of attention. Quarterly Journal of Experimental Psychology, 32, 3–25. Posner, M. I., Nissen, M. J., & Ogden, W. C. (1978). Attended and unattended processing modes: The role of set for spatial location. In H. L. Pick & I. J. Saltzman (Eds.), Modes of perceiving and processing information (pp. 137–157). Hillsdale, NJ: Erlbaum. Rafal, R. D., Calabresi, P. A., Brennan, C. W., & Sciolto, T. K. (1989). Saccade preparation inhibits reorienting to recently attended locations. Journal of Experimental Psychology: Human Perception and Performance, 15, 673– 685. Rayner, K. (1998). Eye movements in reading and information processing: 20 years of research. Psychological Bulletin, 124, 372– 422. Rayner, K., & Fischer, M. H. (1996). Mindless reading revisited: Eye movements during reading and scanning are different. Perception & Psychophysics, 58, 734 –747. Rayner, K., McConkie, G. W., & Ehrlich, S. F. (1978). Eye movements and integrating information across fixations. Journal of Experimental Psychology: Human Perception and Performance, 4, 529 –544. Rayner, K., & Pollatsek, N. (1989). The psychology of reading. Englewood Cliffs, NJ: Prentice Hall. Reichle, E. D., Pollatsek, A., Fisher, D. L., & Rayner, K. (1998). Toward
a model of eye movement control in reading. Psychological Review, 105, 125–157. Reichle, E. D., Rayner, K., & Pollatsek, A. (in press). The E-Z Reader model of eye movement control in reading: Comparisons to other models. Behavioral and Brain Sciences. Reilly, R. G., & Radach, R. (2003). Foundations of an interactive activation model of eye movement control in reading. In J. Hyo¨na¨, R. Radach, & H. Deubel (Eds.), The mind’s eye: Cognitive and applied aspects of eye movements (pp. 429 – 455). Amsterdam: Elsevier Science. Reingold, E. M., & Stampe, D. M. (1997, September). Transient saccadic inhibition in reading. Paper presented at the 9th European Conference on Eye Movements, Ulm, Germany. Reingold, E. M., & Stampe, D. M. (2000). Saccadic inhibition and gaze contingent research paradigms. In A. Kennedy, R. Radach, D. Heller, & J. Pynte (Eds.), Reading as a perceptual process (pp. 119 –145). Elsevier: Amsterdam. Reingold, E. M., & Stampe D. M. (2002). Saccadic inhibition in voluntary and reflexive saccades. Journal of Cognitive Neuroscience, 14, 371–388. Reingold, E. M., & Stampe, D. M. (2003). Using the saccadic inhibition paradigm to investigate saccadic control in reading. In J. Hyo¨na¨, R. Radach, & H. Deubel (Eds.), The mind’s eye: Cognitive and applied aspects of eye movements (pp. 347–360). Amsterdam: Elsevier Science. Remington, R. W. (1980). Attention and saccadic eye movements. Journal of Experimental Psychology: Human Perception and Performance, 6, 726 –744. Remington, R. W., Johnston, J. C., & Yantis, S. (1992). Involuntary attentional capture by abrupt onsets. Perception & Psychophysics, 51, 279 –290. Reuter-Lorenz, P. A., Hughes, H. C., & Fendrich, R. (1991). The reduction of saccadic latency by prior offset of the fixation point: An analysis of the gap effect. Perception & Psychophysics, 49, 167–175. Reuter-Lorenz, P. A., Oonk, H. M., Barnes, L. L., & Hughes, H. C. (1995). Effects of warning signals and fixation point offsets on the latencies of pro- versus antisaccades: Implications for an interpretation of the gap effect. Experimental Brain Research, 103, 287–293. Rizzolatti, G., Buchtel, H. A., Camarda, R., & Scandolara, C. (1980). Neurons with complex visual properties in the superior colliculus of the macaque monkey. Experimental Brain Research, 38, 37– 42. Rizzolatti, G., Riggio, L., Dascola, I., & Umilta, C. (1987). Reorienting attention across the horizontal and vertical meridians: Evidence in favor of a premotor theory of attention. Neuropsychologia, 25, 31– 40. Rizzolatti, G., Riggio, L., & Sheliga, B. (1994). Space and selective attention. In C. Umilta & M. Moscovitch (Eds.), Attention and performance XV: Conscious and nonconscious information processing (pp. 231–265). Cambridge, MA: MIT Press. Schlag-Rey, M., Amador, N., Sanchez, H., & Schlag, J. (1997, November 27). Antisaccade performance predicted by neuronal activity in the supplementary eye field. Nature, 390, 398 – 401. Schneider, W. X., & Deubel, H. (1995). Visual attention and saccadic eye movements: Evidence for obligatory and selective spatial coupling. In J. M. Findlay, R. Walker, & R. W. Kentridge (Eds.), Eye movement research: Mechanisms, processes and applications (pp. 317–324). Amsterdam: Elsevier. Shepherd, M., Findlay, J. M., & Hockey, R. J. (1986). The relationship between eye movements and spatial attention. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 38(A), 475– 491. Smit, A. C., van Gisbergen, J. A., & Cools, A. R. A. (1987). A parametric analysis of human saccades in different experimental paradigms. Vision Research, 27, 1745–1762. Sparks, D. L., & Hartwich-Young, R. (1989). The deep layers of the superior colliculus. In R. H. Wurtz & M. E. Goldberg (Eds.), The neurobiology of saccadic eye movements (pp. 213–256). Amsterdam: Elsevier.
SACCADIC INHIBITION Stampe, D. M., & Reingold, E. M. (2002). Influence of stimulus characteristics on the latency of saccadic inhibition. In J. Hyo¨na¨, D. Munoz, W. Heide, & R. Radach (Eds.), The brain’s eye: Neurobiological and clinical aspects of oculomotor research (pp. 73– 87). Amsterdam: Elsevier Science. Stelmach, L. B., Campsall, J. M., & Herdman, C. M. (1997). Attention and ocular movements. Journal of Experimental Psychology: Human Perception and Performance, 23, 823– 844. Sweeney, J. A., Mintun, M. A., Kwee, S., Wiseman, M. B., Brown, D. L., Rosenberg, D. R., & Carl, J. R. (1996). Positron emission tomography study of voluntary saccadic eye movements and spatial working memory. Journal of Neurophysiology, 75, 454 – 468. Theeuwes, J. (1991). Exogenous and endogenous control of attention: The effects of visual onset and offsets. Perception & Psychophysics, 49, 83–90. Theeuwes, J. (1994). Endogenous and exogenous control of visual selection. Perception, 23, 429 – 440. van Diepen, P. M. J., de Graef, P., & d’Ydewalle, G. (1995). Chronometry of foveal information extraction during scene perception. In J. M. Findlay, R. Walker, & R. W. Kentridge (Eds.), Eye movement research: Mechanisms, processes and applications (pp. 349 –362). Amsterdam: Elsevier. Vitu, F., & O’Regan, J. K. (1995). A challenge to current theories of eye movements in reading. In J. M. Findlay, R. Walker, & R. W. Kentridge
211
(Eds.), Eye movement research: Mechanisms, processes and applications (pp. 381–393). Amsterdam: Elsevier. Vitu, F., O’Regan, J. K., Inhoff, A. W., & Topolski, R. (1995). Mindless reading: Eye movement characteristics are similar in scanning letter strings and reading text. Perception & Psychophysics, 57, 352–364. Walker, R., & Young, A. W. (1996). Object-based neglect: An investigation of the contributions of eye movements and perceptual completion. Cortex, 32, 279 –295. Wurtz, R. H., & Goldberg, M. E. (Eds.). (1989). The neurobiology of saccadic eye movements. Amsterdam: Elsevier. Wurtz, R., & Mohler, C. W. (1976). Organization of monkey superior colliculus: Enhanced visual response of superficial layer cells. Journal of Neurophysiology, 39, 745–765. Yang, S.-N., & McConkie, G. W. (2001). Eye movements during reading: A theory of saccade initiation times. Vision Research, 41, 3567–3585. Yantis, S., & Hillstrom, A. P. (1994). Stimulus driven attentional capture: Evidence from equiluminant visual objects. Journal of Experimental Psychology: Human Perception and Performance, 20, 95–107. Yantis, S., & Jonides, J. (1984). Abrupt visual onsets and selective attention: Evidence from visual search. Journal of Experimental Psychology: Human Perception and Performance, 10, 601– 621.
Received January 3, 2001 Revision received August 8, 2003 Accepted September 5, 2003 䡲
