Free
Article  |   October 2012
Contribution of nonattentive motion to object tracking
Author Affiliations
Journal of Vision October 2012, Vol.12, 28. doi:10.1167/12.11.28
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Hidetoshi Kanaya, Takao Sato; Contribution of nonattentive motion to object tracking. Journal of Vision 2012;12(11):28. doi: 10.1167/12.11.28.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract
Abstract
Abstract:

Abstract  Object tracking has been generally discussed in relation to attention, but it is quite possible that nonattentive low-level motion components are involved. To elucidate this issue, we examined temporal aspects of object tracking by using stimuli comprised of just a single attribute and those comprised of multiple attributes. High-level motion processes supposedly can process cross-attribute motion, while nonattentive low-level motion processes cannot handle such motion. In Experiment 1, we measured the upper temporal limits for within- and cross-attribute object tracking, using stimuli defined by several different attributes (luminance, motion, binocular disparity, flicker, and contrast). It was found that the temporal limits with within-attribute stimuli (4–5 Hz) were much higher than those with cross-attribute stimuli (2–3 Hz). These results suggest that mechanisms involved in within- and cross-attribute object tracking are partially different. We conducted two additional experiments to clarify the nature of this difference. In Experiment 2, we measured the temporal limits for classical apparent motion perception using the same stimulus combinations as for Experiment 1. The temporal limits with within- and cross-attribute stimuli were both between 4 and 5 Hz. These values corresponded to those of within-attribute object tracking but were faster than those of cross-attribute object tracking. In Experiment 3, we measured the temporal limit for voluntary shifts of attention that did not involve motion. Temporal limits quite similar to those for cross-attribute object tracking (2–3 Hz) were obtained. These results suggest that nonattentive motion mechanisms are involved in within-attribute object tracking, whereas attention-based mechanisms mediate cross-attribute object tracking.

Introduction
Verstraten, Cavanagh, and Labianca (2000) reported that the upper temporal frequency limit of “attentive tracking” using ambiguous apparent motion display was 4–8 Hz. They argued that object tracking was mediated by a higher visual process, instead of a first-order motion mechanism (e.g., Adelson & Bergen, 1985), since this limit of object tracking is much lower than that of first-order motion detection (Burr & Ross, 1982; Lu & Sperling, 1995b). 
Verstraten et al. (2000) and Benjamins, Hooge, van der Smagt, and Verstraten (2007) hypothesized that object tracking is realized by actively shifting visual attention from one object to another, and claimed that the 4- to 8-Hz limit they found is the upper temporal limit of this attentional shift. In contrast, Horowitz, Holcombe, Wolfe, Arsenio, and DiMase (2004), using tasks similar to Verstraten et al. (2000), found that the shift of voluntary attention between objects is quite slow (300–500 ms in stimulus onset asynchrony [SOA]), and suggested that object tracking was achieved not by attention-based processes but by a preattentive “object continuity process” such as indexing to target objects (see Pylyshyn, 1989; Pylyshyn & Storm, 1988). 
There is a clear contradiction. Verstraten and his colleagues (Benjamins et al., 2007; Verstraten et al., 2000) argued that object tracking is based on a high-level, feature-based process called attention-based motion process that selects and matches objects by using attention (Cavanagh, 1991, 1992). On the other hand, Horowitz et al. (2004) argued for a preattentive process. However, the nature of preattentive process is not clearly discussed, although they refer to a “object continuity” based on FINST theory (Pylyshyn, 1989; Pylyshyn & Storm, 1988). It is possible that this preattentive process is, at least partially, mediated by ordinary (first- or second-order) motion mechanisms. 
Apparent motion stimuli composed of frames each defined by different attributes such as motion, texture, binocular disparity, or contrast are not detected by relatively low-level first-order or second-order motion mechanisms that detect motion from simple stimuli defined by a single attribute. It is generally assumed that attention is involved in detecting such cross-attribute motions. For example, Lu and Sperling (1995a, 2001) have proposed a mechanism in which each cross-attribute stimulus is allocated on a salience map by selecting a salient feature by using voluntary attention, and then motion is computed on the salience map. Verstraten et al. (2000) mentioned the possibility that attention-based processing such as a feature salience system is involved in object tracking. However, this conjecture is not supported by their own results, since the stimuli they used only involve luminance, which is supposedly processed by a first-order mechanism. Thus, it was not clear from their results whether attention-based, attribute-independent mechanisms were involved. First-order mechanisms cannot detect cross-attribute motion, but the existence of low-level, attribute-specific second-order mechanisms is not really established. However, several reports on second-order reversed-phi phenomenon (Lu & Sperling, 1999; Maruya, Mugishima, & Sato, 2003; Mather & Murdoch, 1999) suggest the existence of a low-level second-order mechanism, and such low-level mechanisms could be attribute specific. 
The main objective of the present study is to clarify the involvement of two types of motion processing, relatively simple low-level and attention-based higher motion processing in object tracking. To this end, in Experiment 1, we examined the temporal limits for object tracking by using stimuli defined by several different visual attributes. The experiment was conducted by using both within- and cross-attribute object-tracking stimuli, that is, both single attribute stimuli, or stimuli with different attributes. If object tracking is mediated by a higher, attention-based mechanism, the temporal limits of object tracking should be about equal to those obtained by Verstraten et al. (2000) regardless of the stimulus types. However, if object tracking with within-attribute stimuli can be mediated by relatively lower motion processing, a dissociation of results between within- and cross-attribute stimuli should appear. In this case, the temporal limits would be higher, and about equal to the results of Verstraten et al. (2000) for within-attribute stimuli, and lower for cross-attribute stimuli. In two additional experiments, we measured temporal limits for simple, classical apparent motion perception using the same stimulus combinations as for Experiment 1 (Experiment 2), and for pure attentional shift that did not involve motion components (Experiment 3), and compared the results to those from Experiment 1 to clarify the contribution of attentional and motion processing to object tracking. 
Experiment 1
We measured and compared upper temporal limits for within- and cross-attribute object tracking by using a stimulus display composed of several different attributes. The stimulus display was similar to those used by Verstraten et al. (2000, experiment 1). We used luminance as a first-order attribute, and motion, binocular disparity, flicker, and contrast as second-order attributes. Motion was generated between stimuli defined either by the same attribute (within-attribute condition) or by luminance and another second-order stimulus (cross-attribute condition; Figure 1). 
Figure 1
 
Examples of stimulus configuration and presentation in Experiment 1. (A) Within-attribute (luminance) stimuli. Two arrays of four rectangular objects defined by luminance were alternately presented. (B) Within-attribute (second-order) stimuli. Two arrays of four objects defined by the same second-order attribute (motion, binocular disparity, flicker, or contrast) were alternately presented. Stimulus with motion-defined objects is shown as an example. White allows indicate motion directions of local dots in objects and the background. Yellow lines indicate borders surrounding objects. (C) Cross-attribute stimuli. Two arrays defined by luminance and one of the second-order attributes were alternately presented.
Figure 1
 
Examples of stimulus configuration and presentation in Experiment 1. (A) Within-attribute (luminance) stimuli. Two arrays of four rectangular objects defined by luminance were alternately presented. (B) Within-attribute (second-order) stimuli. Two arrays of four objects defined by the same second-order attribute (motion, binocular disparity, flicker, or contrast) were alternately presented. Stimulus with motion-defined objects is shown as an example. White allows indicate motion directions of local dots in objects and the background. Yellow lines indicate borders surrounding objects. (C) Cross-attribute stimuli. Two arrays defined by luminance and one of the second-order attributes were alternately presented.
Methods
Observers
Six undergraduate students participated in this experiment. They all had normal or corrected to normal vision and normal binocular vision. They were all well-trained for this type of experiment but unaware of its purpose. 
Apparatus
Stimuli in this experiment were generated by an Apple PowerMac G4 computer (Apple Inc., Cupertino, CA), and presented on a 17-inch CRT monitor (FlexScan T561, EIZO NANAO CORPORATION, Ishikawa, Japan) with a resolution of 1024 × 768 pixels and a refresh rate of 100 Hz. The viewing distance was 57 cm, and the distance was maintained by using a chinrest. Each pixel subtended approximately 1.8 min. A mirror stereoscope was used in binocular disparity conditions. 
Stimuli
The stimulus was generated using MATLAB 5.2.1 and Psychophysics Toolbox extensions (Brainard, 1997; Pelli, 1997) and presented on a CRT screen. Figure 2 shows the stimulus and procedures for this experiment. The stimulus was presented within a field subtending 10.5° (horizontal) × 11.0° (vertical) at the center of the CRT screen. There was a black fixation dot with a diameter of 15 min at the center of the area. Two arrays of four rectangular objects were sequentially presented to generate a circular motion. All objects were placed on a circle whose center was the fixation point. The distance from fixation point to the center of each object was 3.5°. The second array was generated by rotating the first one by 45° while keeping the orientation of objects upright. The two arrays were constantly alternated with no interstimulus interval (ISI) during each stimulus presentation. SOA of the frames varied depending on alternation rate (temporal frequency), which was varied in five steps. 
Figure 2
 
Procedure for object-tracking task in Experiment 1. At the start of a trial, a circular array of eight rectangles appeared and a red rectangular marker was presented at the center of one of the eight objects to designate the object to track (target). Then, the display was switched to a motion display consisted of four rectangles and the marker started to rotate in either clockwise or counterclockwise direction. The marker stayed on the target, i.e., moved together with the target for two complete cycles, or two complete circles, and it disappeared. After the marker disappeared, observers were asked to track the target for 1.8 s. Then, the alternation of apparent motion frames stopped, and a blue rectangular marker appeared on one of eight objects.
Figure 2
 
Procedure for object-tracking task in Experiment 1. At the start of a trial, a circular array of eight rectangles appeared and a red rectangular marker was presented at the center of one of the eight objects to designate the object to track (target). Then, the display was switched to a motion display consisted of four rectangles and the marker started to rotate in either clockwise or counterclockwise direction. The marker stayed on the target, i.e., moved together with the target for two complete cycles, or two complete circles, and it disappeared. After the marker disappeared, observers were asked to track the target for 1.8 s. Then, the alternation of apparent motion frames stopped, and a blue rectangular marker appeared on one of eight objects.
In addition to luminance-defined (first-order) motion, four different types of second-order motion were used in this experiment: motion, binocular disparity, flicker, and contrast. Object and background areas consisted of a dark/bright random-dot pattern with 30% dot density except for contrast-defined stimuli. A 50% dot density was used for contrast-defined stimuli and the luminance-defined stimuli that were paired with contrast-defined stimuli. The visibility of the second-order stimulus was found higher for 30% dot density than for 50% density in pilot experiments. Therefore we adopted 30% for motion-, disparity-, and flicker-defined stimuli. However, for contrast-defined stimuli, the density was set to 50% so that the contrast could be defined by the same amount of increment and decrement of luminance for darker and lighter dots. Each dot consisted of a pixel of the display which subtended 1.8 × 1.8 min. The luminance values for dark and bright dots, except for the luminance stimulus, were 0 and 38.6 cd/m2. Therefore, the dot contrast was 1.0 and the field mean luminance was 12.1 cd/m2
The mean luminance for luminance objects was raised relative to the background by adding a certain value to the original values for both dark and bright dots. This bias value was adjusted for each observer and temporal frequency condition so as to equalize visibilities of luminance-defined (first-order) motion to each second-order motion, using a method described in Procedure. 
The dot density of contrast-defined stimuli was 50%. Object areas had higher contrast (0.7) than the background (0.3). The mean luminance was 23.2 cd/m2. The random-dot patterns for both object and background areas were refreshed every 20 ms. 
The dots within the object for motion-defined stimuli moved in either an upward or downward direction, whereas the dots in the background moved in the opposite direction. The motion direction was fixed within a trial. The motion was generated by shifting each dot by four dot-units every 20 ms. The speed of local dot motion was approximately 360 min/s. For disparity-defined stimuli, two stimulus fields were presented on a CRT screen side by side, and observers viewed them through a mirror stereoscope. The stimulus consisted of a pair of dynamic random-dot stereograms with a refresh rate of 20 ms/frame. Object disparity was 10.8 min and the objects appeared floating against the background. For flicker-defined stimuli, the dots in the object area were refreshed every 20 ms, whereas those in background stayed still for the duration of each object and were refreshed when the alternation of apparent motion occurred. 
Within-attribute apparent motion was generated by alternately presenting two arrays defined by a single attribute with no ISI. The alternation rate was varied in five steps between 2.78 and 5.00 Hz. Cross-attribute apparent motion was generated by alternately presenting two arrays defined by luminance and by another attribute with no ISI. The alternation rate was varied in five steps between 1.67 and 3.57 Hz. 
Procedure
For the object-tracking task, observers viewed one of the object-tracking stimuli while fixating on a fixation point in a dark room. Eye movement was not monitored. At the start of a trial, a circular array of eight rectangles appeared and a red rectangular marker was presented at the center of one of the eight objects to designate the object to track (target). Then, the display was switched to a motion display consisting of four rectangles with ambiguous motion. At the same time, the red marker appeared on the designated target and started to rotate in either a clockwise or counterclockwise direction. It disappeared after making two complete circles. After the marker disappeared, observers were asked to track the target for 1.8 s. Then, the alternation of motion frames stopped, and all eight objects were presented with a blue rectangular marker (probe) on one of them. Observers judged whether the object on which the probe appeared was the tracked target or not with a 2AFC method. In half of the trials, the probe appeared at the correct position where the target was supposed to be, and in the other half, the probe appeared one position before or after the correct one. Thus, the chance level of this task was 50%. 
Temporal frequency of apparent motion was varied in five steps (2.78, 3.13, 3.57, 4.17, 5.00 Hz) for within-attribute conditions and in five steps (1.67, 2.27, 2.78, 3.13, 3.57 Hz) for cross-attribute conditions. In each session, the five temporal frequency conditions were presented 24 times in a randomized order. The attribute condition was fixed within a session. 
For the visibility-matching task, to avoid any artifact resulting from the visibility difference between the luminance-defined and second-order stimuli in the cross-attribute experiments, the visibility for each first-order and second-order stimulus was equalized individually for each observer for each temporal frequency condition using the same stimulus as for the main experiment. In this procedure, luminance-defined motion and one of the cross-attribute motion displays were presented one after another, and observers were asked to match subjective visibility by adjusting the luminance bias for the luminance-defined stimulus. Each unit presentation consisted of eight frames that corresponded to one whole virtual circle. The two types of motion stimuli were alternated until the observer was satisfied. Temporal frequency of apparent motion was varied in seven steps (1.67, 2.27, 2.78, 3.13, 3.57, 4.17, 5.00 Hz). Six trials were conducted in a randomized order. The mean luminance values, calculated for each observer for each temporal frequency, were used for cross-attribute conditions and within-luminance conditions in the main experiment. 
The method we employed to match the visibility of first- and second-order stimulus, a direct matching method, may sound inaccurate compared to more quantitative methods such as a method using a constant increment above detection threshold. However, the major concern we had for this experiment is to avoid underestimation of temporal limits caused by lower visibility for second-order stimuli. From this concern we first set the visibility of second-order stimuli as high as possible and then asked the observers to set the luminance value of first-order stimuli to equalize the visibility. 
Results
For each observer, the percentage of correct responses for the object-tracking task was calculated for each temporal frequency and attribute condition. In Figure 3 (left column), the mean percent correct from the six observers for different within- and cross-attribute conditions was plotted as a function of temporal frequency. In general, the performance decreased as a function of temporal frequency. It should be noted that the performance of cross-attribute conditions reached the chance level around 3 Hz. We defined the upper temporal frequency limit of object tracking as the frequency that corresponds to a 75% correct rate. The limits were calculated by fitting a logistic regression curve on individual tracking performance. Figure 3 (right column) shows the mean upper temporal frequency limits from the six observers for each attribute condition. For within-attribute conditions, the average values ranged from 4 to 5 Hz regardless of attribute. In contrast, in cross-attribute conditions, the average upper temporal limits ranged from 2 to 3 Hz irrespective of attribute paired to luminance. 
Figure 3
 
Results of Experiment 1 (N = 6). Left column: Mean percentage correct for object-tracking task plotted as functions of temporal frequency for (A) motion-defined, (B) disparity-defined, (C) flicker-defined, and (D) contrast-defined objects. Error bars indicate ±1 SE. In each graph, filled rectangles (▪) and open circles (○) show results for within- and cross-attribute stimuli respectively, and filled triangles (▴) show results for corresponding luminance (within) stimuli. The horizontal dotted lines indicate the chance level. Right column: Mean upper temporal limits for six observers. The types of stimuli are as for the left column. In each graph, white and gray bars show results for within- and cross-attribute stimuli respectively. Error bars indicate ±1 SE.
Figure 3
 
Results of Experiment 1 (N = 6). Left column: Mean percentage correct for object-tracking task plotted as functions of temporal frequency for (A) motion-defined, (B) disparity-defined, (C) flicker-defined, and (D) contrast-defined objects. Error bars indicate ±1 SE. In each graph, filled rectangles (▪) and open circles (○) show results for within- and cross-attribute stimuli respectively, and filled triangles (▴) show results for corresponding luminance (within) stimuli. The horizontal dotted lines indicate the chance level. Right column: Mean upper temporal limits for six observers. The types of stimuli are as for the left column. In each graph, white and gray bars show results for within- and cross-attribute stimuli respectively. Error bars indicate ±1 SE.
One-way ANOVA indicated that the differences between attribute conditions were significant in all second-order attributes—motion: F(2,10) = 10.29, p < 0.01; binocular disparity: F(2,10) = 14.53, p < 0.01; flicker: F(2,10) = 13.23, p < 0.01; contrast: F(2,10) = 10.93, p < 0.01. Tukey LSD test indicated that the differences between cross-attribute conditions and within-attribute conditions were significant in all second-order attributes (p < 0.05). 
Discussion
These results indicate that there is a clear difference between the upper temporal limits for object tracking with within-attribute stimuli and with cross-attribute stimuli. The present results for within-attribute stimuli (4–5 Hz) overlap with the 4- to 8-Hz range reported by Verstraten et al. (2000). In contrast, the present results with cross-attribute stimuli (2–3 Hz) are much lower than the temporal limits reported by Verstraten et al. (2000). These results, therefore, suggest that the limit reported by previous studies (Verstraten et al., 2000) do not apply to trackings of cross-attribute stimuli. They apply only to within-attribute stimuli. 
Verstraten et al. (2000) and Benjamins et al. (2007) suggest that the 4- to 8-Hz limit for object tracking reflects the temporal limit of the shift of visual attention between objects, but it is difficult to explain the present results based on their conjecture. None of the present results are related to the feature salience system (Lu & Sperling, 1995a, 2001), since motion detection based on attention shift or feature salience is supposedly independent of attribute. If object tracking as a whole is based on such attribute-free mechanisms, there should not be any difference between temporal limits for within- and cross-attribute stimuli. 
If object tracking with cross-attribute stimuli is mediated by such attribute-free mechanisms, then the present results with within-attribute stimuli should be mediated by some other mechanism and the temporal limits found in the present results reflect the performance of such a mechanism. It is quite plausible that object tracking with within-attribute stimuli is supported by attribute-dependent motion mechanisms. As mentioned in the Introduction, the temporal limit for first-order motion detection is much higher compared to object tracking (Burr & Ross, 1982; Lu & Sperling, 1995b). In addition, first-order motion mechanisms cannot deal with second-order stimuli to begin with. On the other hand, the temporal limit for second-order motion is supposedly lower than that of first-order motion. Lu and Sperling (1995b) reported temporal limits of 3–4 Hz for apparent motion of motion- and disparity-defined gratings. However, very few studies to date have tried to obtain the temporal limit for second-order motion (Hutchinson & Ledgeway, 2006; Lu & Sperling, 1995b). Therefore, in the next experiment, we examined the temporal characteristics of apparent motion per se for all the stimulus combinations used in Experiment 1
Experiment 2
We measured the upper temporal limits for classical apparent motion perception between two isolated rectangular objects using all the attribute combinations deployed in Experiment 1. To compare these results with the results for object tracking obtained in the previous experiment, stimulus parameters were kept constant as much as possible. 
Methods
Methods are generally the same as for Experiment 1 except for the differences regarding stimulus and procedure described below. 
Stimuli
A stimulus field subtending 15.0 (horizontal) × 5.0 (vertical) deg was presented above the fixation point (Figure 4). The distance between the centers of the stimulus field and fixation marker was 3.5°. This distance was the same as the distance from the fixation point to the center of rectangular objects in Experiment 1. The stimulus field was filled with a random-dot pattern in the same way as for Experiment 1. Two-frame apparent motion between two objects was presented in the stimulus field. The object size was 1.0° × 1.0°, and the center-to-center distance between the objects was 2.68°. This separation was the same as the distance between two adjacent objects in Experiment 1
Figure 4
 
Stimulus configuration in Experiment 2. Two-frame apparent motions between isolated objects were presented in the stimulus area 3.5° above the fixation point. The object size was 1.0° × 1.0°, and the spatial separation between the objects was 2.68°. The objects were defined by either luminance, motion, binocular disparity, flicker, or contrast as in Experiment 1. Within-attribute apparent motions were generated by alternately presenting two objects defined by the same attribute for six times with no ISI. Cross-attribute apparent motions were generated by pairing an object defined by luminance and the other defined by one of the second-order attributes and presenting them in the same way.
Figure 4
 
Stimulus configuration in Experiment 2. Two-frame apparent motions between isolated objects were presented in the stimulus area 3.5° above the fixation point. The object size was 1.0° × 1.0°, and the spatial separation between the objects was 2.68°. The objects were defined by either luminance, motion, binocular disparity, flicker, or contrast as in Experiment 1. Within-attribute apparent motions were generated by alternately presenting two objects defined by the same attribute for six times with no ISI. Cross-attribute apparent motions were generated by pairing an object defined by luminance and the other defined by one of the second-order attributes and presenting them in the same way.
As in Experiment 1, the object was defined by one of the following: luminance, motion, binocular disparity, flicker, or contrast. The way each type of object was generated was the same as for Experiment 1. Within-attribute apparent motions were generated by alternately presenting two objects defined by the same attribute six times with no ISI. Cross-attribute apparent motions were generated by pairing an object defined by luminance and another defined by one of the second-order attributes, and presenting them in the same way. Temporal frequency was varied in nine steps (1.67, 2.27, 2.78, 3.13, 3.57, 4.17, 5.00, 6.25, 8.33 Hz). 
Procedure
Observers viewed apparent motion stimuli while fixating on the marker and reported whether they perceived motion or not by pressing one of the two keys. In each session, the attribute condition was fixed, and each of the nine temporal frequency values was repeated 20 times in a randomized order. 
Results and discussion
For each observer, the perception rate of apparent motion was calculated for each temporal frequency and each attribute condition. Figure 5 shows the mean perception rate from six observers. The perception rate generally decreased as temporal frequency increased. It was very high (above 85%) up to 3–5 Hz. Then it gradually decreased and reached almost zero beyond 8 Hz. The perception rate of 50% was obtained at around 5 Hz. This tendency was the same for both within- and cross-attribute conditions regardless of the second-order attribute. All the stimulus combinations produced quite similar performance patterns. Although within-luminance motion yielded the best performance, the difference between this and other within-second-order stimuli, or even between it and cross-attribute stimuli was very small. The cross-attribute stimuli with contrast-defined objects produced exceptionally low performance, although the performance for within-contrast-defined stimuli was comparable to the other within-second-order stimuli. 
Figure 5
 
Results of Experiment 2. Mean perception rates of perceiving motion are plotted as functions of temporal frequency (N = 6) for (A) motion-defined, (B) disparity-defined, (C) flicker-defined, and (D) contrast-defined stimuli. In each graph, filled rectangles (▪) and open circles (○) show results for within- and cross-attribute stimuli, and filled triangles (▴) show results for corresponding luminance (within) stimuli. Error bars indicate ±1 SE.
Figure 5
 
Results of Experiment 2. Mean perception rates of perceiving motion are plotted as functions of temporal frequency (N = 6) for (A) motion-defined, (B) disparity-defined, (C) flicker-defined, and (D) contrast-defined stimuli. In each graph, filled rectangles (▪) and open circles (○) show results for within- and cross-attribute stimuli, and filled triangles (▴) show results for corresponding luminance (within) stimuli. Error bars indicate ±1 SE.
The functions plotted in Figure 5 fit well into the measured temporal frequency range (1.67–8.33 Hz). However, it could be a result of the tendency that observers were homing in on the implicit central value within each range. Such tendency has been reported by Morgan, Watamaniuk, and McKee (2000) for a vernier acuity task. To evaluate such possibility, we repeated the same measurements using three different temporal frequency ranges: 1.67–5.00 Hz, 1.67–8.33 Hz, and 3.57–8.33 Hz. Three observers including one of the authors (HK) participated in this experiment. The stimulus and procedure were the same as for Experiment 2. Figure 6 shows the results from the three temporal frequency ranges in the same way as for Figure 5, and 50% points from these conditions were quite similar to those obtained from Experiment 2. Thus we conclude that the results of Experiment 2 were not affected by the centering tendency. 
Figure 6
 
Results of additional experiment. Mean perception rates of perceiving motion are plotted as functions of temporal frequency (N = 3). In the graph, rectangles (▪), triangles (▴), and circles (○) show the results for 1.67–5.00 Hz, 1.67–8.33 Hz, and 3.57–8.33 Hz range, respectively. Error bars indicate ±1 SE.
Figure 6
 
Results of additional experiment. Mean perception rates of perceiving motion are plotted as functions of temporal frequency (N = 3). In the graph, rectangles (▪), triangles (▴), and circles (○) show the results for 1.67–5.00 Hz, 1.67–8.33 Hz, and 3.57–8.33 Hz range, respectively. Error bars indicate ±1 SE.
The 50% points obtained for motion perception with within-attribute stimuli were quite similar to those for object tracking with within-attribute stimuli in Experiment 1. However, for cross-attribute stimuli, the performance is much higher for motion tasks than for object-tracking tasks. These results suggest, therefore, that the temporal limit of the motion mechanisms could be reflected in the tracking performance for within-attribute stimuli, whereas the tracking performance with cross-attribute stimuli cannot be accounted for by the temporal limits of motion mechanisms that detect motion for cross-attribute stimuli. So the question is: What kind of visual process set the temporal limit for cross-attribute object tracking? Several past studies have argued that visual attention plays an important role in object tracking (Benjamins et al., 2007; Cavanagh & Alvarez, 2005; Verstraten et al., 2000). Verstraten et al. (2000) and Benjamins et al. (2007) specifically argued that the temporal characteristics of attention affect the temporal limit of object tracking. However, considering the agreement between the temporal limits for object tracking and that for apparent motion observed in the present results, the temporal limit for object tracking with within-attribute stimuli is not likely to be determined by the performance of attentional process. However, the contribution of attentional process might be significant for object tracking with cross-attribute stimuli, since the temporal limit for this task was much lower than that for motion perception. To test this possibility, we examined temporal limits for voluntary shifts of attention in the next experiment. 
Experiment 3
The results of Experiment 2 indicate that attentional factors alone cannot account for the dissociation between performances for within- and cross-attribute stimuli. Thus, we conjectured that there is motion involvement as well. To clarify this point, the temporal characteristics of pure attention shift were examined in this experiment by using a stimulus display that involved no motion component. 
Methods
Methods are generally the same as for Experiment 1, except for the differences described below. Five new observers, including one of the authors (HK), participated in this experiment. None of them participated in Experiments 1 and 2. They all had normal or corrected to normal vision. They were unaware of the purpose of this experiment except for HK. 
The stimulus was basically the same as luminance-defined stimuli in Experiment 1 (Figure 7). Either one of the four-object patterns, that is, a pattern with a square or diamond-shape configuration, was used in each trial in a randomized order. This patterned frame was alternated with a homogenous frame with no ISI. Durations of the patterned and homogenous frames were equal. Therefore, only four objects were involved in this experiment, and no motion was involved. A fixed luminance increment value (20.9 cd/m2), instead of an individually adjusted value, was used for all observers to generate luminance-defined rectangles. 
Figure 7
 
Examples of stimulus configuration and presentation in Experiment 3. Only the frame with four luminance-defined rectangles in a square or diamond-shape configuration was used as a patterned frame. This patterned frame was alternated with a homogenous frame with no ISI. Durations of the patterned and homogenous frames were equal. At the beginning of each trial, a red rectangular marker appeared and changed place at every appearance of patterned frame and observers were asked to conduct the object-tracking task in the same way as for Experiment 1.
Figure 7
 
Examples of stimulus configuration and presentation in Experiment 3. Only the frame with four luminance-defined rectangles in a square or diamond-shape configuration was used as a patterned frame. This patterned frame was alternated with a homogenous frame with no ISI. Durations of the patterned and homogenous frames were equal. At the beginning of each trial, a red rectangular marker appeared and changed place at every appearance of patterned frame and observers were asked to conduct the object-tracking task in the same way as for Experiment 1.
At the beginning of each trial, a red rectangular marker appeared and changed place at every appearance of the patterned frame in a rotating fashion until it made two complete circles. Observers were asked to conduct the object-tracking task in the same way as for Experiment 1. The direction of rotation (clockwise or counterclockwise) was randomized between trials. The alternation rate was varied in five steps (2.27, 2.78, 3.13, 3.57, 4.17 Hz). 
Results and discussion
For each observer, the percent correct score for the object-tracking task was calculated for each temporal frequency. In Figure 8, the mean percent correct scores for each temporal frequency were plotted as functions of temporal frequency. The mean percent correct was very high (92.5%) at the lowest temporal frequency (2.27 Hz), and it fell off very quickly as temporal frequency increased. 
Figure 8
 
Results of Experiment 3. Mean percentage correct scores are plotted as functions of temporal frequency (N = 5). Error bars indicate ±1 SE. The horizontal dotted line indicates the chance level.
Figure 8
 
Results of Experiment 3. Mean percentage correct scores are plotted as functions of temporal frequency (N = 5). Error bars indicate ±1 SE. The horizontal dotted line indicates the chance level.
It was found that the upper temporal limit of voluntarily shifting attention in object tracking was about 2–3 Hz. This value is quite similar to the limit of shifting the focus of voluntary attention reported by Horowitz et al. (2004), who measured upper SOA limit for the shift of attention between static placeholders in their “attentional saccade” task. Observers in the experiment were asked to shift the focus of attention between objects in accordance with auditory cues presented with a certain interval. They calculated an SOA that yielded 66.7% correct responses, and reported that attentional shift between objects is slow (300–500 ms of cue SOA). The temporal limit for attention shift found in the present experiment agrees well with their results, and this agreement suggests that the temporal limit for object tracking is restricted by the performance for spatial shifts of attention. 
General discussion
In this study, we measured the upper temporal limit of object tracking to identify higher visual mechanisms involved in the task. In Experiment 1, upper temporal limits for object tracking involving several different within- and cross-attribute stimuli were measured. A clear dissociation was found between the results for within-attribute stimuli and cross-attribute stimuli. The upper limits of object tracking for within-attribute stimuli ranged from 4 to 5 Hz, whereas those for cross-attribute stimuli were between 2 and 3 Hz. In addition, the values for within-attribute stimuli roughly coincide with the results reported by Verstraten et al. (2000) using a similar task but with stimuli defined purely by luminance, although the values for cross-attribute stimuli were much lower than their results. 
These results suggest that the mechanisms involved in object tracking for within- and cross-attribute stimuli are different to some degree. To clarify this point, in Experiment 2 we measured temporal limits for perceiving classical apparent motion by using the same stimulus conditions as in Experiment 1. It was found that all stimuli, whether within- or cross-attribute, yielded very similar temporal limits ranging from 4 to 5 Hz. These values are very close to the limits found for object tracking with within-attribute stimuli in Experiment 1. These results, therefore, suggest that motion mechanisms are involved in object tracking with within-attribute stimuli. However, the slower temporal limit for cross-attribute tracking cannot be accounted for by motion involvement, since it was much slower than the motion limit found in Experiment 2. To identify the process that governs cross-attribute object tracking, we measured the upper temporal limit for voluntary shift of attention using an object-tracking task that does not involve any motion component. The results indicated that the limit was about 2–3 Hz. This value agrees well with the 2- to 3-Hz limits for cross-attribute object tracking found in Experiment 1
This agreement suggests that object tracking with cross-attribute stimuli is mediated by an attentional process, or at least some kind of attentional mechanism is involved in the task. That is, these results suggest that object tracking with cross-attribute stimuli is supported by a process that also supports spatial shift of attention, a process that was assumed by Verstraten et al. (2000) and Benjamins et al. (2007). 
Verstraten et al. (2000) and Benjamins et al. (2007) tried to relate the upper temporal limit they found for their object tracking (4–8 Hz) to the limit of attentional processes. However, it is quite plausible that motion components were involved in their object tracking, since their stimulus was purely luminance based, and it is impossible to separate motion and attentional components with such stimuli. 
Horowitz et al. (2004) argued that object tracking is processed not by attention but by a preattentive “object continuity process” such as indexing to objects (Pylyshyn, 1989; Pylyshyn & Storm, 1988). Their argument is consistent with our results for within-attribute object tracking. The stimulus for object tracking involves ambiguous motion, and Verstraten et al. (2000) argued that attention is required to resolve this ambiguity. However, in the procedure, a marker (the red marker in the present study; Experiment 1) moves without ambiguity. Therefore, the ambiguity may also be resolved by such a motion process that mediates motion inertia (Anstis & Ramachandran, 1987). 
In the object-tracking task in Experiment 1 and 3, it is possible that observers were not tracking the target after the marker disappeared. Rather they might be predicting future location of the target object using motion and velocity of a red marker after its disappearance (e.g., representational momentum phenomenon; Freyd & Finke, 1984; Hubbard, 1995). If this is the case, the lower temporal limits for cross-attribute stimuli were not caused by weaker motion information from cross-attribute stimuli, but it can be interpreted as a result of stronger noise from cross-attribute stimulus that disrupted the accurate prediction of future position. However, this is not very likely since a much lower temporal limit was obtained in Experiment 3 in which the noise effect after the disappearance of marker was supposedly weaker since it was caused by no-motion, single-attribute stimuli. It is more likely that the low limit in Experiment 3 was caused by the difference in the nature of motion rather than by the difference in effectiveness of noise. Therefore, we argue that the difference between the results with within- and cross-attribute stimuli resulted from the difference in motion information available from the two types of stimuli. 
It is very intriguing that stimuli of all types yielded similar temporal limits for seeing apparent motion. Especially, it is noteworthy that purely luminance stimuli (within-luminance) produced results similar to those from all other stimuli. The stimuli other than pure luminance involved a second-order component and cannot be processed by first-order motion mechanisms, while the luminance-defined stimuli could be processed by first-order mechanisms. The temporal limit for first-order mechanisms is known to be fairly fast (above 10 Hz or so; Burr & Ross, 1982; Lu & Sperling, 1995b). Putting these factors together, the present results might suggest that apparent motion of isolated objects with relatively large separation with first-order, second-order, and cross-attribute stimuli are all processed by a common mechanism that first segregate targets from the background by detecting edges and then extract motion probably by using feature matching (Ullman, 1979). 
It is also worth mentioning that, in Experiment 2, contrast-defined stimuli produced similar temporal limits to those from other second-order stimuli. Two possible types of algorithms have been proposed for second-order motion extraction (Sato, 1998; Smith, 1994). One is to extract motion by detectors similar to those for first order (Fourier detectors) after applying some nonlinear preprocessing such as rectification to the input signal (Chubb & Sperling, 1988; Lu & Sperling, 1995b). The other is to extract certain features from the input, then track the features by using a nonlinear process such as feature matching (Ullman, 1979). It is generally assumed that contrast-defined motion is extracted by the former, pseudo-Fourier mechanism (Lu & Sperling, 1995b, 2001). Lu and Sperling (1995b) have reported that, while the upper temporal limits for motion with motion- or disparity-defined stimuli were around 3–4 Hz, the limit for contrast-defined motion was around 12 Hz. The results from Experiment 2 suggest that the same logic might apply to contrast-defined stimuli, and a mechanism slower than the pseudo-Fourier one is involved in object tracking with contrast-defined stimuli. 
In addition, the difference found between the upper temporal limits for object tracking (Experiment 1) and those for voluntary attention shift (Experiment 3) with within-attribute stimuli is quite suggestive. The within-attribute stimuli in Experiment 1 were ordinary second-order stimuli; therefore, these results suggest that there is a nonattentive, or passive, second-order motion process in addition to the second-order mechanism based on active attention (e.g., Cavanagh, 1991, 1992), and that the output from such a nonattentive mechanism is used to guide object tracking for within-attribute stimuli. The existence of a nonattentive second-order motion process has been proposed by several authors based on the existence of second-order reversed-phi phenomenon (Lu & Sperling, 1999; Maruya et al., 2003; Mather & Murdoch, 1999), and the present results support this proposal. 
Conclusion
The results of this study indicated that relatively low-level, nonattentive motion mechanisms as well as higher-order attention-based mechanisms are involved in object tracking. In particular, nonattentive motion mechanisms are involved in within-attribute object tracking, whereas attention-based mechanisms mediate cross-attribute object tracking. 
Acknowledgments
We are grateful to Professor Gayle K. Sato for her help in text editing. This study was supported by a Grant-in-Aid for Scientific Research (21330167) to TS from the Ministry of Education, Culture, Sports, Science and Technology, of Japan. A part of these results was presented at Vision Sciences Society (VSS) 11th annual meeting in 2011. 
Commercial relationships: none 
Corresponding author: Takao Sato. 
Email: lsato@mail.ecc.u-tokyo.ac.jp. 
Address: Department of Psychology, Graduate School of Humanities and Sociology, The University of Tokyo, Japan. 
References
Adelson E. H. Bergen J. R. (1985). Spatiotemporal energy models for the perception of motion. Journal of the Optical Society of America A, 2(2), 284–299. [CrossRef]
Anstis S. M. Ramachandran V. S. (1987). Visual inertia in apparent motion. Vision Research, 27(5), 755–764. [CrossRef] [PubMed]
Benjamins J. S. Hooge I. T. C. van der Smagt M. J. Verstraten F. A. J. (2007). Disengaging attention sets the temporal limit of attentive tracking. Vision Research, 47(8), 1055–1059. [CrossRef] [PubMed]
Brainard D. H. (1997). The psychophysics toolbox. Spatial Vision, 10(4), 433–436. [CrossRef] [PubMed]
Burr D. C. Ross J. (1982). Contrast sensitivity at high velocities. Vision Research, 22(4), 479–484. [CrossRef] [PubMed]
Cavanagh P. (1991). Short-range vs long-range motion: Not a valid distinction. Spatial Vision, 5(4), 303–309. [CrossRef] [PubMed]
Cavanagh P. (1992). Attention-based motion perception. Science, 257, 1563–1565. [CrossRef] [PubMed]
Cavanagh P. Alvarez G. A. (2005). Tracking multiple targets with multifocal attention. Trends in Cognitive Sciences, 9(7), 349–354. [CrossRef] [PubMed]
Chubb C. Sperling G. (1988). Drift-balanced random stimuli: A general basis for studying non-Fourier motion perception. Journal of the Optical Society of America A, 5(11), 1986–2007. [CrossRef]
Freyd J. J. Finke R. A. (1984). Representational momentum. Journal of Experimental Psychology: Learning, Memory, & Cognition, 10(1), 126–132. [CrossRef]
Horowitz T. S. Holcombe A. O. Wolfe J. M. Arsenio H. C. DiMase J. S. (2004). Attentional pursuit is faster than attentional saccade. Journal of Vision, 4(7):6, 585–603, http://www.journalofvision.org/content/4/7/6, doi:10.1167/4.7.6.[PubMed] [Article] [CrossRef]
Hubbard T. L. (1995). Environmental invariants in the representation of motion: Implied dynamics and representational momentum, gravity, friction, and centripetal force. Psychonomic Bulletin & Review, 2(3), 322–338. [CrossRef] [PubMed]
Hutchinson C. V. Ledgeway T. (2006). Sensitivity to spatial and temporal modulations of first-order and second-order motion. Vision Research, 46(3), 324–335. [CrossRef] [PubMed]
Lu Z. L. Sperling G. (1995a). Attention-generated apparent motion. Nature, 377, 237–239. [CrossRef]
Lu Z. L. Sperling G. (1995b). The functional architecture of human visual motion perception. Vision Research, 35(19), 2697–2722. [CrossRef]
Lu Z. L. Sperling G. (1999). Second-order reversed phi. Perception & Psychophysics, 61(6), 1075–1088. [CrossRef] [PubMed]
Lu Z. L. Sperling G. (2001). Three-systems theory of human visual motion perception: Review and update. Journal of the Optical Society of America A, 18(9), 2331–2370. [CrossRef]
Maruya K. Mugishima Y. Sato T. (2003). Reversed-phi perception with motion-defined motion stimuli. Vision Research, 43(24), 2517–2526. [CrossRef] [PubMed]
Mather G. Murdoch L. (1999). Second-order processing of four-stroke apparent motion. Vision Research, 39(10), 1795–1802. [CrossRef] [PubMed]
Morgan M. J. Watamaniuk S. N. J. McKee S. P. (2000). The use of an implicit standard for measuring discrimination thresholds. Vision Research, 40(17), 2341–2349. [CrossRef] [PubMed]
Pelli D. G. (1997). The VideoToolbox software for visual psychophysics: Transforming numbers into movies. Spatial Vision, 10(4), 437–442. [CrossRef] [PubMed]
Pylyshyn Z. (1989). The role of location indexes in spatial perception: A sketch of the FINST spatial-index model. Cognition, 32(1), 65–97. [CrossRef] [PubMed]
Pylyshyn Z. W. Storm R. W. (1988). Tracking multiple independent targets: Evidence for a parallel tracking mechanism. Spatial Vision, 3(3), 179–197. [CrossRef] [PubMed]
Sato T. (1998). Dmax: Relations to low- and high-level motion processes. In Watanabe T.(Ed.), High-level motion processing (pp. 115–151). Cambridge, MA: MIT Press.
Smith A. T. (1994). Correspondence-based and energy-based detection of second-order motion in human vision. Journal of the Optical Society of America A, 11(7), 1940–1948. [CrossRef]
Ullman S. (1979). The interpretation of visual motion. Cambridge, MA: MIT Press.
Verstraten F. A. J. Cavanagh P. Labianca A. T. (2000). Limits of attentive tracking reveal temporal properties of attention. Vision Research, 40(26), 3651–3664. [CrossRef] [PubMed]
Figure 1
 
Examples of stimulus configuration and presentation in Experiment 1. (A) Within-attribute (luminance) stimuli. Two arrays of four rectangular objects defined by luminance were alternately presented. (B) Within-attribute (second-order) stimuli. Two arrays of four objects defined by the same second-order attribute (motion, binocular disparity, flicker, or contrast) were alternately presented. Stimulus with motion-defined objects is shown as an example. White allows indicate motion directions of local dots in objects and the background. Yellow lines indicate borders surrounding objects. (C) Cross-attribute stimuli. Two arrays defined by luminance and one of the second-order attributes were alternately presented.
Figure 1
 
Examples of stimulus configuration and presentation in Experiment 1. (A) Within-attribute (luminance) stimuli. Two arrays of four rectangular objects defined by luminance were alternately presented. (B) Within-attribute (second-order) stimuli. Two arrays of four objects defined by the same second-order attribute (motion, binocular disparity, flicker, or contrast) were alternately presented. Stimulus with motion-defined objects is shown as an example. White allows indicate motion directions of local dots in objects and the background. Yellow lines indicate borders surrounding objects. (C) Cross-attribute stimuli. Two arrays defined by luminance and one of the second-order attributes were alternately presented.
Figure 2
 
Procedure for object-tracking task in Experiment 1. At the start of a trial, a circular array of eight rectangles appeared and a red rectangular marker was presented at the center of one of the eight objects to designate the object to track (target). Then, the display was switched to a motion display consisted of four rectangles and the marker started to rotate in either clockwise or counterclockwise direction. The marker stayed on the target, i.e., moved together with the target for two complete cycles, or two complete circles, and it disappeared. After the marker disappeared, observers were asked to track the target for 1.8 s. Then, the alternation of apparent motion frames stopped, and a blue rectangular marker appeared on one of eight objects.
Figure 2
 
Procedure for object-tracking task in Experiment 1. At the start of a trial, a circular array of eight rectangles appeared and a red rectangular marker was presented at the center of one of the eight objects to designate the object to track (target). Then, the display was switched to a motion display consisted of four rectangles and the marker started to rotate in either clockwise or counterclockwise direction. The marker stayed on the target, i.e., moved together with the target for two complete cycles, or two complete circles, and it disappeared. After the marker disappeared, observers were asked to track the target for 1.8 s. Then, the alternation of apparent motion frames stopped, and a blue rectangular marker appeared on one of eight objects.
Figure 3
 
Results of Experiment 1 (N = 6). Left column: Mean percentage correct for object-tracking task plotted as functions of temporal frequency for (A) motion-defined, (B) disparity-defined, (C) flicker-defined, and (D) contrast-defined objects. Error bars indicate ±1 SE. In each graph, filled rectangles (▪) and open circles (○) show results for within- and cross-attribute stimuli respectively, and filled triangles (▴) show results for corresponding luminance (within) stimuli. The horizontal dotted lines indicate the chance level. Right column: Mean upper temporal limits for six observers. The types of stimuli are as for the left column. In each graph, white and gray bars show results for within- and cross-attribute stimuli respectively. Error bars indicate ±1 SE.
Figure 3
 
Results of Experiment 1 (N = 6). Left column: Mean percentage correct for object-tracking task plotted as functions of temporal frequency for (A) motion-defined, (B) disparity-defined, (C) flicker-defined, and (D) contrast-defined objects. Error bars indicate ±1 SE. In each graph, filled rectangles (▪) and open circles (○) show results for within- and cross-attribute stimuli respectively, and filled triangles (▴) show results for corresponding luminance (within) stimuli. The horizontal dotted lines indicate the chance level. Right column: Mean upper temporal limits for six observers. The types of stimuli are as for the left column. In each graph, white and gray bars show results for within- and cross-attribute stimuli respectively. Error bars indicate ±1 SE.
Figure 4
 
Stimulus configuration in Experiment 2. Two-frame apparent motions between isolated objects were presented in the stimulus area 3.5° above the fixation point. The object size was 1.0° × 1.0°, and the spatial separation between the objects was 2.68°. The objects were defined by either luminance, motion, binocular disparity, flicker, or contrast as in Experiment 1. Within-attribute apparent motions were generated by alternately presenting two objects defined by the same attribute for six times with no ISI. Cross-attribute apparent motions were generated by pairing an object defined by luminance and the other defined by one of the second-order attributes and presenting them in the same way.
Figure 4
 
Stimulus configuration in Experiment 2. Two-frame apparent motions between isolated objects were presented in the stimulus area 3.5° above the fixation point. The object size was 1.0° × 1.0°, and the spatial separation between the objects was 2.68°. The objects were defined by either luminance, motion, binocular disparity, flicker, or contrast as in Experiment 1. Within-attribute apparent motions were generated by alternately presenting two objects defined by the same attribute for six times with no ISI. Cross-attribute apparent motions were generated by pairing an object defined by luminance and the other defined by one of the second-order attributes and presenting them in the same way.
Figure 5
 
Results of Experiment 2. Mean perception rates of perceiving motion are plotted as functions of temporal frequency (N = 6) for (A) motion-defined, (B) disparity-defined, (C) flicker-defined, and (D) contrast-defined stimuli. In each graph, filled rectangles (▪) and open circles (○) show results for within- and cross-attribute stimuli, and filled triangles (▴) show results for corresponding luminance (within) stimuli. Error bars indicate ±1 SE.
Figure 5
 
Results of Experiment 2. Mean perception rates of perceiving motion are plotted as functions of temporal frequency (N = 6) for (A) motion-defined, (B) disparity-defined, (C) flicker-defined, and (D) contrast-defined stimuli. In each graph, filled rectangles (▪) and open circles (○) show results for within- and cross-attribute stimuli, and filled triangles (▴) show results for corresponding luminance (within) stimuli. Error bars indicate ±1 SE.
Figure 6
 
Results of additional experiment. Mean perception rates of perceiving motion are plotted as functions of temporal frequency (N = 3). In the graph, rectangles (▪), triangles (▴), and circles (○) show the results for 1.67–5.00 Hz, 1.67–8.33 Hz, and 3.57–8.33 Hz range, respectively. Error bars indicate ±1 SE.
Figure 6
 
Results of additional experiment. Mean perception rates of perceiving motion are plotted as functions of temporal frequency (N = 3). In the graph, rectangles (▪), triangles (▴), and circles (○) show the results for 1.67–5.00 Hz, 1.67–8.33 Hz, and 3.57–8.33 Hz range, respectively. Error bars indicate ±1 SE.
Figure 7
 
Examples of stimulus configuration and presentation in Experiment 3. Only the frame with four luminance-defined rectangles in a square or diamond-shape configuration was used as a patterned frame. This patterned frame was alternated with a homogenous frame with no ISI. Durations of the patterned and homogenous frames were equal. At the beginning of each trial, a red rectangular marker appeared and changed place at every appearance of patterned frame and observers were asked to conduct the object-tracking task in the same way as for Experiment 1.
Figure 7
 
Examples of stimulus configuration and presentation in Experiment 3. Only the frame with four luminance-defined rectangles in a square or diamond-shape configuration was used as a patterned frame. This patterned frame was alternated with a homogenous frame with no ISI. Durations of the patterned and homogenous frames were equal. At the beginning of each trial, a red rectangular marker appeared and changed place at every appearance of patterned frame and observers were asked to conduct the object-tracking task in the same way as for Experiment 1.
Figure 8
 
Results of Experiment 3. Mean percentage correct scores are plotted as functions of temporal frequency (N = 5). Error bars indicate ±1 SE. The horizontal dotted line indicates the chance level.
Figure 8
 
Results of Experiment 3. Mean percentage correct scores are plotted as functions of temporal frequency (N = 5). Error bars indicate ±1 SE. The horizontal dotted line indicates the chance level.
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×