The dynamic events that capture visual attention: A reply to Abrams and Christ (2005)

We recently demonstrated that, contrary to previous findings, some types of irrelevant motion are capable of capturing our attention (Franconeri & Simons, 2003). Strikingly, whereas sitmulated looming (a dynamic increase in object size) captured attention, simulated receding (a decrease in object size) did not. Abrams and Christ (2003, 2005) have provided a different interpretation of this evidence, arguing that in each case attention was captured by the onset of motion rather than by motion per se. They argued that the only published finding inconsistent with their motion onset account is our evidence that simulated receding motion failed to capture attention. Abrams and Christ (2005) presented a receding object stereoscopically and found that it did capture attention, leading them to conclude that the motion onset account explains existing data more parsimoniously than our account does. Our reply has three parts. First, we argue that evidence of capture by receding motion is interesting but irrelevant to the debate over whether capture by motion requires a motion onset. Second, we show that the original empirical evidence in support of the motion onset claim (Abrams & Christ, 2003) put the motion-only condition at a critical disadvantage. We present a new experiment that demonstrates strong capture by motion in the absence of a motion onset, showing that motion onsets are not necessary for attention capture by dynamic events. Finally, we outline what is known about the set of dynamic events that capture attention.     

Motion onset captures attention: a rejoinder to Franconeri and Simons (2005)

Recently, we have provided evidence that the onset of motion captures attention (Abrams & Christ, 2003, 2005a, 2005b). In the present article, we clarify the motion onset hypothesis, we discuss recent data (Franconeri & Simons, 2005) that, at least on the surface, seem to challenge the hypothesis, and we present results from a new experiment (Christ & Abrams, 2005). Finally, we conclude that, although motion onset does indeed appear to capture attention, motion in the absence of a motion onset might also attract attention under certain circumstances.     

The onset of receding motion captures attention: comment on Franconeri and Simons (2003)

Franconeri & Simons (2003) reported that some but not all types of motion capture attention in a visual search task, ostensibly because some types of motion are behaviorally significant. In the present article, we argue that a more parsimonious explanation of their results is that the onset of motion, but not motion per se, captures attention. This conclusion helps to resolve inconsistencies between the Franconeri and Simons findings and earlier reports from other investigators and is consistent with results that we have recently reported (Abrams & Christ, 2003). The Franconeri and Simons interpretation rests largely on their failure to find attentional capture by one type of motion-simulated receding motion. We report here the results of two experiments that demonstrate that the onset of receding motion does indeed capture attention when the motion is produced using stereo depth cues. As we have argued elsewhere, the capture of attention in displays containing motion appears to be dependent on the onset of the motion, rather than on the mere presence of specific types of motion as suggested by Franconeri and Simons. One possible explanation is that the onset of motion captures attention because it serves as a strong cue to animacy--and the detection of nearby animals can have important consequences for survival.     

Revisiting preview search at isoluminance: new onsets are not necessary for the preview advantage

It has been argued that search performance under preview conditions relies on automatic capture by luminance onsets (Donk & Theeuwes, 2001). We present three experiments in which preview search was examined with both isoluminant and nonisoluminant items (e.g., as defined by luminance onsets). Experiment 1 provided evidence against the automatic capture of attention by onsets. Search benefited when onset previews were followed by new onset stimuli, as compared with a full-set baseline matched for the number of new onsets but in which half the distractors appeared simultaneously at isoluminance. Furthermore, both Experiments 1 and 2 established a preview advantage when isoluminant targets followed onset previews, when compared with appropriate full-set baselines. Experiment 3 replicated this result, while showing that the preview benefit was disrupted by dual-task interference. The data indicate that new onsets are not necessary to generate a preview advantage in search. We discuss the data in terms of search's benefiting from active inhibition of old onset-defined stimuli.     

Is it impossible to inhibit isoluminant items, or does it simply take longer? Evidence from preview search

Visual search can be facilitated when participants receive a preview of half the distractors (the preview benefit in search; Watson & Humphreys, 1997). Donk and Theeuwes (2001) have argued that preview-based benefits are abolished if the display items are isoluminant to a background. This is consistent with the preview benefit being due to onset capture by the new stimuli. In contrast, the present experiments challenge this suggestion and show that preview benefits can occur under isoluminant conditions, providing that they are given enough time to occur. In Experiment 1, we showed that a preview benefit can occur even with isoluminant stimuli, provided that the old items are previewed for a sufficient time. In Experiment 2, we tested and rejected the idea that this advantage is due to low-level sensory fatigue for the preview stimuli. These findings indicate that the preview effect is not caused solely by onset capture.     

Non-transient luminance changes do not capture attention

The processing of luminance change is a ubiquitous feature of the human visual system and provides the basis for the rapid orienting of attention to potentially important events (e.g., motion onset, object onset). However, despite its importance for attentional capture, it is not known whether a luminance change attracts attention solely because of its status as a sensory transient or can attract attention at a relatively high cognitive level. In a series of six experiments, we presented visual displays in which a single object underwent a luminance change that was either visible or obscured by a mask. A target then appeared either at the change location or elsewhere. The results showed that the luminance change attracted attention only in the visible condition. This was even observed with the largest change we could generate (> 75 cd/m(2)). These data suggest that the importance of a luminance change is only in its status as a low-level sensory transient.     

Reflexive orienting to gaze is not luminance dependent

In the investigation of reflexive orienting to cues, two major theories have emerged: One proposes that transients created by the cue trigger attentional shifts, whereas the other argues that object changes are responsible for instigating orienting. In the present study, we examined whether luminance transients produced by the cue can generate reflexive orienting to gaze. Using a temporal order judgment paradigm under luminant or subjectively equiluminant conditions, participants judged which of two peripheral targets onset first. An uninformative gaze cue served to reflexively shift attention toward one object location, thereby temporally prioritizing the target presented there. The results revealed that attention was successfully shifted toward the cued object, as was evidenced by the participants’ selecting the cued object as appearing first significantly more often than the uncued object, even when the two onset simultaneously. Critically, the results were comparable across luminance conditions. Our findings reveal that luminance transients are not necessary for triggering orienting to gaze cues. We suggest that the orienting observed here can be better explained via an object-based hypothesis whereby object changes, not transients, trigger reflexive orienting.     

Visual marking and visual change

Five experiments investigated the types of changes that disrupt the preview effect--the benefit gained in difficult search tasks from presenting some distractors earlier in time. A shape change with or without an overall luminance change at the location of an old item was found to disrupt the preview effect, whereas an equivalent luminance change alone or an isoluminant color change was not disruptive. Results suggest that (a) relatively low-level visual changes may not be sufficient to abolish the benefit, (b) the benefit most likely occurs through inhibition applied to locations within a location master map, and (c) inhibition need not be applied to surface features of objects.     

Constructing isoluminant stimuli for word recognition research: A precautionary study

Isoluminant stimuli are used increasingly often to investigate processes underlying visual word recognition. However, construction of isoluminant stimuli is not straightforward, and inappropriate construction may have the result of misinforming theories that relate word recognition to neurological function. To inform the use of isoluminant stimuli in studies of word recognition, the present article details two experiments in which isoluminant stimuli were constructed using physical onscreen luminance matching and heterochromatic flicker photometry (HFP) with four different stimulus types: disks, squares, rectangles, and letter strings. The findings reveal (1) substantial differences between isoluminance determined by physical onscreen luminance matching and HFP, (2) substantial differences in HFP isoluminance across stimulus types, and (3) substantial differences in HFP isoluminance across participants. These findings indicate that, in contrast to common practice in word recognition research, HFP provides a better indication of isoluminance than physical onscreen matching; but HFP stimuli should match those used in the experiment proper and should be used to assess isoluminance individually for each participant.     

Contingent capture for onsets and offsets: attentional set for perceptual transients

Four experiments were conducted to examine whether attentional set affects the ability of visual transients (onsets and offsets) to capture attention. In the experiments, visual search for an identity-defined target was conducted. In the first 3 experiments, the target display either onset entirely or was revealed by offsetting camouflaging line segments to reveal letters. Prior to the target display, there was a noninformative cue, either an onset or an offset, at one of the potential target locations. Cues that shared the same transient feature as the target display captured attention. The lack of predictable target transients led to attentional capture by all forms of transients. The final experiments with luminance changes without offsets or onsets showed attentional capture when the luminance changes were large. The results suggest that attentional set can be broadly or narrowly tuned to detect changes in luminance.     

Binocular summation of chromatic changes as measured by visual reaction time

We determined visual reaction times to monocular and binocular changes in the luminance of isochromatic stimuli and to monocular and binocular changes in the color of isoluminant stimuli. Two isoluminant color changes were tested: chromatic variations along the red-green axis of Boynton's (1986) two-stage color vision model and chromatic variations along the yellow-blue axis of the same model. The results indicate a greater degree of binocular summation for luminance change than for color change. This result was largely independent of the motor component of reaction time.     

Do isoluminant color changes capture attention?

Four experiments are reported in which the effects of peripheral cues on visual orienting were investigated. In the luminance condition, the cues consisted of a peripheral change in stimulus luminance. In the isoluminance condition, the cues consisted of an isoluminant color change, using the transient tritanopic technique. In Experiments 1 and 2, it was found that peripheral luminance cues captured attention, whereas peripheral isoluminance cues did not. In Experiments 3 and 4, the participants detected a peripheral target that was also isoluminant with the background. Under these conditions, it was found that both luminance and isoluminance cues captured attention. The results are discussed in terms of the roles of the dorsal and ventral streams in visual orienting, and it is concluded that our findings provide partial support for the contingent involuntary orienting hypothesis of C. Folk and colleagues.     

Edges,colour and awareness in blindsight

It remains unclear what is being processed in blindsight in response to faces, colours, shapes, and patterns. This was investigated in two hemianopes with chromatic and achromatic stimuli with sharp or shallow luminance or chromatic contrast boundaries or temporal onsets. Performance was excellent only when stimuli had sharp spatial boundaries. When discrimination between isoluminant coloured Gaussians was good it declined to chance levels if stimulus onset was slow. The ability to discriminate between instantaneously presented colours in the hemianopic field depended on their luminance, indicating that wavelength discrimination totally independent of other stimulus qualities is absent. When presented with narrow-band colours the hemianopes detected a stimulus maximally effective for S-cones but invisible to M- and L-cones, indicating that blindsight is mediated not just by the mid-brain, which receives no S-cone input, or that the rods contribute to blindsight. The results show that only simple stimulus features are processed in blindsight.     

Contingent visual marking by transients

Preview search is a phenomenon in which a set of new items can be searched with seemingly no interference from items already present in the display. The preview effect has been shown to occur only when the presentation of the new items is accompanied by a luminance change (Donk & Theeuwes, 2001). In a series of experiments, we extend the type of transients that can lead to a preview benefit to offsets and motion, and confirm Donk and Theeuwes's finding that equiluminant color changes do not lead to a preview effect. Like Donk and Theeuwes, we find that preview search does not occur when only the old items undergo a transient change, suggesting that the processes responsible for preview search are triggered when the new items undergo a change detectable by the magnocellular system. In addition, we find that irrelevant transients interfere with preview search only when they match the current attentional set (e.g., luminance change or motion). Results suggest that preview search is not the automatic capture of attention by transients, but rather is contingent on top-down control settings.     

Luminance edges are not necessary for visual completion

Luminance edges seem to have an important role in visual feature binding and, more specifically, in visual completion because luminance differences are important for the perception of depth. We investigated this claim in two experiments in which the primed-matching paradigm was used. In Experiment 1, we investigated conditions under which either a partly occluded shape or an occluder was isoluminant with respect to the background. In Experiment 2, the partly occluded shape and the occluder were isoluminant with respect to each other. Evidence was found for visual completion in all cases, and we therefore conclude that luminance edges are not essential for visual completion.     

Attention capture by contour onsets and offsets: No special role for onsets

In five experiments, we investigated the power of targets defined by the onset or offset of one of an object’s parts (contour onsets and offsets) either to guide or to capture visual attention. In Experiment 1, search for a single contour onset target was compared with search for a single contour offset target against a static background of distractors; no difference was found between the efficiency with which each could be detected. In Experiment 2, onsets and offsets were compared for automatic attention capture, when both occurred simultaneously. Unlike in previous studies, the effects of overall luminance change, new-object creation, and number of onset and offset items were controlled. It was found that contour onset and offset items captured attention equally well. However, display size effects on both target types were also apparent. Such effects may have been due to competition for selection between multiple onset and offset stimuli. In Experiments 3 and 4, single onset and offset stimuli were presented simultaneously and pitted directly against one another among a background of static distractors. In Experiment 3, we examined “guided search,” for a target that was formed either from an onset or from an offset among static items. In Experiment 4, the onsets and offsets were uncorrelated with the target location. Similar results occurred in both experiments: target onsets and offsets were detected more efficiently than static stimuli which needed serial search; there remained effects of display size on performance; but there was still no advantage for onsets. In Experiment 5, we examined automatic attention capture by single onset and offset stimuli presented individually among static distractors. Again, there was no advantage for onset over offset targets and a display size effect was also present. These results suggest that, both in isolation and in competition, onsets that do not form new objects neither guide nor gain automatic attention more efficiently than offsets. In addition, in contrast to previous studies in which onsets formed new objects, contour onsets and offsets did not reliably capture attention automatically.     

The glare effect does not give rise to a longer-lasting afterimage

The glare effect is an illusion in which a region appears self-luminous when flanked by gradients that decrease in luminance with distance from the region (Zavagno, 1999 Perception 28 835-838). This region also appears brighter than a surface of the same luminance. We investigated, using the paradigm of afterimages, whether a low-level mechanism at the level of the retina or LGN could account for this apparent brighter sensation. We first replicated the result from the literature that brighter and longer-lasting physical stimuli generate longer-lasting afterimages. We then compared the glare-effect stimuli with their counterpart controls, and found that the glare-effect stimuli did not give rise to longer-lasting afterimages. This suggests that the apparent brighter sensation of the glare effect is not due to a retinal or LGN mechanism, but must have a cortical origin.     

Foveal spatial summation in human cone mechanism

We measured at the fovea the chromatic contrast threshold for stimuli modulated along different chromatic directions in the isoluminant plane of MBDKL colour space, considering the two cardinal axes (L/M) and S/(L + M) and other intermediate non-cardinal directions. This psychophysical determination was conducted as a function of stimulus size. The test stimulus was a foveal isoluminant Gaussian patch with a raised cosinusoidal temporal profile superimposed on a neutral background. The task was performed binocularly. The increment threshold was measured for three observers by a Bayesian adaptive psychometric method (QUEST). The Ricco area of complete spatial summation was estimated from the threshold-versus-area curves. The perceptive fields are smaller for the L/M-cone opponent direction than the S/(L + M)-cone opponent. The perceptive field sizes for the stimuli in non-cardinal chromatic directions and stimuli modulated at the (L/M)-cone opponent direction present similar values. Measurements were made at two luminance levels, 5 and 40 cd m(-2), but the differences found were small. The perceptive field sizes found could be associated with LGN area.     

The binocular combination of chromatic contrast

How is chromatic contrast combined binocularly? One index of binocularity is the binocular contrast summation ratio (BCSR), which is the improvement in contrast sensitivity with binocular rather than monocular presentation. Simmons and Kingdom (1998, Vision Research 38 1063-1071) noted that BCSRs with some red-green isoluminant stimuli were suggestive of full linear summation. This suggestion was investigated further in four subjects by measuring binocular and monocular contrast thresholds for the detection of 0.5 cycle deg(1) isoluminant (red-green) and isochromatic (yellow-black) Gabor patches. These Gabor patches had either vertically or horizontally oriented carrier gratings and were either dichoptically in phase (same coloured bars in binocular correspondence) or in dichoptic anti-phase (opposite coloured bars in binocular correspondence). Full linear summation would be indicated by BCSRs of 2 for the in-phase and close to 0 for the anti-phase conditions. Mean BCSRs at isoluminance were 1.93 and 0.90, respectively, for the in-phase and anti-phase stimuli with horizontal carriers, the former being consistent with full linear summation, but the latter not. Despite these results, BCSRs obtained with isoluminant and isochromatic stimuli under similar conditions were not statistically distinguishable from each other, although there was a tendency for summation at isoluminance with in-phase stimuli to be higher and anti-phase stimuli to be lower. These data fall short of demonstrating full linear summation of chromatic contrast between the eyes under all presentation conditions, but they do indicate that there are strong binocular interactions at red-green isoluminance, which are similar to, and possibly even stronger than, those obtained with luminance stimuli.     

Purely chromatic perception of motion in depth: two eyes as sensitive as one

Motion hyperacuity (phase) thresholds were measured for both lateral and stereoscopic oscillatory motion in both luminance and equiluminant red/green gratings of 2 cycles per degree. Thresholds for lateral chromatic motion did not exhibit the inhibitory fall-off at low temporal frequencies that was found for luminance motion. Phase thresholds for purely chromatic motion were substantially higher than those for luminance gratings, in proportion to the ratio of cone signal modulation, but they could be predicted from the corresponding contrast sensitivities for both types of stimulus. Stereomovement thresholds in luminance gratings showed the stereomovement suppression effect relative to monocular motion sensitivity previously reported for line stimuli, but purely chromatic gratings did not. Together with the lack of an inhibitory fall-off, these results imply that chromatic and luminance motion are processed by different neural pathways, and that the chrominance pathway is capable of supporting a strong percept of stereoscopic motion from purely chromatic gratings.