Feature analysis and the role of similarity in preattentive vision

Texture arrays of line elements at various orientations were used to study three phenomena of preattentive vision. Subjects were asked (1) to discriminate texture areas and to distinguish their form (experiments on texture segmentation); (2) to detect salient or vertical line elements (experiments on pop-out); and (3) to identify configurations of similar or dissimilar targets (experiments on grouping). Within the patterns, line orientation was systematically varied to distinguish the effect of differences between areas from the effect of similarity within areas. In all of the experiments, performance was found to depend on local orientation contrast at texture borders rather than on the analysis of line orientation itself. Texture areas were correctly identified only when the orientation contrast at the border well exceeded the overall variation of line orientation in the pattern. Similarly, only target elements with high local orientation contrast were detected fast and “in parallel.” Targets with an orientation contrast lower than background variation required serial search. Preattentive grouping was found to depend on saliency, as defined by local orientation contrast, but not on the similarity of line elements. In addition to local orientation contrast, which played an important role in all of the visual phenomena studied, influences from the alignment of line elements with the outline of a figure were also seen.     

Feature analysis and the role of similarity in preattentive vision.

Texture arrays of line elements at various orientations were used to study three phenomena of preattentive vision. Subjects were asked (1) to discriminate texture areas and to distinguish their form (experiments on texture segmentation); (2) to detect salient or vertical line elements (experiments on pop-out); and (3) to identify configurations of similar or or dissimilar targets (experiments on grouping). Within the patterns, line orientation was systematically varied to distinguish the effect of differences between areas from the effect of similarity within areas. In all of the experiments, performance was found to depend on local orientation contrast at texture borders rather than on the analysis of line orientation itself. Texture areas were correctly identified only when the orientation contrast at the border well exceeded the overall variation of line orientation in the pattern. Similarly, only target elements with high local orientation contrast were detected fast and "in parallel". Targets with an orientation contrast lower than background variation required serial search. Preattentive grouping was found to depend on saliency, as defined by local orientation contrast, but not on the similarity of line elements. In addition to local orientation contrast, which played an important role in all of the visual phenomena studied, influences from the alignment of line elements with the outline of a figure were also seen.     

Visual search at different spatial scales

Treisman and Gelade's (1980) feature-integration model claims that the search for separate ("primitive") stimulus features is parallel, but that the conjunctions of those features require serial scan. Recently, evidence has accumulated that parallel processing is not limited to these "primitive" stimulus features, but that combinations of features can also produce parallel search. In the experiments reported here, the processing of feature conjunctions was studied when the stimulus features of a combination were at different spatial scales. The patterns in the search array were composed of three cross-shaped or T-shaped (local) elements, which formed an oblique bar (the global pattern) 45 deg or 135 deg in orientation. When the target and distractors differed from each other at one spatial scale only (either in the bar orientation or in the shape of the local elements), target detection was independent of the number of distractors, i.e., the search was parallel. In the conjunction task, in which the target and distractors were defined as the combinations of the bar orientation and the element shape, i.e., both spatial scales were relevant, the detection of the target required slow serial scrutiny of the search array. It is possible that the conjunction search could not be performed in parallel because switches between the two scales (or spatial frequency channels) are linked to attention and the task required the use of both scales in order to find the target.     

Does attention have different effects on line orientation and line arrangement discrimination?

Visual search and texture segregation studies have led to the inference that stimuli differing in the orientation of their component line segments can be distinguished without focal attention, whereas stimuli that differ only in the arrangement of line segments cannot. In most of this research, the locus of attention has not been explicitly manipulated. In the first experiment presented here, attention was directed to a relevant peripheral target by a cue presented near the target location or at the fovea. Effects of attention on orientation discrimination were assessed in a two-alternative forced-choice task with targets that were either: (1) lines that slanted obliquely to the right or left, or were horizontal or vertical, or (2) Y-like targets that had a short arm leading obliquely right or left of a vertical line. In some groups, a four-alternative forced-choice test with lines at 0°, 45°, 90°, and 135° orientations was used. Discrimination of these targets (i.e. targets that differ in the orientation of component line segments) was only minimally facilitated as the time between the onset of the valid cue and the onset of the target (cue-target stimulus onset asynchrony, SOA) was increased from 0 or 17 msec to 267 msec. In contrast, discrimination of targets that did not differ in the orientation of component line segments but differed in line arrangement (T-like characters), was greatly facilitated by longer cue-target SOAs. In Experiment 2, a cue misdirected attention on 20% of the trials. A decrement occurred on incorrectly cued trials in comparison to correctly cued trials for both types of stimuli used (lines and Ts). The results from these experiments suggest that discrimination of line orientation benefits less from focal attention than does discrimination of line arrangement, but that both discriminations suffer when attention must be disengaged from an irrelevant spatial location.     

Preferential processing of target features in texture segmentation

In five experiments, observers were required to detect a texture target and/or identify the orientation of elements composing target and nontarget regions. They were significantly worse at discerning the orientation of nontarget regions than at detecting target presence (Experiment 1). On the other hand, accuracy of identifying target orientation was found to be near 100% (Experiment 2). When observers were required only to identify surround orientation (Experiment 3), accuracy was diminished on target-present trials relative to that on target-absent trials. The superiority of target processing and the interference produced by target presence on surround processing were demonstrated in unpracticed observers (Experiment 4). In Experiment 5, it was found that information regarding target presence is available before information regarding feature values of the target. These findings are consistent with a model of visual attention and search that incorporates a fast generalized difference operator and a slower feature comparison process.     

Interference and dominance in texture segregation: hue, geometric form, and line orientation

Five experiments were designed to test whether (1) lowering the similarity of elements within a region of texture (low-similarity arrays) would interfere with texture segregation, and (2) there would be dominance of one type of property difference over another in determining an observer's choice of boundary in two-boundary (ambiguous) displays. In Experiments 1 and 2, the interference question was assessed using stimuli formed from the dimensions hue and geometric form (circle/square or straight/curved novel shapes). The results indicated that when boundary judgments were based on form differences, segregation was significantly impaired by hue variation. However, hue segregations were not affected by form variation. In Experiments 3-5, the dominance question was examined using stimuli formed from hue and geometric form, as well as those formed from hue and line orientation (horizontal/vertical). Analyses revealed that there was no dominance of one type of property difference over another. Rather, observers' performance was completely predicted by the relative discriminabilities of the two boundaries. These findings support Beck's (1982) model of textural segmentation and call into question traditional notions of the preattentive stage of perceptual processing.     

On the interference of task-irrelevant hue variation on texture segmentation.

Although natural images often include discordant information about object boundaries, the majority of research on texture segmentation has involved variation along a single dimension, e.g. colour, orientation, size. In this study, we examined orientation-based texture segmentation in the presence and absence of task-irrelevant colour variation. Previously, it had been shown that orientation-based texture segmentation was impaired if the elements, normally of one colour, were randomly allocated one of two colours (Morgan et al, 1992 Proceedings of the Royal Society of London, Series B 248 291-295). We found that this interference disappeared, however, when the spatial pattern of the colour variation was regular, as opposed to random, and when the elements were randomly positioned. We consider four models of how relevant and irrelevant texture information might combine to produce the interference effect, with special regard to these new findings. None of the models could account for the dependency of the interference effect on the spatial arrangement of colour and orientation in the texture. We suggest that inter-element separation and spatial-frequency selectivity are critical variables in the interference effect.     

Parallel coding of conjunctions in visual search

Two experiments investigated whether the conjunctive nature of nontarget items influenced search for a conjunction target. Each experiment consisted of two conditions. In both conditions, the target item was a red bar tilted to the right, among white tilted bars and vertical red bars. As well as color and orientation, display items also differed in terms of size. Size was irrelevant to search in that the size of the target varied randomly from trial to trial. In one condition, the size of items correlated with the other attributes of display items (e.g., all red items were big and all white items were small). In the other condition, the size of items varied randomly (i.e., some red items were small and some were big, and some white items were big and some were small). Search was more efficient in the size-correlated condition, consistent with the parallel coding of conjunctions in visual search     

Subset selective search on the basis of color and preview

In the preview paradigm observers are presented with one set of elements (the irrelevant set) followed by the addition of a second set among which the target is presented (the relevant set). Search efficiency in such a preview condition has been demonstrated to be higher than that in a full-baseline condition in which both sets are simultaneously presented, suggesting that a preview of the irrelevant set reduces its influence on the search process. However, numbers of irrelevant and relevant elements are typically not independently manipulated. Moreover, subset selective search also occurs when both sets are presented simultaneously but differ in color. The aim of the present study was to investigate how numbers of irrelevant and relevant elements contribute to preview search in the absence and presence of a color difference between subsets. In two experiments it was demonstrated that a preview reduced the influence of the number of irrelevant elements in the absence but not in the presence of a color difference between subsets. In the presence of a color difference, a preview lowered the effect of the number of relevant elements but only when the target was defined by a unique feature within the relevant set (Experiment 1); when the target was defined by a conjunction of features (Experiment 2), search efficiency as a function of the number of relevant elements was not modulated by a preview. Together the results are in line with the idea that subset selective search is based on different simultaneously operating mechanisms.     

Guided Search: The Effects of Learning

The previous report (Efron & Yund, 1996) offered an interpretation of the results of a number of search experiments within the theoretical context of the guided search model of Cave and Wolfe (1990) and Wolfe (1994). The present report extends this interpretation to the effects of extended practice when subjects search for a target defined by its orientation in the presence of a number of heterogeneous distractor items having differing orientations. Three experiments are described: The first revealed that over the course of 21 experimental sessions extending for a period of 6 weeks there were marked decreases in the magnitude of the reaction time gradient (RTG) and the right visual field superiority observed in the previous experiments. This learning persisted for more than 3 years. The second experiment revealed an interference in the capacity to learn to detect a target of one orientation when subjects had previously learned to detect a target of a different orientation at the same locations. The third experiment revealed that the learning was restricted to the area of the visual field where the target had been presented and that subjects could learn to detect two different targets concurrently. The results of these experiments indicate that the learning is orientation-specific and location-specific and is consistent with a localized increase in the selectivity of the top-down selection mechanism of the guided search model.     

Local and global orientation in visual search

Six experiments investigated the role of global (shape) and local (contour) orientation in visual search for an orientation target. Experiment 1 demonstrated thatsearch for a conjunction of local contours with a distinct global orientation was less efficient than search for a target featurally distinct in terms of both global and local contour orientation. However, Experiments 2 and 4 demonstrated that the presence of a unique line contour was neither sufficient nor necessary to allow efficient search. Experiment 5 found thatsearch for a local orientation difference was strongly impeded by irrelevant variation in global orientation, arguing for a preeminent role for global orientation. Finally, Experiment 6 demonstrated that the orientation search asymmetry holds for the global orientation of stimuli. Taken together, the results are consistent with visual search processes guided predominately by a representation of global orientation.     

Visual discrimination learning and transfer in rats with hippocampal lesions

Two experiments were performed on rats with hippocampal brain damage and on a control group with neocortical lesions. In the first experiment the hippocampal group learned a difficult visual discrimination as promptly as the controls, and neither group was subsequently impaired by adding relevant or irrelevant background cues to the original stimuli. In the second experiment the animals learned a simultaneous visual discrimination in which the stimuli differed in both brightness and orientation. The hippocampal group was impaired relative to the controls on acquisition, and showed poorer transfer to stimuli differing only in brightness or orientation. The results are incompatible with the hypothesis which attempts to explain the effects of hippocampal damage by a widespread reduction in sensory gating, but they are consistent with a more restricted version of the same hypothesis.     

Parallel search for conjunctions with stimuli in apparent motion

A series of experiments was conducted to determine whether apparent motion tends to follow the similarity rule (i.e. is attribute-specific) and to investigate the underlying mechanism. Stimulus duration thresholds were measured during a two-alternative forced-choice task in which observers detected either the location or the motion direction of target groups defined by the conjunction of size and orientation. Target element positions were randomly chosen within a nominally defined rectangular subregion of the display (target region). The target region was presented either statically (followed by a 250 ms duration mask) or dynamically, displaced by a small distance (18 min of arc) from frame to frame. In the motion display, the position of both target and background elements was changed randomly from frame to frame within the respective areas to abolish spatial correspondence over time. Stimulus duration thresholds were lower in the motion than in the static task, indicating that target detection in the dynamic condition does not rely on the explicit identification of target elements in each static frame. Increasing the distractor-to-target ratio was found to reduce detectability in the static, but not in the motion task. This indicates that the perceptual segregation of the target is effortless and parallel with motion but not with static displays. The pattern of results holds regardless of the task or search paradigm employed. The detectability in the motion condition can be improved by increasing the number of frames and/or by reducing the width of the target area. Furthermore, parallel search in the dynamic condition can be conducted with both short-range and long-range motion stimuli. Finally, apparent motion of conjunctions is insufficient on its own to support location decision and is disrupted by random visual noise. Overall, these findings show that (i) the mechanism underlying apparent motion is attribute-specific; (ii) the motion system mediates temporal integration of feature conjunctions before they are identified by the static system; and (iii) target detectability in these stimuli relies upon a nonattentive, cooperative, directionally selective motion mechanism that responds to high-level attributes (conjunction of size and orientation).     

Learning to search for visual targets defined by edges or by shading: Evidence for non-equivalence of line drawings and surface representations

A visual search task was used to test the idea that shaded images and their line-drawn analogues are treated identically from an early stage onwards in human vision. Reaction times and error rates were measured to locate the presence or absence of a target in an array of a variable number of distractors. The target was a cube in one orientation and the distractors cubes in a different orientation. The stimuli were defined by lines alone, shading alone, or lines plus shading. Both the slopes and the intercepts of the search functions (graphs of search time against number of displayed items) were higher for the line drawings than for the stimuli defined by shading. Over six experimental sessions, both the slopes and the intercepts fell for all stimuli, but the relative differences between them were maintained. The data suggest that, at an equivalent stage of practice, line-drawn stimuli are processed more slowly than shaded stimuli in early vision.     

Parallel development of ERP and behavioural measurements of visual segmentation

Visual segmentation, a process in which elements are integrated into a form and segregated from the background, is known to differ from adults at infancy. The further developmental trajectory of this process, and of the underlying brain mechanisms, during childhood and adolescence is unknown. The aim of the study was to investigate the developmental trajectory of ERP reflections of visual segmentation, and to relate this to behavioural performance. One hundred and eleven typically developing children from 7 to 18 years of age were divided into six age groups. Each child performed two visual tasks. In a texture segmentation task, the difference in event‐related potential (ERP) response to homogeneous (no visual segmentation) and checkered stimuli (visual segmentation) was investigated. In addition, behavioural performance on integration of elements into contours was measured. Both behavioural and ERP measurements of visual segmentation differed from adults in 7–12 year‐old children. Behaviourally, young children were less able to integrate elements into a contour than older children. In addition, a developmental change was present in the ERP pattern evoked by homogeneous versus checkered stimuli. The largest differences in behaviour and ERPs were found between 7–8‐ and 9–10‐, and between 11–12‐ and 13–14‐year‐old children, indicating the strongest development between those age groups. Behavioural as well as ERP measurements at 13–14 years of age showed similar results to those of adults. These results reveal that visual segmentation continues to develop until early puberty. Only by 13–14 years of age, children do integrate and segregate visual information as adults do. These results can be interpreted in terms of functional connectivity within the visual cortex.     

Visual search in children and adults: Top-down and bottom-up mechanisms

Three experiments investigated visual search for targets that differed from distractors in colour, size, or orientation. In one condition the target was defined by a conjunction of these features, while in the other condition the target was the odd one out. In all experiments, 6–7- and 9–10-year-old children were compared with young adults. Experiment 1 showed that children's search differed from adults' search in two ways. In conjunction searches children searched more slowly and took longer to reject trials when no target was present. In the odd-one-out experiments, 6–7-year-old children were slower to respond to size targets than to orientation targets, and slower for orientation targets than for colour targets. Both the other groups showed no difference in their rate of responding to colour and orientation. Experiments 2 and 3 highlighted that these results were not a function of either differential density across set sizes (Experiment 2) or discriminability of orientation and colour (Experiment 3). Across all three experiments, the results of both conjunction and odd-one-out searches highlighted a development in visual search from middle to late childhood.     

Visual search in children and adults: top-down and bottom-up mechanisms

Three experiments investigated visual search for targets that differed from distractors in colour, size, or orientation. In one condition the target was defined by a conjunction of these features, while in the other condition the target was the odd one out. In all experiments, 6-7- and 9-10-year-old children were compared with young adults. Experiment 1 showed that children's search differed from adults' search in two ways. In conjunction searches children searched more slowly and took longer to reject trials when no target was present. In the odd-one-out experiments, 6-7-year-old children were slower to respond to size targets than to orientation targets, and slower for orientation targets than for colour targets. Both the other groups showed no difference in their rate of responding to colour and orientation. Experiments 2 and 3 highlighted that these results were not a function of either differential density across set sizes (Experiment 2) or discriminability of orientation and colour (Experiment 3). Across all three experiments, the results of both conjunction and odd-one-out searches highlighted a development in visual search from middle to late childhood.     

Attentional effects on preattentive vision: spatial precues affect the detection of simple features

Most accounts of visual perception hold that the detection of primitive features occurs preattentively, in parallel across the visual field. Evidence that preattentive vision operates without attentional limitations comes from visual search tasks in which the detection of the presence of absence of a primitive feature is independent of the number of stimuli in a display. If the detection of primitive features occurs preattentively, in parallel and without capacity limitations, then it should not matter where attention is located in the visual field. The present study shows that even though the detection of a red element in an array of gray elements occurred in parallel without capacity limitations, the allocation of attention did have a large effect on search performance. If attention was directed to a particular region of the display and the target feature was presented elsewhere, response latencies increased. Results indicate that the classic view of preattentive vision requires revision.     

Attentional selection of objects or features: evidence from a modified search task

Three experiments examined the domain of visual selective attention (i.e., feature-based selection vs. object-based selection). Experiment 1 extended the requirements of the visual search task by requiring a feature discrimination response to target elements presented for short durations (30-105 msec). Targets were embedded in 47 distractor elements and were defined by either a distinct color or a distinct orientation. Observers made a discrimination response to either the target's color or its orientation. When the target-defining feature and the feature to be discriminated were the same (matched conditions), accuracy was enhanced relative to when these features belonged to separate dimensions (mismatched conditions). In Experiment 2, similar results were found in a task in which the target-defining dimension varied from trial to trial and observers performed both color and orientation discriminations on every trial. The results from these two experiments are consistent with feature-based attentional selection, but not with object-based selection. Experiment 3 extended these findings by showing that the effect is rooted in the overlap between target and distractor values in the stimulus set. The results are discussed in the context of recent models of visual selective attention.     

Advantage of dichromats over trichromats in discrimination of color-camouflaged stimuli in humans

This study investigated whether 12 participants with color-vision deficiency had superior visual discrimination of color-camouflaged stimuli shown on a computer screen compared with 12 participants with normal trichromatic vision. Participants were asked to distinguish a circular pattern from other patterns in which textural elements differed from the background in orientation and thickness. In one condition, stimuli were single-colored, green or red; in the other condition, stimuli were color camouflaged with a green and red mosaic overlaid onto the pattern. Color-vision deficient participants selected the correct stimuli in the color-camouflaged condition as quickly as they did in the single-colored condition. However, normal color-vision participants took longer to select the correct choice in the color-camouflaged condition than in the single-colored condition. These results suggest that participants with color-vision deficiency may have a superior visual ability to discriminate the color-camouflaged stimuli.