Perception, 2003, volume 32, pages 645 ^ 656

DOI:10.1068/p3414

Do intersections serve as basic features in visual search?

Jeremy M Wolfeô, Jennifer S DiMase

Brigham and Women's Hospital [ô also Harvard Medical School], 221 Longwood Avenue, Boston,
MA 02115, USA; e-mail: wolfe@[Link]
Received 22 July 2002, in revised form 12 December 2002; published online 6 June 2003

Abstract. The status of `intersection' as a basic feature in visual search tasks has been controversial.
Under some circumstances, a target possessing this attribute (eg a plus) `pops out' of a display
of distractors that lack the attribute (eg Ls). However, those cases may be artifacts of other
features such as relative size or number of line terminators. We report two sets of experiments
with stimuli intended to control for these factors. Search for the presence or absence of intersec-
tions is very inefficient with these stimuli. The results suggest that intersection should not be
included among the list of salient features that support efficient search through visual displays.

1 Introduction
In visual search tasks, observers typically look for a target item in a display with
some number of distracting items. If reaction time (RT) is measured as a function of
the number of items presented (set size), then the slope of the RT6set size function
can be taken as a measure of the efficiency of search (or of the cost of adding a
distractor). In some cases, search is very efficient. For instance, a red item in a field of
green distractors will `pop out' and will be found time-independent of the number
of distractors (Treisman and Gelade 1980). Other stimuli (eg a search for a rotated T
among rotated Ls) will be found inefficiently, with each additional L adding 20 ^ 30 ms
to a search where the target is present and about 40 ^ 60 ms to a search where no
target is present (Treisman and Gelade 1980). Note that this inefficiency is present even
when stimuli are large enough and spaced widely enough to eliminate the need for
eye movements (eg Zelinsky and Sheinberg 1997).
For more than 20 years, researchers have argued over the list of properties that will
support efficient search (for reviews, see Wolfe 1998a, 2000). It is uncontroversial to
assert that features in some dimensions will support efficient search. Examples include
color (Bundesen and Pedersen 1983; Carter 1982; D'Zmura 1991; Farmer and Taylor
1980; Green and Anderson 1956; Moraglia et al 1989; Smith 1962; Van Orden 1993),
size (Bilsky et al 1994; Duncan and Humphreys 1992; Mu«ller et al 1995; Quinlan and
Humphreys 1987; Stuart 1993; Treisman and Gelade 1980), and orientation (Foster
and Ward 1991; Moraglia 1989; Wolfe et al 1992). Other attributes of the visual stimulus
have been more problematic. The form or shape of items has posed a particular problem.
It is clear that it is easy to find some shapes among other shapes (eg Beck 1966a;
Cohen and Ivry 1991; Donderi and Zelnicker 1969; Isenberg et al 1990; Quinlan and
Humphreys 1987; Stefurak and Boynton 1986; Theeuwes and Kooi 1994; Tiana et al
1989; Tsal and Lavie 1988). However, it has not been made clear what shape `features'
might be available to support efficient search. It is not even clear if there are multiple
featural dimensions within a general term like `shape' or `form'.
Good evidence exists for the featural status of line termination (Bergen and Julesz
1983; Julesz and Bergen 1983; Treisman and Gormican 1988). For example, search for
a C among Os is efficient (Treisman and Gormican 1988). It is not obvious whether
646 J M Wolfe, J S DiMase

the relevant features are the terminators on the C or the closure of the O (Elder and
Zucker 1994). Chen has argued for a topological account in which Os are in a different
category (having `holes') than Cs (Chen 1982; Chen and Zhou 1997).
In this paper, we re-examined the status of another candidate shape feature, line
intersection. Intersection has been proposed as a feature on the basis of search data
obtained with stimuli like those shown in figure 1. A plus is easily found amidst Ls
composed of the same vertical and horizontal lines (Bergen and Julesz 1983). More-
over, a region of plusses is readily segmented from a background of Ls (Beck 1966a,
1966b; Bergen and Julesz 1983). Over the years, there has been some controversy about
the reasons for this efficient search and immediate texture segmentation. Treisman and
Gormican (1988) found fairly inefficient search for a plus among the line segments
composing the plus. Spatial-frequency filtering can disrupt the easy discrimination of
plusses and Ls (Bergen and Adelson 1988; but see Julesz and Krose 1988). Nothdurft
(1993) found that masking by spatially bandpass stimuli differentially affected texture
segmentation and the identification of stimuli as plusses or Ls. He argued that the
texture segmentation was not based on an intersection feature. `Effortless' texture
segmentation and `pop-out' in visual search tasks are not the same thing (Wolfe 1992).
However, in the search for basic features, it is an encouraging sign if a property
supports both texture segmentation and efficient search. On the other hand, Kimchi
(1994) argued for a role for configural properties like intersection in the assessment of
the global/local figures used to produce Navon's (1977) `global precedence' effect.

Figure 1. There is no doubt that it is easy to find a plus or a region of plusses among Ls. Does
this mean that line intersection is a basic feature in visual search?

Is intersection a feature that supports efficient visual search? The answer is unclear
in part because research on the status of intersections has suffered from a lack of
variety in the choice of stimuli. Plusses differ from Ls in the presence of an inter-
section. However, there are other confounding differences that complicate analysis.
Plusses have four line terminators while Ls have two. As noted above, line termination
Do intersections serve as basic features in visual search? 647

seems to function as a feature, and search for the item with more of a feature can be
very efficient (Taylor and Badcock 1988; Wolfe 2001). Plusses are radially symmetric
while Ls are not (Olivers and van der Helm 1998; Wolfe and Friedman-Hill 1992).
Plusses look smaller than Ls when their vertical and horizontal segments are the same
length. This difference in `blob' size may be the basis for a number of the filtering
and masking effects. Accordingly, our goal in the present paper is to revisit the role of
intersection in visual search with the use of a new set of stimuli.

2 Experiment 1
2.1 Stimuli
In experiment 1, we created intersection and non-intersection stimuli that were designed
to minimize issues of line termination, symmetry, and size. These are shown in figure 2.
The basic intersection stimulus was a cross (2a)öno longer radially symmetric,
though it is symmetric about one axis. Rather than thinking of this as the intersection
of a vertical and a horizontal line, it can be considered to be composed of four pieces:
a smaller and a larger vertical segment and two horizontal segments of equal size.
Shifting one of the horizontal segments (2b) creates the non-intersection objects (2c).
It can be decomposed into the same four segments. It has the same four line termina-
tors. It lacks an intersection, having instead two T-junctions. Rotating and reflecting
2a and 2c yields four intersection stimuli (2d) and four non-intersection stimuli (2e).

(a) (b) (c)

(d)

(e)
Figure 2. Stimuli with and without intersection used in experiment 1.

Plusses, Ts, and Ls were created by using horizontal and vertical line segments of
equal length for control experiments (see below).
At a viewing distance of 57.4 cm, a 363 element stimulus array subtended a
20.5 deg620.5 deg square centered on the screen. Each item fit within a 5.5 deg65.5 deg
area and was displayed in black against a white screen background. Target and distractor
items were randomly presented in one of four orientations: vertical upright (08), upside-
down (1808), horizontal to the left (908), or to the right (2708).
648 J M Wolfe, J S DiMase

2.2 Subjects
Thirteen na|« ve observers between the ages of 18 and 55 years served as subjects. Data
for three subjects were discarded because of failure to complete the task. Each observer
passed the Ishihara test for color blindness and had 20=25 corrected vision or better.
All gave informed consent and were paid for their time.
2.3 Methods
Data were collected from each subject for six different search tasks. Four of these were
control conditions intended to replicate standard search results. Plusses among Ls and Ls
among plusses should replicate the standard finding of efficient search for stimuli of
this sort. Ts among Ls, and Ls among Ts provide a benchmark for `inefficient' search.
The critical conditions present our intersecting `cross' stimuli among stimuli without
intersection and vice versa. The order in which these six conditions were presented was
counterbalanced across subjects. Each block consisted of 30 practice and 300 test trials.
Before the start of each block, subjects were told what the target and distractors
would be. On each trial, items were presented at 3, 6, or 9 randomly chosen loci within
the array. Set size, positions of target and distractors, and the presence or absence of the
target item were random across trials.
At the beginning of each trial, a tone was sounded as the stimuli appeared. Search
stimuli remained visible until the subject responded by pressing either a ``yes'' key if
the target was detected or a ``no'' key if not. Feedback was provided after each trial
in the form of text on the screen and a beep if the response was incorrect. There was
an intertrial interval of 400 ms. Exactly half of the trials included a target. Subjects
were instructed to respond as quickly as possible while minimizing errors. Figure 3a
shows a demonstration version of the search for the new intersection stimuli. A larger
set size is used in figure 3b, which contains two intersection targets in order to give
the reader a feeling for the difficulty of the task.

(a) (b)
Figure 3. Find the `cross' stimuliöthe items with the putative intersection feature. (a) A typical
trial for the intersection search condition of experiment 1. (b) The same search in a set size
much larger than those used in this study. This permits the reader to gain an intuitive feel for
the difficulty of the task.

2.4 Results
As will be intuitively clear from figure 3, the new intersection stimuli do not support
efficient search. Figure 4 illustrates the average reaction time against set size for all
ten subjects for each condition. Results for the plus among L and L among plus search
Do intersections serve as basic features in visual search? 649

1600
Cross versus no intersection
96x ‡ 541
1400
Reaction time=ms

1200
No intersection versus cross
1000 45x ‡ 510
L versus T
33x ‡ 474
800 T versus L
14x ‡ 499
600 L versus plus
1:2x ‡ 492
Plus versus L
400 3:3x ‡ 442
(a)
Cross versus no intersection
212x ‡ 496
2400

2000
Reaction time=ms

No intersection versus cross

1600 114x ‡ 490
L versus T
89x ‡ 408
1200
T versus L
59x ‡ 418
800 L versus plus
8:3x ‡ 509
Plus versus L
400 7:7x ‡ 476
3 6 9
Set size
(b)
Figure 4. Reaction time plotted against set size for (a) target-present trials and (b) target-absent
trials. Results and the best-fitting linear regression lines are shown for each of the six conditions
of experiment 1. Condition labels give slopes and intercepts.

replicate the standard finding that led to the idea that intersection was a feature.
Searches are very efficient. Results for the T among L and L among T tasks produce
typical inefficient search with slopes of 14 and 33 ms=item, respectively, on the target-
present trials and target-absent slopes that are more than twice that steep (see Wolfe
1998b). In contrast, the new stimuli produced slopes of 96 ms=item when the inter-
section stimuli were the targets and 45 ms=item when the intersection stimuli were the
distractors. This is not only less efficient than the plus and L results; it is less efficient
than the inefficient T and L results. Error rates are given in table 1.
Table 1. Error rates for experiment 1.

Condition Misses=% False alarms=%

Plus versus L 1.7 0.7

L versus plus 1.5 0.5
T versus L 1.3 0.6
L versus T 4 0.6
Cross versus no intersection 6.5 0.3
No intersection versus cross 2.6 0.5
650 J M Wolfe, J S DiMase

It seems almost superfluous to note that these results are statistically reliable. An
ANOVA on the target-present slopes reveals a main effect of condition (F5, 9 ˆ 57:462,
p 5 0:0001). A posteriori tests show that slopes for the new stimuli are significantly
steeper than those for the traditional plus and L stimulus ( p 5 0:0001) and also steeper
than the inefficient T versus L slopes ( p 5 0:0001). The sole exception is that the search
for the non-intersection targets was not significantly less efficient than the search for
an L ( p 5 0:09). Similar results were found for target-absent trials.
There were reliable search asymmetries in the data (Treisman and Gormican 1988;
Treisman and Souther 1985) with search for a T among Ls being reliably more efficient
than for an L among Ts (paired t-test on slopes, t9 ˆ 2:482, p 5 0:035) and search
for the non-intersection target being more efficient than search for the cross target
(t9 ˆ 6:628, p 5 0:0001). Similar results were found for target-absent trials.
While not directly relevant to the topic of this paper, the asymmetry in the T
versus L searches is interesting. Our subjects found it easier to find a T among Ls than
vice versa. This may reflect the presence of an extra line terminator on a T. More of a
feature is generally easier to find than less.
2.5 Discussion and control experiments
2.5.1 Eye movement controls. One of the hallmarks of a basic feature in visual search is
that it can be presented in a number of ways and still produce efficient search. Thus,
for example, orientation is a basic feature whether it is defined by texture, luminance,
color, etc (Cavanagh et al 1990). Similarly, almost any moving stimulus `pops out'
from amidst any set of stationary stimuli. In experiment 1, a perfectly clear intersection
is very hard to find amidst non-intersection distractors. A second hallmark of a feature
in visual search is that the presence of the feature is easier to detect than its absence
(Treisman and Gormican 1988; Treisman and Souther 1985). In experiment 1, the data
go in the other direction. While both searches are inefficient, the target without the
intersection is easier to find than the target with the intersection.
Indeed, the new stimuli produce slopes that are so steep that one worries about
an eye-movement artifact. Any search can be made to be extremely inefficient if
the stimuli are designed so as to require subjects to fixate each stimulus or group of
stimuli. To eliminate this explanation for the present results, ten subjects performed
an identification task on briefly flashed presentations of the new intersection stimuli.
On each trial, subjects saw one item: either the cross/intersection stimulus or the non-
intersection stimulus. It was flashed in one of the nine stimulus locations used in
experiment 1. Stimulus position varied randomly from trial to trial. The stimulus was
present for 150 ms, a time too brief to permit subjects to make a voluntary fixating
eye movement. Stimuli and other conditions were otherwise identical to those of
experiment 1. The measure of interest in this case is accuracy. Under these circum-
stances, subjects were 83% correct showing that they could reliably identify the stimuli
without the need to fixate them.
In a second replication of experiment 1, set size was reduced to 1, 2, or 3 items, and
items were enlarged. Each stimulus fit within a 9.5 deg69.5 deg area in a 262 element
array. Observers were told to identify the presence or absence of the cross/intersection
stimulus among non-intersection distractors. Reaction times for search among these
large, uncrowded items are similar to those for the smaller stimuli of experiment 1
(figure 5).
2.5.2 Similarity controls. Perhaps similarity between targets and distractors accounts for
the difficulty of search with the stimuli of experiment 1. It is well known that search
slopes increase as target ^ distractor similarity increases and as distractor ^ distractor
similarity decreases (Duncan and Humphreys 1989). Indeed, the stimuli of experi-
ment 1 were created so that targets and distractors would be similar in size, number of
Do intersections serve as basic features in visual search? 651

1000 Target absent

150x ‡ 438
900
Reaction time=ms

800 Target present

103x ‡ 417

700

600
Figure 5. Intersection stimuli are difficult to
500 detect among non-intersection distractors
even when items are large and set sizes are
400 small.
1 2 3
Set size

line terminators, and every other feature except intersection. Moreover, distractors are
presented in different orientations, decreasing distractor ^ distractor similarity (Duncan
and Humphreys 1989). Of course, if intersection was a convincing preattentive feature,
its presence shouldöby definitionögreatly reduce any target ^ distractor similarity.
Consider that targets of one color can be readily found among distractors of another,
even if the stimulus items are identical in every other way. Still, in the present case,
it might be imagined that `intersection' is an aspect of form, and the targets and
distractors are otherwise fairly similar in form. In order to address this issue, observers
were presented with the stimuli shown in figure 6. These stimuli are identical save for
two minor changes. A pixel by pixel comparison would find these stimuli to be more
similar than the intersection and non-intersection stimuli of the main experiment. Of
course, by splitting the item in figure 6a into three pieces, we expected to find efficient
search for this segmented target (figure 6b) among the intact distractors (figure 6a).
The task was to detect the presence or absence of the segmented stimulus among the
original non-intersection stimuli. All other aspects of the experiment were equivalent
to those in experiment 1. Ten observers were tested. Observers produced the expected
efficient search for these items. The average slope was 1.8 ms=item when the target was
present, and 5.9 ms=item when the target was absent.

Figure 6. Stimuli for the similarity control experiment.

The original non-intersection stimulus is shown in (a);
deletion of a few black pixels produced (b).
(a) (b)

In addition to search efficiency and asymmetry, a typical feature will support easy
texture segmentation. In figure 7, there is a 462 item region composed of intersection
stimuli among non-intersection stimuli. It is clear that this region does not effortlessly
segment from the non-intersection background.
In sum, the results of experiment 1 argue against the status of intersection as a
feature. However, before striking intersection from the list of candidate features,
it would be desirable to have some converging evidence obtained with a different set of
stimuli. Perhaps more of an intersection feature would be easier to find. In experi-
ment 2, we asked subjects to detect a target with many intersections among distractors
with no intersections.
652 J M Wolfe, J S DiMase

Figure 7. It is difficult to distinguish the area of crosses embedded within this array of non-
intersecting items.

3 Experiment 2
3.1 Methods and stimuli
The stimuli for experiment 2 are shown in figure 8.
In experiment 2, subjects discriminated between squares containing multiple inter-
sections among squares without intersections (or vice versa); these `brickwall' stimuli
are shown in figure 8a. In control conditions, as shown in 8b, subjects could be asked
to search for line terminators among non-terminator stimuli (or vice versa). Stimuli
were designed to keep target and distractor differences to a minimum while retaining
the differences in either line termination or intersection. Individual items subtended
5.5 deg65.5 deg at the 57.4 cm viewing distance. All other aspects of the stimuli and
the procedure were identical to those in experiment 1. Ten subjects were tested.
3.2 Results
Figure 9 shows the average RT data as a function of set size for all subjects. Confirming
introspection (see figure 8), search for the target with six intersections among distrac-
tors with no intersections is not efficient (figure 8a). It is somewhat easier to find the
target with no intersections (but not significantly, paired t9 ˆ 1:028, p 5 0:34). Search
for line terminators (figure 8b) produced reasonably efficient slopes of 8 ms=item,
while the more difficult task of search for the absence of terminators produced a
significantly steeper average slope of 47 ms=item. This typical search asymmetry was
significant (paired t9 ˆ 5:412, p 5 0:0004). Error rates are shown in table 2.
Do intersections serve as basic features in visual search? 653

(a) (b)
Figure 8. Sample stimuli from experiment 2. Subjects could be asked to look for (a) intersections
or their absence, or (b) for terminators or their absence.

1250

1125
No terminator versus terminator
47x ‡ 549
Reaction time=ms

1000
Intersection versus no intersection
36x ‡ 588
875 No intersection versus intersection
26x ‡ 603
750

625 Terminator versus no terminator

8x ‡ 502
500

2000 Intersection versus no intersection

132x ‡ 566
No terminator versus terminator
1750
131x ‡ 490
No intersection versus intersection
Reaction time=ms

1500 95x ‡ 557

1250

1000 Terminator versus no terminator

40x ‡ 552

750

500
3 6 9
Set size
Figure 9. Reaction time6set size results for the four conditions of experiment 2.
654 J M Wolfe, J S DiMase

Table 2. Error rates for experiment 2.

Target Misses=% False alarms=%

Intersections 2.0 0.8

No intersections 2.0 0.7
Terminators 1.1 0.6
No terminators 2.4 0.8

3.3 Discussion
The results of experiment 2 do not support the argument that intersection serves as a
basic feature in visual search. In this case, a stimulus with multiple intersections failed
to be found efficiently among distractors with no intersections. The data from the
terminator version of the experiment can be used to demonstrate that stimuli of this
sort can produce reasonably efficient feature search. The presence of line termination
was readily detected. In keeping with the typical results for feature search, the absence of
that feature was more difficult to detect. Thus, these results replicate the well-established
finding that terminators serve as basic features in visual search while undermining the
claims of intersection to similar status.

4 Conclusions
Some researchers have questioned the entire enterprise of cataloging `preattentive'
features (DiLollo et al 2001; Nakayama and Joseph 1998), but this seems to be an
unfortunate case of throwing out a useful baby with the theoretical bath water. The
primary objection of the opponents of preattentive features is to the notion of a separate
`preattentive' stage of visual processing (Neisser 1967; Treisman and Gelade 1980). This
is not the place for an extended discussion of the usefulness of the preattentive/attentive
distinction, though it is worth noting that the idea that you can direct attention to
an object implies that there was a time prior to the time when you attended to that
object. If any visual processing occurs during that time, it might reasonably be called
preattentive without assuming other theoretical entanglements.
Even if one completely abandoned the notion of preattentive processing, the effort
to define basic features in visual search would be worthwhile. It is beyond doubt that
search for some targets among some distractors can be accomplished in a time that is
essentially independent of the number of distractors. It is equally clear that not all
stimulus attributes can support this type of efficient searchöeven if those attributes
are quite salient (eg facesöNothdurft 1993; Purcell et al 1996; Suzuki and Cavanagh
1995; but see, perhaps, Eastwood et al 2001). Therefore, it seems reasonable to ask
what the set of basic features might be and to suppose that the set is not arbitrary.
If we knew the set, we might better understand the earlier stages of visual processing.
The results of the experiments presented here lead us to the conclusion that `intersection'
should not be on this list of basic features.
Acknowledgments. This research was supported by a grant from NIH:NEI (EY05087). We thank
Todd Horowitz and Anne Treisman for useful discussions.
References
Beck J, 1966a ``Effect of orientation and shape similarity on perceptual grouping'' Perception
& Psychophysics 1 300 ^ 302
Beck J, 1966b ``Perceptual grouping produced by changes in orientation and shape'' Science 154
538 ^ 540
Bergen J R, Adelson E H, 1988 ``Early vision and texture perception'' Nature 333 363 ^ 364
Bergen J R, Julesz B, 1983 ``Rapid discrimination of visual patterns'' IEEE Transactions on Systems,
Man, and Cybernetics SMC-13 857 ^ 863
Do intersections serve as basic features in visual search? 655

Bilsky A A, Wolfe J M, 1995 ``Part ^ whole information is useful in size6size but not in orientation
6orientation conjunction searches'' Perception & Psychophysics 57 749 ^ 760
Bundesen C, Pedersen L F, 1983 ``Color segregation and visual search'' Perception & Psychophysics
33 487 ^ 493
Carter R C, 1982 ``Visual search with color'' Journal of Experimental Psychology: Human Perception
and Performance 8 127 ^ 136
Cavanagh P, Arguin M, Treisman A, 1990 ``Effect of surface medium on visual search for
orientation and size features'' Journal of Experimental Psychology: Human Perception and
Performance 16 479 ^ 492
Chen L, 1982 ``Topological structure in visual perception'' Science 218 699 ^ 700
Chen L, Zhou W, 1997 ``Holes in illusory conjunctions'' Psychonomic Bulletin and Review 4 507 ^ 511
Cohen A, Ivry R B, 1991 ``Density effects in conjunction search: Evidence for coarse location
mechanisms of feature integration'' Journal of Experimental Psychology: Human Perception
and Performance 17 891 ^ 901
DiLollo V, Kawahara J-i, Zuvic S M, Visser T A W, 2001 ``The preattentive emperor has no
clothes: a dynamic redressing'' Journal of Experimental Psychology: General 130 479 ^ 492
Donderi D C, Zelnicker D, 1969 ``Parallel processing in visual same ^ different decisions''
Perception & Psychophysics 5 197 ^ 200
Duncan J, Humphreys G W, 1989 ``Visual search and stimulus similarity'' Psychological Review
96 433 ^ 458
Duncan J, Humphreys G W, 1992 ``Beyond the search surface: Visual search and attentional
engagement'' Journal of Experimental Psychology: Human Perception and Performance 18
578 ^ 588
D'Zmura M, 1991 ``Color in visual search'' Vision Research 31 951 ^ 966
Elder J, Zucker S, 1994 ``A measure of closure'' Vision Research 34 3361 ^ 3369
Eastwood J D, Smilek D, Merikle P M, 2001 ``Differential attentional guidance by unattended
faces expressing positive and negative emotion'' Perception & Psychophysics 63 1004 ^ 1013
Farmer E W, Taylor R M, 1980 ``Visual search through color displays: Effects of target ^ background
similarity and background uniformity'' Perception & Psychophysics 27 267 ^ 272
Foster D H, Ward P A, 1991 ``Asymmetries in oriented-line detection indicate two orthogonal
filters in early vision'' Proceedings of the Royal Society of London, Series B 243 75 ^ 81
Green B F, Anderson L K, 1956 ``Color coding in a visual search task'' Journal of Experimental
Psychology 51 19 ^ 24
Isenberg L, Nissen M J, Marchak L C, 1990 ``Attentional processing and the independence of
color and shape'' Journal of Experimental Psychology: Human Perception and Performance 14
869 ^ 878
Julesz B, Bergen J R, 1983 ``Textons, the fundamental elements in preattentive vision and percep-
tions of textures'' The Bell Systems Technical Journal 62 1619 ^ 1646
Julesz B, Krose B, 1988 ``Features and spatial filters'' Nature 333 302 ^ 303
Kimchi R, 1994 ``The role of wholistic/configural properties versus global properties in visual form
perception'' Perception 23 489 ^ 504
Moraglia G, 1989 ``Display organization and the detection of horizontal line segments'' Perception
& Psychophysics 45 265 ^ 272
Moraglia G, Maloney K P, Fekete E M, Al-Basi K, 1989 ``Visual search along the color dimension''
Canadian Journal of Psychology 43 1 ^ 12
Mu«ller H J, Heller D, Ziegler J, 1995 ``Visual search for singleton feature targets within and across
feature dimensions'' Perception & Psychophysics 57 1 ^ 17
Nakayama K, Joseph J S, 1998 ``Attention, pattern recognition and popout in visual search'',
in The Attentive Brain Ed. R Parasuraman (Cambridge, MA: MIT Press) pp 279 ^ 298
Navon D, 1977 ``Forest before the trees: The precedence of global features in visual perception''
Cognitive Psychology 9 353 ^ 383
Neisser U, 1967 Cognitive Psychology (New York: Appleton, Century, Crofts)
Nothdurft H-C, 1993 ``Faces and facial expressions do not pop out'' Perception 22 1287 ^ 1298
Olivers C N L, van der Helm P A, 1998 ``Symmetry and selective attention: A task comparison''
Perception & Psychophysics 60 1101 ^ 1116
Purcell D G, Stewart A L, Skov R B, 1996 ``It takes a confounded face to pop out of a crowd''
Perception 25 1091 ^ 1108
Quinlan P T, Humphreys G W, 1987 ``Visual search for targets defined by combinations of color,
shape, and size: An examination of the task constraints on feature and conjunction searches''
Perception & Psychophysics 41 455 ^ 472
Smith S L, 1962 ``Color coding and visual search'' Journal of Experimental Psychology 64 434 ^ 440
656 J M Wolfe, J S DiMase

Stefurak D L, Boynton R M, 1986 ``Independence of memory for categorically different colors

and shapes'' Perception & Psychophysics 39 164 ^ 174
Stuart G W, 1993 ``Preattentive processing of object size: Implications for theories of size percep-
tion'' Perception 22 1175 ^ 1193
Suzuki S, Cavanagh P, 1995 ``Facial organization blocks access to low-level features: An object
inferiority effect'' Journal of Experimental Psychology: Human Perception and Performance
21 901 ^ 913
Taylor S, Badcock D, 1988 ``Processing feature density in preattentive perception'' Perception &
Psychophysics 44 551 ^ 562
Theeuwes J, Kooi J L, 1994 ``Parallel search for a conjunction of shape and contrast polarity''
Vision Research 34 3013 ^ 3016
Tiana C, Lennie P, D'Zmura M, 1989 ``Parallel search for color/shape and color/motion conjunc-
tions'' Investigative Ophthalmology & Visual Science, Supplement 30 252
Treisman A, Gelade G, 1980 ``A feature-integration theory of attention'' Cognitive Psychology 12
97 ^ 136
Treisman A, Gormican S, 1988 ``Feature analysis in early vision: Evidence from search asymmetries''
Psychological Review 95 15 ^ 48
Treisman A, Souther J, 1985 ``Search asymmetry: A diagnostic for preattentive processing of
separable features'' Journal of Experimental Psychology: General 114 285 ^ 310
Tsal Y, Lavie N, 1988 ``Attending to color and shape: The special role of location in selective
visual processing'' Perception & Psychophysics 44 15 ^ 21
Van Orden K F, 1993 ``Redundant use of luminance and flashing with shape and color as high-
lighting codes in symbolic displays'' Human Factors 35 195 ^ 204
Wolfe J M, 1992 `` `Effortless' texture segmentation and `parallel' visual search are not the same
thing'' Vision Research 32 757 ^ 763
Wolfe J M, 1998a ``Visual search'', in Attention Ed. H Pashler (Hove, East Sussex: Psychology Press)
pp 13 ^ 74
Wolfe J M, 1998b ``What do 1,000,000 trials tell us about visual search?'' Psychological Science 9
33 ^ 39
Wolfe J M, 2000 ``Visual attention'', in Seeing 2nd edition Ed. K K De Valois (San Diego, CA:
Academic Press) pp 335 ^ 386
Wolfe J M, 2001 ``Asymmetries in visual search: An introduction'' Perception & Psychophysics 63
381 ^ 389
Wolfe J M, Friedman-Hill S R, 1992 ``On the role of symmetry in visual search'' Psychological
Science 3 194 ^ 198
Wolfe J M, Friedman-Hill S R, Stewart M I, O'Connell K M, 1992 ``The role of categorization
in visual search for orientation'' Journal of Experimental Psychology: Human Perception and
Performance 18 34 ^ 49
Zelinsky G J, Sheinberg D L, 1997 ``Eye movements during parallel/serial visual search'' Journal
of Experimental Psychology: Human Perception and Performance 23 244 ^ 262

ß 2003 a Pion publication

ISSN 0301-0066 (print) ISSN 1468-4233 (electronic)

[Link]

Conditions of use. This article may be downloaded from the Perception website for personal research
by members of subscribing organisations. Authors are entitled to distribute their own article (in printed
form or by e-mail) to up to 50 people. This PDF may not be placed on any website (or other online
distribution system) without permission of the publisher.