Professional Documents
Culture Documents
Melissa N. D’Angelo
Abstract
The purpose of this study is to see if limited stimuli conditions will still result in the same
findings as Treisman and Gelades study. Target combinations of colors and symbols are used to
determine if reaction time is affected by serial search and array size. Psychology 211 students
were told to complete this task as a course requirement. The first two trials sought to find the
effect of parallel search, and the last trial tested reaction time with a conjunction target. Feature
based-search functions such as color and shape, showed no significant differences in reaction
time. Conjunction-based targets had a larger effect, and took significantly longer to find than
color and shape. Results indicated in Figure 1 indicate an increasing trend in reaction time as
array size increases. Conjunction based search pulled all average means up in this trend. Serial
search, in accordance with more items causes an increase in reaction time. These findings
suggest that looking for conjunction is resource demanding while parallel is not. It requires more
attention for us to pick out objects that “blend-in” to a setting. Other evidence suggests attention
may not be responsible for this though, unlike the Feature-Integration Theory.
EFFECT OF CONJUNCTION AND ARRAY SIZE 3
To test how visual attention operates, Michael Posner explored the use of cues (Posner,
1980). Participants were directed to select which side they see the object on. However, cues
would pop up in the opposite direction of where the target would be. He predicted each time cues
showed up on the wrong side, that response time would increase. His results indicated longer
reaction times when the cues were in different spots, compared to those on the same side (Posner,
1980). Therefore, once attention is focused on a particular area, it will take us longer to reattach
Shortly after the release of Posner’s work, Treisman and Gelade did additional work on
visual and focused attention. Through prior knowledge and research, they developed the
feature-integration theory. Treisman and Gelade in 1980 said that the Feature-Integration Theory
hypothesized that attention is the glue that gives us the ability to recognize objects(1980). Serial
and Parallel search helped explain our limit for attention from the theory. Parallel search is
present in the preattentive stage of attention. It suggests that objects that are defined by a
distinguishable feature will take less of our focused attention to identify. Thus, it will require
minimal focused attention in the preattentive stage. On the other hand, a conjunction target, or a
target with multiple features together may require more time and attention to distinguish from
distractors. This is better known as serial search, or looking for multiple things. This level of
attention passes the ability to be recognized in the preattentive stage, thus requiring more time to
identify. When attention is overloaded, which was hypothesized to happen because of serial
search, illusory conjunctions would occur. This is when we instinctively combine features, and
believe it’s a target (Treisman, & Schmidt, 1982). It was hypothesized therefore that it would
take longer for individuals to process and correctly identify conjunction targets. Due to the fact
EFFECT OF CONJUNCTION AND ARRAY SIZE 4
that it would be harder to focus on a similar looking object to those in the array, they
hypothesized that attention would have to be refocused(as Posner found) and applied heavier to
To test this theory on focused attention, as well as it’s aspects, they conducted nine total
experiments. One study in particular tested the pop-out effect, to see if participants could identify
a stimulus or target that is meant to stand-out from the rest in the array. These target stimuli
included color-based, shape-based and conjunction based targets. Two rounds of the experiment
focused on feature-based searches(color and shape). These targets were purposefully meant to
differentiate heavily from distractor symbols. However, in the final round they were asked to find
a conjunction target that didn’t “pop out” in an array. Treisman and Gelade’s findings suggest it
took participants significantly longer in the final round to locate conjunction targets in arrays.
They also found that the time taken to respond increased as the number of items in the array
increased. Parallel or feature-based search deemed to take less time for participants to complete
compared to serial search. From this experiment, they had concluded that attention was the
reason that serial processing took longer to induce than parallel. Additional research done by
Treisman and Schmidt brought about the idea of illusory conjunctions in 1982. Assuming that
conjunction-based search was deemed serial processing, the thought of attention having to focus
more on other aspects caused illusory conjunctions(Treisman, & Schmidt, 1982). Therefore, this
This experiment is a partial replication of Treisman and Gelade. The purpose of this study
is to determine if reaction time will increase with conjunction and array size with limited
conditions. A smaller range of colors, and number of arrays will be used to determine if the
results from Treisman and Gelad’s hypothesized theory of feature integration will apply. It is
EFFECT OF CONJUNCTION AND ARRAY SIZE 5
expected that feature-based targets, or targets with distinct features will take less time to identify
compared to conjunction-based search. The number of items in an array are also expected to
Method
Participants
Convenience sampling was used to gather participants for this study. The participant pool
Methods 211. The task was completed by the students as a course requirement, thus there was
low motivation for performance. Students must have had normal or corrected vision in order to
participate.
Stimuli
Randomized arrays of symbols, generated by the computer software were used. The
number of symbols present within each image included 1, 5, 15 and 30. The arrays consisted of
sometimes the target conjunction, and distractors. Stimuli used were the symbols N, X, and O, in
colors green brown and blue. Conditions of the targets included feature-based search, which
focused on finding a specific color or shape. Targets in feature-based search included (in order of
use in the experiment), a blue letter, a green or brown “O” and a green “N”. Testing serial search,
the other condition was Conjunction-based search. This focused on finding objects similar to
distractors. The target used in conjunction-based search was a green “N”. Distractors in these
arrays were the Green X’s and N’s, designed to distract participants from the target combination.
Not all the arrays included a target image, or distractors. An asterisk symbol also popped up in
the center of each array for the first second that the image popped up. There were a total of 72
Procedure
All students participating were over the age of 18, thus did not need consent from parents
or legal guardians. A within-subjects design was used to conduct this experiment, where all
participants were exposed to all conditions. Lab computers provided by the university allowed
students access to the software needed to complete the task given to them. This program is
specially designed to collect reaction times during the activity created. Participants were given
instructions to use the software on their computers to access the task. Once achieved, a set of
instructions popped up on the screen. The participants were made aware of the task they would
be enduring through this set of instructions. Participants were informed that the experiment
would take less than 20 minutes, consisting of 3 blocks with 24 trials each.
Similar to Treisman and Gelade, two of the blocks tested for parallel feature search, and
the last tested serial search.. In the first round, participants were asked to identify a blue letter
amongst the array. When testing the shape condition in the second round, participants were asked
to find a brown or green letter “O”. The last round tested the conjunction of color and shape,
asking participants to find a green N. By using the green N, participants should struggle to
differentiate this target from the distractor symbols in the array, increasing response time. It was
asked of them to find a specific conjunction target amongst each array. If the target was in that
array, they were asked to click the “Y” key on their keyboard. If they did not see the target
indicated, they were asked to press the “N” key. They were also informed that before every trail,
an asterisk would be seen in the center of the computer screen. Participants were shown the
arrays in a random order and each image was only shown once. Each participant took about
10-15 minutes in total, taking 3-4 minutes for each block. This experiment was self-paced, but
participants were encouraged to answer as quickly as possibly to ensure accuracy of the data.
EFFECT OF CONJUNCTION AND ARRAY SIZE 7
Results
Reaction time rates in accordance to array size and condition are located on Figure 1.
Reported analysis was only based on valid trials. Mean response times for each condition suggest
that conjunction-search was on average slower than feature-based search conditions. There also
seemed to be an increasing trend in mean response time as the number of items increased.
Multiple statistical analyses such as 3x4 repeated measures analysis of variance, Bonferroni
Pairwise comparisons, and Greenhouse Geisser corrections were calculated. A 3x4 repeated
measures analysis of variance between conditions, (Color shape and conjunction) and array size
(1, 5, 15, 30), was used to find if there was a difference between groups. The Bonferroni Test or
Pairwise comparisons were then used to find which variable included that significant difference.
Anywhere where we violated assumptions of sphericity, greenhouse geiser was used to correct it.
Color-valid and Shape-valid showed similar as well as shorter average response time regardless
of array size. The condition effect showed F(1.650, 59.411)=55.759 with a p value less than
0.001. There was also a large effect of .608 somewhere in conditions. Pairwise comparisons
found that these significant differences and effects lied under the conjunction condition. The test
shows there to be a p value less than .001 for conjunction search. Meanwhile, color and shape
conditions showed no significant difference among groups, with a p value of .967. A p value that
large suggests no effect of color and shape on response time. This shows that the only condition
that had a large enough effect for a difference of response time was conjunction.
As the number of items in the array increased, there was a tendency for response time on
average to increase. Looking at figure 1, conjunction search seems to increase overall response
times for each item size. Regardless of condition, there were slight increases based on the
EFFECT OF CONJUNCTION AND ARRAY SIZE 8
number of items in the array. Significant differences were found somewhere with a p value of
0.001 and F(2.191, 78.880)=33.997. There seemed to be an effect of .486. Pairwise comparisons
showed 1 array item to be faster than 5, which was faster than 15 and 30 items. The last two
showed no difference, however they were greater than the smaller array sizes. It was found that
at 15 and 30, there was an equivalent p value of 0.41, therefore it was not statistically reliable.
However, the main difference was found to be between 1, 5, and 15 items since there was a
Conditions and items also were confirmed to have an effect on reaction time together.
Effects of the array size depends on which search condition we are looking at. Feature-based
search conditions with small array sizes tended to reflect smaller reaction times, regardless of
search condition. However, Conjunction-based with larger array sizes reflected slower response
times. The interaction effect reflected an F statistic of (2.856, 102.8) =21.028, and a p value
below 0.001. There was significant difference somewhere, as well as an effect of 0.369. In
pairwise comparisons, the p value for array size for conjunctions was less than or equal to .041.
Therefore, showing that there is a significant interaction for array size with conjunction .
Discussion
The findings from this study confirm our hypothesis stating that reaction time showed to
increase with conjunction and array size. Even with limited conditions compared to the Treisman
and Gelade experiment, the results remained the same. In fact, this study could be seen as even
more significant. When looking at Feature-Based Search, Treisman and Gelade’s conditions of
array size had a small effect size, while this experiment had minimal to none. The effect of array
size was much more drastic compared to thiers. Results can contribute to the idea that it requires
more attention to differentiate similar objects, compared to objects that “pop-out”. Shape and
EFFECT OF CONJUNCTION AND ARRAY SIZE 9
color can be identified at a quick glance. However, attention needs to reach outside of the
Though our research was consistent with Treisman and Gelade’s hypothesis, it refutes
their Feature-Integration Theory. It was believed that attention is responsible for binding these
features together to recognize objects, which in turn results in making illusory conjunctions
(Treisman, A. & Schmidt, 1982). Additional research done by Townshend implies that it is likely
that conjunction-search may not even be considered serial. Though conjunction-search shows a
that sensory overload from the amount of arrays may be what’s slowing down response time,
instead of the item you are searching for (Townshend, 1990). Conjunction search is very resource
demanding, therefore it’s hard to tell if attention causes the inability to properly recognize
targets. When you have lots of arrays, you are more likely to get overloaded and get more wrong,
but we don't know if that is due to attention or serial processing. We can’t assume one search is
parallel, and that the other is serial(Townshend, 1990). Donk also questioned the existence of
illusory conjunctions. She gathered data that suggested that people may become confused on
what their target is when searching in an array with similar symbols. Therefore, the
misconception of targets causes the belief that a target is present instead of illusory conjunctions.
When there's also a large array size, participants may feel overwhelmed or in a time
crunch(Ashby et.al,1996). This anxiousness may cause them to guess instead of truly looking for
a target (Ashby et.al,1996). This too diminishes the idea of illusory conjunctions.
The research conducted could have some possible limitations. Before the start of the task,
participants were not asked if they wore any type of eyewear for corrected vision. It is possible
that some who participated did not have the vision correction they needed. These possibilities
EFFECT OF CONJUNCTION AND ARRAY SIZE 10
could have inevitably skewed the results. However, the probability that some participants who
performed this task did not have the proper eyewear is very unlikely, due to the severity of our
results. The effect of array size and conjunction was extremely large, and there was a whopping
difference between parallel and serial search. Another possible limitation could be that the colors
used little of a difference in pigment. Therefore, it would be harder for individuals to recognize
differences in targets faster. The harder it is to find a difference, the more likely participants will
take longer to respond. This may cause some bias since it suggests increased reaction time. Since
there was a huge difference between conditions though, it is unlikely that the use of colors closer
in pigment would have drastically changed the effect. It is additionally unaccurate to assume that
parallel is feature-based search, and serial is conjunction search. As depicted, it is unknown how
to prove if something is serially based processing. Just because there was a large effect, does not
For future replications of this experiment, participants should be asked if they have
corrected vision before they begin the task. Participants should also be asked if they have any
color blindness. Even if it seemed to not matter much in this experiment, it is likely to invalidate
results of another replication. If they do not have the correct materials needed for them to see
properly, they should be asked to wear them. Anyone with colorblindness should be advised not
to participate. They may not be able to correctly identify certain colors, invalidating results.
Colors that are less close in pigment to other colors can also be used. This may help participants
find target conjunctions easier, resulting in more accurate results. Along those lines, replicators
can use different variations of colors and symbols. Varying the stimuli will further validate the
findings from this experiment across multiple different colors and shapes. The more research
confirming this hypothesis with multiple variations of stimuli, the more applicable to a real
EFFECT OF CONJUNCTION AND ARRAY SIZE 11
world setting. Other stimuli such as animals, foods, etc can also be used instead of letters. Doing
References
Ashby, F. G., Prinzmetal, W., Ivry, R., & Maddox, T. (1996). A formal theory of feature
binding in object perception. Psychological Review, 103, 165-192.
Townshend, J. T. (1990). Serial vs. parallel processing: Sometimes they look like
Tweedledum and Tweedledee but they can (and should) be distinguished. Psychological
Science, 1(1), 46-54
Figure 1:
Mean Search Times and Standard Errors as a Function of Array Size and Search Condition.