Reaction time differences in single feature search
and conjunction feature search
The world around us is full of objects. The ability to find one definite object amongst many others
distracting objects is a part of our routine visual behavior. This ability to perform “visual search” has been
a very interesting subject of research in the past years. In our experiment we investigate some of the
factors that affect human`s ability to find a certain target among distracters and now the reaction time vary
in different conditions, depending on the search condition – single feature or conjunction feature search,
the display size and whether target is present or absent. In our experiment we investigate how the
reaction time (RT) is affected in the three conditions .
Introduction:
Reaction time is measured as a function of the display size and this is the time that
participants use to answer whether a target is present or absent amongst a number of
distracters. (Wolfe, 1994).The difficulty of the task is measured by the relationship between the
RT and the total number of items in the display. For example if the target is defined only by one
feature (single feature search), the target could be found easily and the increasing of the
distracters` number will have little effect, but if a target is defined by two or more features
(conjunction feature search), then it would be more difficult to find the target and the number of
distracters will affect the RT. This is called “Visual Paradigm”. The feature integration theory
deals with it in deeper context. The feature integration theory is proposed by Treisman and
Gelade (1980). It suggests that there are two main stages of visual processing-parallel and
serial. The first stage (parallel visual processing) establishes feature maps, which provide
information on object`s features like color, shape, orientation, size, etc. These maps cannot
provide information about a feature`s location or about the identity of other feature on the same
position. This kind of visual processing is appropriate for single feature search tasks (SFS),
where the object is defined only by one feature. If, the object is defined by two or more features
we need the second stage (serial visual processing).At this stage a master map of locations is
used .It connects to all feature maps and codes the object`s location in the visual field. This
stage is used for conjunction feature search tasks (CFS), where the object is defined by two or
more features. The theory proves that SFS is much easier than CFS and as in the first case
attention needs to be applied only to one feature; The (RT) is faster, more accurate and not
affected by the display size and the target presence/absence. The CFS requires more time and
this is affected by the number of distracters on the display and the target`s presence/absence. It
is proved that if the target is absent, an extra distracter will always need attention, which
increases the reaction time, however, in a target present condition, half of the time the target will
be found, before an extra distracter is checked. This leads to 2:1 approximate ratio of absent:
present search slopes.
In our experiment ,we are investigating the differences between the reaction times and the
search slopes of the two main types of visual search-single feature search (SFS) and
conjunction feature search (CFS).Our hypothesis is that the results will support the feature –
integration theory and will prove that SFS is easier and not RTs are not affected by the number
of distracters and the presence/absence of the target, while the CFS is more difficult and the
RTs are affected by the display size and the presence/absence of the real target. We will explain
how the RT does changes in the different conditions and how the RTs in the different conditions
compare to each other.
Method:
Participants:
The participants of this experiment were 82 undergraduate psychology students from The
University of Warwick.
Procedure:
Materials:
Participants used a computer to conduct the experiment.
Task:
The participants conducted the experiment individually. They were asked to report whether a
blue vertical target is present or not. There were 3 conditions: a SFS, where the target is defined
by color-vertical blue item amongst green vertical items; SFS where the item is defined by its
shape –vertical blue target amongst blue horizontal items and a CFS ,where the target is
defined by a combination of color and shape –a blue vertical item amongst green vertical and
blue horizontal distracters. The three conditions were presented on computer, using a stimulus
presentation program. Within the conditions each search display contained of either 4,8 or 16
display items and the target was either present or absent. Each combination of target
present/absent was repeated 20 times.
Participants were asked to set up the experiment themselves. First they had to assign the trial
type-demo, practice and full experiment. The full experiment consisted of 120 trial in each
condition. After that they set up the first condition according to their birthday on the following
principle: if they divide their date of birth by 3 and there is nothing left, they should start with the
SFS –color condition, if there is 1 left, with the SFS-shape condition and if there is 2 left with the
CFS condition. The last part was setting up the key assignment. Due to counterbalancing
reasons the participants with odd date of birth were asked to press “Z” key during the three
conditions if the target is absent and “M” key if the target is present. Those who had even date
of birth were asked to do the opposite.
Results:
The results did not fully support the hypothesis. The SFS condition supports the hypothesis.
On Graph 1 ,where the blue line indicates RT, with target absent and the green line indicates a
target present RT, we can see that there is no difference the RT no matter of the display size
and the presence/absence of the target, because the search slopes are flat. In Table 1 are
shown the numerical values of the slopes in the three conditions, with the target present/absent.
In SFS-color condition, when the target is present, each extra distracter adds 0,501 ms to the
total RT, and if the target is absent, each distracter adds 0.497 ms to the RT. They are almost
equal and this condition fully supports the feature integration theory. In the second SFS
condition, where shape matters, we have a contradiction with the theory. The blue line in Graph
2 is rather steep, so the RT is affected by the display size and the target`s absence, presence.
The RT of target absent increases linearly, with the increase of the display size, while in target
present condition search slopes are is almost flat. According to Table 1 there is significant
difference between the search slopes of target present/absent conditions, and the RT values are
higher that the RT values in the previous condition. Each extra distracter adds 1,751 ms if the
target is present and 5,972 ms if the target is absent. This condition rejects the hypothesis. The
CFS confirmed our expectations. Graph 3 shows how the display size affects the RT in target
present/absent condition. In CFS, RTs and Error rates are significantly much higher than in SFS
conditions. The RT in target present condition is affected by 15,933 (refer to Table 1) ms per
item added, while in target absent condition RT is affected by almost twice this value-27.248ms.
The approximate ratio absent: present is approximately 2:1.This result fully supports our
hypothesis and the feature –integration theory.
Discussion:
Reaction times in SFS are faster than in CFS. However, not always the SFS presents flat
search slopes. In our experiment, we proved the feature integration theory for the SFS-color and
the CFS, but not the SFS-shape. The search slopes in SFS-color are not affected neither by the
display size not by the target`s presence/absence. They are flat. This is because the feature
“color” pops out and it is easy to notice the target with less attention. In CFS the theory is fully
supported .The search slopes become steeper, when the display size is increased and the RT in
target-absent condition is twice slower than the RT in target present condition. This is because
in a display set each item requires attention, which increases the time to respond whether the
target is present or not and if the target is present half of the times we would find it without
checking extra distracters. That is why the ratio absent: present is 2:1.The contradiction comes
with the SFS-shape. If the target is present the RT is almost not affected by the display size and
the search slopes are almost flat. It is just a little more higher than the one in SFS-color.
However if the target is absent in this case, RT increases significantly, when the display size is
increased and the search slopes are steeper. This can be due to the similarity of the targets and
non-targets(same size blue vertical and horizontal lines).Search slopes become steeper ,when
target-non target similarity is increased (Duncan & Humphreys,1992).according to Duncan &
Humphreys (1989) ,the RT and search slope`s steepness depend on the target`s and
distracter`s shape. If targets are not similar to distracters, then it would be easier to find the real
item and the search slopes will be flat, if, however, target and non-target have similar shapes,
then the search is more difficult, requires more time and the search slopes are steeper,
especially in a target-absent condition. This is why it is not surprising that the search slopes are
steeper even in a SFS condition.
References:
Duncan, J. & Humphreys. W.(1989). Visual search and stimulus similarity. Psychological
Review, 96,433-458
Duncan J. & Humphreys. W. (1992). Beyond the search surface: Visual search and attentional
engagement. Journal of Experimental Psychology: Human Perception and Performance, 18,
578-588
Treisman, A. &Sato, S.(1990). Conjunction search revisited. Journal of Experimental
Psychology: Human, Perception and Perforamnce, 16,459-478.
Treisman, A. M. & Gelade, G.(1980).A feature integration theory of attention .Cognitive
Psychology,12,97-136.
Wolfe, J.M.(1994)Guided search2.0:A revised model of Visual Search .Psychonomic Bulletin &
Review,1,202-238.
Appendix
Table 1
Search slopes in SFS and CFS condition
Presen
t
Absent
Graph 1:
Color
Shape
Conjunctio
n
0.501
0.497
1.751
5.972
15.933
27.248
Graph 2:
Graph 3: