These results suggest that preattentive processing may allow perceivers to
get individual features correct but, without focused attention, they are at risk
for creating illusory conjunctions.
Illusory conjunctions also arise with more naturalistic stimuli. In one study,
researchers used a slide projector to present subjects for 10 seconds with draw-
ings of faces (Reinitz et al., 1994). Half of the subjects were put in a situation of
divided attention:they were asked to count dots that appeared superimposed on
the slide of each face. Later, both groups of subjects were asked to look at an-
other series of slides and determine which of the faces they had seen before and
which were new. The subjects in the divided attention condition were success-
ful at recognizing the individual features of the faces—but they were inatten-
tive to recombinations of those features. Thus, if a ‘‘new’’ face had the eyes
from one ‘‘old’’ face and the mouth from another, they were as likely to say
‘‘old’’ as if the relations between the features had stayed intact. This result
suggeststhatextractingfacialfeaturesrequireslittleornoattention,whereas
extracting relationships between featuresdoesrequire attention. As a conse-
quence, subjects who suffered from divided attention could remember what
features they had seen but not which whole faces they belonged to!
If you make so many mistakes when putting the features together without
attention in the laboratory, why don’t you notice mistakes of this type when
your attention is diverted or overloaded in the real world? Part of the answer is
that you just might notice such mistakes if you start to look for them. It is
common, for example, for eyewitnesses to give different accounts of the way
the features of a crime situation combined to make the whole. Two witnesses
might agree thatsomeonewas brandishing a gun but disagree on which of a
team of bank robbers it was. Another part of the answer is provided by a lead-
ing researcher on attention,Anne Treisman.Treisman argues that most stimuli
you process are familiar and sufficiently different from one another so that
there are a limited number of sensible ways to combine their various features.
Even when you have not attended as carefully as necessary for accurate inte-
gration of features, your knowledge of familiar perceptual stimuli allows you to
guess how their features ought to be combined. These guesses, or perceptual
hypotheses, are usually correct, which means that you construct some of your
Figure 7.15
Combinations of features.
Perception 157