the Thatcher Illusion relates to face recognition processing.
Face inversion effect
The face inversion effect was introduced by Yin in 1969 and was defined as the larger
decrease in face recognition performance than for other mono-oriented objects when they
are presented upside down. Since people are used to the way they perceive a form, for
example seeing faces in an upright orientation, recognition of an inverted face would
be difficult.
Yin (1969) reported that participants of his experiment were unable to perceive a
general impression of the whole image for inverted faces. When a face stimulus is
presented at an upright orientation, the processing of facial features (e.g. eyes, mouth,
nose etc.) would affected by alterations to the identity or the position of one or several
other features of the face (Tanaka & Farah, 1993). Fig. 1 shows an inverted face of a
famous talk show host which we may hardly tell his expression or even to identify him at
the first sight.
It was explained that recognizing inverted faces is difficult for us because we are
relatively insensitive to the spatial relationships between the features of inverted faces.
Human faces are similar (with features such as eyes and mouth) but unique (with distinguish
spatial relationships). Thus those spatial relationships carry much of the information about
personal identity which helps us identifying others. As stated above, orientation affects how
we perceive objects; therefore we would find it more difficult to recognize upside-down faces
than normally upright faces.
The introduction of the idea of face inversion illusion leads to the interest on face
inversion recognition; and Thompson in 1980 found a powerful inversion illusion in face
recognition: the Thatcher Illusion.
Thatcher Illusion
Thatcher Illusion (Thompson, 1980) was a vivid illusion of Rock’s (1973) work and it
supported previous findings that inversion interferes with the face recognition processing.
It was named Thatcher Illusion or Margaret Thatcher Illusion because Margaret Thatcher’s
face was first used to demonstrate the effect, see fig, 2 and fig. 3. The effect can definitely be
applied to other faces though; and faces in the Thatcher illusion stimuli are generally called
“thatcherized faces”.
Images on the right in both fig. 2 and fig. 3 were thatcherized, i.e. mouth and eyes
were inverted. People typically found both images in fig. 2 looks normal. Although some of
them might find the right image looks a bit unnatural because the shadow around the mouth
and eyes are strange, they overall thought the two images were abnormal. However, when the
thatcherized faces were shown upright (fig. 3), people were alarmed and realized the
grotesqueness in the image on the right.
Rise of Thatcher Illusion
There are three competitive hypotheses explaining why the Thatcher illusion arises.
The first hypothesis is called the expression hypothesis: (a) the grotesque appearance of a
Thatcherized face is due to its expression, (b) inversion impairs the encoding of expression,
and therefore (c) inversion disrupts the perception of grotesqueness of a thatcherized face
(Bartlett & Searcy 1993: 284). This hypothesis suggested that we have difficulties when
encoding expressions on inverted faces, so we do not realize the problem with the
expressions unless the faces are right-side up.
The second hypothesis suggested that the illusion is related to reference frames. It was
found that we use two reference frames when we see an object such as a face. One of which
is based on the object itself, and the other is based on our contextual or egocentric sense of
orientation (e.g. what is supposed to be on the top and what at the bottom). When
thatcherized faces were inverted, the top of the mouth and eyes are different for the two
reference frames. However, once the images were in their usual orientation, the two reference
frames were then in agreement, and the face became grotesque.
The third hypothesis is based on the binary process theories of facial perception (Rossion
& Boremanse, 2008). It is assumed that we process faces by considering individual features
(e.g., eyes, mouth and nose) and their configuration (i.e. how they are organized and related
to each other). It is difficult for us to process the “configural” information when the faces are
inverted. Thus we will rely on the “local features” rather than perceiving faces as a whole. In
this case, thatcherized faces will not look strange since we are only focusing on individual
features. However, when the faces are right-side up, they look grotesque because both the
configuration and the features are clearly shown. This explanation is the most determined
among the three hypotheses and is related to face recognition processes; it will be elaborated
in detail in the following section.
Thatcher Illusion and faces recognition process
Thatcher illusion indicated that faces recognition could be depend on the face holistically
or on individual facial features (e.g. eyes, nose and mouth). According to the holistic
hypothesis, upright faces are stored as unparsed perceptual wholes in which individual
components are not explicitly represented (Tanaka & Farah, 1993). The “featural” hypothesis
on the other hand suggested that face recognition is characterized by a high sensitivity to
information about local features. Thatcher illusion showed that both hypotheses occur in
faces recognition process, while they involve in different situations.
As illustrated in the previous section, people found both the original and thatcherized
faces normal when they were inverted. It can be explained that we are only able to process
individual features in inverted images since it is unusual for us to perceive inverted faces.
Therefore, the right-side up eyes and mouth in the thatcherized image looks actually more
familiar and people typically think that both inverted pictures are usual.
On the other hand, we can easily perceive the images holistically once they are in their
normal orientation (i.e. in the orientation that we are familiar with). When recognising up-
right faces, “both component and configural information will be combined into a single
holistic representation” (Schwaninger, Carbon, Leder, 2003: 84). This means faces are now
seen as wholes, the relationship between local features as well as the entire representation
would be considered in the process of face recognition. In this situation, the upside-down
eyes and mouth on the thatcherized image are completely out of place on an otherwise
recognizable face, this is the reason people are alarmed by thatcherized faces.
Conclusion
The title of this paper invites discussing and analyzing the demonstration and causes of
Thatcher Illusion and explaining the relation between the Thatcher Illusion and face
recognition processing. Thatcher Illusion was introduced by Thompson in 1980. However it
was not elaborate in details in Thompson’s work, thus the Face Inversion Effect (Yin, 1969)
was first discussed since it helps explaining the Thatcher Illusion.
The theory of face inversion effect suggested that we are relatively insensitive to the
spatial relationships between the features of inverted faces, so recognizing inverted faces
would be difficult for us. Hence in the case of Thatcher Illusion, our ability in nterpreting
inverted thatcherized face is impaired thus we may not realize its grotesqueness. When
the thatcherized face is back into its normal orientation, we can then perceive both the
featural and holistic information of the image and found the face looks grotesque.
There are mainly three approaches explaining why the Thatcher Illusion arises:
expression hypothesis which suggested that we do not realize the problem with the
expressions when encoding expressions on inverted faces; the second argument was the
interpretation of reference frames which stated that we use two reference frames when
interpreting an object. One of them is based on the object, and the other is based on our
perceptual sense of orientation. Last but not least the binary process theories of facial
perception is the most powerful and logical explanation of the Thatcher illusion.
The third hypothesis, which also explains the face recognition process, suggests that
we process faces by considering both individual facial features and their configuration.
By elaborating this hypothesis, we can suggest some crucial ideas about the process of
face recognition: faces recognition could depend on perceiving the face holistically or
focusing on individual facial features; and these two ways involve in different situations.
Importantly though, face recognition process is a very complicated mechanism and
there will never be too many clarifications. Our knowledge in this field has been
improved by previous studies; however further research is essential so as to provide more
evidence on how people process faces.
Reference
Bartlett, J.C. & Searcy, J. (1993). Inversion and configuration of faces. Cognitive
Psychology, 25, 281-316
Bruce, V. & Young, A.W. (1998). In the eye of the beholder: The science of face
perception. Oxford:Oxford University Press, P. 158 – 160.
Rock, I. (1973). Orientation and form. New York: Academic Press.
Rossion, B., Boremanse, A. (2008). Nonlinear relationship between holistic processing of individual faces and picture-plane rotation: evidence from the face composite illusion. Journal of Vision, 8, 1-13.
Schwaninger, A., Carbon, C.C., & Leder, H. (2003). Expert face processing:
Specialization and constraints. In G. Schwarzer & H. Leder, Development of face
processing (pp. 81-97), Göttingen: Hogrefe.
Tanaka, J.W. & Farrah, M.J. (1993). Parts and wholes in face recognition. Quarterly Journal of Experimental Psychology, 46A, 225-245
Thompson, P. (1980). Margaret Thatcher: A new illusion. Perception, 9, 483-484
Yin, R. K. (1969). Looking at upside-down faces. Journal of Experimental Psychology, 81, 141-145.