Review of Psychology,

2013, Vol. 20, No. 1-2, 29-36 UDC 159.9

The role of context in Müller-Lyer illusion: The case of negative Müller-Lyer illusion

PAVLE VALERJEV and TANJA GULAN

The Müller-Lyer illusion is one of the most prominent geometrical-optical illusions that have been the subject
of experimental investigation throughout a century. Like most optical illusions the Müller-Lyer illusion is caused
by an appropriate context - inward or outward fins that act in a specific manner. These fins either prolong or shorten
the central line placed between the fins. In this study, we manipulated the context by varying the presentation of
the illusion parts, which led to the negative effect of Müller-Lyer illusion. Here we disassociated the context, i.e.,
the fins that create the illusion, from the lines the length of which needed to be assessed. Firstly, the fins were pre-
sented alone for 10 seconds, than the line would appear alone for 100 ms. In such situations the phenomenon of
illusion emerges in an opposite direction: the line that temporally succeeded the inward fins seemed longer, while
the line succeeding the outward fins seemed shorter. An experiment with three experimental situations was carried
out. Twenty participants took part in the experiment, in three different experimental situations. The size of the illu-
sion was measured using the method of constant stimuli, which was also used to determine the point of subjective
equality. The analysis confirmed the described effect which was found to be significant when compared to two other
situations: the control situation and the standard Müller-Lyer illusion situation. The negative Müller-Lyer illusion
is possibly caused by a kind of after effect, which occurred by prolonged gazing at the fins and/or by fatigue of the
appropriate selective angle sensitive cells. Such findings implicate that angle sensitive cells might be active in the
emergence of the standard Müller-Lyer illusion.

Key words: Müller-Lyer illusion, negative Müller-Lyer illusion, temporally disassociated presentation, figural after
effect, angle selective cells

The majority of optical illusions are the result of the spe- with single fins (e.g., Greene & Nelson, 1997). Further-
cific context manipulation of the critical part of the stimuli. more, arches, squares, or circles replacing the fins will lead
There are different examples of how context distorts figures
in a way that they might seem different in size, shape, or
shade. In the particular case of the Müller-Lyer (ML) il-
lusion, the critical part of the stimulus is the central line
(shaft), while the fins represent the context that can modify
the perceived length of the central line. The fins usually
come in two forms: inward (<>), also called arrowhead, and
outward (><), featherhead fins. The length of the central
shaft, i.e., the line is overestimated when it ends in outward
fins, or is underestimated when it ends in inward fins. The
example of the ML illusion can be seen in Figure 1.
The ML illusion comes in several different variations.
This illusion will also emerge if the double fins are replaced

Pavle Valerjev, Department of Psychology, University of Zadar, Croatia.

E-mail: valerjev@unizd.hr (the address for correspondence);
Tanja Gulan, Linguistics Department, University of Zadar, Croatia. E-mail: Figure 1. The example of Müller-Lyer illusion with typical cen-
tanja.gulan@gmail.com. tral shaft and outward and inward fins.

29
 VALERJEV and GULAN, Müller-Lyer illusion in context, Review of Psychology, 2013, Vol. 20, No. 1-2, 29-36

to the same effect (e.g., Howe & Purves, 2004). The illusion argue that the interpretation of image patterns is preceded
will manifest itself even if the central line is omitted: The by a step where image features such as lines, intersections of
space between outward fins will seem greater than the space lines, or local image movement must be derived. In that pro-
between inward fins. Research on the topic of ML illusion cess noise from different sources is present, and that noise
date back to the beginning of the twentieth century. For in- causes bias. In return that bias is responsible for alterna-
stance, Judd (1902) demonstrated the effect of practice on tions in image perception. Furthermore, the authors argue
the ML illusion. His participants were asked to estimate the that this bias is always present, and is part of uncertainty in
size of the central line in 980 trials over the period of a few our visual system. The illusory patterns are such that this
days. The results indicated that the strength of the ML illu- bias is prominent.
sion was reduced, and this the author attributed to practice. In this paper we are not dealing with figural after effect
Although the ML illusion is aroused on simple principle, explanation, but it seems that this phenomenon might play
the interaction between central line and fins, that type of some role in the negative ML illusion emergence. Contem-
interaction is not yet understood. porary knowledge claims that different visual after effects
The negative ML illusion was first identified phenome- are the result of the fatigue of specialized cells in the visual
nologically. The authors noticed that when one is exposed to cortex while, simultaneously, the spontaneous activation of
looking at the fins of the ML illusion for a prolonged period other specialized cells create the negative effect. In the per-
of time (e.g., 10 seconds or more) it can affect the percep- ception of figures, feature detectors might play an impor-
tion of the lines whose presentation succeeds the presenta- tant role in the origins of after effects. Köhler and Fishback
tion of the fins (see Figure 4). Interestingly, this effect seems (1950a, 1950b) could not predict the mechanism of feature
to act in the opposite direction than the one in the stand- detectors because the revolutionary work of Hubel and Wie-
ard ML illusion, which is the reason why this is labelled as sel (1959) on the subject of specialized cells in the visual
“negative”. cortex was published almost ten years later. Some of these
A review of the literature showed no record of the nega- detectors react to simple stimuli, lines and dots, while others
tive ML illusion except the work by Köhler and Fishback react to more complex ones (angles, curves, and crosses).
(1950a, 1950b), but the effect of the negative ML illusion Regardless of the theoretical explanation of the figural
was never studied in detail, or measured. Köhler and Fish- after effect mechanism, this seems like a suitable explana-
back were intrigued by Judd’s’ work (1902) and wanted to tion for ML illusion weakening in condition of prolonged
find an alternative explanation for the reduction or possible observing. If this is the case the figural after effect should
destruction of the ML illusion. The authors concluded that have a similar effect on other illusions if the illusion con-
practice was not the reason for the ML illusion destruction text is manipulated in the same way (temporally disassoci-
as stated by Judd (1902). They explained this phenomenon ated presentation). However, the authors have carried out
in terms of figural after effect, which was already of par- a phenomenological analysis on several other illusions and
ticular interest to Köhler (e.g., Köhler & Wallach, 1944). At these observations showed that the after effect of temporally
that time, one group of theories whose main objective was disassociated illusions result in the positive effect for the
the explanation of figural after effect applied to illusions Ponzo and Hering illusion, and have no effect on Sanders
emerged (Ganz, 1966; Köhler & Wallach, 1944). In these and Poggendorff illusion. This raises the question as to why
theories interference between nearby lines occurs because this negative effect is manifested in the ML illusion only.
of satiation in the cortex or lateral inhibition processes. The phenomenon of negative ML illusion could offer a
Figural after effect can be manifested in the shape and posi- new perspective on this illusion. In his work, Gregory (1968)
tion distortions of the figures when the observer gazes long listed numerous explanations for the ML illusion. Some of
enough at the preceding figure. Köhler and Wallach (1944) them have great historical but little scientific value. Gregory
offered an explanation for the figural after effect in accord- himself recommends depth theory as the best solution. Ac-
ance with gestalt teaching, i.e., in terms of electrical field cording to his theory, fins create the illusion of depth that
theory, which they termed “satiation theory”. According to makes the central line appear either closer or more distant
them, the concept of satiation is explained by changes in the from the observer (Fischer, 1967; Gregory, 1963). Because
so-called electro-tonus. While gazing at the stimulus, these of the size-constancy effect, the more distant line appears
changes are manifested in the different electrical conductiv- to be greater, as in the case of Ponzo illusion. However, it
ity of the visual cortex. Osgood and Heyer (1952; in Os- should also be noted that this explanation does not explain
good, 1953) criticized this explanation and offered a “statis- the case when the ML illusion is created using squares or
tical theory” that explains figural after effect in terms of the circles instead of fins, or in the case when the central line
differences in activity distributions that occur when com- is absent. Other explanations of the ML illusion include the
bining the activity distribution of observed stimuli and the confusion theory that claims that confusion about the place
previous stimulus. Fermüller and Malm (2004) reemployed of the beginning and the end of central line is the reason
statistics of visual computation in order to use mathematical for illusion emergence, as well as averaging theory (Erle-
models for prediction of geometrical optical illusions. They bacher & Sekuler, 1969; Pressey, 1970) that claims that the

30
 VALERJEV and GULAN, Müller-Lyer illusion in context, Review of Psychology, 2013, Vol. 20, No. 1-2, 29-36

perceived length of the central line is defined as the aver- Stimuli

age of the real length and the length between the fin tips.
The receptive field models (Walker, 1973) say that the size The stimuli were arranged in order to represent the
of proximal figures should determine the magnitude of dis- psychophysical method of constant stimuli. The standard
tortions. Among these models emerges the idea that there stimulus (SS) was always presented on the left side of the
are detectors for line orientation in the visual cortex. These computer screen, and on the left of the fixation cross that
detectors measure any orientation as the ratio of vertical to was presented on the center of the screen. The length of the
horizontal extent, which causes overestimation or underes- standard line was set to 50 mm in all experimental condi-
timation of the outward or inward fins (Caelli, 1977). Later, tions. The variable stimulus (VS) was always presented on
that idea of oriented receptive fields is modified into idea the right side of the computer screen, on the right of the
of special frequency filters that is responsible for distortion fixation cross, and varied in size. In experimental conditions
in line length perception. Kawabata (1976) and Ginsburg with the standard ML illusion (see Figure 2) the illusion
(1984, 1986) offered an explanation for ML illusion that in- consisted of the usual parts: two lines, one with the inward
cluded filtering processes caused by lateral inhibition and fins and one with the outward fins. The size of the right vari-
which produce a certain amount of blurring of the retinal able stimuli, that is, the size of the line varied from 24-76
picture. However, this explanation failed to satisfy predic- mm for 2 mm. In half of all situations, the standard stimulus
tion in illusion effect as some other theories did, such as was with inward fins, and in second half of all situations
depth theory. with outward fins. The preparation stimuli between the two
Another group of theories is related to feedback from experimental conditions lasted for 1500 ms and consisted
efferent commands for eye movement (e.g., Judd, 1905; only of the fixation cross. After 1500 ms the target stimuli
Kaufman & Richards, 1969). These theories suggest that appeared and lasted for 100 ms.
perceptual distortions might occur because of inappropriate In the control situation the stimuli consisted of two black
tendency to fixate the center of gravity of contextual fig- shafts that were in line with one another, one to the left of
ures when attempting to fixate the end of the focal shaft. the fixation cross and one to the right of the fixation point.
However, results from different research suggest that simple The standard left line was also set to 50 mm in all experi-
straightforward explanation is not satisfactory and that there mental conditions. The size of the variable line was set to
may be more than one simple mechanism contributing to il- vary in length from 40 to 60 mm, with a step of 2 mm. The
lusion emergence. The aim of this study was to demonstrate length of the variable line was based on the results of pre-
the phenomenon of the negative ML illusion and to measure liminary research that showed that lines shorter or longer
the magnitude of the effect of the negative ML illusion. than 40 and 60 mm were not necessary, i.e., a line of 60
mm was estimated longer then line of 50 mm in all cases.

METHOD

Participants

Twenty psychology students (16 females) from the Uni-

versity of Zadar participated in this study. Their age span
ranged from 18 years to 21 years, with a median of 19 years.
All participants had normal or corrected vision.

Design

The experiment consisted of three experimental situa-

tions, a) standard ML illusion, b) a temporally disassociated
ML illusion and, c) a control condition with two shafts that
were in line with each other. Experimental situations were
rotated using a Latin square rotation, one third of the partici-
pants began with the standard ML illusion, a second third of
the participants began with the temporally disassociated ML Figure 2. A presentation of the standard Müller-Lyer illusion. The
illusion, etc. All participants took part in every experimental left shaft was of fixed length and the right shaft varied in size. In
situation. The order of stimuli in every experimental situa- half of all situations, the left stimulus was with inward fins, and in
tion was randomized for each participant. second half of all situations with outward fins.

31
 VALERJEV and GULAN, Müller-Lyer illusion in context, Review of Psychology, 2013, Vol. 20, No. 1-2, 29-36

Materials and procedure

The stimuli were presented on a CRT 17” monitor with

the refresh rate of 80 Hz. The procedure followed the meth-
od of constant stimuli for measuring the differential thresh-
old. The participants’ task was to press either the right or
the left button depending on which shaft they considered to
Figure 3. Control situation: The left line was of fixed length, and be the longer one. The size of variable stimuli changed ran-
the right line varied in size. domly. There were 20 series for control and the standard ML
situation and 16 series for the temporally disassociated ML
illusion. As mentioned earlier, the second experimental situ-
ation required the prolonged gazing at fins that might cause
eye fatigue, therefore a series of 16 situations was consid-
ered optimal. Testing lasted for 90 minutes with pauses.

RESULTS

From the series of answers the point of subjective equal-

ity (PSE) for each participant in every situation was calcu-
lated. The size of the illusion effect can be expressed using
the PSE value. This is the value where the participant sub-
jectively equalizes the variable stimulus with the standard
stimulus. It was calculated using the equation from the Ex-
perimental Psychology of Sensory Behavior (Corso, 1967).

^ Sa - Sbh^C - pbh
PSE = Sb +
pa - pb

PSE – point of subjective equality

Figure 4. Temporarily disassociated presentation of the Müller- Sa, Sb – stimulus value immediately above the upper
Lyer illusion. In half of all situations, the left stimulus was with and below threshold
inward fins, and in second half of all situations with outward fins.
The left shaft was of fixed length, and the right shaft varied in size. C – the proportion of judgments defining the PSE crite-
rion (C = .5)
pa , pb – the proportion of “higher” and “lower” judg-
ment for the stimulus value immediately above or below the
corresponding threshold
Moreover, the research would take too long to complete and
In data analysis both the standard and the negative ML
the participants might experience fatigue. The preparations
illusion were compared to the control condition, separately
stimuli with the fixation cross lasted for 1500 ms, and after-
for the inward and the outward fins condition, which makes
wards the critical stimuli appeared for 100 ms (see Figure
a total of five conditions. All five conditions were tested in
3).
one-way repeated measures ANOVA, which showed a sta-
The third experimental condition that was crucial for tistically significant effect, F(4, 76) = 406.77, p < .001. The
this experiment dealt with the temporally disassociated ML post hoc analysis (Fisher LSD) showed significant differ-
illusion (see Figure 4), i.e., the shaft and fins were not pre- ences between all five conditions (p < .05). It is important to
sent on the screen simultaneously. Firstly, the fins would ap- note that all four experimental conditions are significantly
pear on the left and the right hand side of the screen for a different from the control condition, and that the effects ob-
period of 10 seconds, and later only two shafts would appear tained in the negative ML conditions are opposite to those
that fitted the previously presented fins. The shafts were pre- in the standard ML condition. These effect directions can be
sented for only 100 ms where the left shaft was fixed to the seen in Figure 5 when comparing graph values to the dotted
length of 50 mm and the right shaft varied form 24 to76 line that indicates the magnitude of the standard stimulus.
mm. In order to avoid the expectation of line length only In the negative ML illusion conditions with inward fins, the
shafts varied in length, while the fins were held constant and shaft was perceived to be longer, and in the condition with
were always in an equal distance of 50 mm. outward fins it was perceived to be shorter. On the contrary,

32
 VALERJEV and GULAN, Müller-Lyer illusion in context, Review of Psychology, 2013, Vol. 20, No. 1-2, 29-36

Figure 5. The comparison of the control and four experimental

conditions. Two of them illustrate the effect of the negative Mül-
ler-Lyer illusion (nML), while the last two illustrate the classical Figure 6. A comparison of the negative and the standard Müller-
Müller-Lyer (ML) illusion. The variable stimulus (VS) was always Lyer illusion in two conditions, with the inward fins or the outward
compared to the standard stimulus (SS). Inw. = inwards fins; Outw. fins as a variable stimulus (VS) compared to the standard stimulus
= outward fins. (SS) of 50 mm.

the inward fins caused shaft to be perceived as shorter, and standard ML illusion is 43.28% (SD = 8.22). Using one-way
the outward fins caused it to be perceived as longer in the repeated measures ANOVA, both results were analyzed in
standard ML condition. As seen, this effect is in the opposite contrast to the average error in the control situation (2.34%,
direction of the expected standard ML illusion effect. Thus, SD = 1.60). The tested difference between situations was
this indicates the existence of the negative ML illusion. significant, F(2, 38) = 426, p < .001. The post hoc analy-
The stimuli with inward fins work in a direction of de- sis (Fisher LSD) showed significant differences between all
creasing the central line, while stimuli with outward fins three tested situations (p < .01). The absolute effect of the
work in a direction of increasing the central line. In the up- standard ML illusion is significantly greater than the nega-
coming analysis we will present a comparison of the stand- tive ML illusion and the control situation. The absolute ef-
ard and the negative ML illusion effect, as well as the effect
size of each illusion.
The two-way repeated measures ANOVA showed a sig-
nificant interaction effect, F(1, 19) = 610.38, p < .001, for
the ML illusion type, and for the fins orientation (see Figure
6). Moreover, the post-hoc analysis (Fisher LSD) showed
significant differences between all four situations (p < .001).
This finding demonstrates not only the significance of the
effect on both illusions, but also the significance of the op-
posite effect for the negative and the standard ML illusion.
In the last part of the analysis, absolute values for the
effects size of the ML illusion, the standard and the nega-
tive one, were calculated. For calculating the size effects
we used the equation proposed by Bruno, Bernardis, and
Gentilucci (2008). According to procedure, absolute effects
of the illusions were calculated, and were expressed as the
percentage of the basic line that was defined by a standard
stimulus of 50 mm. The sizes of the illusions were com-
pared (see Figure 7). The average value of the negative ML Figure 7. The absolute magnitude of the negative and the standard
illusion is 6.63% (SD = 2.92), while the average value of the Müller-Lyer illusion.

33
 VALERJEV and GULAN, Müller-Lyer illusion in context, Review of Psychology, 2013, Vol. 20, No. 1-2, 29-36

fect of the negative ML illusion is significantly greater than ML illusion is apparently based on the mechanism of after
the control situation. effect. It is possible that gazing at the fins for a prolonged
period of time causes the after effect that leads to the delu-
sional effect which is in the opposite direction to the length
DISCUSSION of the central line. Furthermore, the fins or arrows in the ML
illusion can be seen as angles that have a corresponding ori-
The main goal of this study was to demonstrate and
entation and position. By using the technique of single cell
measure the size of the negative ML illusion, as well as to
recording on macaque monkeys it has been found that the
compare it to the standard ML illusion. The phenomenon
representation of stimuli complex features begins in the vis-
of the negative ML illusion has been proved and measured.
ual cortex in the V2 area (Ito & Komatsu, 2004; Kobatake &
The effect of the ML illusion was significant when com-
Tanaka, 1994), and continues to spread via the ventral visual
pared to the control situation, and was also less than the
pathway to the V4 area. These complex visual features in-
standard ML illusion. Moreover, the effect of the opposite
clude angles of different sizes and different orientations. Ito
direction compared to the standard ML illusion was proven.
and Komatsu (2004) found that selective cells in the V2 area
In order to elicit the negative ML illusion one must ma- are not so selective, meaning that even though these cell are
nipulate the context in a different manner. The negative ML activated by angles, they also became activated when look-
illusion manifests itself in the “temporally disassociated” ing at single lines of the same orientation that form a part of
condition of the standard ML illusion, where the context is angle. In the V4 area more selective cells were found that
temporally disassociated from the critical part of the stimuli. react to complex features of stimuli, including the angles
This results in distortion in the line length in the observer’s (Kobatake & Tanaka, 1994). These cells are more special-
perception. However, this distortion is in the opposite di- ized and more sensitive to the whole angle orientation, but
rection than one would expect in the standard ML illusion. are not sensitive to orientation of single lines, which means
The line that was placed where inward fins were previously that these cells represent angles more elaborately. As in the
placed seemed to be longer, while the line that placed where case of other after effects mechanisms, it is possible to put
outward fins were previously placed seemed shorter. specialized cells in a temporary state of fatigue. When this
The contemporary research on the topic of ML illusion occurs the spontaneous activation of other cells is greater
usually use ML illusion as a tool to investigate models of than when compared to those in a state of fatigue, which
human vision, in particular proposed division between vi- causes the opposite effect. There are many known exam-
sion-for-action (identified with the V1-PPT dorsal stream) ples. For instance, adaptation to red color stimuli will cause
and vision-for-perception (the V1-IT ventral stream). This a green after effect. Furthermore, adaptation to angled lines
model is good at predicting broad range of behavioral and causes an after effect that makes seemingly vertical lines
neuropsychological data, but what remains controversial angled in a different direction. Adaptation to the movement
is why visually guided action is immune from visual illu- of stimuli in one direction will cause the after effect of the
sions (Bruno et al., 2008; Thompson & Westwood, 2007). movement of still stimuli in the opposite direction (see
In another words, researchers are more interested in illusion Mather, Verstraten, & Anstis, 1998). Adaptation to a specific
destruction especially when the action is involved. Further spatial frequency grid will cause a lower sensitivity to the
debate on this topic aims to isolate motor responses from corresponding spatial frequency grid (McCollough, 1965).
conscious perception (Bruno, Knox, & de Grave, 2010). In accordance with this, one can assume that the adapta-
This resulted in revival of the theories of eye movement tion to angles of specific orientation might cause an illusory
as a source of the illusion emergence, as well as efferent greater activity of the opposite angle orientation detectors.
theory that describes a dynamic interaction of the responses Such activity can cause short-termed distortion of the line
on stimulus perception (Honda, 1985, 1990). However, in length in the same way as actual angles do in the standard
their meta-analysis Bruno et al. (2010) showed that the size ML illusion.
of the illusion effect on saccades showed a large variability, The evidence suggests that the underlying cause of the
ranging from 30% to less than 10%. This indicates that there ML illusion does not lie in multiple processes, as stated in the
are other factors modulating the size of the illusion besides theory of perspective/depth (Gregory, 1963; Fischer, 1967).
mere responses mode (perceptual or motor). Moreover, the As depth theory has been proved as an unsuitable explana-
theory of eye movement, and occulomotory feedback is ap- tion for the ML illusion, later findings (Dragoi & Lockhead,
plicatory in this research because of stimuli presentation, 1999) suggest a population model of orientation detectors in
where participants were instructed to gaze at fixation cross, the visual cortex that explain their obtained results for ML
and afterwards lines were presented in tachitoscopic man- illusion as a byproduct of the orientation and distance effect
ner. of long-range horizontal cortical connections. Their starting
The effect of negative ML illusion elicited in this study point is the idea that human visual system decomposes an
points to several things that might prove crucial for a better image using local filters tuned for stimulus features, such as
understanding of the ML illusion. First of all, the negative spatial frequency or orientation. Furthermore, psychologi-

34
 VALERJEV and GULAN, Müller-Lyer illusion in context, Review of Psychology, 2013, Vol. 20, No. 1-2, 29-36

cal evidence suggests that the local filters are not independ- associated illusions, distortion had a positive, not a negative
ent, but they receive input from neighborhood spatial fre- effect. It can be assumed that the illusions that emerge from
quency or orientation filters. Dragoi and Lockhead (1999) the perceptive processes that are based in the impression of
argue that this network of long-range interconnections may depth are not subject to this after effect. Furthermore, the
serve as substrate for context dependency. In other words, ML illusion is influenced by the negative after effect and
context dependency means that the perceived features of the therefore probably does not emerge from the same mecha-
stimulus depend on context surrounding it. In their research nisms as the Ponzo illusion does.
the confirmations for such suggestion was obtained. In conclusion, this study demonstrates the existence of
However, we suggest that feature detectors specialized the negative ML illusion that has the opposite effect when
for angles are the one that may play an important role in compared to the standard ML illusion. This effect is a short-
the ML illusion. These features detectors are activated in termed and weaker one than in the standard version of this
the standard ML illusion and their after effect is active in illusion, but is still significant. The underlying cause of the
the negative ML illusion. These specialized cells can be re- negative ML illusion is probably some kind of negative af-
ferred to as neural filters. However, the mechanism how the ter effect, and it is possible that the direction of this illusion
standard ML illusion changes into the negative one is still emerges because of the fatigue of the cells specialized for
on the level of speculation and requires further systematic angles of specific orientation. If this is the case, than it can
investigation. be assumed that the same process is the underlying cause of
Furthermore, it is possible that the position of the angle the standard version of this illusion, i.e., this illusion could
in the visual field is perceived by an imaginary placed posi- be the result of the activation of the cells that are specialized
tion that is located inside the angle, rather than one placed for selective angles. The same explanation might be appli-
on the starting point of the angle. This assumption is in line cable to variation of the ML illusions when the central line
with the claim of uncertainty that arouses in visual system is not placed between angles, but between arches, squares,
due to bias (Fermüller & Malm, 2004). However, this as- or circles.
sumption also requires further investigation. If this proves
correct, the imaginary position in the visual field that rep-
resents the angle can make a move in the angle focus on REFERENCES
the line in the inward condition and behind the line in the
Bruno, N., Bernardis, P., & Gentilucci, M. (2008).Visually
outward condition. Also, this might be the reason why the
guided pointing, the Müller-Lyer illusion, and the func-
inward fins make the line seem shorter, and the outward fins
tional interpretation of the dorsal-ventral split: Conclu-
make it seem longer. This proposed explanation is consist-
sions from 33 independent studies. Neuroscience and
ent with the confusion theory of the ML illusion emergence,
Biobehavioral Reviews, 32, 23–437.
which suggests that the perceptual system miscalculates the
location of the arrowhead vertex, displacing it toward the Bruno, N., Knox, P., & de Grave, D. (2010). A metaanaly-
concave side (Chiang, 1968). Furthermore, Chiang’s (1968) sis of the effect of the Müller-Lyer illusion on saccadic
theory applies to patterns in which lines running close to- eye movement: No general support for a dissociation
gether affect one another. Two close lines influence each between perception and oculomotor action. Vision Re-
other’s location and become one when the sum of their dis- search, 50, 2671-2682.
tribution of activation on retina forms a single peak. This Caelli, T. (1977). Is perceived length affected by interac-
leads to an overestimation of acute angles, and provides ex- tions between orientation detectors? Vision Research,
planation of the ML illusion, as well as the Poggendorff and 17, 837–841.
Zöllner illusion. Confusion theory can explain most of the Chiang, C. (1968). A new theory to explain geometrical il-
known variations of the ML illusion, e.g., when the lines are lusions produced by crossing lines. Perception and Psy-
bordered with squares of circles. In the last two examples of chophysics, 8, 174-176.
the ML illusion the position of the bordered object is placed
Corso, J. H. (1967). The experimental psychology of sen-
within that object.
sory behavior. New York: Holt, Rinehart & Winston.
One should not neglect the possibility that the ML illu-
Dragoi, V., & Lockhead, G. R.. (1999). Context-Dependent
sion emerges because of a more general figural after effect
Changes in Visual Sensitivity Induced by Muller-Lyer
as discussed by Köhler and Wallach (1944) or Osgood and
Stimuli. Vision Research, 39, 1657-1670.
Heyer (1952; in Osgood, 1953). But then it remains unclear
when considering all „temporally disassociated” illusions Erlebacher, A., & Sekuler, R. (1969). Explanation of the
only the ML illusion shows a pattern of this negative effect. Müller-Lyer illusion: Confusion theory examined.
On the other hand, the Ponzo and Hering illusion are two Journal of Experimental Psychology, 80, 462-467.
illusions that most probably emerge because of the expe- Fermüller, C., & Malm, H. (2004). Uncertainty in visual
rience of perspective, i.e., depth. In the phenomenological processes predicts geometrical optical illusions. Vision
effect verification of the Ponzo and Hering temporally dis- Research, 44, 727-749.

35
 VALERJEV and GULAN, Müller-Lyer illusion in context, Review of Psychology, 2013, Vol. 20, No. 1-2, 29-36

Fischer, G. H. (1967). A common principle relating to the Kaufman, L., & Richards, W. (1969). Spontaneous fixation
Müller-Lyer and Ponzo illusion. American Journal of tendencies for visual forms. Perception and Psycho-
Psychology, 80, 626-631. physics, 5, 85–88.
Honda, H. (1985). Spatial localization in saccade and pur- Kawabata, N. (1976). Mathematical analysis of the visual
suit-exe-movement conditions: A comparison of per- illusion. IEEE Transactions on Systems, Man and Cy-
ceptual and motor measures. Perception and Psycho- bernetics, 6, 818–824.
physics, 38, 41-46.
Kobatake, E., & Tanaka, K. (1994). Neuronal selectivities to
Honda, H. (1990). The extraretinal signal from the pursuit- complex object features in the ventral visual pathway of
eye-movement system: Its role in the perceptual and the the macaque cerebral cortex. Journal of Neurophysiol-
egocentric localization systems. Perception and Psy- ogy, 71, 856–867.
chophysics, 48, 509-515.
Köhler, W., & Fishback, J. (1950a). The destruction of the
Ito, M., & Komatsu, H. (2004). Representation of Angles
Müller-Lyer illusion in repeated trials: An examina-
Embedded within Contour Stimuli in Area V2 of Ma-
caque Monkeys. The Journal of Neuroscience, 24(13), tion of 2 theories. Journal of Experimental Psychology,
3313–3324. 40(2), 267–281.
Ganz, L. (1966). Mechanisms of the F.A.E.’s. Psychological Köhler, W., & Fishback, J. (1950b). The destruction of the
Review, 73, 128-150. Müller-Lyer illusion in repeated trials: Satiation pat-
terns and memory traces. Journal of Experimental Psy-
Ginsburg, A. P. (1984). Visual form perception based on bi-
chology, 40(3), 398–410.
ological filtering. In L. Spilman & G. R. Wooten (Eds.),
Sensory experience, adaptation and perception (pp. 53- Köhler, W., & Wallach, H. (1944). Figural after-effects: An
72). Hilsadale, NJ: Erlbaum. investigation of visual processes. Proceedings of the
Ginsburg, A. P. (1986). Spatial filtering and visual form per- American Philosophical Society, 88, 269-357.
ception. In K. R. Boff, L. Kaufman, & J. P. Thomas Mather, G., Verstraten, F., & Anstis, S. (1998). The motion
(Eds.), Handbook of perception and human perfor- after effect: A modern perspective. Cambridge, Mass:
mance, 34 (pp. 1-41). New York: Wiley. MIT Press.
Greene, E., & Nelson, B. (1997).Evaluating Müller-Lyer ef- McCollough, C. (1965). Adaptation of edge-detectors in the
fects using single fin-set configurations. Perception & human visual system. Science, 149, 1115–1116.
Psychophysics, 59(2), 293-312.
Osgood, C. E. (1953). Method and theory in experimental
Gregory, R. L. (1963). Distortion of visual space as inappro-
priate constancy scaling. Nature, 199, 678-680. psychology. New York: Oxford University Press.
Gregory, R. L. (l968). Perceptual illusions and brain mod- Pressey, A. W. (1970). The assimilation theory applied to
els. Proceedings of Royal Society, B 171, 179-296. a modification of the Müller-Lyer illusion. Perception
and Psychophysics, 8, 411–412.
Howe, C. Q., & Purves, D. (2004). The Müller-Lyer illusion
explained by the statistics of image-source relationship. Thompson, A. A., & Westwood, D. A. (2007). The hand
Proceedings of the National Academy of Sciences, 102, knows something what eye does not: Reaching move-
1234-1239. ments resist the Müller-Lyer illusion whether or not the
Hubel, D. H., & Wiesel, T. N. (1959). Receptive fields of target is foveated. Neuroscience Letters, 426(2), 111-
single neurons in the cat’s striate cortex. Journal of 116.
Physiology, 148, 574–591. Walker, E. H. (1973). A mathematical theory of optical illu-
Judd, C. H. (1902). Practice and its effects on the perception sions and figural aftereffects. Perception and Psycho-
of illusions. Psychological Review, 9(1), 27-39. physics, 13, 467–486.

36