Perception, 2005, volume 34, pages 847 ^ 856

DOI:10.1068/p5273

The roles of inducer size and distance in the Ebbinghaus

illusion (Titchener circles)

Brian Robertsô, Mike G Harris, Tim A Yates

School of Psychology, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK;
e-mail: [email protected]
Received 23 April 2004, in revised form 19 October 2004; published online 15 June 2005

Abstract. Although the Ebbinghaus illusion is commonly used as an example of a simple

size-contrast effect, previous studies have emphasised its complexity by identifying many factors
that potentially influence the magnitude of the illusion. Here, in a series of three experiments,
we attempt to simplify this complexity. In each trial, subjects saw a display comprising, on one
side, a target stimulus surrounded by inducers and, on the other, an isolated probe stimulus.
Their task was to indicate whether the probe appeared larger or smaller than the target. Probe
size was adjusted with a one-up, one-down staircase procedure to find the point of subjective
equality between probe and target. From these experiments, we argue that the apparent effects of
inducer size are often confounded by the relative completeness of the inducing surround and
that factors such as the similarity of the inducers and target are secondary. We suggest a simple
model that can explain most of the data in terms of just two primary and independent factors:
the relative size of the inducers and target, and the distance between the inducers and the target.
The balance between these two factors determines whether the size of the target is underestimated
or overestimated.

1 Introduction
The Ebbinghaus illusion (Ebbinghaus 1902; see Wundt 1898, page 142, figure 49) was
popularised in the English-speaking world by Titchener (1901, page 169, figure 3) as part
of a laboratory exercise on visual illusions in his handbook of experimental psychology.
In this illusion, the apparent size of a central target is changed by a ring of surround-
ing inducers. Figure 1a illustrates its most popular form, as it most often appears in
general textbooks. In this form, it is typically used to illustrate a simple size-contrast
effect, in which large inducers make the target appear smaller whilst small inducers
make it appear larger (eg Obonai 1954). Regrettably, like many well-established geo-
metrical illusions, the Ebbinghaus illusion is not quite so simple and probably arises
from a number of processes, all contributing a small amount to a large overall effect.
Likely factors include:
. A size-contrast effect between target and inducers of the type described above.
For a given distance between the target and the inducers, the magnitude of the
Ebbinghaus illusion is governed by the relative size and number of the inducers
(eg Massaro and Anderson 1971).
. The distance between the central target and the inducers. Depending on this
distance, perceived space might be compressed or expanded, or neighbouring con-
tours might attract or repel each other, resulting in a distortion of the perceived
size of the elements. If contours are important, then only the inner or outer edge
of the inducing ring might be relevant, or each might make separate contributions
from the same spatial function (eg Oyama 1960; Weintraub and Schneck 1986).
This type of spatial interaction might also account for the Delboeuf illusion
(Delboeuf 1892), which is closely related to the Ebbinghaus illusion (eg Zigler 1960;
Girgus et al 1972). In this illusion, illustrated in figure 1b, the central target appears
larger when the inducing ring is close and smaller when it is distant.
ô Current address: Psychology, School of Life and Health Sciences, Aston University, Aston Triangle,
Birmingham B4 7ET, UK; e-mail: [email protected]
848 B Roberts, M G Harris, T A Yates

(a)

Figure 1. (a) The Ebbinghaus illusion. (b) The

Delboeuf illusion.
(b)

. The completeness of the inducing annulus. In most studies of the Ebbinghaus

illusion (eg Girgus et al 1972) the number of inducers is kept constant while their
size is varied, so that smaller inducers provide a less complete and possibly less
effectual ring. Removal of this confound may be particularly important in under-
standing the relationship between the Ebbinghaus and Delboeuf illusions.
. The similarity of the inducers to themselves and to the target. Similarity might
range from simple structural features such as luminance, polarity, and shape
(eg Jaeger and Grasso 1993; Ehrenstein and Hamada 1995), to more complex
conceptual relationships when recognisable and meaningful elements are used
(eg Coren and Miller 1974; Coren and Enns 1993). Note, however, that there is
some recent evidence that such apparent semantic relationships may be explicable
in simpler structural terms, namely similarity of the perimeters of the target and
the inducers (Choplin and Medin 1999). Furthermore, Rose and Bressan (2002)
have shown that the similarity effect does not generalise across shapes; rather the
magnitude of the Ebbinghaus illusion depends upon the shapes of both the target
and the inducers.
Comprehensive reviews of research on the Ebbinghaus illusion until the 1970s can be
found in Robinson (1972) and in Coren and Girgus (1978). Since then, the Ebbinghaus
illusion has been used primarily as a research tool (eg Aglioti et al 1995; Kwok and
Braddick 2003) to provide evidence for separate representational processing streams
(Milner and Goodale 1995). In the few recent papers on the illusion per se, researchers
have tended to isolate and study one or sometimes two of the potential factors listed
above. This work has provided useful information about some of the pieces of the puzzle,
but leaves the impression that the overall jigsaw may be too complex to solve. Our
approach here is rather different; we seek the simplest possible model of the illusion
that, though not necessarily complete, provides a simple, coherent, and plausible frame-
work encompassing all of the principal factors.
In experiment 1 we confirm that both inducer size and inducer distance are important
in determining the magnitude of the illusion and we go on, in experiments 2 and 3,
to examine each of these individual factors in detail, and the roles of similarity and
Simple model of the Ebbinghaus illusion 849

completeness as additional factors. We develop a simple algebraic model incorporating

only size and distance as explicit, and possibly independent, factors in which the
roles of other factors, such as similarity and completeness, can be understood as
secondary effects.

2 Experiment 1
Girgus et al (1972) reported that varying the relative size of the inducers in the
Ebbinghaus illusion produces changes in the apparent size of the target consistent
with a size-contrast effect, but that increasing inducer distance causes a decrease
in apparent target size irrespective of inducer size. Indeed, these authors were the first
to demonstrate for this illusion that inducers smaller than the target can reduce its
apparent size. We began by attempting to replicate these basic effects. Since pilot work
confirmed that filled and open figural elements produce qualitatively and quantitatively
similar results, we chose to use filled elements throughout.
2.1 Method
Stimuli were generated and displayed on a standard Macintosh computer with a screen
resolution of 10246768 pixels and a frame rate of 75 Hz. A typical stimulus is shown
in figure 2. Target and probe stimuli were drawn with standard QuickDraw routines
without anti-aliasing. On each trial, the subject was instructed to fixate the central
cross and the stimulus was presented until the subject indicated by a mouse click
whether an isolated probe stimulus appeared larger or smaller than a target stimulus
surrounded by eight equally spaced inducers. The probe, target, and inducers were
white (139 cd mÿ2 ) against a mid-grey background (71 cd mÿ2 ) subtending 54.25 deg
horizontally by 40.75 deg vertically.

Figure 2. Exemplar stimulus used in experiments 1 ^ 3.

The centres of the probe and target stimuli were separated horizontally by 27.00 deg.
On each trial, the probe was randomly assigned to either the left or right of the dis-
play. Target size was fixed at a radius of 0.95 deg and probe size was varied from trial
to trial under the control of a standard one-up, one-down staircase. Five different
inducer sizes (0.32, 0.63, 0.95, 1.27, and 1.58 deg radius) and three different inducer
distances (2.64, 4.22, and 8.45 deg, measured from the centre of the target to the centre
of the inducers) were investigated. In each session, five of the resulting fifteen con-
ditions were chosen randomly without replacement and each was presented by two
staircases, one beginning with the probe a random amount smaller than the target, the
other beginning with the probe a random amount larger than the target. Each session
also included two staircases investigating a control condition in which the target was
presented without inducers.
 850 B Roberts, M G Harris, T A Yates

The twelve staircases within a session were randomly interleaved from trial to trial.
Each began with a probe-diameter step size of 0.211 deg (4 pixels). Step size was halved
at each reversal down to a minimum step size of 0.053 deg (1 pixel). Once this minimum
step size had been reached, each staircase continued for a further ten reversals and the
average of these ten reversal points was taken as an estimate of the point of subjective
equality (PSE). Staircases were removed from the pool as soon as ten reversals had
been completed with them at the minimum step size.
The experiment was performed in a darkened room. Three postgraduate students
took part. Each had normal vision and was unaware of the purpose of the study.
They viewed the display binocularly from a distance of 27.5 cm, with their heads
supported by a chin-and-forehead rest. Each subject completed one practice session
and six experimental sessions, providing four estimates of illusion magnitude in each
of the fifteen experimental conditions and twelve estimates in the control condition.
2.2 Results and discussion
Figure 3 shows, separately for each inducer distance, the average illusion magnitude,
expressed as the difference in probe-target radius, as a function of inducer radius.
The results were analysed with a two-factor repeated-measures ANOVA. In this and all
subsequent statistical analyses, the data for each subject were expressed as the difference
from the perceived size in the control condition.
near inducers
medium distance
0.05 1.05
far inducers
control (no inducers)
Relative illusion magnitude

0.00 1.00
Illusion magnitude=deg

ÿ0.05 0.95

ÿ0.10 0.90

ÿ0.15 0.85

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

Inducer radius=deg
Figure 3. Results of experiment 1: illusion magnitude as a function of inducer radius and distance.
Error bars show 95% confidence limits for the near and far inducers; those for the medium inducers
are similar in size but have been omitted for clarity. Vertical dotted line indicates target radius.

The main effect of inducer size was very reliable (F4, 8 ˆ 7:90, p ˆ 0:007). However,
the most obvious finding is that, in all the experimental conditions, the effect of the
inducers was to reduce the apparent size of the target. This effect clearly emphasises
the oversimplification of the Ebbinghaus illusion as primarily a size-contrast effect.
Inducer size has a relative effect so that, if large and small inducer conditions were
paired, as in figure 1a, it would appear that small inducers increased the apparent
size of the target and vice versa. In fact, on the basis of figure 3, it would be more
appropriate to conclude that inducers generally reduce apparent target size and that small
inducers are simply less effectual in doing this.
Simple model of the Ebbinghaus illusion 851

Girgus et al (1972) did not specify size and distance in degrees and did not provide
the viewing distance required to compute them. However, our data can be compared
with theirs by expressing size in terms of the inducer/target ratio of radii and distance
in terms of the ratio of the target radius and the distance between the target and
inducer centres. Such a comparison would lead one to predict from the data of Girgus
et al (1972) that the target should have been seen as bigger than its actual size for
the smallest and closest inducers that we used. The most likely factor underlying this
discrepancy is the greater number of inducers in the surround in the current study
(8 versus 4). This issue is considered further in experiments 2 and 3.
Inducer distance also has an effect, so that the reduction in target size tends to
be more pronounced at greater distances (Girgus et al 1972; Jaeger and Grasso 1993).
The main effect of inducer distance approached reliability (F2, 4 ˆ 4:97, p ˆ 0:082). There
is also some suggestion that the effect of inducer size is reduced at greater distances,
but the interaction between size and distance was not reliable (F8, 16 ˆ 1:26, p ˆ 0:330).

3 Experiment 2
In experiment 1, there were always 8 inducers in the experimental conditions so that,
at smaller inducer sizes, the surround was less complete. The studies of Girgus et al
(1972) and Jaeger and Grasso (1993) also used a fixed number of inducers. This con-
found was investigated in experiment 2 by increasing the number of smaller inducers
so that, in each experimental condition, the inducers always occupied a fixed propor-
tion of the surround circumference. The inducer distance was fixed at the largest
value used in experiment 1, but the range of inducer sizes was extended to encompass
both smaller and larger radii. In addition, as a simple manipulation of similarity,
conditions in which the target and inducer luminance polarity were the same or differ-
ent were compared. This approach has the advantage of allowing salient differences
between the target and inducers without introducing differences in shape (see eg Coren
and Miller 1974) or in lightness contrast with the background (see eg Jaeger and Grasso
1993) that might also affect the results.
3.1 Method
The display, viewing conditions, and procedure were similar to those of experiment 1.
Target size was fixed at 1.27 deg radius and inducer distance was fixed at 8.45 deg.
There were thirteen variations of inducer size. The first nine consisted of 8 inducers
with radii of 0.05, 0.32, 0.63, 0.95, 1.27, 1.58, 1.90, 2.22, or 2.53 deg. In the remain-
ing four conditions, the number of smaller inducers was increased so that a constant
proportion (about 0.75) of the surround circumference was occupied (12860.053 deg,
6460.316 deg, 3260.633 deg, and 1661.267 degöall dimensions in degrees). In addition,
a control without inducers was included. The resulting fourteen conditions were pre-
sented with same (white ^ white, black ^ black) and different (white ^ black, black ^ white)
target ^ inducer polarities against the mid-grey background. The luminance of the black
stimuli was 2 cd mÿ2 ; the luminance of the white stimuli and grey background was the
same as in experiment 1.
Seven randomly chosen conditions were investigated in each session, each controlled
by two staircases. Three postgraduate students and one of the authors (MGH) completed
two full replications of the experiment, giving four estimates of the PSE in each of the
twenty-eight conditions.
3.2 Results and discussion
The results of the 8-inducer conditions are similar to those of experiment 1. At this
inducer distance, all inducers make the target appear smaller, and smaller inducers are
less effectual in doing this (F8, 24 ˆ 19:51, p 5 0:0001). There is some suggestion that
inducers are more effectual when they have the same luminance polarity as the target,
852 B Roberts, M G Harris, T A Yates

though neither the main effect of polarity (F1, 3 ˆ 1:59, p ˆ 0:296) nor its interaction
with inducer size (F8, 24 ˆ 1:32, p ˆ 0:281) is reliable. Although luminance polarity is a
very simple manipulation, its effect on the general appearance of the stimulus is quite
profound. Nonetheless, it is clearly not a primary factor in understanding the basic
Ebbinghaus illusion.
The most interesting new result is shown in the conditions in which the number
of inducers was increased to compensate for their decreased size. Since there was no
main effect or interaction involving polarity, these conditions are averaged across
polarity in figure 4 for clarity. Note that the four conditions shown occupy the same
proportion of the surrounding annulus as the 8-inducer condition with a radius of
2.53 deg (mean, pooled across polarity, indicated by the open square in figure 4). The
main effect of inducer size was not reliable (F3, 9 ˆ 1:48, p ˆ 0:285); surrounds contain-
ing small inducers are, by and large, as effectual in reducing apparent target size as
those containing larger inducers. It seems that completeness of the surround, unlike
inducer ^ target polarity, is an important factor that is likely to be misinterpreted as a
size effect in many studies of the Ebbinghaus illusion.
0.08 1.08 8 inducers, same target ±
inducer polarity
0.04 1.04 8 inducers, different
target ± inducer polarity
number of inducers
0.00 1.00 adjusted to occupy
constant proportion of
Relative illusion magnitude

the annular circumference

Illusion magnitude=deg

ÿ0.04 0.96 (averages of the two

polarity conditions)
average of same and dif-
ÿ0.08 0.92 ferent polarity 8-inducer
conditions in largest
inducer condition
ÿ0.12 0.88 control (no inducers)

ÿ0.16 0.84

ÿ0.20 0.80

ÿ0.24 0.76
ÿ0:5 0 0.5 1.0 1.5 2.0 2.5 3.0
Inducer radius=deg
Figure 4. Results of experiment 2: illusion magnitude as a function of inducer radius and distance.
Error bars show 95% confidence limits. The fine curve is the best fit of the model developed and
described in experiment 3 to the (solid and open) squares; see text for details. Vertical dotted line
indicates target radius.

4 Experiment 3
In this experiment, we attempted to disentangle the effects of surround completeness
and inducer size and to study the effect of inducer distance in more detail. Only two
inducer sizes were used öone was smaller than and the other was equal to the size of
the target. This upper limit on inducer size allowed the range of inducer distances
tested to be extended to smaller values than were used in experiment 1.
4.1 Method
The display, viewing conditions, and procedure were similar to those of experiments
1 and 2. Target size was fixed at 1.27 deg radius. Three types of inducer were studied.
The first was a complete annulus, similar to that used in the Delboeuf illusion (see
figure 1b) with a thickness of 0.63 deg (equivalent to continuously overlapping inducers
Simple model of the Ebbinghaus illusion 853

with a radius of 0.32 deg). As in experiments 1 and 2, the other two conditions used
circular inducers, here with radii of 0.32 or 1.27 deg. The inducers were presented at
distances of 1.90, 2.53, 2.85, 3.17, 3.80, 4.44, 5.07, or 8.45 deg, although, for the larger
inducers, the shortest distance could not be included. At each distance tested, the
number of inducers was adjusted so that, in all cases, they occupied an approximately
constant proportion (about 0.75) of the surround circumference (14, 19, 21, 24, 28, 33,
38, and 63 for the small; 5, 5, 6, 7, 8, 9, and 16 for the larger inducers). Luminance
polarity (target ^ inducer same or different) and control (no inducer) conditions were
also included, giving a total of forty-eight conditions.
Eight randomly chosen conditions were investigated in each session, each driven
by two staircases. Three postgraduate students and one of the authors (MGH) each
completed two full replications of the experiment, providing four estimates of the PSE
in each condition.
4.2 Results and discussion
A three-factor, repeated-measures ANOVA, omitting the smallest inducer distance,
revealed reliable main effects of inducer size/type (F2, 6 ˆ 192:12, p 5 0:001) and dis-
tance (F6, 18 ˆ 43:97, p 5 0:001), and a reliable interaction between them (F12, 36 ˆ 2:40,
p ˆ 0:021). The main effect of polarity, both of its first-order interactions, and the second-
order interaction, was not reliable. A second three-factor, repeated-measures ANOVA,
on only the complete annulus and smaller inducers at all 8 distances, confirmed the
reliable main effect of inducer distance (F7, 21 ˆ 20:40, p 5 0:001). However, here the main
effect of inducer type was not reliable (F1, 3 ˆ 1:94, p ˆ 0:258), although the interaction
between distance and type approached reliability (F7, 21 ˆ 2:35, p ˆ 0:061). Again, none
of the terms involving polarity was reliable.
Since, as in experiment 2, polarity produced no reliable main effects or interactions,
data were pooled across the same and different conditions (see figure 5). As confirmed
by the ANOVAs, the results for the complete annulus and the small inducers are very
similar. At distances greater than about 3.5 deg, the effect of the surround is to reduce
apparent target size. However, at close distances, the surround has the opposite effect
and tends to increase apparent target size (Oyama 1960; Girgus et al 1972). The larger
(same size as the target) inducers reduce apparent target size at all distances.
As in experiment 2, there is relatively little effect of inducer size per se at large
distances, because the number of inducers has been adjusted so that they occupy a
similar proportion of the surround circumference in all conditions. However, despite
this compensation, size does seem to matter at shorter distances, where larger inducers
consistently decrease apparent target size but smaller inducers and complete annuli con-
sistently increase apparent target size.
What is the simplest possible model of these findings? First, comparison of the
effects of complete annuli and small inducers strengthens the suggestion (eg Zigler
1960; Girgus et al 1972) that the Delboeuf illusion and this configuration of the
Ebbinghaus illusion are different versions of the same effect.(1) This effect can best be
described as an interaction between the target and inducer that depends upon the
distance between them. Visual space may be distorted or neighbouring contours may
attract or repel each other (eg Weintraub and Schneck 1986) so that, when target and
inducer are close, the target appears larger and, when they are distant, the target
appears smaller. We can obtain a good estimate of the spatial form of this effect by
combining the data from the complete annuli and small inducers. The bold solid line

(1) Note that we have confined our consideration of the Delboeuf illusion to the effect of the

annulus (inducer) on the perception of the inner circle (target). In fact, the inner circle will also
have affected the perception of the annulus (eg Oyama 1960). A similar bidirectional effect may
also apply to the inner circle and surrounding inducers in the Ebbinghaus illusion.
 854 B Roberts, M G Harris, T A Yates

0.15 1.15 complete annular inducer

(the Delboeuf illusion)
small circular inducers
0.10 1.10 large circular inducers
control (no inducers)

Relative illusion magnitude

0.05 1.05
Illusion magnitude=deg

0.00 1.00

ÿ0.05 0.95

ÿ0.10 0.90

ÿ0.15 0.85
1 2 3 4 5 6 7 8 9
Inducer distance=deg
Figure 5. Results of experiment 3: Illusion magnitude as a function of inducer distance. Note
that the number of inducers has been adjusted so that large and small elements occupy approx-
imately the same (0.75) proportion of the annular circumference. Error bars show 95% confidence
limits. Modelling: bold solid line, illusion magnitude ˆ ÿ 0:7883 ‡ 0:37616 exp ( ÿ x=2:3076), where
x ˆ inducer distance; r ˆ 0:97. Fine solid line: best linear fit of the above function, f(x), to the data
for large inducers [magnitude ˆ a ‡ b f(x), a ˆ ÿ0:06913, b ˆ 0:53382, r ˆ 0:90]. Fine dashed line
(almost obscured by fine solid line): example of a more complex model in which inner and outer
perimeters of the large inducers make separate contributions from the above spatial function f(x),
and the contribution of each is weighted by circumference. The improvement in fit is marginal
(r ˆ 0:91), suggesting that such subtleties do not offer a more compelling explanation of the illusion.

in figure 5 shows the best least-squares fit to these combined data of a simple decay
function of the form a ‡ b exp ( ÿ x=c), where x is the inducer distance. This provides
a good, and very convenient, account of the data (r ˆ 0:97).
Can this simple spatial function also account for the effects of inducer size? For
example, since larger inducers inevitably produce a larger overall figure for a fixed
centre-to-centre distance, are the effects observed with larger inducers just a horizontally
shifted version of those observed with smaller inducers? Or perhaps the inner and
outer edges of the inducer each contribute a separate distortion from the same spatial
function and the resulting illusion is the sum of these contributions. However, given
that the basic spatial interaction seems to be exponential in form (see Appendix),
either of these accounts could in practice produce a change only in the amplitude of
the function rather than a change in shape or vertical position. It is clear from figure 5
that this could not provide a good account of the data for the larger inducers, which
also seem to require a downward vertical shift. Nonetheless, as demonstrated by the
fine solid line in figure 5, the important point is that the data for the large inducers
are well described by fixing the space constant of the exponential term and simply
scaling and vertically shifting it (r ˆ 0:91). This confirms that the same basic spatial
function can reasonably be used to describe all the results from experiment 3. This
same basic function, appropriately scaled and shifted, also provides a reasonable fit
(r ˆ 0:85) to its equivalent conditions in experiment 2 (square symbols and associated
curve in figure 4), which included different subjects, though here a simple linear function
would fit just as well (r ˆ 0:88).
Simple model of the Ebbinghaus illusion 855

In summary, the data can be understood mainly in terms of just two components.
The first is a pure spatial effect of inducer distance, well described by a decaying
exponential, that results in increased perceived target size when the target and inducers
are close and decreased perceived target size when the target and inducers are distant.
This component alone can account for the Delboeuf illusion and for the Ebbinghaus
illusion with small inducers. The second component is a size effect that modulates
this basic function, shifting it downwards and reducing its amplitude. The reduction
in amplitude can, in principle, be explained as a secondary effect öfor example, as the
sum of separate contributions from the inner and outer edges of the inducers (see
Appendix). As our goal is to find the simplest plausible description, we prefer this inter-
pretation and adopt the stance that the distance and size components are effectually
independent.

5 General discussion
We do not wish to claim that our simple model provides a complete account of the
Ebbinghaus illusion. For example, it seems unlikely that the distance function should
continue to infinity (and much more likely that the sum of the total expansion and
compression should be zero). Nor have we characterised the size term or studied
systematically the effects of similarity. We do believe, however, that we have demon-
strated that an apparently complex jigsaw may be solvable. Taking the simplest possible
approach at every alternative allows us to suggest that:
. The completeness of the inducer ring (surround) is a primary factor and, unless
controlled, may complicate interpretations by masquerading as an apparent effect
of inducer size.
. Once completeness is controlled, a common, exponentially decaying function of
distance can explain not only the general form of the Ebbinghaus illusion but also
the Delboeuf illusion.
. The Ebbinghaus illusion also involves a size-contrast component that acts indepen-
dently of the distance factor, determining only the perceived size of the target in a
particular configuration and not the way that its perceived size varies with inducer
distance. It is not true that small inducers will inevitably lead to overestimates of
target size (ie a simple size-contrast effect) but, rather, that the perceived size
of the target will depend upon both the size and distance of the inducers.
. The polarity of the inducers and target is not a primary factor in determining the
form of the illusion. Even though this single manipulation produces targets and
inducers that are physically very different, we cannot, of course, generalise to all
the ways in which the target and inducer might differ. Indeed, other studies of the
Ebbinghaus and Delboeuf illusions have reported effects of, for example, hue
(eg Oyama 1962), contrast (eg Weintraub et al 1969; Weintraub and Cooper 1972;
Jaeger and Pollack 1977), and shape (eg Rose and Bressan 2002). Nonetheless,
such effects are relatively small and we suspect that the perceived similarity of the
target and inducers is not a major factor in understanding the Ebbinghaus illusion.
Acknowledgment. Part of this work was supported by the British Society for Putting Things Near
To Other Things.
References
Aglioti S, DeSouza J F, Goodale M A, 1995 ``Size-contrast illusions deceive the eye but not the
hand'' Current Biology 5 679 ^ 685
Choplin J M, Medin D L, 1999 ``Similarity of the perimeters in the Ebbinghaus illusion'' Perception
& Psychophysics 61 3 ^ 12
Coren S, Enns J T, 1993 ``Size contrast as a function of conceptual similarity between test and
inducers'' Perception & Psychophysics 54 579 ^ 588
Coren S, Girgus J S, 1978 Seeing Is Deceiving: The Psychology of Visual Illusions (Hillsdale, NJ:
Lawrence Erlbaum)
856 B Roberts, M G Harris, T A Yates

Coren S, Miller J, 1974 ``Size contrast as a function of figural similarity'' Perception & Psychophysics
16 355 ^ 357
Delboeuf J L R, 1892 ``Sur une nouvelle illusion d'optique'' Acadëmie Royale des Sciences, des Lettres
et des Beaux-Arts de Belgique. Bulletins 24 545 ^ 558
Ebbinghaus H, 1902 Grundzu«ge der Psychologie volumes I and II (Leipzig: Verlag von Viet & Co.)
Ehrenstein W H, Hamada J, 1995 ``Structural factors of size contrast in the Ebbinghaus illusion''
Japanese Psychological Research 37 158 ^ 169
Girgus J S, Coren S, Agdern M, 1972 ``The interrelationship between the Ebbinghaus and Delboeuf
illusions'' Journal of Experimental Psychology 95 453 ^ 455
Jaeger T, Grasso K, 1993 ``Contour lightness and separation effects in the Ebbinghaus illusion''
Perceptual and Motor Skills 76 255 ^ 258
Jaeger T, Pollack R H, 1977 ``Effect of contrast level and temporal order on the Ebbinghaus
circles illusion'' Perception & Psychophysics 21 83 ^ 87
Kwok R M, Braddick O J, 2003 ``When does the Titchener Circles illusion exert an effect on
grasping? Two- and three-dimensional targets'' Neuropsychologia 41 932 ^ 940
Massaro D W, Anderson N H, 1971 ``Judgemental model of the Ebbinghaus illusion'' Journal of
Experimental Psychology 89 147 ^ 151
Milner A D, Goodale M A, 1995 The Visual Brain in Action (Oxford: Oxford University Press)
Obonai T, 1954 ``Induction effects in estimates of extent'' Journal of Experimental Psychology 47 57 ^ 60
Oyama T, 1960 ``Japanese studies on the so-called geometrical-optical illusions'' Psychologia 3 7 ^ 20
Oyama T, 1962 ``The effect of hue and brightness on the size-illusion of concentric circles''
American Journal of Psychology 75 45 ^ 55
Robinson J O, 1972 The Psychology of Visual Illusion (London: Hutchinson Education)
Rose D, Bressan P, 2002 ``Going round in circles: Shape effects in the Ebbinghaus illusion'' Spatial
Vision 15 191 ^ 203
Titchener E B, 1901 Experimental Psychology: A Manual of Laboratory Practice volume I (London:
MacMillan)
Weintraub D J, Cooper L A, 1972 ``Coming of age with the Delboeuf illusion: Brightness contrast,
cognition, and perceptual development'' Developmental Psychology 6 187 ^ 197
Weintraub D J, Schneck M K, 1986 ``Fragments of Delboeuf and Ebbinghaus illusions:
Contour/context explorations of misjudged circle size'' Perception & Psychophysics 40 147 ^ 158
Weintraub D J, Wilson B A, Greene R D, Palmquist M J, 1969 ``Delboeuf illusion: Displacement
versus diameter, arc deletions, and brightness contrast'' Journal of Experimental Psychology
80 505 ^ 511
Wundt W, 1898 ``Die geometrisch-optischen Ta«uschungen'' Abhandlungen der Sa«chsischen Akademie
der Wissenschaften zu Leipzig 24 55 ^ 178
Zigler E, 1960 ``Size estimates of circles as a function of size of adjacent circles'' Perceptual and
Motor Skills 11 47 ^ 53

APPENDIX
A sideways shift, s, of an exponential function is equivalent to a change in the magnitude
of the function because:
eÿ…x‡s† ˆ eÿx eÿs
where x is inducer distance.
Less obviously, any model in which the inner and outer edges of the inducing
stimulus evoke separate responses from the same exponential function is also equivalent
to a change in the magnitude of the function:
Ieÿx ‡ Oeÿ…x‡s† ˆ …I ‡ Oeÿs † eÿx
where I is the distance of inner inducer edge; and O is the distance of outer inducer edge.
Note that this property also means that the annular inducer used in experiment 3,
although thicker than ideal, should nonetheless provide an accurate estimate of the
spatial form of the exponential function, if not its magnitude.

ß 2005 a Pion publication

ISSN 0301-0066 (print) ISSN 1468-4233 (electronic)

www.perceptionweb.com

Conditions of use. This article may be downloaded from the Perception website for personal research
by members of subscribing organisations. Authors are entitled to distribute their own article (in printed
form or by e-mail) to up to 50 people. This PDF may not be placed on any website (or other online
distribution system) without permission of the publisher.