doi:10.1016/j.cub.2006.11.046


Current Biology Vol �7 No �
R8

fusiform gyrus. This relatively 
posterior locus of injury 
buttresses the view of acquired 
prosopagnosia as a disorder 
of visual processing. Although 
people with developmental 
prosopagnosia have no obvious 
lesion in the fusiform gyrus, 
recent work suggests that there 
is a subtle alteration of the white 
matter connections in this region 
of the brain. Neurons in this area 
have been shown to respond 
vigorously, and selectively, to the 
visual image of faces.

Aren’t faces just harder to 
recognize? Or the subject 
of greater experience? In 
fact, several such alternative 
explanations have been proposed. 
Many are variants on the 
‘individuation’ account, which 
holds that special processing 
for faces is not specific to 
this visual stimulus category. 
Proponents of this view note that 
faces are a special category of 
object for which we constantly 
identify individual exemplars, 
whereas such ‘subordinate-level’ 
processing is seldom necessary 
for recognition of other classes of 
object. In line with this view, some 
prosopagnosics demonstrate 
impairments in individuation 
of other object classes, such 
as specific animals. But this 
within- category impairment 
may reflect damage to adjacent 
but separate cortical areas. 
Furthermore, some experiments 
have still found specific 
impairment for faces versus other 
object classes in prosopagnosic 
subjects when the difficulty of 
individuation for these stimulus 
sets is matched. The issue of 
the specificity of the deficit in 
prosopagnosia remains an active 
area of debate. 

Where can I find out more?
Barton, J.J. (2003). Disorders of face 

perception and recognition. Neurology 
Clinic 21, 52�–548.

Moscovitch, M., Winocur, G., and Behrmann, 
M. (�997). What is special about face 
recognition? Nineteen experiments on 
a person with visual object agnosia and 
dyslexia but normal face recognition. J. 
Cognitive Neurosci. 9, 555–604.

Department of Neurology, University 
of Pennsylvania, Philadelphia, 
Pennsylvania �9�04, USA.  
E-mail: aguirreg@mail.med.upenn.edu
Correspondences

Mediaeval 
artists: Masters 
in directing the 
observers’ gaze

Ute Leonards1, Roland 
Baddeley1, Iain D. Gilchrist1,  
Tom Troscianko1, Patrick Ledda2 
and Beth Williamson3

The gold leaf in early 
Renaissance paintings such as 
Duccio’s “The Annunciation” 
(�3��) appears to glow when lit 
by candles as the artist would 
have expected. Subjectively, 
the candle-lit painting appears 
drastically different from the 
same painting illuminated with 
diffuse daylight typical of modern 
art galleries. By analysing 
the density of observers’ eye 
fixations when looking at this 
painting under these two lighting 
conditions, we found objective 
differences in where in the 
painting observers attended: 
specifically, the glow of the gold 
induced shifts in fixations to 
symbolically important regions of 
the painting.

To investigate the effects of 
illumination on the perception 
of mediaeval paintings, we first 
constructed an area of gold 
leaf, using the techniques of the 
Italian Renaissance painters, 
and measured its reflectance 
properties. We also measured 
the spectral characteristics 
of beeswax candle light and 
daylight illumination. This 
allowed us to mimic the effects 
of such light sources on gold. 
On the basis of these data, 
we then used photorealistic 
computer graphics to render 
a high resolution digital scan 
of the original painting under 
both beeswax candle illuminant 
and daylight illuminant (see 
Supplemental experimental 
procedures in the Supplemental 
data available on-line with 
this issue). In the behavioural 
experiment, we had two groups 
of human participants view one 
of the two rendered pictures 
while we measured their eye 
movements. Figure � shows 
how the nature of the illuminant 
affected where in the picture the 
participants fixated in the picture.

The gold leaf, which is used 
so extensively in paintings of 
this era, creates a dramatic 
glow effect when lit by candles, 
which would have been the 
contemporary illuminant for these 
paintings. This glow effect leads 
the eye to fixate in a different part 
of the image than when ordinary 
diffuse daylight illumination is 
used. In the case of Duccio’s 
Annunciation, observers look 
less at the faces of the Angel 
and the Virgin and more towards 
the Virgin’s hand. One might 
speculate that this is the eyes 
being directed away from the 
faces (looking at them directly 
might have been considered 
irreverent). Alternatively, the 
eyes could have been directed 
towards the Virgin’s hand, which 
she uses to grasp her veil and 
to gather her mantle around 
herself protectively. Her gesture 
has been claimed to reflect 
a state of mind somewhere 
between disquiet and reflection, 
consistent with the early stages 
of the Angelic Colloquy or the 
interaction between the Angel 
and the Virgin [�]. The methods 
used by this great Italian painter 
appear to exploit the effect 
of ‘glowing’ gold to direct the 
viewer’s eyes. 

Our results raise two important 
scientific issues. First, why does 
the subtle induction of ‘glow’ 
have such a significant effect 
on eye movements? Sources of 
illumination are often the most 
perceptually visible regions 
in a visual scene. Traditional 
low- level salience models of 
eye movement control [2] would 
predict a high fixation probability 
of these regions. In the current 
experiment, however, eye 
movements were not directed 
toward the brightest parts of the 
image, such as the gold- covered 
garments or windows under 
candle-light conditions, as 
predicted if fixations were driven 
by either brightness or the global 
effect in which saccades are 
directed to the centre of gravity 



Magazine
R9
Figure �. Lighting conditions can affect the significance of parts of a painting as 
revealed by the pattern of observers’ fixations.

The top two panels show Duccio’s “The Annunciation” rendered either using syn-
thetic diffuse illumination (A), or synthetic candle light (B). When observers view 
such paintings, they make about three eye movements a second, and where they 
choose to fixate will be determined by both the visual characteristics of the picture 
and their interest. (C) The density of eye fixations from observers, estimated using 
a new kernel density estimator specifically designed to estimate the density of eye 
movements (see Supplemental data). The density of fixations to the diffuse illumina-
tion is shown in green; the fixations to the candle light are shown in red, and hence 
regions fixated in both conditions are shown as yellow. As can be seen, the vast 
majority of eye movements were concentrated around the faces of the angel and 
Virgin. (D) The differences in fixation patterns between candle lit and diffusely lit 
paintings, analysed using a spatial permutation test [7], where regions coloured red 
show areas significantly more fixated when viewed using synthetic candle light, and 
regions coloured blue show locations viewed under synthetic diffuse light. As can be 
seen, simply changing the lighting changes the pattern of emphasis from the faces  
to the Virgin’s hand.
of a number of discrete objects 
[3]. Instead, fixations were more 
likely on the less visible hands. 
The apparent avoidance of 
the illuminant fits with recent 
computational studies showing 
that extremes of low-frequency 
brightness predict low fixation 
probability [4] and neuroimaging 
studies demonstrating that the 
visual system shows differential 
activation of light-emitting 
and light-reflecting objects in 
the occipito-temporal cortex 
[5]. Moreover, light sources 
are mostly behaviourally 
unimportant.

Second, by measuring 
fixation density across different 
conditions for the same visual 
scene and using rigorous 
statistical methods developed 
for the analysis of brain images, 
we are now in a position to 
objectively identify the factors 
that determine what is fixated. 
As we have shown here with 
Duccio’s painting, the nature 
of the lighting has the effect of 
significantly altering the way 
in which observers look at 
paintings. The current practice 
in the exhibition of mediaeval 
paintings is to use lighting 
that maximises visibility and 
discriminability [6]. However, 
these criteria might disrupt 
the sophisticated techniques 
available to the artist to direct the 
observer’s gaze. 
Supplemental data
Supplemental data including 
 experimental procedures are available 
at http://www.current-biology.com/cgi/
content/full/�7/�/R8/DC�

Acknowledgments
This work was supported by Reverse 
Engineering the VERtebrate Brain 
(April 2005–March 20�0) EPRSC 
GR/S�9639/0�. We thank Dr. David 
 Saunders and the National Gallery, 
 London, for providing a hyper-spectral 
scan of Duccio’s “Annunciation”.

References
 �.  Baxandall, M. (�972). Painting and 

Experience in Fifteenth-Century Italy. 
(Oxford/NewYork: Oxford University 
Press), pp. 48–56.

 2.  Itti, L., Koch, C., and Niebur, E. (�998). 
A model of saliency-based visual 
attention for rapid scene analysis. IEEE 
Trans. Pattern Analysis Mach. Intell. 20, 
�254–�259.

 3.  Coren, S., and Hoenig, P. (�972). Effect 
of non-target stimuli on the length of 
voluntary saccades. Percept. Motor Skills 
34, 499–508. 

 4.  Baddeley, R.J., and Tatler, B.W. (2006). 
High frequency edges (but not contrast) 
predict where we fixate: A Bayesian 
system identification analysis. Vision Res. 
46, 2924–2833. 

 5.  Leonards, U., Troscianko, T., Lazeyras, F., 
and Ibanez, V. (2005). Cortical distinction 
between the neural encoding of objects 
that appear to glow and those that do 
not. Cogn. Brain Res. 24, �73–�76.

 6.  Linhares, J.M.M., Carvalhal, J., 
Nascimento, S.M.C., Regalo, M.H., and 
Leite, M.C.V.P. (2005). Estimating the best 
illuminant for art paintings by computing 
chromatic diversity. Perception 34 
(Suppl), 88.

 7.  Nichols, T.E. and Holmes, A.P. (2002). 
Nonparametric permutation tests for 
functional neuroimaging: a primer with 
examples. Human Brain Mapping 15, 
�–25. 

�Department of Experimental 
Psychology, 2Department of Computer 
Science and 3Department of History 
of Art, University of Bristol, �2a Priory 
Road, Bristol BS8 �TU, UK.  
E-mail: ute.leonards@bristol.ac.uk

The editors of Current Biology 

welcome correspondence on 

any article in the journal, but 

reserve the right to reduce 

the length of any letter to be 

published. All Correspondence 

containing data or scientific 

argument will be refereed. 

Queries about articles for 

 consideration in this format 

should be sent by e-mail to 

cbiol@current-biology.com

http://www.current-biology.com/cgi/content/full/17/1/R8/DC1
http://www.current-biology.com/cgi/content/full/17/1/R8/DC1
mailto:ute.leonards@bristol.ac.uk
mailto:cbiol@current-biology.com

	Mediaeval artists: Masters in directing the observers’ gaze
	Supplemental data
	Acknowledgments
	References