D. Bo6et, J. Vauclair/Beha6ioural Brain Research 109 (2000) 143-165
161
tion task of human faces according to sex; the birds
solved the task by using textural information, i.e. sur-
face properties (average intensity of images, and other
properties not yet totally identified, such as colour of
the skin, vertical intensity gradient, local contrast, etc.)
rather than by using shape. The pigeons chose the
textural information despite the fact that the shape
contained information more useful for this task than
texture (a preference which is reversed in humans).
When considering the factors influencing picture
recognition, some experiments can be particularly inter-
esting, such as those in which the only difference be-
tween the presented stimuli is the presence or absence
of the third dimension. For example, Klopfer’s experi-
ment [57] (and see above) compared the effect of 3-D
decoys and 2-D images of ducks: the lack of the third
dimension was sufficient to cause a decrease in atten-
tion or in responses to variation in features. The same
was true for Vandenheede and Bouissou’s [100,101]
experiments; ewes showed fear reactions to a real hu-
man or a human-like model but not to a colour full-size
human slide. Thus, the absence of the third dimension
appears to be the unique factor which could explain the
lack of fear in ewes. In contrast, Ryan and Lea [86]
showed that pigeons could easily discriminate individ-
ual conspecifics when the stimuli were live pigeons, but
not when they were photographs or stuffed pigeons;
what seemed to be important for the pigeons for recog-
nising individuals was not the third dimension, but the
presence of movement or/and vocalisations. Similarly,
Kodric-Brown and Nicoletto [58] presented female fish
with three conditions: (1) live males, (2) live males
behind one-way glass (to prevent any interaction), and
(3) images of males displayed on videotapes. The results
indicate that a live male was more attractive only when
interactions were possible; when interactions were not
possible, a videotaped male was as attractive as a real
male. However, this type of experiment (allowing the
comparison between two situations with or without
3-D) is quite rare and it is generally more difficult to
detect a single factor which could influence the subjects’
behaviour.
At this point, it might be useful to envision that at
least three stages could be considered with respect to
the level of precision and the nature of the relationship
between the object and its picture.
6.3. Stages in the relations between real objects and
their pictures
One can postulate that the first and minimal step for
picture perception implies an ability on the subject’s
part to discriminate one or a few salient visual fea-
ture(s) on the picture (e.g. a form, a colour or any other
relevant information) which is necessary and sufficient
to assess its recognition. Many examples were provided
in this review to indicate that animals belonging to
different phyla may react to a picture as they would
react to the real object (for example by displaying
adequate social responses). Given the very nature of
pictures (e.g. their bi-dimensionality, the fact that they
do not necessarily show all visual cues provided by real
objects, such as depth or motion), the next step would
imply that the subjects establish some correspondence
between objects and their photographed representa-
tions. A criterion for assessing such a correspondence
was suggested by Wilkie et al. [108] (and see above), in
their study on perception and memory for places in
pigeons; this criterion is called ‘transfer of influence’
and refers to the fact that knowledge gained with the
real stimulus (or its picture) affects subject’s reactions
with the picture (or the real object). Again, several
cases were mentioned in this survey in favour of the
existence of such a correspondence and let us now
consider a final example. In an investigation of individ-
ual recognition in budgerigars, Trillmich [96] trained
one bird with colour slides of other budgerigars as
sample stimuli and this discrimination showed generali-
sation to live bird models. Further, discrimination
transfer was also found in the other direction; namely,
with live budgerigars serving as sample stimuli and
picture slides of those same birds as test stimuli.
The final level to be considered is that of the equiva-
lence; in this case, subjects might form a true equiva-
lence between the real object and some or all of the
dimensions presented on the picture. Some criteria can
be proposed to differentiate object-picture correspon-
dence from object-picture equivalence. A quite obvious
criterion for equivalence refers to the ability demon-
strated by a subject for bi-directional discrimination
transfer from real objects to their pictures and from
pictures to real objects. A second and more stringent
criterion would be met if the subject showed an ability
to acknowledge some identity between an object and its
picture when the picture carries only some visual di-
mensions of the real stimulus, or/and that the visual
information is altered in some ways; cross-modality
matching experiments (see above) using not only pho-
tos but drawings, outlines, etc. of the real objects, offer
valuable approaches to test the stability and the limits
of the equivalence established between these connected
aspects of objects.
6.4. The problem of the confusion between objects and
their pictures
An important question related to the issue of picture
recognition concerns the interpretation of the data as
revealing either true recognition or a mere confusion
between a stimulus and its picture. This issue of decid-
ing between the two kinds of processing is especially
relevant in the context of the experiments that investi-