It is still often assumed that learning is not necessary in order to interpret highly iconic pictures. A reason for this belief is that several studies have shown that an in-fant can recognise familiar three-dimensional objects in two-dimensional photo-graphs from a very early age. But, as have been stressed in this text several times al-ready, recognition does not equal understanding pictures as pictures (e.g. DeLoache
13 They seem to have grasped twice as much at the edge of the oval than at the black-and-white ink drawings, but less than at colour ink drawings for example.
& Burns, 1994). If the latter ability is to be demonstrated, the child must show that it is able to relate pictorial information to the real world, but without confusing the two.
In development, perception of pictorial information precedes conceptual under-standing of what a picture is (Pierroutsakos & DeLoache, 2003). One must learn that a picture is both similar and different from what it depicts. Developmental psy-chologists agree that experience leads to the concept of picture, a concept that ac-cording to DeLoache and colleagues includes that a picture has a double nature and that there are culturally appropriate uses of pictures. The former requirement is dubbed the dual-representation hypothesis. “To understand and use a symbol, dual representation is necessary - one must mentally represent both facets of the symbol’s dual reality, both its concrete characteristics and its abstract relation to what it stands for” (DeLoache, 2003). This notion is inspired by the “duality of pictures” á la Gibson (1979) and the “double reality of pictures” from Gregory (1970), both as referred to in Pierroutsakos and DeLoache (2003).14
DeLoache and colleagues assume that it is difficult for the child to hold these two aspects in mind simultaneously: that a picture is an object in itself, and that it is about something else. Does the dual-representation hypothesis explain reference?
No, only what is necessary: an ability to simultaneously entertain two views of the
“symbolic” artefact which can otherwise compete with each other. They believe that the physical aspects can obscure the referential part, which make infants’ interest in depictions as real objects an initial problem for true pictorial competence. This is attributed to a hierarchy of prominence in e.g. Sonesson (1989). In virtue of being in closer connection to the perceiving system some objects are more attention grab-bing than others. A piece of paper, or a model, as objects in themselves can be very prominent indeed in relation to a referent removed in time and/or space, not to say the specific relation between these two entities.
DeLoache first became aware of the problem of dual representation when she found that 3-year-old but not 2.5-year-old children could realise the relation between a scaled down model room and a real room (DeLoache, 1987). The task had been to first observe a hiding event in a scale model, or one being indicated on a picture, and then find the full-sized equivalent to the object in the real room, or a larger scale model. That it was a problem of dual representation became apparent when it was found that 2.5-year old children who failed in the scale-model task could succeed in a similar task when utilising photographs instead (DeLoache, 1987; DeLoache &
Burns, 1994). Photographs have less object properties than do a model room with miniature furniture. The latter affords play and investigation in its own right.
2-year-olds failed also in the photograph condition despite various modifications to simplify the task, forcing the researchers to conclude that “24-month-old subjects did not interpret pictures as representation of current reality” (DeLoache & Burns, 1994). The reason they added “current reality” was that children were surprisingly good at placing toys in a real room in locations that had been pointed out to them in a photograph. These experiments are thus not a black-and-white test of referential understanding of pictures, but about the nature and use of this reference.
14 Haber (1980) calls the same phenomenon the “dual reality of pictures.”
A retrieval task involving a picture or model requires much more of the subject than just realising that there is a relation (DeLoache & Burns, 1994). As with the simpler task, recognition of the objects and their spatial relationships is necessary, as well as the formulation of this relationship into a proposition (“the object is behind x”). This should then be stored in memory.15 The subject must then discriminate between picture/model and reality, and also understand the relation between the two spaces. This relationship can involve many things, of which understanding that the picture depicts current reality in an adjacent room seemed to be specifically prob-lematic in the retrieval tasks of DeLoache and Burns (1994). In addition the task requires more than global reference from model to real space. There is also a need for specific mappings between points in the scale model or photographs, and points in the room referred to. It is thus a task that requires e.g. parsing of part-whole rela-tionships, as well as memory for those relationships.
A similar suspicion can be raised regarding the performance in tests where chil-dren have been asked to place stickers on a picture or a doll to indicate where they have stickers on their own body (Smith & DeLoache, 1996). Children 2.5 to 3.5 years old place stickers in the right place on a photograph only 55% of the time, with younger children making the most mistakes. In the model-search task they might thus have understood the overall reference between the two spaces, but been unable to recognise the significance of local mappings. Recognising such mappings might have been more salient in the photographic condition.
Finally, but crucially, the subject must be able to act on the knowledge gained above without interference from conflicting memories.
Despite all the above factors, the main reason for the failures with scale models for the 2.5 year old children was hypothised to be that the model was just too much an object in itself to serve a referential function: i.e. a problem with dual representa-tion. DeLoache (2000) report that when the scale model was placed behind glass or when children were not allowed to interact with it the function of object-in-itself diminished and the referential nature became more apparent, leading to more suc-cessful searches by the children. This discovery could also be helped by instruction, focusing on the intent behind the scale model (DeLoache et al., 1999).
The workings of dual representation also became apparent for video stimuli in Troseth and DeLoache (1998). A hiding event taking place on video could not be solved in real life by most 2-year-olds but by most 2.5-year-olds. However, when seeing the same event through a window all 2-year-olds were successful at later re-trieving the object. Furthermore, when being led to believe that they saw the event through a window, when in reality they were watching a video tape, performance approached the window condition.
Similarly, when infants are made to believe that an actual room has been shrunk to a miniature model by a magical shrinking machine, and then subsequently enlarged again, children (2.5 years) are more than able to find an item in a room that was hidden as a miniature item in the model (i.e. shrunken room) (DeLoache, 2004).
DeLoache and her colleagues have in the research described above clearly shown that it is the correspondence between two entities that seem to be the problem for 2
15 Delays are detrimental also to 3-year-olds’ attempts at object retrieval (Uttal et al., 1995).
year old children. By 2.5 years they seem to have overcome this for pictures but not for scale models. Interestingly, a reason Troseth and DeLoache (1998) expected also younger children to be able to successfully retrieve an object hidden on video was that nonhuman primates had seemed able to (i.e. Menzel et al., 1978). However, the data from the children supports the interpretation that the apes succeeded precisely because they took the video events to be some kind of events through a window (see section 9.3).
Suddendorf (2003) has pushed back the age for succeeding with both photo-graphs and video cues in retrieval tasks to 2 years. Instead of four trials on the same room he has used single trials on four rooms. According to Suddendorf the reason for the low performance in earlier studies was that small children are subject to strong perseveration effects, which means that they tend to search at the location where they last found a hidden object. When counteracting this by only letting them do one search per room they perform as well as 2.5-year-olds. However, the respective experiments were conducted in different laboratories, which is always a complicating factor. Sharon & DeLoache (2003), however, maintain that persevera-tion errors are a consequence of failing to adequately realise the symbol-referent rela-tionship, since it in their study is apparently not a consequence of poor inhibitory control. Further research is clearly needed to settle this.
But succeeding in a search task does not mean that all types of iconic reference is understood. In a simple test by Callaghan (1999) children were shown a drawing of an object and were then required to put a similar object down a tube. But the task was framed in an ambiguous way. Given a choice of the referent object or a second picture of the object, 2, but not 3-year-olds, often put pictures down the tube. The youngest children in Callaghan’s study also failed with the simplest types of dis-criminations, that of a straight line versus a circle representing a stick or a ball.
Callaghan (2000) even insists that the appreciation of the “symbolic nature” of pictures (and objects like scale models) per se, does not actually appear until the middle of the third year of life, and for some mediums and domains even later. She suspects that the performance in DeLoache and others’ studies were biased by the fact that they used pictures of familiar objects with verbal children, and that verbal labelling would affect the use of pictures in guiding action. It was rather the sym-bolic function of language, and not the symsym-bolic function of pictures, that guided these children.
There are further findings that suggest that the referential nature of pictures can be grasped much earlier than at 2.5 years of age. Preissler and Carey (2004) report that after associating a new word with a picture of an object, subjects as young as 18 months choose the object rather than the associated picture when asked to indicate the novel entity in a choice between the two. This is contrary to the result of putting pictures down tubes in Callaghan (1999). They had learnt that the new word re-ferred to the object in the picture and not the picture itself. This is argued to show that the children understood the referential nature of the picture. Importantly, these were not photographs, but black-and-white line drawings, albeit prototypical views of simple objects. This would preclude that the children worked in reality mode and just chose the three-dimensional versions because they were seen as more proper
exemplars of the same thing. Note, however, that this is a weak test of iconicity. No interpretation of novel pictures at the expense of other novel pictures is required.
And again, language is an integrated part of the task.
A follow-up study by Preissler and Bloom (2007) address both these issues. In a setup where 2-year-olds were shown two novel objects and two novel drawings, children generalised from drawing to object if the drawing had been named (e.g.
“this is a whisk, can you find another one?”), but from drawing to drawing if no label had been used (i.e. “can you find me another one of this?”). Depending on context the children thus attended to either the drawing or to the picture as a piece of paper. Thus, the dual nature of pictures, as well as interpretation of simple line drawings, can be grasped by 2-year-olds if the problem is properly framed (by lan-guage).