Children's Understanding of the Relation Between Delayed Video Representation

and Current Reality: A Test for Self-Awareness?



Thomas Suddendorf

Department of Psychology

University of Auckland

Private Bag 92019, Auckland, New Zealand

fax: 64 9 3737450






MS # 97-55 – 2nd Revision

Re-submission date: Mai. 15th, 1998



Children's Understanding of the Relation Between Delayed Video Representation

and Current Reality: A Test for self-awareness?




While children from about 18 months on can use a mirror to show self-recognition through the retrieval of a covertly placed mark from their forehead, Povinelli, Landau, and Perilloux (1996) showed that four- but not two-year-olds pass a similar surprise-mark test involving three-minute-old videos of themselves. The authors argued that this marks the emergence of what they called the "proper self". The current study (n=40) supported the claim for a developmental asynchrony by showing that three- and four-year-olds who fail the delayed video version of the surprise-mark task pass the mirror version. However, the same performance pattern was observed in an analogous task involving the introduction of an object in the room that was visible only on the video screen (surprise-object task). Surprise-mark and surprise-object tasks were positively correlated. Thus, young children’s problems seem to reflect general difficulties in reasoning from the unexpected information in the video to the current situation, rather than particular deficits in self-awareness.



Children's Understanding of the Relation Between Delayed Video Images

and Current Reality: A Test for Self-Awareness?


Mirror self-recognition, operationalized as the use of a mirror to locate and inspect an otherwise invisible, covertly placed mark ("surprise-mark" task), has often been discussed as an indicator of self-awareness in both animals (e.g., Gallup, 1970; Lethmate & Dücker, 1973; Patterson, 1991) and humans (e.g., Amsterdam, 1972; Field & Adamiak, 1990; Lewis & Brooks-Gunn, 1979; see Parker, Mitchell, & Boccia, 1994, for a recent review). Children generally pass these tests by about 18 to 24 months. Whether or not performance on these surprise-mark tests is evidence for self-awareness, however, has been hotly debated (e.g., Epstein, Lanza, & Skinner 1981; Heyes, 1994; Mitchell, 1996).

The argument that mirror self-recognition indicates some new level of a self-awareness has been supported by the claim that passing the surprise-mark task develops after children learn how to use the mirror to locate objects that are outside their direct visual field (Anderson, 1984; Brooks-Gunn & Lewis, 1984; Gallup, 1983). The reasoning, then, is that although infants know how the mirror works they pass the surprise-mark task only once they have a certain awareness of self 1. However, data on this purported developmental progression are inconsistent (e.g., Field & Hogg, 1992; Loveland, 1986; Robinson, Connell, McKenzie, & Day, 1990). Field and colleagues (Field & Adamiak, 1990; Field & Hogg, 1992), for example, reported that children performed better on the surprise-mark task than on a parallel task in which the mirror was used to locate objects in space. Perhaps performance on the surprise-mark task should after all be regarded mainly as evidence for understanding mirrored representation and other forms of direct feedback such as live video (cf., Menzel, Savage-Rumbaugh, & Lawson, 1985), shadows (cf., Boysen, Bryan, & Shreyer, 1994) or simultaneous imitation (cf., Asendorpf & Baudonniere, 1993). It has been suggested that children younger than 18 months may not grasp the direct relationship between these feedbacks and the self because they cannot yet form secondary representations (Asendorpf, Warkentin, & Baudonniere, 1996; Suddendorf, in press)2.

Even if one grants that the emergence of mirror self-recognition is a significant marker in the development of self-awareness, which I believe is a reasonable assumption, the extent to which it indicates anything more than recognition of what one looks like from the outside remains debatable (Hart & Fegley, 1994; Powell & Jordan, 1993). Self-awareness about internal states is not addressed by the surprise-mark task. Furthermore, the entire temporal dimension of one’s awareness about self cannot be tested with this paradigm. But the definition of who one is derives largely from considering one's past, present and potential future (Humphrey, 1986; Suddendorf, 1994; Suddendorf & Corballis, 1997). It has been argued that mental time travel, that is the construction of autobiographical memory and personal future, does not mature until around age four (Suddendorf, 1994; Suddendorf & Corballis, 1997; Wheeler, Stuss, & Tulving, 1997). It is only then, two years after the child shows mirror self-recognition, that the mental prerequisites (including an understanding of the representational nature of mind) appear to be sufficiently installed for the development of a personal identity through time.

Independently, Povinelli (1995) suggested that an initial system of self-representation (the present self) restricted to the here-and-now is measured by the mirror surprise-mark task (although he argues that it implies more than a concept of what one looks like). This system later develops to include temporal differentiation (the proper self). It is only this temporal differentiation that can bring about the integration of one’s self into a sense of continuity, regardless of change. Povinelli, Landau and Perilloux (1996) set out to devise an extension of the mirror surprise-mark test that they thought might capture the emergence of the proper self. In the first experiment two to four-year-olds were covertly marked with a large sticker while being videotaped. The tape was played back to each child some three minutes after this procedure. None of the two-year-olds, 25% of three-year-olds and 75% of four-year-olds reached up to remove the sticker from their forehead after seeing it on the video replay.

The second experiment used two polaroid photographs that depicted the child being marked with the sticker on their forehead. The developing time of the polaroids is about three minutes, a delay similar to that operating in the first experiment. After several leading questions ("who is that?" - pointing to child in picture - "what is that?" - pointing to sticker) the child was asked where the sticker really is (or "where is it right now?"). Again the same significant age difference was observed: 37% of the three-year olds and 87% of four-year-olds reached up to the sticker. The three-year-olds continued to perform poorly in spite of the leading questions and although 85% of them passed the mirror self-recognition test. Further, there was an association between failing to reach up to the sticker and referring to one’s image by dissociative phrases such as "his head". Finally, a third experiment used two-and-a-half to three-and-a-half year-olds to contrast live versus delayed video feedback as an experimental condition. A significant difference between the groups was observed, with 37% of the delayed and 67% of the live group passing the test.

In sum, while two-year-olds generally pass the test in front of a mirror, only by about age four do the majority of children pass three-minute-delayed versions (although mirror and delayed video performance have not been directly contrasted). These results appear to fit in nicely with the prediction that a proper self or a personal identity through time emerges at around age four (Povinelli, 1995; Suddendorf, 1994; Suddendorf & Corballis, 1997). The first possible explanation for younger children’s difficulty might thus be that they have not yet developed such a proper self. If this were the only reason for younger children’s failure on the delayed video version of the surprise-mark task, then one would predict that children can use delayed video feedback to locate objects in space before they can use it to locate marks on their own body.

In apparent support for this, there is some evidence suggesting that even two-and-a-half-year-olds can use live video of a hiding event to locate an object in another room (Troseth & DeLoache, 1997). To test the present prediction, however, delayed video feedback has to be investigated. This has as yet not been examined other than in the context of the sticker task. If children can use this feedback to locate an object in space at an earlier age than to locate a sticker on their forehead, then this would support the claim that changes in self-awareness are measured by Povinelli et al.’s task. Such a result would be analogous to the supposed developmental asynchrony between using a mirror for object localisation and mirror self-recognition. Alternatively, however, if there is no such asynchrony in the delayed video task, then this points to a general problem in reasoning from surprising information in delayed video representations to present reality rather than to a deficit specific to self-awareness.

It could still be argued that a general problem with delayed video tasks is indicative of an immature capacity for mental time travel and thus for a proper self. Indeed, Povinelli et al. (1996) go even further by suggesting that the same basic representational skills might be involved in both understanding the causal connection between past, present, and future and understanding videos as representations. Many recent studies have shown that there is a dramatic change in children's understanding of representation between age three and four (e.g., Flavell, Green, & Flavell, 1986; Gopnik & Graf, 1988; Gopnik & Astington, 1988; Perner, 1991; Suddendorf & Fletcher-Flinn, 1997; Wimmer & Perner, 1983; Zaitchik, 1990). Although the interpretation of these findings is debated, the key developmental achievement of four-year-olds appears to be the ability to represent representations as representations. Such an ability to metarepresent is thought to be evident when children show an understanding of false beliefs (e.g., Wimmer & Perner, 1983). With the ability to metarepresent, the child is in a position to simultaneously represent several conflicting representations of the same object or event (e.g., "true" and "false"; "current" and "outdated" representations; see Astington, 1994; Carruthers & Smith, 1996; Flavell, 1993; Gopnik, 1993; Perner, 1991). Thus, younger children may not yet have developed a temporally extended sense of self (or of any other changing object for that matter) because conflicting representations of past and present self cannot yet be simultaneously entertained as aspects of the same personal identity (Povinelli, 1995; Suddendorf, 1994). At the same time, metarepresentational skill might also be necessary to understand the representational nature of the video. Younger children have problems appreciating that a picture (Robinson, Nye, & Thomas, 1994) or television image (Flavell, Flavell, Green, & Korfmacher, 1990) is both an object in its own right and a representation of something else. This may make questions such as "where is the sticker really?" ambiguous. The emergence of metarepresentation may thus be important in considering the relationship between past and present, symbol and reality, and combinations of those dichotomies. Thus, the second potential explanation for young children’s difficulty with Povinelli et al.’s (1996) tasks is that they have not yet developed metarepresentational thinking. As with the development of other skills that are purported to depend on metarepresentation (see for example Gopnik & Astington, 1988; Perner & Ruffman, 1995; Suddendorf & Fletcher-Flinn, 1996; Suddendorf, Fletcher-Flinn, & Johnston, 1996), one would expect to find significant correlations between these tasks.

But younger children may have problems with more basic aspects of the delayed video information. Povinelli et al. (1996) reported poorer self-recognition with delayed relative to live video feedback, and this might be due to the lack of contingency in the former. However, they did not test explicitly whether the children appreciated the correspondence between the images of objects in the delayed video and the actual objects. DeLoache (e.g., 1991; DeLoache & Burns, 1994) has researched children’s growing understanding of the fact that one thing (e.g., picture, map, model, video image) can correspond to another (e.g., a room) and thus carry information about aspects of the other thing. Her results show that three-year-olds have a basic understanding of such symbolic relations. It is still possible that three-year-olds’ problems with the delayed self-recognition task are due to a failure to recognise the correspondence between objects in the delayed video and in present reality. This possibility can be tested with tasks employing analogous procedures to the delayed self-recognition task but without the surprise element of incongruent information coming from the video (i.e., sticker in hair). This also provides the opportunity to test children’s understanding of the test questions in the surprise-mark task.

There is yet an even simpler alternative explanation, which is that recognition of one’s self on video may depend on experience with video cameras and their properties. Povinelli et al. (1996) did not report any attempts at controlling or measuring the subjects' knowledge about videos and their degree of past experiences with cameras, although their third experiment contrasted live with delayed feedback and it might be argued that differential experience cannot account for differences in performance in this case. While even young children generally have had ample opportunity to learn about the properties of mirrors, only few will have had experience with video cameras. Indeed, most children have had plenty of experience with TV images that are not at all related to themselves or to their immediate environment. Mirror exposure is necessary for the development of mirror self-recognition. So one might expect that experience with video cameras is necessary for understanding delayed video playback. It is therefore necessary to investigate the possibility that children perform better on the mirror than on the delayed video surprise-mark task because they have not had enough opportunity to learn and understand the properties of video playback.

The present study investigated these issues resulting from Povinelli et al.’s (1996) findings. First of all, three and four-year-old children’s ability to recognise themselves in delayed videos was directly compared with their ability to recognise themselves in mirrors. Based on the previous findings it was predicted that the delayed video version of the surprise-mark task would be passed by the majority of four but not three-year-olds. In contrast, both three and four-year-old children who failed on the delayed video version were expected to pass the mirror version (i.e., show mirror self-recognition). Four potential explanations for the expected late development of delayed self-recognition were investigated. First, an analogous task involving the hiding of an object in the room ("surprise-object" task) was designed to test whether children’s problems are specific to information about the self. Second, two tests for metarepresentation were included to assess whether there is an association with the video surprise tasks. Third, several tests were used to insure that children understand the test questions and the basic correspondence between video and present reality. Finally, information about previous experience with video feedback was collected and the study addressed the possibility of learning effects.



Twenty three-year-olds (mean age: 42 months; SD=3.5; range: 36-47) and twenty four-year olds (mean age: 54; SD=3.6; range: 48-59) participated. Twenty three boys and 17 girls who were recruited from four creches in Auckland, New Zealand. Consent from parents, staff and children was obtained before testing.


The basic set up involved a small table and chair for the child, a video-camera on a tripod two meters opposite the table and a 51 cm TV color monitor on a 1.5 m high trolley adjacent to the camera. The false-belief task involved two 20 cm sized dolls, a 12 x 12 cm basket with lid, a round tin (13 cm diameter) and a marble. In the sticker task a variety of colourful stickers were hidden under one of four brightly coloured plastic cups. Three colourful juggling balls and a 35 cm soft toy orang-utan served as temporal markers in the video feedback tasks. In the surprise-mark task, a 7.5 x 4 cm red sticker with a yellow "smily" in the centre was placed in the subject's hair. In the analogous surprise-object task, a 20 cm brown teddy bear was placed in a 50 x 15 x 25 shoe carton in front of the table. A 45 X 120 cm mirror was used when the children failed to retrieve the sticker or teddy in the video playback version.


Parents were asked to fill in a brief questionnaire when signing the consent form required by the Ethics Committee. Demographic information was gathered regarding the child's name, age, sex, whether there was a video recorder at home, and whether the child had any experience with video cameras.

The children were tested in a quiet room in the creche. Children were invited individually to play with two experimenters. Both experimenters spent some time in the creche to allow the children to get familiar with their presence. The child was seated at a table and told that some games were going to be played. Pointing to a video-camera and a TV screen, the main experimenter told the child that the other experimenter was going to record them and that they would be watching what was recorded after each game was finished.


Insert Table 1 about here


The tasks were given to the children in the order presented in Table 1. First, the false-belief task was conducted and the video recording was played back to the children. A second task was designed to familiarize the children with the delayed feedback testing procedure and to investigate the children's ability to recognize the correspondence between video playback and reality. After these two sets of playback the delayed feedback experiment commenced. The children were assigned to either one of two matched groups (according to age and sex). One group received the surprise-mark task first and then the surprise-object task, and the other group was given the tests in the reverse order.

False-belief task

The main experimenter told the child the story of the dolls Sally and Anne (after Prior, Dahlstrom & Squires, 1990). In this classic metarepresentation task the child is told that Sally puts her marble into her box and goes outside to play. Naughty Anne takes the marble out of the box, puts it into her basket, and closes both lids. When the child correctly recalls where Sally put the marble and where the marble is now he or she is asked the crucial false-belief question: When Sally comes back, where will she look first for her marble (following Siegal & Beattie, 1990)? After the children pointed to either location they were told that they could now watch what they had just done on the television.

Video playback I

The television was next to the camera slightly elevated on a trolley, so that the children could not see their own reflection in the screen. The second experimenter rewound the tape and started the playback where the Sally-Anne story began. In order to help the children make a connection they were asked whether they remembered this. At the end of the story the video was put on hold so that the picture was clearly visible. The main experimenter moved to the screen and, pointing to the child on the screen, asked: `Who is this?'. Then the experimenter pointed to the box on the screen and asked: `What is that?', `Where is that box really?', and `Can you find it?'. The box and the basket of the Sally-Anne story were still on the table in front of the child and it was recorded whether the child recognized the correspondence between video image of the box and its real counterpart. The material was then removed from the table (giving the child further opportunity to realize that what happens in the room, while corresponding to the video, does not alter the video images directly), the video was turned off, and the child was invited to play a second game.

Sticker game and video playback II

Four colourful cups were placed on the table in front of the child. The experimenter told the child: "I will hide a beautiful sticker under one of these cups and you can see me doing it on the TV and then try and find it. Okay, Lea (the second experimenter) turns the camera on and you have to turn around so I can hide the sticker." The child was then given a book to look at while the main experimenter placed a sticker under the red cup. The video was rewound (to the point where the child turns around to look at the book) and played back. "Now have a look at the video. You can see where I have hidden the sticker. It is under the same cup." The video is put on hold so that the hiding is clearly visible. Then the child was asked to lift up the cup which he or she thought contained the sticker. If the child failed to chose the correct cup he or she was allowed to search under the remaining cups. When the sticker was found it was placed on a white sheet of paper and the child was told that he or she could keep all the stickers found during the games.

The experimenter then went to the television where the cups were still visible (video on hold) and asked "If I stuck my hand down in here (places hand on top of TV), could I pick up a cup and take a sticker out?" This question was taken from Flavell et al. (1990), who found that three-year-olds tended to confuse television images and their referent objects.

Delayed feedback experiment

Surprise-mark task. The child was invited to play some more find-the-sticker games. These were simply guessing games that did not provide data but were used to keep the child’s attention during the experiment (see below). The child was informed: "The video camera over there is recording us and we will see what we have done on the TV after the game". In order to support children's recognition of when the action that they were to see on the TV took place, a distinctive temporal marker was introduced. The child was introduced to a soft toy. "Look who I have got here. This is Otis the orang-utan. Let's wave to the camera, all three of us. I put Otis down over here [behind the child, visible to the camera] while we play the sticker game."

The experimenter hid a sticker under one of the four cups while the child was turned around to look at a book. The child then took a guess about which cup the sticker was hidden under until it was retrieved. When the sticker was found the experimenter praised the child and in the process patted her or his forehead. This was done to accustom the child to the experimenter touching his or her head. After the second trial experimenter used the praise as the opportunity to covertly place a large red sticker in the child's hair. A third trial was then conducted to control whether or not the child has noticed the marking. Four children reached up to the sticker and were therefore excluded from analysis and replaced with new subjects.

The child was then told: "Let's have a look at what we have just done. Lea just rewinds the tape. Okay here we go." The tape was replayed beginning with the temporal marker (i.e., Otis, the child and the experimenter waving) and stopped at trial three at a position where the sticker on the child's forehead was clearly visible. If the child had not reached up to retrieve the sticker while watching the video (which showed the marking event), the following questions were asked while experimenter was pointing to the respective places on the screen (the experimenter provided the answers given in brackets if the child did not respond): a) Who is that? (It is you, "name of the child"); b) What is that? (It is a sticker); c) Where is that sticker really? d) Can you find it?

If the child failed to reach for the sticker, a mirror was placed in front of the TV and the same questions were repeated.

Surprise-object task. This procedure was virtually identical to the surprise-mark task but featured a teddy that was visible to the child only on the video. The child was supposed to discover something new about the room rather than something new on his or her own body. The experimenter juggling three colourful balls served as the temporal marker for this test. The balls were taken out of a shoe carton, and the children were asked whether there was anything more in the carton. When they replied in the negative, the experimenter placed the carton in front of the table so that it was open and visible to the camera but outside the visual field of the child. After a brief period of juggling the balls were put away and the child was invited to another round of find-the-sticker guessing games.

On the second trial, while the child had turned around to look at the book, a teddy bear was put into the box visible to the camera. After the third trial the game was stopped and the child was told about watching the video as in the delayed self-recognition task. After rewinding, the tape was started from the point the balls were taken out of the carton. The video was put on hold at trial three where the teddy bear was clearly visible. If the child did not spontaneously get up and retrieve the teddy he or she was asked the following questions and answers were given while pointing to the teddy on the screen: a) What is that? (It is a teddy); b) Where is the teddy really? c) Can you find it? If the child was unsuccessful the same procedure was repeated with a mirror that was placed in front of the TV.


There were no significant sex-differences in performance on any of the tasks and the results reported are therefore collapsed across sex. In the delayed video version of the surprise-mark task it was found that 75% of four-year-olds and 55% of three-year-olds passed the test (see Table 2). The difference between the two age groups was not statistically significant. Of those children who passed the task, 59% retrieved the sticker spontaneously while watching the video of the marking event. As was predicted, all those who failed the video task, except for one three-year-old, retrieved the sticker when looking at a mirror (mirror self-recognition) rather than at the delayed video image.

Children were asked to label the person (themselves) they saw on the video. They either used the personal pronoun ‘me’ (52.5%), their proper name (37.5%), or did not respond (10%). Significantly more four- (70%) than three-year-olds (35%) used the personal pronoun (chi-square=4.91; df=1; p<.05). The use of the proper name was, however, not found to be significantly associated with failing the surprise-mark task (see Table 3).

To further investigate whether performance difficulties on the surprise-mark task are due to developments specific to self awareness, performance on the surprise-object task were compared. Eighty percent of four-year-olds versus 50% of three-year-olds passed the delayed video version of the surprise-object task. This difference between the age groups was significant (chi-square=3.95; df=1; p<.05). In this condition, where the children had to get up from their chair to retrieve the object, 93% of those who passed did so only after the experimenter’s prompting. All but three three-year-olds passed the mirror version of the task after failing the video version. The overall pattern of performance was very similar to the one in the surprise-mark task (see Table 2).

Each delayed video surprise task was passed by 65% of the subjects. Half the subjects passed both tasks and 20% failed on both. Fifteen percent passed the surprise-object task but failed the surprise-mark task, while another fifteen percent displayed the reverse pattern. A chi-square test confirmed that performance on the surprise-object task was significantly correlated with that on the surprise-mark task (Chi-square=4.64; df=1; p<.05). These results show that children’s problems with using surprising information from delayed video feedback is not specific to information about self. This suggests the creation of a new variable combining the scores of both surprise tasks into a single measure of children’s ability to use surprising information in the video feedback.


Insert Table 2 about here


Table 2 shows the differential performance of three and four-year-olds on this compound measure of performance on the surprise tasks. Four-year-olds were found to be significantly more likely to pass the surprise tasks than were three-year-olds (chi-square = 6.2; df=2; p<.05). In fact, only one four-year-old failed on both measures. This inter-individual age difference as well as the intra-individual difference between passing mirror and video tasks require explanations. Three potential explanations were investigated.

One possible explanation is that children perform better on the mirror than on the video task because they have not had enough opportunity to learn and understand the properties of video playback. In addition to clear instructions and the benefits of two trials with video feedback prior to the surprise tasks, there were two ways in which this issue was addressed. First, parents were asked about their child's experience with videos and video cameras. Eighty percent of the children had videos at home and 45% were reported to have had prior experience with video cameras. Neither children who had a VCR at home nor those who had experience with cameras performed significantly better on the surprise tasks than those who did not. Second, half the subjects (matched by age and sex) received the surprise-mark task before the surprise-object task, while the other half were given the tasks in reverse order. The virtually identical design of the two tests would have given children ample opportunity to learn and therefore improve their performance on the second trial. (Note that failing the first task resulted in exposure to a mirror and subsequent retrieval of the sticker or teddy.) However, there was no significant improvement on the second trial. Thirteen children passed the surprise-mark tasks when it was given first and 13 passed it when it was given last. Similarly, 12 children passed the object-mark task when it was given first and 14 when it was given last. Hence there is no evidence to suggest that children’s difficulties with the surprise tasks are due to lack of experience.

Another possibility is that the ability to appreciate the basic correspondence between the delayed video image and the real world is responsible for younger children’s performance difficulties. Most participants showed a general understanding of this correspondence. All but four children were able to label their delayed video image correctly (either by proper name or personal pronoun). Asked where the box really is and whether they can find it (while experimenter pointed to the image on the screen), 82.5% of subjects correctly pointed to the box in front of them. Note that the same phrasing of questions was used as in surprise tasks. The difficulty with the surprise tasks therefore cannot simply be attributed to a misunderstanding of the test questions. Finally, 87.5% of children retrieved the sticker from the correct cup in their first attempt after seeing the baiting on the video. These children seem to have understood that the video image carries information about the current situation in the room. Taken together, these results show that the majority of participants understood the basic correspondence between image and referent. None of these measures, nor any combination of them, was found to be significantly correlated with the surprise tasks (see Table 3). Thus, there is no evidence to suggest that children’s difficulties with the surprise tasks are due to not understanding the correspondence between images of objects in video playback and actual objects in the immediate environment or to not understanding the test question.

The final potential explanation was that a general improvement in representational capacity (i.e., metarepresentation) might be responsible for the improved performance of four-year-olds. Two relevant tasks were given to the subjects. The false-belief task was passed by 75% of four and 25% of three-year-olds (chi-square=10.00, df=1, p<.005). Following Flavell et al. (1990), children were also asked a question about TVs (if I stuck my hand down in here could I pick up a cup and take a sticker out?) that may depend on the understanding of TVs as representational systems (i.e., on metarepresentation). Sixty-five percent of four-year-olds versus only 15% of three-year-olds answered correctly. The difference between the age groups was significant (chi-square=10.42, df=1, p<.005). The idea that both tasks are based on the same metarepresentational understanding is supported by a significant point biserial correlation between them (see Table 3).


Insert Table 3 about here


Yet, neither of these metarepresentation measures (nor a combination of both) was found to be significantly correlated with performance on the surprise tasks (with or without controlling for age). In fact, performance on the false-belief task was not at all associated with the surprise-mark task. Thus, there is no evidence to suggest that metarepresentational skill is responsible for the performance on the surprise tasks.

Interestingly, however, the question about TVs was found to be significantly correlated with the other measures related directly to understanding the delayed video image in relation to current reality: Identifying the box on the table as the one depicted in the video and choosing the correct cup after seeing the baiting on the video (see Table 3). Yet, not even a new compound variable of all three of these tasks was found to be associated with performance on the surprise tasks.


This study has shown a profound difference between young children’s ability to use delayed video feedback and direct mirror feedback to locate previously unknown objects. Children who failed to retrieve the mark in their hair with the help of three minute-old-videos were able to locate that mark with the help of a mirror. Although Povinelli et al. (1996) did not directly compare delayed video with mirrored feedback, this result is consistent with their findings.

However, the present study also found this performance pattern for an analogous task involving an unexpected object in the room. The surprise-object task was passed by the same proportion of children as the surprise-mark task and again children’s problems disappeared when a mirror was used instead of the video. Thus, there was no evidence for a developmental asynchrony between surprise-mark and surprise-object task that would resemble the proposed asynchrony in the development of mirror object-localisation and mirror self-recognition. The video surprise tasks were equally difficult.

Furthermore, there was evidence for some intra-individual consistency in performance across both tasks. This was confirmed by the significant positive correlation between performance on the surprise-object and the surprise-mark tasks. Altogether, then, younger children appear to have some problems with reasoning from the surprising new information in the video playbacks to the present situation, rather than a problem specifically related to developments in self-awareness. These problems seem to dissipate by about age four.

Significantly more four- than three-year-olds passed these video surprise tasks. Indeed, only one four-year-old failed on both tasks. Thus, by age four most children appear to have some understanding of the relation between delayed video representation and current reality. This understanding seems particularly evident in those children who retrieved the mark or object spontaneously while watching the replay, because they must have construed the video as immediately relevant to the current situation. There was a difference between the two conditions in regard to spontaneous retrieval. In the surprise-mark task the majority of children who passed did so spontaneously, while in the surprise-object task prompting was needed in most cases. But this difference is probably due to the task setting. While the children can immediately reach for the sticker on their forehead, the teddy is out of reach in a box behind the table and the children had to get up from their chair and walk around the table in order to retrieve it. Without encouragement (questioning) from the experimenter, they are not likely to break out of the setting and investigate parts of the room (after all, they were asked to sit and watch the video). The children who did retrieve the mark or the object spontaneously did not differ systematically from the other children on any other measure (see Table 3) and thus did not provide a new clue in the search for why younger children fail these tasks.

One potential reason for the difficulty with the video surprise tasks might have been inexperience. However, the current study did not find a significant association between having had experience with video cameras (or video recorders) and performance on the surprise tasks. But this issue may warrant further scrutiny in future research. The question was only whether the children had experience with such devices. More detailed analysis of children’s past experiences with live video and video playback would be desirable. However, the current design itself provided the subjects with some experience. The first two video playbacks gave the children some opportunity to learn about the properties of this feedback. It might indeed be due to this experience that three-year-olds in the present study performed better than those in Povinelli et al’s (55% versus 25% in Exp.1 and 37% in Exp. 2 & 3) (But this difference might also be due to the fact that the three year-olds in the present study were three months older). However, there was no sign of a learning effect on the second compared to the first surprise test. Thus, the data do not support an explanation entirely based on lack of experience.

Children also did not fail because they could not appreciate the correspondence between the delayed video image and present reality. The majority of participants acknowledged their image either with their name or personal pronoun. Most children also had no difficulty retrieving a sticker from a correct cup after seeing the baiting on video. This task is particularly interesting because it is similar to other tasks that show young children’s difficulty with the basic understanding that one thing (e.g., a picture) can correspond to, and thus carry information about, another thing (e.g., DeLoache & Burns, 1994). Finally, the majority of children did not appear to have problems finding the referent of the video image when asked about an image of a box in the video (using the same questions: "Where is it really? Can you find it?"). Thus, neither failure to understand the correspondence between video image and present reality, nor to understand the test questions, can be responsible for children’s difficulties. Indeed, performance on none of these tasks was found to be significantly related to performance on the surprise tasks.

They were however associated with the question of whether one can put a hand into the TV and take something out of the image. None of the children who failed the correspondence tasks answered this question correctly. These children appear to confuse image and referent (cf., Flavell et al., 1990). Surprisingly, children with camera experience did worse on the TV question than those without. Understanding the representational nature of the TV image might be the result of the emergence of the ability to metarepresent. Indeed, a positive correlation between false-belief task performance (a classic measure of metarepresentation) and answering this question about the TV image was found. This is an interesting finding in support of the claim that representational understanding does not develop independently in specific domain, but draws on the same general skill (Perner, 1991; Suddendorf, in press; Suddendorf & Fletcher-Flinn, in press).

Metarepresentation was expected to be related to performance on the surprise tasks not only because it appears necessary for understanding the relation between image and referent, but also because it was proposed to be necessary for mental time travel and the development of a "proper self", that is, the ability to integrate past instances of self (autobiographical memory) as causally connected aspects of one’s changing but enduring self. Povinelli et al. (1996) designed the video version of the surprise-mark task with the express purpose of measuring the development of the "proper self". They found, and the present results confirm, that the age of passing these tasks coincides with the normal time for the emergence of autobiographical memory, the decline of infantile amnesia and the emergence of metarepresentation (e.g., Nelson, 1992; Perner, 1991; Povinelli, et al., 1996; Suddendorf & Corballis, 1997). Yet, the present study did not find an association between metarepresentation (as measured by the false-belief task and the TV question) and performance on the surprise tasks. These results suggest that some children fail the surprise tasks although they have metarepresentational understanding and others pass these tasks although they have not yet developed metarepresentational thinking. Performance on Povinelli et al.’s (1996) surprise-mark task might be quite unrelated to what it was designed to measure.

Some children may have retrieved the sticker (or the teddy) without metarepresentational thought, autobiographic memory and a "proper self". They may have simply used their understanding of feature correspondence between video image and current reality and, disregarding the lack of direct contingency between events in the video and reality, concluded from the surprise element in the video to investigate the current situation. Some children in the present study were indeed very surprised when they actually found the sticker in their hair or the teddy in the box. Direct reasoning from image to the present situation might have completely side-stepped the issue of temporal displacement.

On the other hand, some children may have failed the surprise tasks even though they had metarepresentational thought, autobiographic memory and a "proper self".

The key development that researchers have emphasized as the link between metarepresentation, autobiographic memory and self is that the four-year-old child becomes able to re-experience past events as events that happened to ‘me’ (e.g., Perner & Ruffman, 1995). But the surprise tasks did not involve re-experiencing the past event. Indeed, the marking and the placing of the teddy were done so that the child did not experience it in the first place. Instead, the children have to appreciate that something that they did not experience and that was thus not part of their autobiographical memory, might actually have taken place.

These considerations put some serious doubts on Povinelli et al.’s (1996) proposed explanation and even on the basic assumption that one particular capacity or skill is measured by the video version of the surprise-mark task. Children may pass or fail for various reasons, which might be why the current study found no support for any one of the potential explanations tested. But negative results are difficult to interpret.

However, the research did produce telling positive results. Children have the same difficulty with the surprise-object task as with the surprise-mark task. This parallel pattern of results as well as the positive correlation between performance on these tasks strongly suggests that both tasks require the same skills. So even if all young children fail on the tasks for the same reason, failure appears to reflect general deficits in reasoning from the unexpected information in the delayed feedback to unknown aspects of the current situation, rather than particular deficits in self-awareness. By age four, however, most children can use video feedback of a previous event to draw some inference about the current situation - whether self related or not.



Amsterdam, B.K. (1972). Mirror self-image reactions before age two. Developmental Psychobiology, 5, 297-305.

Anderson, J.R. (1984). The development of self-recognition: A review. Developmental Psychobiology, 17, 35-49.

Asendorpf, J.B. and Baudonniere, P.M. (1993). Self-awareness and other-awareness: Mirror self-recognition and synchronic imitation among unfamiliar peers.Developmental Psychology, 29, 88-95.

Asendorpf, J.B., Warkentin, V., and Baudonniere, P.M. (1996). Self-awareness and other-awareness II: Mirror self-recognition, social contingency awareness, and

synchronic imitation. Developmental Psychology, 32, 313-21.

Astington, J.W. (1994). The child's discovery of the mind. London: Fontana Press.

Baron-Cohen, S., Leslie, A., & Frith, U. (1985). Does the autistic child have a "theory of mind"? Cognition, 21, 37-46.

Boysen, S.T., Bryan, K.M., & Shreyer, T.A. (1994). Shadows and mirrors: Alternative avenues to the development of self-recognition in chimpanzees. In S.T. Parker, R.W. Mitchell, & M.L. Boccia (Eds.), Self-awareness in animals and humans (pp. 227-240) Cambridge: Cambridge University Press.

Brooks-Gunn, J., & Lewis, M. (1984). The development of early visual self-recognition. Developmental Review, 4, 215-239.

Carruthers, P., & Smith, P.K. (1996) Theories of theories of mind. Cambridge University Press.

Dawson, D., & McKissick, F.C. (1984). Self-recognition in autistic children. Journal of Autism and Developmental Disorders, 14, 383-394.

DeLoache, J.S. (1991). Symbolic functioning in very young children: Understanding of pictures and models. Child Development, 62, 736-752.

Deloache, J.S., & Burns, N.M. (1994). Early understanding of the representational function of pictures. Cognition, 52, 83-110.

Epstein, R., Lanza, R.P., & Skinner, B.F. (1981). "Self-awareness" in the pigeon. Science, 212, 695-696.

Field, J., & Adamiak, J. (1990). Use of mirror reflection for self and non-self search during the second year. New Zealand Journal of Psychology, 19, 58-62.

Field, J., & Hogg, V. (1992). Young children's ability to find objects reflected in mirrors. Australian Journal of Psychology, 44, 9-11.

Flavell, J.H. (1986). The development of children's knowledge about the appearance-reality distinction. American Psychologist, 41, 418-425.

Flavell, J.H. (1993). The development of children's understanding of false belief and the appearance-reality distinction. International Journal of Psychology, 28, 595-604.

Flavell, J.H., Green, F.L., and Flavell, E.R. (1986). Development of knowledge about the appearance-reality distinction. Monographs of the Society for Research in Child

Development, 51, (1, Serial No. 212).

Flavell, J.H., Flavell, E.R., Green, F.L., & Korfmacher, J.E. (1990). Do young children think of television images as pictures or real objects? Journal of Broadcasting and Electronic Media, 34, 399-419.

Gallup, G.G., Jr. (1970). Chimpanzees: self recognition. Science, 167, 86-87.

Gallup, G.G., Jr. (1983). Toward a comparative psychology of mind. In R.L. Mellgren (Ed.), Animal cognition and behavior (pp. 473-510). New York: North-Holland.

Gopnik, A. (1993). How we know our minds: The illusion of first-person knowledge of intentionality. Behavioral and Brain Sciences, 16, 1-16.

Gopnik, A., & Astington J.W. (1988). Children's understanding of representational change and its relation to the understanding of false belief and the appearance-reality distinction. Child Development, 59, 26-37.

Gopnik, A., & Graf, P. (1988). Knowing how you know: Young children's ability to identify and remember the source of their belief. Child Development, 59, 1366-1371.

Hart, D., & Fegley, S. (1994). Social imitation and the emergence of a mental model of self. In S.T. Parker, R.W. Mitchell, & M.L. Boccia (Eds.), Self-awareness in animals and humans (pp. 149-165). Cambridge: Cambridge University Press.

Heyes, C.M. (1994). Reflections on self-recognition in primates. Animal Behavior, 47, 909-919.

Humphrey, N.K. (1986). The inner eye. London: Faber & Faber.

Lethmate, J., & Dücker, G. (1973). Untersuchungen zum Selbsterkennen im Spiegel bei Orang-Utans und einigen anderen Affenarten. Zeitschrift für Tierpsychologie, 33, 248-269.

Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self. Plenum, New York.

Loveland, K.A. (1986). Discovering the affordances of a reflecting surface. Developmental Review, 6, 1-24.

Menzel, E.W., Savage-Rumbaugh, E.S., and Lawson, J. (1985). Chimpanzee (Pan

troglodytes) problem solving with the use of mirrors and televised equivalents of mirrors. Journal of Comparative Psychology, 99, 211-17.

Mitchell, R.W. (1996). Self-recognition, methodology and explanation: a comment on Heyes (1994). Animal Behavior, 51, 467-469.

Nelson, K. (1992). Emergence of autobiographical memory at age 4. Human Development, 35, 172-177.

Parker, S.T., Mitchell, R.W., & Boccia M.L. (Eds.). (1994). Self-awareness in animals and humans. Cambridge: Cambridge University Press.

Patterson, F. (1991). Self-awareness in the gorilla Koko. Gorilla, 14, 2-5.

Pepperberg, I.M., Garcia, S.E., Jackson, E.C., & Marconi, S. (1995). Mirror use by African Grey Parrots. Journal of Comparative Psychology, 109, 182-195.

Perner, J. (1991). Understanding the representational mind. Cambridge, MA: MIT Press.

Perner, J. (1995). The many faces of belief: reflections on Fodor's and the child's theory of mind. Cognition, 57, 241-69.

Perner, J., & Ruffman, T. (1995). Episodic memory and autogenetic consciousness: developmental evidence and a theory of childhood amnesia. Journal of Experimental Child Psychology, 59, 516-548.

Povinelli, D.J. (1989). Failure to find self-recognition in Asian elephants (Elephans maximus) in contrast to their use of mirror cues to discover hidden food. Journal of Comparative Psychology, 103, 122-131.

Povinelli, D.J. (1995). The undublicated self. In P. Rochat (Ed.), The self in early infancy (pp. 161-192). Amsterdam: North-Holland-Elsevier.

Povinelli, D.J., Landau, K.R., & Perilloux, H.K. (1996). Self-recognition in young children using delayed versus live feedback: Evidence for a developmental asynchrony. Child Development, 67, 1540-54.

Powell, S.D., & Jordan, R.R. (1993). Being subjective about autistic thinking and learning to learn. Educational Psychology, 13, 359-370.

Prior, M., Dahlstrom, B., & Squires, T.L. (1990). Autistic children's knowledge of thinking and feeling states in other people. Journal of Child Psychology and Psychiatry, 51, 587-601.

Robinson, J.A., Connell, S., McKenzie, B.E., & Day, R.H. (1990). Do infants use their own images to locate objects reflected in a mirror? Child Development, 61, 1558-1568.

Robinson, E.J., Nye, R., & Thomas, G.V. (1994). Children's conceptions of the relationship between pictures and their referents. Cognitive development, 9, 165-91.

Siegal, M., & Beattie, K. (1990). Where to look first for children's knowledge of false belief. Cognition, 38,, 1-12.

Suddendorf, T. (1994). Discovery of the fourth dimension: Mental time travel and human evolution. Master's thesis, University of Waikato, Hamilton, New Zealand .

Suddendorf, T. (in press). The rise of the metamind. In M.C. Corballis and S. Lea (Eds.) The descend of mind: Psychological pserspectives on human evolution. London: Oxford University Press.

Suddendorf, T., & Corballis, M.C. (1997). Mental time travel and the evolution of the human mind. Genetic, Social and General Psychology Monographs, 123, 133-167.

Suddendorf, T., & Fletcher-Flinn, C.M. (in press). Children’s divergent thinking improves when they understand false beliefs. Creativity Research Journal.

Suddendorf, T., Fletcher-Flinn, C.M., & Johnston, L. (in press). Pantomime and theory of mind. Journal of Genetic Psychology.

Suddendorf, T., & Fletcher-Flinn, C. (1996). Some correlates of the emerging theory of mind. Australian Journal of Psychology, 48, (sup.), 26-7.

Suddendorf, T., & Fletcher-Flinn, C.M. (1997). Theory of mind and the origins of divergent thinking. Journal of Creative Behavior, 31, 169-179.

Suddendorf, T., & Fletcher-Flinn, C.M. (in press). Children’s divergent thinking improves when they understand false beliefs. Creativity Research Journal.

Suddendorf, T., Fletcher-Flinn, C.M., & Johnston, L. (in press). Pantomime and theory of mind. Journal of Genetic Psychology.

Troseth, G. L. & DeLoache, J.S. (1997). The medium can obscure the message: Young children’s understanding of video. Unpublished manuscript.

Welch-Ross, M.K. (1995). An integrative model of the development of autobiographical memory. Developmental Review, 15, 338-365.

Wheeler, M.A., Stuss, D.T., & Tulving, E. (1997). Toward a theory of episodic memory: The frontal lobes and autonoetic consciousness. Psychological Bulletin, 121, 331-354.

Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and constraining function of wrong beliefs in young children's understanding of deception. Cognition, 13, 103-128.

Zaitchik, D. (1990). When representations conflict with reality: The preschooler's problem with false beliefs and "false" photographs. Cognition, 35, 41-68.



Author Note

Thomas Suddendorf, Department of Psychology, University of Auckland, New Zealand.

I thank the children, parents and teachers of the participating creches. I am grateful to Leah Johnston for her work as experimenter and her support throughout the data gathering period. Michael Corballis, Daniel Povinelli, Jeff Field and three anonymous reviewers provided valuable comments on earlier versions of this paper.

Correspondence concerning this article should be addressed to Thomas Suddendorf, Department of Psychology, University of Auckland, Private Bag 92019, Auckland, New Zealand.



1 This line of reasoning finds even stronger support from comparative research. While several species such as African grey parrots (Pepperberg, Garcia, Jackson, & Marconi, 1995) and elephants (Povinelli, 1989) have been shown to be capable of mirror-mediated spatial locating, only our closest relatives, the great apes, can pass the classic surprise-mark task (Gallup, 1970; Lethmate & Dücker, 1973; Patterson, 1991).

2 Secondary representations might be required because the mirror image, imitator or shadow have to be considered not merely as primary representations (objects in their own right) but also as secondary representations of something else (e.g., of oneself). This correspondence between image and actual entity is only appreciated from about 18 months onwards. This capacity is also reflected in the child's emerging fascination with picture books - for infants with only primary representations pictures are merely colourful pieces of paper, but with secondary representation they become stories of magical attraction to the child.

Table 1. Steps of the procedure of the video feedback experiments. Note that half of the subjects received the tasks in the order 1, 2, 4, 3.


Video-recorded tasks Questions asked after video has been played back and put on hold


1) False-belief task a) who is this? (child)

b) what is that? (box)

c) where is the box really?

d) can you find it?


2) Sticker game a) under which cup is the sticker?

b) If I stuck my hand down in here (placing hand on top of TV), could I pick up a cup and take a sticker out?

3) Surprise-mark task

temporal marker (Otis the orangutan)

3 trials of guessing game; on the 2nd trial

a sticker is covertly placed in child’s hair a) what is that? (sticker in hair)

b) where is that sticker really?

c) can you find it?

4) Surprise-object task

temporal marker (E juggling)

3 trials of guessing game; on the 2nd trial

a teddy is covertly placed into the box a) what is that? (teddy in box)

b) where is that teddy really?

c) can you find it?



Table 2. Performance on the video versions of the surprise tasks.


surprise-mark task surprise-object task Compound surprise


age fail pass fail pass none one two


3 9 11 10 10 7 5 8

4 5 15 4 16 1 7 12




Table 3. Correlation coefficients between the measurements


1 2 3 4 5 6 7 8 9 10


1. Surprise compound score

2. Surprise-mark .82**

3. Surprise-object .82** .34*

4. Age group .32* .21 .31*

5. Personal pronoun .17 .14 .14 .35*

6. Camera experience .10 .24 -.07 -.20 -.05

7. Identifying box .18 .08 .21 .07 .09 -.11

8. Selecting correct cup -.05 .04 -.12 .38* .09 -.23 .03

9. False-belief task .13 .00 .21 .50** .15 .10 .20 .23

10. TV question .21 .17 .17 .51** .37* -.33* .38* .31* .41*

11. Spontaneous retrieval .64** .63** .42** -.05 .01 .14 .13 -.13 -.05 -.08


* p<.05; **p<.005