Do Young Children of the “Selfie Generation” Understand Digital Photos as Representations?

Abstract

In research from the 1990s, very young children failed to use pictures as representations of real events. Today, many children in the “selfie generation” are constantly photographed by their families using smartphones. While family photos are created, children are exposed to live video on the phone screen that, with a screen touch, becomes an instant photo. Children also revisit these family photos in the phone’s photo library. This study explored whether toddlers growing up around smartphone photography are more successful in applying information from a photo to a real event, compared to children in the earlier research. Sixty 2-year-old children (23.0 to 26.2 months; M = 24.5 months) were asked to use pictures of a toy’s hiding place (printed photographs or digital photos on an iPhone) to search for the hidden toy in 5 conditions. Toddlers were not successful with printed or digital images, whether the digital photos were accessed from the phone photo library or the researcher took the photos during the study. However, after children collaborated with the researcher to create digital photos to help an adult confederate, they were significantly more likely to use photos themselves to solve the search task. Children who experienced this scaffolding with printed photos were somewhat more successful than those without training. As with traditional symbolic media, young children’s learning from emerging technologies needs the support of an adult who co-views the medium and helps clarify the symbolic relation between screen and world.

Many young children are frequently photographed by their families using smartphones (Caron, 2020; St-Esprit, 2018), which are ubiquitous in developed countries. For instance, in a nationally representative US survey, 98% of parents of children birth to age 8 years reported having a smartphone (Rideout & Robb; 2020), as did 98% of families of children under age 5 surveyed in Korea (Change et al., 2018) and 99% of parents of children under age 3 in a UK study (Taylor et al., 2018). While family photos are created, children are exposed to live video on the phone screen that becomes an instant photo. Many children also view family photos in the phone’s photo library. As mature users of representational (symbolic) media, adults may assume that the connection between pictures that were just taken and the familiar people, pets, and events that are depicted must be completely obvious to young children. However, research indicates that the symbolic relation between an image and what it stands for is not always clear to young children (Callaghan et al., 2004; DeLoache & Burns, 1994; Troseth, 2003a; Troseth & DeLoache, 1998). In many cases, children need explicit scaffolding or active mediation from older people to understand digital and analog representations (Callaghan, 1999; McClure et al., 2018; Myers et al., 2018; Strouse & Troseth, 2014; Strouse et al., 2018).

Children’s understanding of pictures is an important issue, particularly with emerging technology offering children a panoply of screen images and the opportunity to learn from them. Educational smartphone apps for children rely on pictures to convey information, and watching videos on parents’ phones or tablets is a common childhood activity. Children are first exposed to many object categories through representational media, particularly digital and printed pictures (DeLoache & Burns, 1994). For example, they may see a picture of a zebra in an ABC book or video before encountering one at the zoo. Realistic pictures have the universal ability to communicate information without the need for language (Gombrich, 1974). Even nonhuman animals can interpret the contents of pictures: monkeys recognize group mates in photographs (Pokorny & de Waal, 2009) and dogs sniff the hindquarters of other dogs depicted in life-size photos (Cabe, 1980, as cited in Thomas et al., 1994).

In an early study to explore whether picture perception in human children was automatic, Hochberg and Brooks (1962) attempted to keep their infant son away from all pictures until he was 19 months of age. Despite this lack of experience, when first exposed to pictures, the child immediately recognized and named photos and line drawings of familiar objects and people. Other studies demonstrate that very young infants can recognize objects and people depicted in pictures and discriminate depicted objects from actual objects (e.g., Barrera & Maurer, 1981; DeLoache et al., 1979; Dirks & Gibson, 1977; Rose, 1977; Slater et al., 1984). An initial tendency of young infants to try to physically manipulate objects depicted in pictures or video (e.g., by attempting to grasp 2-D pictured objects) is replaced by pointing at the images and talking about them by 19 months (DeLoache et al., 1998; Pierroutsakos & Troseth, 2003); that is, toddlers adopt the way that adults behave toward pictures as objects of contemplation rather than action (Callaghan et al., 2004; Werner & Kaplan, 1963).

This beginning of “pictorial competence” eventually leads to what DeLoache (1987) calls dual representation: thinking about a picture as an object with particular qualities (e.g., a flat, colorful, patterned surface on a screen or paper) as well as considering what the picture stands for or represents—necessary because “[a] picture is both a surface in its own right and a display of information about something else” (J. Gibson, 1979, p. 282). To use a picture as a representation requires generalizing meaning beyond the 2D image. For example, when a very young child sees a picture that is labeled by someone older (e.g., “This is a zebra”), they might simply form an association between the word and the picture, or they might understand that both the picture and the label represent an entity that exists beyond the picture, in the world. Preissler and Carey (2004) clarified what toddlers (18- and 24-month-olds) understand about pictures by teaching them the word whisk using a line drawing of a whisk. Afterwards, they presented children with the line drawing and the actual kitchen utensil and asked them to identify the whisk. Both age groups were more likely to choose the 3D whisk than the drawing from which they had learned the word; a control condition ensured children were not biased to always choose any real object over a photo. These results suggest some understanding of representation that generalized beyond the picture (also see Ganea et al., 2008, 2009). Other studies show that this early understanding is not universal, but depends on supportive experience with adults showing, labeling, and discussing pictures (Callaghan et al., 2011; Walker et al., 2013).

Although toddlers display some representational understanding in the picture labeling task, they have trouble realizing how pictures relate to (and inform viewers about) changing situations in the real world. For instance, in research by Harris and colleagues (1997), children watched an adult squirt ketchup on a toy pig and were then shown three photos (the clean pig, the pig with ketchup, and the pig with a white spot). Children under age two did not reliably identify which picture best resembled the toy’s current state. Thus, although very young children recognize the contents of pictures, they lack a full understanding of how pictures relate to ongoing reality (also see Robinson et al., 1994; Zaitchik, 1990).

Children of this age also were challenged by the request to use pictures to solve a problem. In DeLoache and Burns’ (1994) classic object retrieval task, when 24-month-olds were shown a photo of a piece of furniture and told that a toy was hidden there, they did not reliably use this information to find the toy in an adjoining room. DeLoache and Burns carried out a series of follow-up experiments to try to help toddlers grasp the representational relation between the photo and the hidden toy’s location, but no experimental manipulation succeeded: children’s correct search across trials did not improve when they were shown photos of one hiding place at a time, or photos where the toy could be seen secreted in its hiding place, or when the researcher pointed to a piece of furniture in a wide-angle photograph depicting the whole room. They did not seem to realize that the photos provided information relevant to the task at hand. However, children were successful when told verbally where to search, indicating that at this age, they could complete the basic task of searching without perseverating (returning to the previous hiding location) when they knew they possessed information to guide their search.

Compared to mastering this search task, children may have a relatively easy time using pictorial representations in Preissler and Carey’s (2004) labeling task because the procedure echoes what many parents do while reading picture books: they label a picture using the same language they would use for the referent, saying (for instance), “This is a zebra” rather than “This is a picture” (Deloache & Burns, 1994; Gelman et al., 1998; Thomas et al., 1994). Because children in picture-rich societies are socialized how to respond to pictorial representations (Callaghan et al., 2011; Gelman et al., 2005; Szechter & Liben, 2004; Tomasello, 1999; Walker et al., 2013), the familiar format of the labeling task likely supported children’s understanding. The search task, on the other hand, requires toddlers to use photos in a novel way, to gain information about an event in real time that they did not witness directly.

One condition in DeLoache and Burns’ (1994) research indicates that toddlers can map between a picture and its referent when a speaker refers to an object by pointing to its picture (rather than stating its name) as part of a directive sentence. Across trials, children reliably succeeded when the researcher simply asked them to place a Big Bird toy “right there in his room” [pointing to a depicted furniture item]. To do so, children needed to identify the pictured object, realize that it corresponded to a particular piece of furniture in the room, and understand the researcher’s verbal directions. In the search task, children needed to do these things plus realize that the picture specified a current situation. Based on the picture and the researcher’s words, they had to mentally construct an event they had not directly seen (the researcher hiding the toy in the room next door). To do so, they needed to realize the researcher was using the picture to give them information to solve the problem. Each time the researcher indicated a new hiding place, children needed to update their mental image of the toy’s location based on the new picture. This task is challenging because after the first trial, children must choose between conflicting pieces of information: their memory of where the toy was the last time they searched (i.e., information from their direct experience), and new information from a picture (Suddendorf, 2003; Troseth, 2003b). Yet using photos to update one’s knowledge and keep informed of current events is an important aspect of pictorial competence, something that older children and adults regularly do (DeLoache et al., 1996).

At the time DeLoache and Burns’ research was carried out in the 1990s, young children’s pictorial experience consisted of exposure to pictures in books, pictures on their clothes and highchairs, and family photos in frames and albums. Such pictures typically “provide no immediate real-world information” (Ittelson, 1996, p. 173); when a parent reads a book to a child before bed, the zebra on the page is unrelated to conditions in the child’s bedroom. In the 1990s, family photos depicted familiar people, but represented past events. Most children rarely if ever saw pictures representing current situations.

Today, this is a common experience. Digital cameras allowing instant photo taking and viewing were introduced during the 2000’s. Initially, these cameras were bulky and expensive. Now almost all mobile phones have instant picture capabilities, and people of all socioeconomic backgrounds possess smartphones (Canadian Radio-television and Telecommunications Commission, 2016; Kabali et al., 2015; Pew Research Center, 2016), including 95% of lower-income US families with young children (Rideout & Rob, 2020). Toddlers in Southern India use smartphones an average of 45 minutes a day (Varadarajan et al., 2021) and children ages 3 to 6 from a medium-to-low-SES urban Chinese community average more than an hour a day using a smartphone or iPad (Chen et al., 2020). Saudi children (1 to 3 years) spend an hour a day on mobile devices (Alroqi et al., 2021).

Built-in webcams and front facing cameras (on smartphones, tablets, and other computers) allow children to see themselves in real time on the screen during the taking of photo or video “selfies.” Due to this experience with pictures that clearly show the present, young children today might better understand the connection between photos and real events. Alternatively, children may still need scaffolding from adults to understand how a 2D image can relate to reality.

Besides “selfie” photographs, other kinds of pictures in emerging digital technology may be confusing for young children, compared to understanding traditional media (Troseth et al., 2019). Photos in books and photo albums might represent a specific individual (a story character, a family pet) or a generic category (zebras). Video often represents the past (e.g., a home video, a documentary) but video chat represents the present moment. Computer generated images can look completely realistic, yet depict imaginary things (e.g., a boy riding a dragon). Augmented Reality apps and games (e.g., Snapchat, Pokémon Go) and Virtual Reality blend the real with the fictional (Bailey & Bailenson, 2017). Adults may need to support young children’s understanding of how these new kinds of pictures relate to the real world.

Research with video has shown the value of adult scaffolding (active mediation or “co-viewing”) to promote children’s understanding of pictorial images. After 24-month-olds watched on live video as a researcher hid a toy in an adjoining room, they failed to use this information to find the toy (Troseth & DeLoache, 1998). In one study, a door between the rooms was left open, allowing children to watch and compare the real event to its video representation (Troseth 2003a). Children succeeded on four training trials while they could see the actual hiding event, yet they failed to search correctly once the door was closed and they had to rely only on the video. However, when an adult co-viewer explicitly highlighted the representational relation between video and reality, toddlers were able to use video images to solve problems and to learn words (Strouse & Troseth, 2014; Strouse et al., 2018; Troseth 2003b). Naturally occurring everyday experience with live video (on parents’ flip-screen video cameras; on security monitors in stores) also related to success in the search task (Troseth, Casey, et al., 2007).

The current research builds upon prior work by exploring how photography on mobile phones might affect children’s understanding of photos, using the object retrieval search task paradigm with 24-month-old children. We examined what children would need in the way of picture format, scaffolding, and experience to demonstrate a representational understanding of pictures. In a preliminary study carried out in the early 2000s (before the wide use of digital photography), we first explored whether experience with Polaroid or digital photography would help children gain insight into pictures representing current reality. In our main experiment, we investigated whether children today more easily use pictures shown on a digital device (an iPhone) for information than traditional printed photos, and considered how their use of pictures compared to that found in earlier research. We also examined whether today’s “selfie generation” needed specific experience that emphasized the representational role of pictures or if they came to the lab task possessing insight into the connection between pictures and current reality from everyday experience with their parents’ mobile phones.

Preliminary Study

In the 1990s, Polaroid “instant” pictures were rare instances when printed pictures represented events in real time—specifically, events that occurred about a minute before. DeLoache and Burns (1994) used an instant Polaroid camera in one condition of their study; they demonstrated taking pictures of a toy sitting on several pieces of furniture during an orientation, with the camera rigged to immediately eject a fully developed (pre-taken) Polaroid photo. Then the researcher took the Polaroid camera with her while hiding the toy, and returned to show children a Polaroid photo of the hiding place. Two-year-old children did somewhat, but not significantly, better (27%) than with standard printed photos (6–23%), reflecting that a couple of children used the Polaroid photos for information. Children’s lack of experience with and understanding of Polaroid cameras could account for the failure of these instant photos to clarify the representational connection for most of the toddlers.

In a small preliminary study in the early 2000s, we gave 24-month-old children from 18 families photo experience and looked for improvements in their use of pictures in the search task. First, all children were tested in the lab for their baseline success using printed photos. Six families owned early digital cameras; they were asked to take digital pictures of their children engaging in distinctive play activities (e.g., blowing bubbles) at least 5 times during the next 2 weeks and to immediately show their children the photos, pointing out the connection between these temporally-relevant pictures and current reality.

The other 12 families were randomly assigned to use either inexpensive disposable cameras or Polaroid cameras. (We supplied the cameras). Parents using Polaroid cameras were asked to show their children each photo as soon as it developed (about 1 minute later) and to make connections to the just-past pictured event. Families given disposable cameras served as a control group: parents took pictures of their children in the same way, but the pictures were not immediately available to view since the film needed to be developed and the pictures printed. Thus, one difference between conditions was whether children were shown pictures that depicted current reality (digital photos), just-past reality (Polaroid photos), or no exposure to such photos. Additionally, parents in the two “instant” photography groups were asked to scaffold their children’s understanding of the connection between pictures and reality. Because those in the digital photography group possessed an expensive item of relatively new technology, it was possible that they differed from the other families in important ways, such as having access to more resources. However, average family income (ranging from $30,000-$50,000 to $75,000-$100,00 for the study participants) was somewhat lower in the group with digital cameras (M = $50,000-$75,000) than in the other groups (M’s = $75,000-$100,000).

At the end of the 2 weeks, families returned to the lab and children again participated in the object retrieval task with traditional printed pictures. To ensure fidelity of implementation of the home activities, we printed out the photos and checked parent diaries of the photo-taking activities. Families in both the Polaroid and digital groups took an average of 12.5 photos, whereas families with the disposable cameras averaged 17.5 photos.

Children randomly assigned to the disposable camera control group exhibited the expected low search performance on lab visit 1 (13% errorless retrievals); they did slightly better on visit 2 (17%). Children randomly assigned to get experience with Polaroid photos had 25% baseline performance on visit 1because one child succeeded on all 4 trials. After receiving experience with instant Polaroid photos at home, the children found the toy on 33% of trials (two children were successful). The 6 children whose parents had digital cameras were unsuccessful on visit 1 (8%). However, after being exposed to pictures that showed current reality for 2 weeks, they improved substantially in their search performance on visit 2 (61%). This increase in success using printed photos was significant in a non-parametric (due to non-normal data) Related-Samples Wilcoxon Signed Rank Test, p = .043.

This small study suggested that toddlers’ exposure to instant photos that displayed a just-prior event that they witnessed might help them recognize how a picture could provide information relevant to a current situation. Due to the small number of children tested, and changes in how children are currently exposed to digital images, further research was warranted. It was possible that children today, being regularly exposed to digital photography on smartphones, would not need special researcher-assigned exposure to photos representing current reality to immediately use pictorial information to solve a real-world problem.

The main study below was based on the hypothesis that children’s experiences with modern smartphone picture technology might impact their understanding and use of photos. We examined whether the ability of digital photographs to provide temporally relevant information about real situations might help 24-month-old children today use information from photographs to solve the object retrieval task, compared to their counterparts back in 1994.

Further, if children’s concept about the function of pictures (Troseth et al., 2004) is tied to experience with the device used to take and display such photos, young children today might relate a photo shown on a smartphone to current reality with greater ease than the same photo in printed form, displayed in a picture frame. On the other hand, it could be that young children still require scaffolding by an adult highlighting the relevance of the photo. This notion is supported by a previous study using a search task with a scale model, in which children benefited from a researcher describing the intentional origins and function of the model to stand for the search room (Sharon, 2005; also see Peralta et al., 2013). Therefore, 24-month-olds participated in one of five conditions in which a researcher used printed or iPhone photos to give them information about the location of a toy. In two of the conditions, children received additional training to highlight the representational role of photos.

Experiment 1

Method

Participants

Sixty children from a city in the southeastern United States (33 males; 27 females) participated in one of five conditions, with a similar average age range (23.0 to 26.2 months; M = 24.5 months) as in previous research. Participant families were recruited by phone using state birth records and randomly assigned to condition with age, sex, and parent socioeconomic status (self-reported income and education level) approximately matched across condition. Parents reported that their children were White (80%), Black or African American (8.3%), Asian (3.3%), or other racial background (1.7%); race was not reported for 6.7% of children. Three children were Hispanic or Latino (5%). Family income ranged from $15,000-$30,000 to over $150,00 (median = $100,000-$150,000). Parent education ranged from high school diploma to doctoral degree; median joint parental education (numerical average of both parents’ highest education level) fell between “some college/associate degree” and “doctoral degree”. All children were learning English as their first language. An additional 11 children did not complete the task due to fussiness; their data were dropped from analyses. Data were collected in 2016–2017, before any recent changes in digital media use due to the Covid-19 pandemic.

Materials

The experiment took place in two adjoining rooms: a waiting room with chairs for the child and parent, and the search room (6.8m x 3.1m), which included a small round table with a green tablecloth, a sofa with a red blanket draped on it, a blue upholstered chair, a wooden end table with a straw basket on top, and a red floor pillow. In some conditions, 4in x 6in (10.16 × 15.24cm) printed color photos of the furniture (see Figure 1) were displayed in plastic frames as in previous research. In other conditions, digital photos of the furniture were displayed on an iPhone 5S with a 2in x 3in (5.08 × 7.62cm) screen (either accessed from the camera’s photo library or taken in real time, depending on condition). The object to be hidden was a 20 cm-tall Piglet stuffed animal toy (see Figure 1). The entire session was recorded with two digital video cameras (one in the search room and one in the waiting room), with output recorded in a picture-in-picture format for later coding. Parents completed a questionnaire regarding the child’s exposure to digital media (e.g., taking pictures with smartphones, video chatting).

Figure 1.

Photos of the hiding locations and the toy that was hidden.

Study design

In the five conditions, we varied the kinds of photos children were shown and the support they were given. Table 1 outlines the planned comparisons between conditions based on our a priori hypotheses. The first condition was nearly identical to that used by DeLoache and Burns (1994): the researcher hid Piglet and then showed children a printed photo of the hiding location to tell them where to search. The photos were the same dimensions as those used in the earlier research. In the iPhone photo condition, the researcher showed children the same pictures on an iPhone screen (accessed from the phone’s photo library). Due to the size of the phone screen, these pictures were only half as big as the printed photos. Although this size difference put children in the iPhone conditions at a slight perceptual disadvantage, we hypothesized that they might spontaneously understand the role of smartphone photos in conveying information, due to experience with family members taking and instantly viewing digital photos with them.

Table 1.

The 5 Photo Conditions and Planned Comparisons

Condition	Brief Description	Planned Comparison
Printed photo	Similar to DeLoache & Burns, 1994, Experiment 5
iPhone photo	Same photos displayed on an iPhone	Printed photo
Live iPhone photo	Researcher takes iPhone photos in “real time”	iPhone photo
Live iPhone photo + training	Training: Take Live iPhone photos to help someone else	Live iPhone photo
Printed photo + training	Training: Show printed photos to help someone else	Printed photo

Because our participants were very young, it also was possible that their cumulative experience with smartphone photos might not have given them insight into the connection between photos and what they represent. In the third condition, we attempted to scaffold children’s understanding by highlighting the temporal immediacy of the photos and their relation to current events in the lab, capitalizing on the ability of smartphones to take and view photos in real time. On each trial in the Live iPhone Photo condition, the researcher told children that they took the photo right after hiding Piglet. Taking and immediately showing photos might echo children’s prior smartphone experience, so children might make better use of these pictures for finding Piglet compared to pictures from the phone’s photo library.

On the other hand, toddlers might need the connection of photos to current reality to be explicitly demonstrated. In previous research with video, scaffolding by an adult helped children to recognize and to learn from the connection between video image and reality (Strouse & Troseth, 2014; Troseth, 2003b). To give toddlers the optimal chance to understand the informative role of a photo, we had children participate in creating photos for someone else. During four training trials, children helped the researcher hide Piglet and take an iPhone photo of the hiding location for another adult (the assistant) who then used the photo to retrieve the toy. Taking and showing the photo to the adult highlighted for children the temporal contingency between the photo and the hiding event. We expected this training to significantly boost children’s later use of photos as information to retrieve Piglet, compared to children in the Live iPhone condition who used photos supposedly just taken on the iPhone, but without the prior experience of creating photos to help someone else.

To probe how important the “instant” nature of digital photography was in the training, children in a final condition helped the researcher hide Piglet in a location that matched one of the printed photos. Then the researcher and child showed the photo to the assistant, who searched for and found Piglet. This training emphasized the connection between photos and hiding events, but without the temporal contingency of making instant photos. In both training conditions, the shared activities highlighted the intentional creation or selection of a photo to convey information—as Sharon (2005) did with a scale model—and provided children a successful example of the task solution (use of the photo for information) by the assistant.

Procedure

Lab sessions lasted approximately 30 minutes and included a warm-up, an orientation phase, a placement trial, and 4 search trials. First the child and researcher spent a few minutes playing with blocks while an assistant described the experiment to the parent and obtained informed consent. Next, the researcher introduced the child to Piglet and invited the child and parent into “Piglet’s room” for the orientation. In the first three conditions, the child continued with the test trials directly following the orientation. In the latter two conditions, the child took part in four training trials and then proceeded to the test trials.

Orientation phase.

The researcher named each furniture item and explicitly pointed out the connection to its photograph. In the conditions using printed photos and pre-taken digital photos (from the iPhone’s photo library), the researcher said: “This is a picture of Piglet’s [furniture item]. And look, here is Piglet’s [actual furniture item]! See, they’re the same!” In the conditions that involved taking real-time iPhone photos, the researcher said: “I’m going to take a picture. [Takes photo.] Let’s look at the picture I just took,” then made the identical comparison between picture and furniture. The last item compared to its photo was always the floor pillow. The researcher then accompanied the child and parent back to the waiting room.

Placement trial.

This phase was designed to emphasize how a picture can be used to give information. In all conditions, the researcher showed the child the photo of the floor pillow, stating: “Piglet wants to go to his room and sit right here. You take him and help him sit right here.” If the child did not follow the direction, the researcher gave hints, re-emphasizing the relation between the photo and the actual pillow. After the child placed Piglet on the pillow (with or without help), the child and researcher returned to the waiting room.

The parent was asked to stay in the waiting room and work on filling out the questionnaires. If the child needed the parent to be in the search room for comfort, the parent was instructed to not provide any verbal or nonverbal hints during the test trials.

Search task.

On each of four trials, the researcher showed the child either a printed photo of a hiding place or the same photo on the iPhone (depending on condition), saying, “I’m going to hide Piglet right here in his room, and when I come back you can go and find him!” In the first two conditions, the researcher then handed the photo/iPhone to the assistant and went into Piglet’s room, closing the door. The assistant pointed to the hiding location on the photo and reminded the child, “[Researcher] is hiding Piglet right here in his room!” Hiding places were not labeled, so children had to rely on the pictures rather than on verbal cues for information.

The “Live” iPhone condition was the same, except the researcher told the child, “I’m going to hide Piglet in his room and take a picture of where I hide him. Then, I’ll come back and show you the picture of where I hid Piglet, and you can go and find him there!” While the researcher was in Piglet’s room, the assistant reminded the child, “[Researcher] is hiding Piglet right now in his room. She is taking a picture to show us where she hid him.” (The researcher actually selected the image from the iPhone’s photo library for consistency across conditions.)

In all conditions, the researcher then returned to the waiting room, pointed to the photo, and said, “Piglet is hiding right [in/under] here in his room. Go find him!” If the child went to an incorrect location, the researcher provided a series of increasingly explicit prompts, actually helping the child to retrieve Piglet if necessary. The order of hiding locations was counterbalanced across participants.

Training phase.

In the two “training” conditions, before the search trials, children helped the researcher hide Piglet for an adult assistant to find, while taking a photo (iPhone + Training condition) or using the printed photo (Printed + Training condition) of the hiding place. Then they returned to the waiting room to show the assistant the photo. The researcher’s comments to the assistant matched the language directed to the children in the test phase for the respective condition. The child accompanied and helped the researcher hide Piglet; the researcher either took a photo of the hiding location with the iPhone camera in real time, or pointed out to the child the connection between the printed photo they had brought with them and Piglet hiding in that location. After returning to the waiting room, the researcher addressed the assistant following the same, condition-specific script described in the test phase. On each of the four training trials, the assistant demonstrated understanding by looking carefully at the photo, nodding, and immediately retrieving Piglet. After the fourth training trial, the researcher told the child that now it was their turn to search for Piglet, and the test phase began.

Measures

Children’s search behavior was scored during the session by the assistant, and later coded from a video recording by a different trained research assistant. In the rare case of a disagreement, the videos were re-scored and an agreement was reached. Consistent with the criteria in previous research, only errorless retrievals on the first search were considered correct. Children received a score from 0 to 4 across the four search trials, with 4 representing 100% errorless retrievals. First trial performance was also analyzed separately from trials 2 through 4, resulting in a score ranging from 0 to 3 across the last three trials.

Results

Prior media exposure

According to parent responses to the media questionnaire (missing for one participant), 95% of the toddlers had seen themselves on a screen (phone, tablet, computer, camera) during the previous month while someone filmed a video or took a photo of them, with 42% experiencing this daily or multiple times a week. Nearly half (44%) of the toddlers had taken a photo or video “selfie” themselves during the previous month. Parents reported that 90% of the toddlers engaged in video chat at least once a month, with 17% video chatting daily or multiple times a week, and another 27% video chatting weekly. The average age of first exposure to video chat was around seven months of age (M = 7.22 months, SD = 5.77), and parents believed that children were actually engaging in the video chat experience from around their first birthday (M = 12.8 months, SD = 6.10). Besides these different kinds of live video and instant photo exposure, 97% of the children had seen themselves on a screen while watching the replay of a video or display of a photo that was taken at an earlier time, with 44% viewing such playback of images daily or multiple times a week. See Table 2 for media exposure across condition.

Table 2.

Media Exposure Across Condition

	Screen View During Photography/Videos		Views Photos/Video Replays		Makes Selfie Photos/Videos		Video Chat		Average Frequency Across Media
Condition	M	SD	M	SD	M	SD	M	SD	M	SD
Printed photo	2.45	0.82	2.36	1.03	1.27	1.56	2.27	1.01	2.09	1.20
iPhone photo	2.00	1.20	2.08	1.22	1.00	1.16	1.36	0.98	1.62	1.23
Live iPhone photo	2.42	1.01	2.50	0.74	0.83	1.07	1.70	0.68	1.87	1.11
Live iPhone photo + training	2.00	0.76	2.00	0.98	0.75	1.17	1.36	0.52	1.51	1.02
Printed photo + training	2.64	1.03	2.55	1.09	0.73	0.97	1.82	1.09	1.93	1.25

Note. Parent responses to questions: “In a typical month, approximately how frequently has your child [had this media exposure]?”

0 = never; 1 = once a month; 2 = once a week; 3 = several times a week; 4 = daily

In this small sample, we found no significant correlations between the amount of prior digital video/photo experience (which was uniformly high) and children’s success on the search task. There were no significant condition differences in prior experience; however, children randomly assigned to the iPhone and Live iPhone with Training conditions happened to have somewhat lower exposure from their daily lives compared to children in the other conditions. Thus, if anything, less prior exposure worked against our hypothesis of better search performance in these supportive conditions.

Search performance

In each planned comparison across conditions, we looked at overall performance across the 4 search trials, and then separately examined children’s success on the last 3 trials, when children’s memory of where they found the toy (with or without help) on the previous trial competed with information from a photo they were just shown in directing their search. This latter analysis provides a stringent test of children’s understanding that the picture provides the relevant, current information. A summary of children’s mean search score and standard deviation across the conditions is presented in Table 3. Percentages are reported in the text to allow comparison to previous research. On Trial 1 search performance, there were no significant condition differences; we report mean percentages for comparison.

Table 3.

Mean Errorless Retrievals and Standard Deviations Across Condition

	Trials 1–4		Trials 2–4
Condition	M	SD	M	SD
Printed photo	0.42	1.00	0.25	0.62
iPhone photo	0.92	1.56	0.67	1.15
Live iPhone photo	0.92	0.79	0.33	0.49
Live iPhone photo + training	1.83	1.19	1.33	0.98
Printed photo + training	1.33	1.55	1.08	1.31

Printed photo vs. iPhone photo.

On the first planned comparison of children’s use of traditional printed photos or the same images on an iPhone screen, the condition differences across all 4 trials (Printed photo, 11% correct; iPhone photo, 23%) and on the last 3 trials (Printed photo—8.3%; iPhone photo—22%) were not significant. On the first trial, with no memory of a previous search trial competing with information offered by the picture, 17% of toddlers searched in the place shown in the printed photo, compared to 25% with iPhone photos. For most of the 24-month-olds, merely seeing a photo that the researcher displayed on an iPhone did not help them realize that the picture told them where to find the hidden toy.

iPhone photo vs. Live iPhone photo.

We next examined whether children were significantly more successful if the photos they were shown supposedly were taken by the researcher using the iPhone right after hiding Piglet, rather than being photos from the iPhone photo library. No significant condition difference emerged across all four trials (iPhone photo—23%; Live iPhone photo—23%) nor on the challenging trials 2–4 (iPhone photo—22%; Live iPhone photo—11%). On Trial 1, 58% of toddlers who were told the researcher had just taken the iPhone photo were successful, compared to 25% with the pre-taken iPhone photo (not a significant difference); this short-lived “improvement” was not sustained across trials.

Live iPhone photo vs. Live iPhone photo + Training.

Toddlers who received experience helping the researcher make photos to inform the assistant achieved significantly more errorless retrievals across their own four search trials (46% success rate) compared to toddlers in the Live iPhone Photo condition (23%; t(22) = 2.22, p = .037, d = 0.95). Across the two conditions, children’s success rate was similar on Trial 1 (50% with training vs. 58% without) but not on subsequent trials: toddlers with the training experience continued finding the toy on 44% of the trials, whereas searching by toddlers without this helpful experience dropped to only 11% correct, t(22) = 3.15, p = .005, d = 1.34.

Printed photo vs. Printed photo + Training.

Toddlers who received the extra training with printed photos performed marginally better (33% errorless retrieval rate) than toddlers who used printed photos without the training (11%), t(22) = 1.716, p = .10. Similarly, the errorless retrieval rate for trials 2–4 was 36% for children with training compared to 8.3% for those without, a marginally significant difference with a large effect size, t(15.7) = 1.99, p = .064, d = 0.81 (degrees of freedom correction from 22 to 15.7--Levene’s test for equality of variances).

Individual performance across conditions.

Many 24-month-olds found the search task with pictures challenging. Looking at individual data, a few children were quite successful (3 or 4 trials correct): 1 child using Printed photos, 2 with iPhone photos; 2 with Live iPhone photos plus training, and 3 with Printed photos plus training. Additionally, a number of children were correct on 1 or 2 trials: Printed: 1 child; iPhone: 2 children; Live iPhone: 8 children; Live iPhone + Training: 11 children; Printed + Training: 3 children. The condition in which the most children (11 of 12) had some success was after receiving the training with Live iPhone photos.

Discussion

This research sheds light on very young children’s learning from picture images on smartphone screens, and shows the need for social support in the form of adult scaffolding for toddlers to recognize the connection between digital pictures and what they represent. The default for parents of the “selfie generation” is to have a smartphone, and therefore a camera, with them throughout the day, making it easy for them to snap pictures of their growing children (Caron, 2020; Rideout & Robb, 2020; St-Esprit, 2018). Despite frequent exposure to the taking and viewing of instant pictures, toddlers in the current study did not reliably use digital pictures on an iPhone nor printed pictures as a source of information to solve a search problem, with scores similar to children’s success rates in the 1990s (6% – 23%), before the advent of smartphone photography (DeLoache & Burns, 1994; Troseth, 2003b).

In the Live iPhone photo condition, when the researcher emphasized the temporal contingency between a photo and a current event by taking and displaying instant photos of the hiding place during the search task (rather than using images from the phone photo library), children were initially more successful (on trial 1), but they did not continue to use information from the instant photos on later trials. One potential interpretation is that, following right after the orientation (where instant pictures were taken and compared to the furniture/future hiding places), this experience merely helped children detect the correspondence between a photo and a piece of furniture in the search room, but not to recognize the photos as representations intended to convey information (Callaghan et al., 2004). Or using live iPhone photos may have helped children to achieve “representational insight” into the symbolic relation (Suddendorf, 2003; Troseth, Bloom Pickard, et al., 2007), but this insight was fragile, only sufficient to guide their search on trial 1, when they had no conflicting memory of previously finding the toy in the room (Suddendorf, 2003). Suddendorf reported that toddlers could use a photo to search correctly in a single place in each of four different rooms in succession (in essence, on four “first trials” untainted by a conflicting, outdated memory), but not on four trials in the same room. In general, very young children are challenged by situations where they must activate and maintain currently relevant information in the face of conflicting information, particularly if they previously marked the needed information as irrelevant (Chevalier & Blaye, 2008). In the case of using representational media, for a child of this age to choose information from a photo over a memory from direct experience requires a deeper awareness of the symbolic relation.

In the current study, only when the researcher gave the child a brief “apprenticeship” in making iPhone photos to help another person search did the children maintain the same level of success across trials, often using the representation (the photo) as the basis for their search rather than being guided by their memory of a past successful search. In this training condition with the live iPhone photos, helping the researcher take a picture further emphasized the temporal connection between image and hiding event. Children who received this experience were significantly more successful in using photos to guide their searching, in the same way toddlers benefitted by getting experience with instant digital photographs of themselves in the preliminary study reported here. A crucial part of the symbolic experience in both cases was an adult (the parents in the preliminary study; the researcher in the main study) explicitly pointed out the connection between the instant photo and the real situation that was depicted. Similarly, when parents in an earlier study showed toddlers live video of themselves on their home television for two weeks, pointing out the connection between image and real event, children succeeded in search tasks in the lab with video and photos (Troseth, 2003b). Parent scaffolding helps young children figure out how pictures, video, and screen images during video chat relate to reality (McClure et al., 2018; Myers et al., 2018; Strouse & Troseth, 2014; Strouse et al., 2018).

Additionally, the training trials may have helped children to understand what was happening when the researcher went into the adjoining room “to hide Piglet” and what the adult intended when they returned and displayed the photo. Making the connection between a photo and an event not directly seen (the hiding event in the room) may be a challenging idea for very young children--even for a just-past event in a familiar space (the search room). In a training study by Peralta and Salsa (2008), children watched a doll being hidden in a room, went to another room with the researcher, and then were asked to point on a picture of the room where the doll was hidden. Children could do so because, “the mental image the children needed to guide their behaviour [pointing on the photo] is based on their direct experience with reality. On the other hand, in the search task the picture represents a source of new knowledge about a current unseen reality” (p. 10). Connecting between a seen event and a photo in the point task seems to have helped toddlers subsequently to understand that the researcher showed them photos in the search task to convey information, and they succeeded.

Finally, the live iPhone training condition may have supported children’s understanding by revealing the researcher’s communicative intentions behind making a picture to inform the confederate, who used it to guide their behavior. In a study by Peralta and colleagues (2013), when the researcher stressed her intention behind showing photos, emphasizing that the pictures were helpful sources of information to find the toy in the room, this helped children 6 months older than our participants to reliably use the photos. Similarly, when the maker of a different representation--a scale model of a room--highlighted her intentions for the model to stand for a larger room, this helped preschoolers to apply information from the model to find a toy hidden in the room (Sharon, 2005). In essence, the experience of helping the researcher make photos for someone else in the current study may have helped children get some beginning understanding that a photograph is “an intentional selection and instantiation of some referent” that emerges from “the picture-maker’s role and intentions” (Liben, 2005).

There was no significant difference (but a large effect size) between toddlers’ success using printed photos with and without the additional training. This suggests that all aspects of the training with live iPhone photos, including the temporal contingency between the hiding event and image, and being involved in taking the photo, were important. Although children who had the extra training with printed photos did not start off successful, across trials their success improved somewhat, suggesting that the training helped them to gradually recognize that the researcher was giving them information with photos. In a microgenetic study of problem solving in which toddlers displayed gradual improvements after scaffolding across multiple trials, Chen and colleagues (2000) reasoned that instruction “can create a structure for interpreting subsequent experience” so that children learn from their later attempts at a task (p. 19).

Even with support, it was still a challenge for 24-month-old children to apply information from pictures to solve a search problem, and performance was lower than has been found with older children (DeLoache & Burns, 1994; DeLoache, 1991). For mature symbol-users, one of the main functions of photos and video is to learn about events we do not experience directly because of distance in time or location (e.g., Dorothea Lange’s photograph of a migrant mother during the Great Depression of the 1930s; video of the Mars rover being lowered onto the distant planet) or our perceptual limitations to detect an event (e.g., Edgerton’s photo of a drop of milk hitting a surface--Liben, 2003). More mundane examples of learning about absent things from pictures abound (e.g., pictures of distant friends’ growing babies posted on social media). Yet this understanding is not automatic in very young children.

Because many children receive substantial experience with parents labeling and discussing pictures in books, they seem to understand relatively early that pictures can represent categories (e.g., zebras) as verbal labels do (Preissler & Carey, 2004; Ganea et al. 2008, 2009). In the US, 46% of children under age 2 are read to daily, and 73% are read to at least weekly (Rideout & Robb, 2020). Young children (age 3 to 6) in medium-to-low SES families in China were exposed to books (i.e., shared reading with parents and/or read themselves) at least multiple times a week (Chen et al., 2020). The most common parent-child co-use of digital technology reported for 3- to 6-year-olds in Hong Kong involved sharing on-screen pictures and photos (Wu et al., 2014). Cross-cultural research shows the connection between adult support and children’s picture understanding: middle-class Canadian and American children learned the novel names of pictured objects and generalized those labels to the actual objects at a younger age than children did in Peruvian, Indian (Callaghan et al., 2011), and Tanzanian (Walker et al., 2013) village cultures where parents do not typically offer scaffolded picture experience.

However, few parents in any culture may habitually point out the connection between pictures and current events, making the idea that pictures can update one’s real-world knowledge a later development. For instance, young children in research studies did not correctly pick which photo depicted a toy’s current condition (a pig toy that they watched being squirted with ketchup; Harris, Kavanaugh, & Dowson, 1997). When children have conflicting information such as prior direct experience, understanding the relation between a photo and a current situation is a particular challenge (Zaitchek, 1990).

A full understanding of how and what pictures represent develops across childhood (DeLoache et al., 1996). Besides figuring out how pictures relate to the past and present, children must learn when pictures represent specific entities (e.g., video of a new-born baby zebra at the local zoo) and when they stand for generic classes (“zebras” in an ABC book; DeLoache & Burns, 1994). They also must learn what a drawing requires to be a good representation (Callaghan, 1999) and they continue to learn how a picture-maker’s intention is related to depiction in images (Bloom, & Markson, 1998; Gelman & Ebeling, 1998). Additionally, they must grapple with which properties pictures share with depicted referents; for example, children eventually distinguish that an actual, physical glass of milk would spill if turned upside down, but a photo of a glass of milk would not (Flavell et al., 1990).

Limitations and Future Directions

Among the American participants who volunteered for the current research, there was relatively little ethnic/racial or socioeconomic diversity. Because there are income-based and cultural differences in media exposure that might affect children’s learning from digital images, research in this area warrants future replication in other groups and cultures. For instance, in Rideout and Robb’s 2020 U.S. population survey, lower-income families of young children (birth to age 8) have less access to a home computer (63%), tablet (59%), and high-speed internet (74%) at home compared to more affluent families (95% have home computers, 83% tablets, and 94% high-speed internet). Another difference is the amount of time spent daily with screen media: children from lower-income homes averaged 1 hour more than children in middle-income families, and 2 hours more than children in higher-income families. Looking at this comparison by race/ethnicity, Black children averaged 2 hours more spent with screen media than White children, and 1 hour more than Latinx children. Before the pandemic, video chat was used very little by young children: Black children video chatted for 2 minutes per day, Latinx children for 1 minute, and White children for less than 1 minute. In our sample, media exposure was not related to children’s success in using digital or analogue pictures to solve a search task, but it is possible that different exposure to smartphones, to video chat, to other screen media, or to reading picture books with parents, might have changed the results.

In some ways, American families are becoming more homogenous in media exposure (Rideout & Robb, 2020): access to smartphones with picture- and video-taking capabilities was consistently high across SES groups in this survey (95% for lower-income families, 99% for higher-income families). Additionally, the greatest increases in time spent reading to young children in the last decade were made by Black families and lower-income families—the only subgroups doing a substantial amount of daily reading on electronic devices. Minutes spent reading daily by Black families (48 minutes) and by lower-income families (43 minutes) were significantly higher than reading times reported by White and Latinx, and by middle- and higher-income families (ranging between 29 and 32 minutes per day).

Although access to smartphones is becoming prevalent across large parts of the world (e.g., Chang et al., 2018; Taylor et al., 2018), there are cultural differences in how much children are exposed to such emerging technologies. For instance, Korean parents (96% of whom owned smartphones) reported that only 12% of their 2- to 5-year-old children had daily exposure to smartphones, and 39% to television--much less than the American children in Rideout and Robb’s survey. Across the African continent, mobile phones are quickly becoming a means of economic development and family connection across distances, but children’s exposure is much less common in rural areas than in the cities (Porter et al., 2012). Therefore, differences in how children respond to and understand digital pictures, based on experience, might well occur.

Supporting children’s engagement with emerging technology

An aspect of pictorial competence that is important for children’s engagement with emerging technology is determining which pictures represent reality, and which pictured entities are imaginary (Troseth et al., 2019). For instance, computer generated images may be indistinguishable from photographs, and live video of real people is blended with fictional features and events in augmented reality. As children (with their limited knowledge of the world) use emerging technology, they need adults to help them navigate the varying relation between depictions and reality.

The growing literature documenting the benefit of social scaffolding of young children’s use of digital media (e.g., McClure et al., 2018; Myers et al., 2018; Strouse et al., 2018) has shaped children’s media policy (AAP, 2016) and points to the importance of support for children’s use of emerging technology. Pictures are ubiquitous in current and emerging digital media. To adults, the meaning of these images may seem obvious and simple. However, as DeLoache noted in 1995, “There are no fully transparent symbols” (p. 113). The need to scaffold children’s understanding and use of digital images will remain an important principle for developing effective emerging technology for young children.

Biography