Computers in Human Behavior 56 (2016) 56e64

Contents lists available at ScienceDirect

Computers in Human Behavior

journal homepage: www.elsevier.com/locate/comphumbeh

Full length article

Young children's transfer of learning from a touchscreen device

Brittany Huber*, Joanne Tarasuik**, Mariana N. Antoniou, Chelsee Garrett, Steven J. Bowe 1,
Jordy Kaufman, The Swinburne Babylab Team2
Swinburne University of Technology, PO Box 218, Hawthorn, Victoria 3124, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Because young children are devoting increasing time to playing on handheld touchscreen devices, un-
Received 5 July 2015 derstanding children's ability to learn from this activity is important. Through two experiments we
Received in revised form examined the ability of 4- to 6-year-old children to learn how to solve a problem (Tower of Hanoi) on a
29 September 2015
touchscreen device and subsequently apply this learning in their interactions with physical objects. The
Accepted 10 November 2015
results were that participants demonstrated signicant improvement at solving the task irrespective of
Available online xxx
the modality (touchscreen vs. physical version) with which they practiced. Moreover, children's learning
on the touchscreen smoothly transferred to a subsequent attempt on the physical version. We conclude
Keywords:
Children
that, at least with respect to certain activities, children are quite capable of transferring learning from
Multimedia touchscreen devices. This result highlights the limitations of generalizing across screen-based activities
Touchscreen (e.g., screen time) in discussing the effects of media on young children's development.
Human-computer-interaction 2015 Elsevier Ltd. All rights reserved.
Transfer

1. Introduction
Plato famously opined, Enforced learning will not stay in the
mind. So avoid compulsion and let your children's lessons take the
form of play (Plato, 1955). Indeed, importance of play in education
and development has been appreciated from Plato's time through
the beginnings of developmental science, and continues today (e.g.,
Golinkoff, Hirsh-Pasek, & Singer, 2006; Frost, Wortham, & Reifel,
2012; Ginsburg, 2007). Given the perceived importance of play, it is
understandable that new and unfamiliar changes to the nature of
children's play tend to elicit societal unease. Instances of this can be
seen with the decline of outdoor play in the last century (Frost,
2010), and as toys with media branding and electronic features
began to saturate the market in the 1970s and 1980s (e.g., Varney,
1999).
While electronic toys and video games were introduced to
children's worlds as early as 1972, the release of the iPad in 2010
precipitated a dramatic new shift toward digital gaming by young
* Corresponding author.
** Corresponding author.
E-mail addresses: bhuber@swin.edu.au (B. Huber), jtarasuik@swin.edu.au
(J. Tarasuik), cgarrett@swin.edu.au (C. Garrett), sbowe@deakin.edu.au (S.J. Bowe),
jkaufman@swin.edu.au (J. Kaufman).
1
Now at Deakin University and Swinburne University of Technology.
2
Swinburne Babylab Team: Mark Finn, Renee Rowsell, Leila Dafner, Jemma
Richards, Lucy Sommers and Anthony Bartel. babylab@swin.edu.au
children. Apple's iPad and similar mobile touchscreen devices have
made interactive media accessible to much younger child-
rendlargely because the ne motor skills needed to use traditional
computers and video games are not necessary (Scaife & Bond, 1991;
Vatavu, Cramariuc, & Schipor, 2014). A recent review (Hirsh-Pasek
et al., 2015) details that the most popular category in Apple's iTunes
App Store is the educational category with over 80,000 apps. This
nding is consistent with survey results from Common Sense Me-
dia, which reported that more than half of parents had downloaded
apps specically for their children (Rideout, 2013).
As with earlier shifts in children's play, the popularity of
touchscreen use is raising a new set of worries related to children's
cognitive, social, emotional, and physical development (e.g., Carson,
Clark, Berry, Holt, & Latimer-Cheung, 2014; Hernandez, 2014;
Mascheroni, 2014). Consistent with (and contributing to) these
worries is the position statement by the American Academy of
Pediatrics (AAP, 2010) on children's media use, which recommends
strict limits on screen time for all young children and discourages
any screen time for children less than 2 years of age. The AAP's
statement appears in many journalistic pieces that cover this topic,
often failing to convey that their recommendations are largely
based on passive television-viewing research and that there is little
peer-reviewed published research on interactive touchscreen me-
dia and young children (e.g. Petersen, 2013; Shapiro, 2014; Wright1
Consulting, 2014).
Despite the fact that the AAP's statement has since evolved to

http://dx.doi.org/10.1016/j.chb.2015.11.010
0747-5632/ 2015 Elsevier Ltd. All rights reserved.
 B. Huber et al. / Computers in Human Behavior 56 (2016) 56e64
indicate the suggested limits are particularly relevant to recrea-
tional (as opposed to educational) content (American Academy
of Pediatrics, 2013; Christakis, 2014), many unanswered questions
must still be resolved to evaluate how interactive media promotes
or hinders learning in young children, and whether following the
AAP's suggestions would have real benets. For example, how and
why should we discriminate between educational and recrea-
tional digital apps when play in the physical world is often
assumed to be both educational and entertaining? More specif-
ically, how should we treat interactive touchscreen apps that are
essentially digital analogues of physical games and activities?
Clearly many such forms of entertainment, such as playing chess,
checkers or card games, require important working memory and
executive functioning skills that we want to foster in young chil-
dren. But are children as able to learn from such activities on
touchscreens as they are in real life? Moreover, when such skills
are learned using digital apps, can young children readily show
apply, or transfer, these skills in their subsequent interactions with
physical objects?
These are important questions to answer for a number of rea-
sons. From a practical perspective, parents and educators should
know if they can safely replace or augment physical games with
these digital analogues. We note that this question is not just a
hypothetical, as parents, educators, and journalists have raised it
with us repeatedly over the past several yearsdtypically in the
form of the question: Does [activity x] count as screen time?
Additionally, answering this question will help clarify which as-
pects of interactive digital play interfere with or foster learning.
Specically, does the screen itself or the lack of rich haptic feedback
somehow make learning less likely for young users?
As such, the key aims of the present investigation were: 1) to
determine if children improve at a problem-solving task after
practicing with an isometric task on touchscreen; 2) to explore the
extent to which practice modality (e.g. touchscreen or physical
version) affects performance; and 3) to determine if the benets of
touchscreen practice require prior experience solving the physical
version of the task.
To achieve these aims, it was important to select a problem-
solving task with an isometric touchscreen version that allowed
for meaningful comparisons of children's performance across mo-
dalities. Consequently, we carried out two experiments using the
Tower of Hanoi puzzle. There were ve main reasons for this choice.
First, unlike a game of checkers or chess, there is a single optimal
solution path that does not depend on the performance of a com-
petitordthus vastly reducing the inter-game variability in perfor-
mance. Second, regardless of whether one tackles the puzzle using
the standard physical disks or using a touchscreen app, the solution
to the puzzle and the scoring methods for prociency are identical,
which allowed us to make straightforward comparisons across
modalities. Third, prior work with the puzzle indicated that it can
be used to assess problem solving in children as young as 4 years of
age (e.g., Bull, Espy, & Senn, 2004; Lillard & Peterson, 2011) and we
wanted to include preschool children in our sampledas this is a
group that until recently has little experience using computing
devices (Scaife & Bond, 1991; Vatavu et al., 2014). Fourth, the Tower
of Hanoi (ToH) puzzle is also used to assess the same set of cognitive
and executive functioning abilities (e.g. planning) that are used in
games like checkers or chess (Klahr & Robinson, 1981; Welsh,
Friedman, & Spieker, 2006)dthus allowing us to draw conclu-
sions about the extent to which children can learn from digital
analogues of these types of games. Finally, pre- and post-
intervention administrations of the ToH task have previously
been used to successfully investigate other factors that can affect
learning (e.g., sleep; Ashworth, Hill, Karmiloff-Smith, & Dimitriou,
2014).
57
In Experiment 1 we addressed the question of whether ToH
practice on a touchscreen device improved performance when
children returned to the physical version of the same task (aim 1).
We also compared performance of those who received touchscreen
practice to those that practiced on the original physical version of
the task (aim 2). In Experiment 2, we addressed the question of
whether touchscreen practice has benets even if the child has had
no recent experience with the physical version of the problem (aim
3). The general methodology of both experiments is illustrated in
Table 1.
2. Experiment 1
In a baseline trial, 4- to 6-year-old children attempted to com-
plete the Tower of Hanoi puzzle in its standard, physical three-
dimensional (3D) form. Children then received two practice trials
on either the same 3D physical version or on a two-dimensional
(2D) touchscreen version of the same puzzle. Following these
practice trials, the children received a nal test trial on the original
3D physical version. We compared how the two types of practice
affected performance on this nal trial. We reasoned that an ability
to transfer problem-solving skills gained in the touchscreen prac-
tice trials would manifest in a signicant improvement from
baseline.
2.1. Method
2.1.1. Participants
Participants were 50 children (27 female, 23 male) aged 4e6
years (M 5.1 years, SD 0.8) recruited from the university's
greater metropolitan area. Each was assigned to one of two con-
ditions, Transfer (n 21, M age 5.2 years, SD 0.8) or No-transfer
(n 29, M age 5.1 years, SD 0.7). There was no statistically
signicant difference in mean age of the condition groups,
t(48) 0.365, p .717. An additional 10 children were recruited
but were not included in this data set for the following reasons:
failure of recording equipment (n 7; all were in the Transfer
condition as the failure related to screen recordings of the iPad),
refusal by the child to attempt the task (n 2), and procedural error
(n 1). All parents provided informed consent for their child's
participation, approved by the host institution's research ethics
committee.
2.1.2. Materials
The ToH puzzle has been used extensively with children and
adults to assess problem solving, planning ability, and executive
functioning. Novel to most children, the task requires self-control
and working memory of three rules to inhibit making an invalid
move. We used the three-disk version of the ToH, which consisted
of three wooden pegs and three wooden disks, each a difference
color and size (small, medium, and large) that can travel across any
of three pegs. To solve the puzzle, the three disks must be moved
from their starting peg to the third peg while abiding by three
rules: 1) only one disk can be moved at a time, 2) a larger disk
cannot be placed on a smaller disk, and 3) the disks must always be
placed on one of the three pegs (i.e., they cannot be put on the
ground or table).
Touchscreen ToH trials were completed on a commercially
available, 2D iPad app (Extra Tower of Hanoi by Morard Dany)
played on an iPad 2. The traditional 3D model of the ToH is
described above. The child's ToH set was situated on a table directly
in front of the child. An additional set sat across the table in front of
the experimenter. This set was used to depict the goal state, and
was visible to the child throughout the task. This goal state
set always had the same modality as the child's set (i.e., when the
58 B. Huber et al. / Computers in Human Behavior 56 (2016) 56e64

Table 1
Methodology for experiments 1 and 2.

Experiment 1 Experiment 2

Transfer No-transfer No-pre-exposure

Trial 1 3D 3D 2D
Trial 2 2D 3D 2D
Trial 3 2D 3D 2D
Trial 4 3D 3D 3D

3D physical version of the problem solving task. 2D touchscreen app version of the problem solving task.

participant attempted the puzzle on the iPad, the other iPad

depicted the goal state).
We modied the instructions of the task to accommodate
younger children by incorporating a cover story about monkeys
based on Klahr (1978). Participants were continuously recorded
using an unobtrusively placed camera. App activity was recorded
using a series of software tools, such as Reector (AirSquirrels.com),
a receiver that allowed us to wirelessly mirror app activity onto the
screen of an unseen computer running a screen-recording
application.

2.1.3. Design and procedure

Pilot work by our lab conducted prior to this experiment sug-
gested that many 4- to 6-year-old children persistently struggle
with the complexity of the standard 3-disk, 7-move, ToH puzzle
(i.e., they failed to solve the puzzle in any trial during the experi-
mental session). To overcome their difculty in the current study,
all children were initially presented with a 5-move version of the
puzzle in which the task is set up for the child with the rst two
correct moves towards the goal state completed. The initial state
and goal state of both versions are illustrated in Fig. 1. Those who
solved the 5-move version within 2 min moved on to the harder
version and were presented with the 7-move version for Trials 1
through 4. The remainder of the children used a 5-move puzzle
throughout the experiment; the pretrial served as their baseline
trial (Trial 1; See Fig. 2). Overall, two-thirds of the participants Fig. 2. Schematic of trial progression in Experiments 1 and 2 with respect to the easier
and harder ToH puzzle.
(n 33) completed the seven-move version of the task. Of the 17
children who completed the task on the 5-move version, 76%
(n 13) were under 5 years of age. benet from the touchscreen practice, children must be able to
Prior to the 5-move trial, children were randomly assigned to transfer their learning to the 3D version of the puzzle in the test
one of two conditions: a Transfer condition (n 21) or a No- trial. In the No-transfer condition, all four trials were instead con-
transfer condition (n 29). In the Transfer condition, children ducted with the same 3D modeldrequiring no learning transfer
performed an initial baseline trial (Trial 1) with the 3D model, across modality in order to demonstrate improvement at test.
followed by two 2D practices (Trials 2 and 3) on the iPad, and n- Each trial (including the pretrial) lasted up to 5 min. At the 5-
ishing with a test trial (Trial 4) on the 3D model. Thus, in order to

Fig. 1. Initial state and goal state of the 5-move and 7-move ToH puzzle.
 B. Huber et al. / Computers in Human Behavior 56 (2016) 56e64
min mark or when the participant solved the puzzle (whichever
occurred rst), the trial ended and the task was reset by an
experimenter in preparation for the next trial. If at any time the
child made a move that constituted a rule break, they were
reminded of the rule by the experimenter. In the 3D trials, if the
child did not immediately correct a rule break, the experimenter
guided the disk back to its previous position. In 2D trials, the app
automatically returned the disk to its pre-violation position.
2.1.4. Coding
From the video and screen recordings of testing sessions, we
coded all disk moves in each of the four trials and calculated the
total number of complete moves that each child made. A complete
move was dened as occurring when the child lifted a disk from a
peg and placed it on another or the same peg. When a rule break
occurred, both the original rule break and the subsequent correc-
tion counted as a separate complete move. We also recorded the
total amount of time taken by the child to complete a trial either by
solving the puzzle or reaching the 5-min time limit. In the latter
case, the time recorded was 5 min.
From these measures, two main dependent variables were
computed. The rst was Extra-moves, which denoted the total
number of complete disk moves made by the child beyond the
minimum required to solve the puzzle (either 5 or 7). Thus, the
minimum Extra-moves score for any single trial was zerodmean-
ing that the child solved the puzzle in the minimum number of
moves. The second measure was Time-per-move, which was
computed as trial time divided by total number of moves made. We
reasoned that this measure would reect the child's condence at
solving the task and skill at moving the disks. Both of these mea-
sures have been used in previous studies using the Tower of Hanoi
(e.g., Schiff & Vakil, 2015).
Six missing practice trials (i.e., Trials 2 and 3) were treated as
missing data in our analyses. A single observer coded the majority
of the recordings. To assess coder reliability a second observer
coded four trials of a randomly selected subset of participants
(n 36 trials). A weighted Cohen's Kappa score of 0.94 was
computed for ranked data on completed moves (i.e., each trial was
assigned a ranking of 1e4 as a function of complete moves made).
2.2. Results
2.2.1. Results summary
The key aims of Experiment 1 were to determine if 2D practice
would result in learning that was transferrable to the 3D puzzle;
and to determine if practice on the 2D model was comparable to
practice with the original 3D model in terms of nal trial perfor-
mance. Overall, the results showed similar levels of effectiveness of
2D and 3D practice: both groups showed downward trends in the
number of excess moves and time per move. These analyses are
described below.
2.2.2. Extra-moves analysis
Extra-moves data was evidently skewed to the right on all trials
(see Fig. 3). These outcome measures were therefore analyzed with
negative binomial regression for panel data, using the xtnbreg
procedure in Stata12.1, accounting for over-dispersion in the data
and allowing for random intercepts.
The analysis of Extra-moves (Table 2) revealed that children
required more moves to solve the task in the initial baseline trial
than either of the subsequent practice trials (Trials 2 and 3) or the
nal test trial (Trial 4). While there was a clear signicant effect of
Trial, c2 (3) 71.7, p < .001, there was no evidence of a signicant
Trial by Condition interaction, c2 (3) 5.08, p .167, suggesting
that this improvement occurred regardless of practice modality.
59
Within-modality comparison of Trial 1 to Trial 4 veried this
nding: Transfer: c2 (1) 24.27, p < .0001; No-transfer: c2
(1) 21.73, p < .0001.
Additionally, analyses did not reveal any signicant interactions
involving age: Age by Trial, c2 (3) 4.68, p .197; Age by Condition,
c2 (1) 0.83, p .362; Age by Trial by Condition, c2 (3) 6.00,
p .111.
We repeated this analysis on a subset of our data that included
only 4-year-old children (n 26). This analysis revealed that the
same overall pattern of results applies for the youngest children in
our sample. There was a signicant effect of Trial, c2 (3) 46.5,
p < .001, with these preschoolers generally improving with
repeated practice. The lack of a Trial by Condition interaction, c2
(3) 3.22, p .359, suggested that practice modality did not affect
performance at test. Individual comparisons of Trials 1 to 4 in each
condition veried this nding: Transfer: c2 (1) 16.40, p < .0001;
No-transfer: c2 (1) 18.27, p < .0001.
In the 4-year-old performance analysis, we included the task
difculty variable as a factor, 5 moves (n 13) versus 7 moves
(n 13), because a high proportion of the 4-year-old participants
(50%) participated using the 5-move version (compared with only
16% of the older children).3 We found a statistically signicant
interaction involving task difculty and Trial, c2 (3) 12.96,
p .005, but no signicant interaction of task difculty and Con-
dition, c2 (1) 0.31, p .576. There was also no statistically sig-
nicant 3-way interaction for Trial, Condition and task difculty, c2
(3) 1.50, p .683.
2.2.3. Time-per-move analysis
Likewise to the Extra-moves data, Time-per-move data was
likewise skewed to the right on all trials (see Fig. 3) and therefore
also analyzed with negative binomial regression for panel data. Our
analysis of Time-per-move (Table 2) revealed that children moved
the disks more quickly as they became more practiced, regardless of
practice modality, c2 (3) 97.0, p < .001. Unsurprisingly, age was a
signicant predictor of Time-per-move, demonstrating that older
children were faster at making moves than younger children. Our
analysis of Time-per-move did not reveal any interactions involving
age and Trial, c2 (3) 3.65, p .302; Age and Condition, c2
(1) 0.28, p .596, or Trial and Condition, c2 (3) 4.70, p .195,
nor was there a statistically signicant 3-way interaction
(Age  Condition  Trial), c2 (3) 3.55, p .315.
2.3. Discussion
The key result of Experiment 1 is that children readily trans-
ferred learning from their practice with the tablet-based Tower of
Hanoi puzzle to the physical version. This nding was robust and
remained valid across the two main dependent measures: Extra-
moves and Time-per-move. Notably, acquiring problem-solving
skills from screen-based media and successfully applying them to
the physical world is an unusual nding in the broader literature
(c.f., Chen & Siegler, 2013), the implications of which are described
in detail in the General Discussion.
The 3D baseline trial (Trial 1) in Experiment 1 was included to
control for inter-participant variability; however, we were aware
that this baseline trial could have introduced a confounding factor.
3
Note that we also performed a series of alternative analyses, which included (a)
separate analysis of the 5-move and 7-move puzzle participants, (b) an analysis
with age analyzed discretely (4 vs. 5 vs. 6 years) rather than continuously, and (c) an
analysis of a recoded version of the data where rule breaks did not count as two
separate moves. In all of these analyses, the pattern of results was very similar to
those reported here. There was signicant improvement regardless of practice
modality across task difculty versions and age group.
60 B. Huber et al. / Computers in Human Behavior 56 (2016) 56e64

Fig. 3. Box-and-Whisker Plots of Extra-moves and Time-per-move measures: (a) Transfer condition (Experiment 1); (b) No-transfer condition (Experiment 1); (c) No-pre-exposure
condition (Experiment 2). The Extra-moves and the Time-per-move variables both illustrate improvement in performance across all conditions.

Table 2
Negative binomial regression model of extra-moves and time-per-move in Experiment 1.

Predictor N Extra-moves Time-per-move

b 95% CI p-value b 95% CI p-value

Trial
1 50 Ref Ref Ref Ref Ref Ref
2 48 0.40 [0.70, 0.09] <0.010* 0.35 [0.52, 0.18] <0.001*
3 47 1.30 [1.69, 0.92] <0.001* 0.66 [0.85, 0.47] <0.001*
4 50 1.38 [1.77, 0.99] <0.001* 0.90 [1.10, 0.70] <0.001*
Age 0.06 [0.26, 0.14] 0.553 0.24 [0.40, 0.08] 0.004*
Condition
No-transfer Ref Ref Ref Ref Ref Ref
Transfer 0.06 [0.25, 0.37] 0.700 0.05 [0.20, 0.30] 0.705
Constant 1.01 [0.80, 2.10] 0.069 4.17 [3.17, 5.18] <0.001
*
denotes statistical signicance at p < 0.05.
 B. Huber et al. / Computers in Human Behavior 56 (2016) 56e64
Specically, it is possible that some children learned how to solve
the puzzle using the 3D version after a single trial, and thus per-
formed well on the 3D puzzle in Trial 4dunrelated to the middle
practice trials. Similarly, children's experience with the 3D version
may have provided them with a requisite understanding of the
puzzle onto which they could map their 2D practice experi-
encesdthus they were only able to benet from 2D practice
because they had prior experience with the 3D analogue.
For these reasons, the third aim of our study was to determine
whether recent practice with the physical version of the puzzle is
necessary for young children to benet from subsequent 2D prac-
tice. It was reasoned that if either of these baseline-related issues
were responsible for the pattern of results, then removing the
initial 3D baseline trial should prevent children from demon-
strating transfer from the touchscreen to the physical modality.
Experiment 2 was designed to test this hypothesis.
3. Experiment 2
To address this outstanding issue, we conducted another
experiment that did not include a 3D baseline trial. In this experi-
ment the participants were given three attempts to complete the
ToH puzzle with the 2D version before attempting the puzzle with
the 3D version for the test trial. Because children were not exposed
to the ToH 3D model prior to attempting the puzzle on the 2D
model, improved performance from Trial 1 to Trial 4 in Experiment
2 would indicate that young children can learn problem-solving
skills from an activity on a touchscreen without referencing an
analogous 3D experience.
3.1. Method
Participant recruitment and experimental procedures in
Experiment 2 were identical to those of Experiment 1, with the
exception that all children completed Trial 1 on the 2D model
rather than the 3D model. Eighteen 4- to 6-year-old children
(M 5.4 years, SD 0.9) participated in Experiment 2 (9 male, 9
female), with an additional ve children omitted from analysis
because of video recording errors. All participants performed the
2D version of the ToH puzzle for the rst three trials, and the 3D
version of the ToH puzzle for Trial 4. As in Experiment 1, all children
completed a 5-move pretrial; however, the pretrial was completed
on the 2D version of the ToH puzzle instead of the 3D version.
3.2. Results
Consistent with Experiment 1, Extra-moves and Time-per-move
were found to be over dispersed (see No Pre-exposure panel in
Fig. 3) and were therefore analyzed with negative binomial
regression for panel data (see Table 3). The key result of Experiment
2 was that in terms of Extra-moves children showed signicant
improvement from baseline (Mdn 12.5, SD 13.9) to test
Table 3
Negative binomial regression model of extra-moves and time-per-move in Experiment 2.
Predictor N Extra-moves
b 95% CI
Trial
1 18 Ref Ref
2 17 0.34 [0.83, 0.14]
3 16 1.27 [1.92, 0.62]
4 18 0.84 [1.38, 0.29]
Age 0.28 [0.55, 0.01]
Constant 2.19 [0.57, 3.80]
* denotes statistical signicance at p < 0.05.
61
(Mdn 6.0, SD 9.0), despite having never been exposed to the 3D
puzzle prior to Trial 4. Although there was a signicant effect of
Trial on Time-per-move, c2 (3) 14.10, p .003, this effect
appeared to be driven by improvement in the 2D modality only, as
there was no signicant difference in Time-per-move when
comparing baseline (Mdn 4.5, SD 5.7) to test (Mdn 4.5,
SD 1.9).
Age was a signicant predictor of Extra-moves, with older
children performing signicantly better than younger children,
though age did not signicantly affect Time-per-move.
3.3. Discussion
Children were signicantly better at solving the puzzle in Trial 4
than they were in Trial 1, even though Trial 4 was the rst time that
they had encountered the 3D ToH puzzle. This nding illustrates
that young children's ability to learn problem-solving skills on a
touchscreen device and apply this knowledge in their interactions
with physical objects, was not dependent on (nor obviously
enhanced by) prior exposure to the physical object.
4. General discussion
Relating to the original aims of the study, the key ndings of this
work were that:
1) Four- to 6-year-old children improved at a problem-solving task
through practice on a touchscreen device.
2) The extent of this improvement was similar to that of children
who practiced only with a physical version. That is, there was no
evidence that children who practiced using the 3D puzzle per-
formed any better on the nal 3D trial than children who
practiced using the 2D model.
3) The learning benets of touchscreen practice did not require an
initial exposure to the 3D version of the puzzle.
We therefore conclude that children are capable of improving at
3D problem-solving tasks, even when practice occurs on a
touchscreen device. While we used a Tower of Hanoi task, we
expect that these results would generalize to other games requiring
planning and executive functioning (e.g., chess).
Our ndings have important implications for our understanding
of children's ability to learn from interactive media; and are
particularly pertinent in the evolving societal debate regarding the
proper role of new media in the lives of young children (e.g.,
Earnest et al., 2014; Goggin, 2013). These implications are discussed
below.
4.1. Learning to solve problems from interactive media
Our ndings offer an important counterexample to claims that
preschoolers struggle to learn problem-solving strategies from
Time-per-move
p-value b 95% CI p-value
Ref Ref Ref Ref
0.166 0.38 [0.70, 0.06] 0.020*
<0.001* 0.66 [1.02, 0.30] <0.001*
0.003* 0.26 [0.56, 0.05] 0.101
0.042* 0.03 [0.26, 0.20] 0.804
0.008 4.28 [0.56, 8.01] 0.024
62 B. Huber et al. / Computers in Human Behavior 56 (2016) 56e64
screen-based media. Notably, the predominant view is based on
research about children and non-interactive video. For example, a
review by Schmidt and Vandewater (2008) described various
studies of children's educational television programming in which
school-age children struggled to apply skills taught via television
when minor changes were made to the task or context (e.g.,
Hodapp, 1977). In contrast, our nding that both 2D and 3D training
resulted in signicant improvement with the test indicates that
preschool children are capable of learning to solve problems using
an interactive touchscreen application. The change of modality
(from 2D to 3D) clearly did not represent a signicant change for
the participants in terms of task demands or task context. Indeed, it
is well known that young children struggle to transfer problem-
solving skills to new contexts (for a review, see Clerc, Miller, &
Cosnefroy, 2014), and yet we found no compelling evidence of
any such struggle here.
Our participants successful learning regardless of training mo-
dality is consistent with work on the advantages of active versus
passive learning. A large body of research on toddlers, children and
adults demonstrates that learning is enhanced when the learner is
actively engaged in the learning process (Hirsh-Pasek et al., 2015).
For example, Sommerville, Hildebrande and Crane (2008) found
that 10-month-old infants better learned to identify a goal and
solve a tool-use problem when they had received practice manip-
ulating the tool, compared with observing another person use the
tool. Similarly a study of adults learning to tie knots from a video
demonstration revealed that those who could control the video
playback learned more efciently than those who were required to
simply watch the video passively (Schwan & Riempp, 2004).
In fact, just presenting a task as a problem to be solved can
encourage children to attend to and encode video in ways that
promote learning. In a recent study by Chen and Siegler (2013),
toddlers were able to imitate a series of actions seen over video,
provided that these steps were shown as sequential steps to ach-
ieve a clearly dened goal. The results of this work contrast with
much of the work revealing children succumbing to a video
decit. For example, young children are often unable to correctly
retrieve objects from various locations after seeing the hiding
events occur via video (Troseth, 2003). Similarly, they also struggle
to imitate a series of actions presented on video or even on a
touchscreen device (e.g., Moser et al., 2015), when the actions are
not clearly presented as a problem-solving task. As Chen and Sie-
gler point out, creating 2D content that conveys a clear goal such
that children see the task as a problem to be solved can be chal-
lengingdespecially for younger children who are still learning to
identify others' goals. Furthermore, as the number of steps to
achieve a goal increases (taxing the viewer's memory), video be-
comes a less valuable medium for learning (Barnett, 2014; Barr,
2010). With increasing complexity, learning to solve problems
increasingly depends on the participant's engagement in the task
(e.g., Bauer & Mandler, 1992).
For these reasons, it is apparent why learning to solve problems
using an interactive touchscreen app is effective. App use, almost by
denition, requires consistent engagement, and because it is active
in nature, the requirement to appreciate solutions through passive
viewing is removed (Geist, 2014). The current study revealed
smooth transfer from the touchscreen device to the physical puzzle,
which helps to demonstrate the limited value of screen time as a
concept. Passive viewing is clearly very different from active
engagement, yet discussions of the effects of screen time on chil-
dren tend not to discriminate on this important factor.
We acknowledge that questions remain relating to exactly what
problem-solving skills were acquired through the practice trials. As
with a game like checkers or chess, a variety of cognitive skills are
important in succeeding at the ToHdincluding long-term memory
(of goals and rules), working memory, and inhibition (Zelazo,
Carter, Reznick, & Frye, 1997). With practice, the children in our
study may have reduced the total number of moves they made by
improving any combination of these skills.
4.2. Social contingency versus contingency
It is well established that learning is enhanced when the
learning context involves elaboration through social interaction
(e.g., Chi, 2009). Although standard video is not responsive to
viewer behavior, many young children will respond to prompts by
familiar, socially meaningful onscreen characters. Prior research
has revealed that programming that incorporates this type of par-
asocial interaction positively affects learning outcomes in young
children (Richert, Robb, & Smith, 2011; Gola, Richards, Lauricella, &
Calvert, 2013). Consequently, parasocial contingency has been
identied as an important feature to incorporate when trying to
maximize children's learning from passive television viewing
(Linebarger & Walker, 2005) and digital media more generally
(Hirsh-Pasek et al., 2015).
Despite the benets of parasocial video interaction, no work to
date has specically examined if the benets of contingent media
must be social in nature (Kirkorian & Pempek, 2013). Our nding of
smooth transfer across the modalities suggests that social contin-
gency may be important for television programming not because it
is social, but because it is the only known method involving any
form of contingency in that restrictive medium. In contrast, with
touchscreen technology, the inherent contingency in performing
actions and witnessing their effects may be enough to foster
transferable learning. This prediction is well founded in the liter-
ature on children's learning, as there are many examples of even
young infants quickly learning from nonsocial contingencies (e.g.,
Kaufman, Gilmore & Johnson, 2006; Wang et al., 2012). Moreover,
recent evidence suggests that the most profound benet of
including social elements into children's digital experiences, is not
on learning to solve problems per se, but on maintaining social
relationships when physical presence is not possible (Tarasuik,
Galligan, & Kaufman, 2011, 2013).
4.3. Executive functioning and problem solving
Our results are notable in the context of recent research sug-
gesting that screen-media use negatively affects subsequent exec-
utive functioning. Lillard and Peterson (2011) for example, reported
impaired executive functioning in 4-year-olds after watching 9 min
of a fast-paced cartoon. The ToH task that we used to assess
learning transfer is also commonly used as a measure of executive
functioning (as it was in the aforementioned study), yet most
children in our study showed improved performance after repeated
practice in the touchscreen modality. Our ndings demonstrate
that while some types of media consumption may negatively
impact executive functioning, other types may instead offer effec-
tive opportunities to practice and hone executive functioning skills.
4.4. Conclusion: generalizing to other digital learning experiences
One of the reasons for carrying out this research was to address
concerns about the generalizability of knowledge and skills learned
on a touchscreen device: If skills can be learned from a touchscreen,
can these skills transfer to the real world (Kirkorian & Pempek,
2013)? Obviously the task for children in our experiments was
somewhat different from what children generally experience with
apps in their everyday lives. In particular, our digital task was iso-
metric to the physical task. As such, future research should evaluate
the extent to which problem-solving strategies learned digitally
 B. Huber et al. / Computers in Human Behavior 56 (2016) 56e64
can be applied to new physical (and digital), non-isometric prob-
lems. Until this is done, we are only able to draw strong conclusions
about the how digital replacement activities can foster learning in
young children. Notwithstanding this limitation, it is notable that
generalizability across modalities in the opposite direction is
largely accepted. For example, the Cambridge Neuropsychological
Test Automated Battery (CANTAB) is a computerized assessment of
cognitive function and is pervasively used in research with both
healthy and clinical populations, including young children (e.g.,
Luciana & Nelson, 2002). If we expect screen-based assessments to
accurately reect knowledge, skills, and abilities gained in the
physical world, it seems likely to expect transfer in reverse direc-
tion, which our results suggest is the case for 4- to 6-year-old
children.
Acknowledgments
A Google Faculty Research Award to Jordy Kaufman, Mark Finn
and Anthony Bartel supported this work. We thank the many par-
ents and children who gave their time to participate in this
research. We also thank the Swinburne Babylab interns for their
help in carrying out this research. This work was facilitated through
the generous support of Prof. George Collins.
References
American Academy of Pediatrics. (2010). Media education. Pediatrics, 126,
1012e1017.
American Academy of Pediatrics. (2013). Children, adolescents and the media. Pe-
diatrics, 132, 958e961. http://dx.doi.org/10.1542/peds.2013-2656.
Ashworth, A., Hill, C. M., Karmiloff-Smith, A., & Dimitriou, D. (2014). Sleep enhances
memory consolidation in children. Journal of Sleep Research, 23, 304e310.
http://dx.doi.org/10.1111/jsr.12119.
Barnett, S. M. (2014). Virtual to real lifedassessing transfer of learning from video
games. Learning by Playing: Video Gaming in Education, 15.
Barr, R. (2010). Transfer of learning between 2D and 3D sources during infancy:
informing theory and practice. Developmental Review, 30(2), 128e154. http://
dx.doi.org/10.1016/j.dr.2010.03.001.
Bauer, P. J., & Mandler, J. M. (1992). Putting the horse before the cart: the use of
temporal order in recall of events by one-year-old children. Developmental
Psychology, 28, 441e452. http://dx.doi.org/10.1037/0012-1649.28.3.441.
Bull, R., Espy, K. A., & Senn, T. E. (2004). A comparison of performance on the towers
of London and Hanoi in young children. Journal of Child Psychology and Psy-
chiatry, 45, 743e754. http://dx.doi.org/10.1111/j.1469-7610.2004.00268.x.
Carson, V., Clark, M., Berry, T., Holt, N. L., & Latimer-Cheung, A. E. (2014).
A qualitative examination of the perceptions of parents on the Canadian
sedentary behaviour guidelines for the early years. International Journal of
Behavioral Nutrition and Physical Activity, 11, 1e8.
Chen, Z., & Siegler, R. S. (2013). Young children's analogical problem solving: gaining
insights from video displays. Journal of Experimental Child Psychology, 116,
904e913. http://dx.doi.org/10.1016/j.jecp.2013.08.00.
Chi, M. T. H. (2009). Active-constructive-interactive: a conceptual framework for
differentiating learning activities. Topics in Cognitive Science, 1, 73e105. http://
dx.doi.org/10.1111/j.1756-8765.2008.01005.
Christakis, D. A. (2014). Interactive media use at younger than the age of 2 years:
time to rethink the American Academy of Pediatrics guideline? JAMA Pediatrics,
168, 399e400.
Clerc, J., Miller, P. H., & Cosnefroy, L. (2014). Young children's transfer of strategies:
utilization deciencies, executive function, and metacognition. Developmental
Review. http://dx.doi.org/10.1016/j.dr.2014.10.002.
Ernest, J. M., Causey, C., Newton, A. B., Sharkins, K., Summerlin, J., & Albaiz, N.
(2014). Extending the global dialogue about media, technology, screen time,
and young children. Childhood Education, 90(3), 182e191. http://dx.doi.org/
10.1080/00094056.2014.910046.
Frost, J. L. (2010). A history of children's play and play environments: Toward a
contemporary child-saving movement. London: Routledge. http://dx.doi.org/
10.1080/21594937.2012.698461.
Frost, J. L., Wortham, S. C., & Reifel, S. (2012). Play and child development (4 ed.).
Boston, MA: Pearson.
Geist, E. (2014). Using tablet computers. Young Children, 69(1), 58e63.
Ginsburg, K. R. (2007). The importance of play in promoting healthy child devel-
opment and maintaining strong parent-child bonds. Pediatrics, 119, 182e191.
http://dx.doi.org/10.1542/peds.2006-2697.
Goggin, G. (2013). Youth culture and mobiles. Mobile Media & Communication, 1(1),
83e88. http://dx.doi.org/10.1177/2050157912464489.
Gola, A. A. H., Richards, M. N., Lauricella, A. R., & Calvert, S. L. (2013). Building
meaningful parasocial relationships between toddlers and media characters to
63
teach early mathematical skills. Media Psychology, 16, 390e411. http://
dx.doi.org/10.1080/15213269.2013.783774.
Golinkoff, R. M., Hirsh-Pasek, K. A., & Singer, D. G. (2006). Play Learning: a
challenge for parents and educators. In D. Singer, R. M. Golinkoff, & K. Hirsh-
Pasek (Eds.), Play Learning: How play motivates and enhances children's
cognitive and social-emotional growth. New York, NY: Oxford University Press.
Hernandez, A. (2014). Toddlers and tablets: emerging apps take cues from learning
science. Education Next, 14, 94.
Hirsh-Pasek, K., Zosh, J. M., Golinkoff, R. M., Gray, J. H., Robb, M. B., & Kaufman, J.
(2015). Putting education in Educational apps: lessons from the science of
learning. Psychological Science in the Public Interest, 16, 3e34. http://dx.doi.org/
10.1177/1529100615569721.
Hodapp, T. (1977). Children's ability to learn problem-solving strategies from
television. Alberta Journal of Educational Research, 23(3), 171e177.
Kaufman, J., Gilmore, R. O., & Johnson, M. H. (2006). Frames of reference for
anticipatory action in 4-month-old infants. Infant Behavior and Development, 29,
322e333.
Kirkorian, H. L., & Pempek, T. A. (2013). Toddlers and touch screens: potential for
early learning? Zero to Three, 33, 32e37.
Klahr, D. (1978). Goal formation, planning, and learning by preschool problem
solvers or: my socks are in the dryer. In R. Siegler (Ed.), Children's thinking:
What develops (pp. 181e212). New Jersey, USA: Laurance Erlbaum Associates,
Inc.
Klahr, D., & Robinson, M. (1981). Formal assessment of problem-solving and plan-
ning processes in preschool children. Cognitive Psychology, 13(1), 113e148.
http://dx.doi.org/10.1016/0010-0285(81)90006-2.
Lillard, A. S., & Peterson, J. (2011). The immediate impact of different types of
television on young children's executive function. Pediatrics, 128, 644e649.
http://dx.doi.org/10.1542/peds.2010-1919.
Linebarger, D. L., & Walker, D. (2005). Infants' and toddlers' television viewing and
language outcomes. American Behavioral Scientist, 48, 624e645.
Luciana, M., & Nelson, C. A. (2002). Assessment of neuropsychological function
through use of the Cambridge neuropsychological testing automated battery:
performance in 4-to 12-year-old children. Developmental Neuropsychology,
22(3), 595e624.
Mascheroni, G. (2014). Parenting the mobile Internet in Italian households: parents'
and children's discourses. Journal of Children and Media, 8, 440e456.
Moser, A., Zimmerman, L., Dickerson, K., Grenell, A., Barr, R., & Gerhardstein, P.
(2015). They can interact, but can they learn? Toddlers' transfer learning from
touchscreens and television. Journal of Experimental Child Psychology, 137,
137e135.
Petersen, A. (2013, October 28). Pediatricians set limits on screen time. The Wall
Street Journal.
Plato. (1955). The republic. Harmondsworth, Middlesex: Penguin Books.
Richert, R. A., Robb, M. B., & Smith, E. I. (2011). Media as social partners: the social
nature of young children's learning from screen media. Child Development, 82,
82e95. http://dx.doi.org/10.1111/j.1467-8624.2010.01542.x.
Rideout, V. (2013). Zero to eight: Children's media use in America 2013. Common
Sense Media.
Scaife, M., & Bond, R. (1991). Developmental changes in children's use of computer
input devices. Early Childhood Development and Care, 69, 19e38. http://
dx.doi.org/10.1080/0300443910690103.
Schiff, R., & Vakil, E. (2015). Age differences in cognitive skill learning, retention and
transfer: the case of the Tower of Hanoi Puzzle. Learning and Individual Differ-
ences, 39, 164e171.
Schmidt, M. E., & Vandewater, E. A. (2008). Media and attention, cognition, and
school achievement. The Future of Children, 18, 63e85. http://dx.doi.org/
10.1353/foc.0.0004.
Schwan, S., & Riempp, R. (2004). The cognitive benets of interactive videos:
learning to tie nautical knots. Learning and Instruction, 14, 293e305. http://
dx.doi.org/10.1016/j.learninstruc.2004.06.005.
Shapiro, J. (2014, August). Screen time that's valuable for young kids. Mind/Shift.
Retrieved from http://blogs.kqed.org/mindshift/2014/08/screen-time-games-
learning-thats-valuable-for-young-kids/.
Sommerville, J. A., Hildebrand, E. A., & Crane, C. C. (2008). Experience matters: the
impact of doing versus watching on infants' subsequent perception of tool-use
events. Developmental Psychology, 44, 1249. http://dx.doi.org/10.1037/a0012296.
Tarasuik, J., Galligan, R., & Kaufman, J. (2011). Almost being there: video commu-
nication with young children. PLoS One, 6(2). http://dx.doi.org/10.1371/
journal.pone.0017129.
Tarasuik, J. C., Galligan, R., & Kaufman, J. (2013). Seeing is believing but is hearing?
Comparing audio and video communication for young children. Frontiers in
Psychology, 4(FEB). http://dx.doi.org/10.3389/fpsyg.2013.00064.
Troseth, G. L. (2003). Getting a clear picture: young children's understanding of a
televised image. Developmental Science, 6(3), 247e253. http://dx.doi.org/
10.1111/1467-7687.00280.
Varney, W. (1999). Toys, play and participation. In B. Martin (Ed.), Technology and
public participation (pp. 15e36). Wollongong, Australia: University of
Wollongong.
Vatavu, R. D., Cramariuc, G., & Schipor, D. M. (2014). Touch interaction for children
aged 3 to 6 years: experimental ndings and relationship to motor skills. In-
ternational Journal of Human-Computer Studies. http://dx.doi.org/10.1016/
j.ijhcs.2014.10.007.
Wang, Q., Bolhuis, J., Rothkopf, C. A., Kolling, T., Knopf, M., & Triesch, J. (2012). In-
fants in control: rapid anticipation of action outcomes in a gaze-contingent
64 B. Huber et al. / Computers in Human Behavior 56 (2016) 56e64

paradigm. PloS One, 7, e30884. http://dx.doi.org/10.1371/journal.pone.0030884.
Welsh, M., Friedman, S. L., & Spieker, S. J. (2006). Executive functions in developing
children: current conceptualizations and questions for the future. Blackwell
Handbook of Early Childhood Development, 167.
Wright1 Consulting. (2014, April 16). Part 2: Kid-tech trends and what you need to
know [Web log post]. Retrieved from http://www.wright1consulting.com/2014/
04/part-2-kid-tech-trends-need-know/.
Zelazo, P. D., Carter, A., Reznick, J. S., & Frye, D. (1997). Early development of ex-
ecutive function: a problem-solving framework. Review of General Psychology,
1(2), 198. http://dx.doi.org/10.1037/1089-2680.1.2.198.