Developmental Psychology Copyright 2006 by the American Psychological Association
2006, Vol. 42, No. 6, 1206  1219 0012-1649/06/$12.00 DOI: 10.1037/0012-1649.42.6.1206
Cognitive Engagement and Story Comprehension in Typically Developing
Children and Children With ADHD From Preschool
Through Elementary School
Elizabeth P. Lorch, Richard Milich, Clarese C. Astrin, and Kristen S. Berthiaume
University of Kentucky
The present study examined children s cognitive engagement with television as a function of the
continuity of central or incidental content and whether this varied with age and clinical status. In
Experiment 1, 9- to 11-year-old children s response times on a secondary task were slower the later a
probe occurred in a sequence of central events, and response times predicted recall. Experiment 2
extended these results to 6- to 8-year-old children. Experiment 3 revealed that children with attention-
deficit/hyperactivity disorder (ADHD) failed to show the pattern consistently observed for comparison
children. The results support the hypothesis that typically developing children build a representation
during viewing that reflects the causal structure of the televised story but that this skill is deficient in 4-
to 9-year-old children with ADHD.
Keywords: ADHD, cognitive engagement, secondary probe, attention
Supplemental data: http://dx.doi.org/10.1037/0012-1649.42.6.1206.supp
Knowledge of children s story comprehension has been in- gagement with a televised story vary as a function of the contin-
formed by theories of the comprehension process. A theme com- uation of plot-relevant content? Second, does this pattern differ as
mon to most theories is an emphasis on the causal and enabling a function of age? Finally, do children with documented attention
relations between events in a story, which define plot-relevant problems and story comprehension difficulties (i.e., diagnosed
events and give the story coherence (Ackerman, 1986; Black & with attention deficit/hyperactivity disorder [ADHD]) differ in
Bower, 1980; Graesser & Clark, 1985; Mandler & Johnson, 1977; cognitive engagement from comparison children?
Trabasso, Secco, & van den Broek, 1984; van den Broek, 1997).
Characteristics of the causal structure of stories have been found to
Influences on Children s Attention to Television
predict a product of children s comprehension that is, memory
for story events (e.g., Trabasso et al., 1984; van den Broek, 1989;
Several theoretical viewpoints on children s television viewing
van den Broek, Lorch, & Thurlow, 1996) with effects becoming
concur that children, from an early age, are active viewers whose
stronger with age (Nezworski, Stein, & Trabasso, 1982; Schmidt &
visual attention to television is guided by ongoing comprehension,
Paris, 1983; van den Broek et al., 1996). However, much less
expectations, and purposes for viewing (Anderson & Lorch, 1983;
research has investigated the online processes children may engage
Huston & Wright, 1983; Salomon, 1983). For example, Anderson
in to detect and use causal relations between events to build a
and Lorch (1983) proposed that for young children, a look at the
representation of a story (Ackerman, Paine, & Silver, 1991; Tra-
television may begin in response to any of several possible rea-
basso & Nickels, 1992; Trabasso, Stein, Rodkin, Munger, &
sons, including stimuli that elicit orienting responses; formal fea-
Baughn, 1992). One avenue of research that may contribute to
tures that signal informative, appealing, child-relevant content
knowledge of children s online story comprehension processes is
(Alwitt, Anderson, Lorch, & Levin, 1980); and cues derived from
the study of systematic variations in children s attention to tele-
the behavior of other children (Anderson, Lorch, Smith, Bradford,
vised stories (Anderson & Lorch, 1983; Huston & Wright, 1983).
& Levin, 1981). Once a look has begun, its continuation primarily
The present series of studies builds on this line of research by
depends on the child s ongoing judgments of whether program
using a measure of moment-to-moment cognitive capacity usage to
content is comprehensible (Anderson, Lorch, Field, & Sanders,
investigate several questions. First, does children s cognitive en-
1981; Lorch, Anderson, & Levin, 1979; Pingree, 1986).
Huston and Wright (1983) discussed similar factors as influ-
ences on visual attention but conceptualized a series of decisions
children make about whether to continue a look at the television.
Elizabeth P. Lorch, Richard Milich, Clarese C. Astrin, and Kristen S.
At the beginning of a look, decisions are most heavily a function
Berthiaume, Department of Psychology, University of Kentucky.
of formal characteristics, once again those that are indicative of
This research was supported by National Institute of Mental Health
informative, interesting, child-relevant content (Calvert, Huston,
Grant MH47386.
Watkins, & Wright, 1982; Huston et al., 1981). If a look continues,
Correspondence concerning this article should be addressed to Elizabeth
more extensive, deeper cognitive processing takes place, such that
P. Lorch, Department of Psychology, University of Kentucky, Lexington,
KY 40506-0044. E-mail: [e-mail] a subsequent decision is affected by both current comprehension
1206
COGNITIVE ENGAGEMENT WITH TELEVISION
1207
and initial expectations about content (Campbell, Wright, & Hus- from one 3-s interval to the next (after removal of variance
ton, 1987; Rolandelli, Wright, Huston, & Eakins, 1991). As further resulting from overall attention, age, and tape version). Consistent
cycles occur, later decisions are increasingly influenced by deeper with the results of Lorch and Castle (1997), Hawkins et al. (1991)
processing and more elaborated expectations about story content found much higher stability in attention late in normal stories than
(Hawkins, Tapper, Bruce, & Pingree, 1995; Rolandelli et al., in edited stories. Taken together, the findings of Hawkins et al.
1991). In addition, Huston and Wright (1983) proposed that there (1991), Lorch and Castle (1997), and Meadowcroft and Reeves
are changes in these cycles of decisions as development advances. (1989) suggest that children s attention is responsive to the plot
For younger children, the continuation of a look at the television relevance of story content and becomes increasingly engaged as a
may be more dependent on superficial features of the program than meaningful narrative structure develops.
for older children, who are more likely to make elaborated deci- The major purpose of the current studies is to extend investiga-
sions on the basis of deeper processing of content (Rolandelli et tion of school-age children s online cognitive processing of tele-
al., 1991). vision story content by examining whether children s cognitive
These perspectives and their associated empirical support indi- engagement with a television program increases as sequences of
cate that children s visual attention to television varies systemat- story content that are central to the plot continue or decreases as
ically in relation to program characteristics. Huston and Wright s sequences of content that are incidental to the plot continue. In
(1983) perspective also suggests a context for the online process- addition, Experiment 2 investigates whether there are developmen-
ing of television story structure. To the extent that children build tal differences in children s cognitive engagement as a function of
an ongoing representation of the interrelations among story events the continuity of central or incidental content.
during viewing, measures of attention may reveal effects of event Finally, Experiment 3 examines the cognitive engagement of
centrality, causal structure, and plot development. children with ADHD, whose problems in sustaining attention and
Several investigations suggest such indications of children s comprehending causal relations have been well documented
online processing of structural features of television stories. (Lorch, Eastham, et al., 2004; Milich, Lorch, & Berthiaume,
School-age children have been found to look at the television more 2005). Two studies have investigated directly the role of cognitive
during presentation of material that adults rated as important (i.e., engagement in story comprehension among children with ADHD.
central) to the plot of the story than during material rated as low in Lorch, Eastham, et al. (2004) used a television viewing method-
importance (Baer & Lorch, 1990; Lorch & Baer, 1997). Mead- ology to test the hypothesis that differences in cognitive engage-
owcroft and Reeves (1989) obtained a related finding using a ment would account for group differences in recall of causal
secondary task technique to assess cognitive engagement with the relations (i.e.,  why questions) when toys were present during
television program. The secondary task technique is based on the viewing. The results of three different analytic strategies con-
assumptions that mental processing requires time and that central verged to support the hypothesis that greater cognitive engage-
processing capacity is limited (Basil, 1994; Kahneman, 1973). ment, as indexed by long looks (i.e., longer than 15 s) at the
Thus, the more attentional capacity is engaged by a continuous television, enabled comparison children to form a more complete
primary task, such as viewing a television program, the less is representation of the relations among story events, thereby ac-
available for a secondary task, such as a keypress in response to an counting for differences between the comparison and ADHD
occasional auditory tone (Britton, Graesser, Glynn, Hamilton, & groups in understanding causal relations questions. Thus, the find-
Penland, 1983; Britton & Tesser, 1982; Thorson, Reeves, & ings from Lorch, Eastham, et al. (2004) constitute the first com-
Schleuder, 1985). Meadowcroft and Reeves (1989) found that 5- to pelling evidence that the amount of time spent in deeper cognitive
8-year-old children who had tested high in the development of processing during long looks helps explain the differential patterns
story schema skills showed greater cognitive engagement (i.e., of recall in children with ADHD and comparison children.
slower responses to secondary probes) with central content pre- The results of Lorch, Eastham, et al. (2004) suggest that varia-
sented in a normal story structure than with the same content tions in cognitive engagement may account for the problems that
presented out of order and thus in the absence of a coherent story children with ADHD have with causal relations questions. How-
structure. ever, the degree of cognitive engagement was inferred from time
Two additional studies also compared children s responses to spent in long looks. The secondary task may offer a more direct
normally structured, coherent stories and stories in which scenes way to assess cognitive engagement. To date, only one study has
had been edited out of sequence, via existing edit points (Hawkins, used this procedure with an ADHD sample (Whirley, Lorch,
Kim, & Pingree, 1991; Lorch & Castle, 1997). The use of existing Lemberger, & Milich, 2003). In this study, participants were 22
edit points allowed the edited stories to be locally comprehensible boys with ADHD and 36 comparison boys, ranging in age from 9
but lacking in a coherent story structure. Both studies used the to 11 years. Boys with ADHD responded significantly slower than
same types of Sesame Street stories, and in both studies children the comparison boys, a common finding in reaction time studies.
showed high visual attention to normal stories and to edited stories. More important, the patterns of response times (RTs) across the
Both studies, however, revealed indications that children s atten- central sequences differed between the two groups of boys. The
tion was engaged more systematically by the normally structured comparison boys showed the predicted pattern of longer RTs (i.e.,
stories. Lorch and Castle (1997), using the secondary task meth- increased cognitive engagement) the longer into a central sequence
odology, found that 5-year-old children showed greater cognitive the probes appeared. In contrast, the boys with ADHD actually
engagement during the second half of normal stories than during showed shorter RTs from the beginning to the middle of the central
the first half, but the authors did not observe this difference for the sequences, which suggests decreased cognitive engagement as the
stories shown in edited form. Hawkins et al. (1991) investigated central sequences progressed. It was only for probes occurring late
the predictability in 3- to 6-year-old children s visual attention in the central sequences that the RTs of the boys with ADHD
LORCH, MILICH, ASTRIN, AND BERTHIAUME
1208
showed the expected increase. In Experiment 3, we use a devel- had ever been referred for any attentional or behavioral difficulties. Children
were paid $5 for their participation, which lasted about 45 min. Data were lost
opmental perspective to compare the cognitive engagement of
from 2 participants as a result of equipment malfunction.
children with ADHD and their comparison peers, using an age
range that included younger children (i.e., 4 to 9 years) than the
previous secondary task studies.
Materials
To address whether children s cognitive engagement is related
to the development of content that varies in plot relevance, pro-
Each participant viewed one episode of the situation comedy Growing
grams were selected that contained strong narrative structures,
Pains. This program was chosen specifically because its content is suitable
with story lines revolving around child characters, and that enabled and interesting for children. Furthermore, the plot of this specific episode
( Dad s Birthday ) centered on the activities of one of the child characters
the identification of continuous sequences (approximately 15 s or
in the family. A synopsis of the plot appears in the Appendix, which is
more) of events that were central or incidental to the plot. Central
available on the Web. With all commercials removed, the episode is 23 min
sequences are those that are crucial to plot development, whereas
in length. A detailed audiovisual script of the program was created as part
incidental sequences are ones that could be removed without
of a previous study (Lorch et al., 2000). The script was divided into 407
affecting the coherence of the story. Central and incidental se-
individual units of meaning, with each unit representing a single idea or event.
quences of events were operationalized in terms of the combina-
In accordance with procedures defined by Trabasso and van den Broek
tion of two criteria: whether events in the sequence were part of the
(1985), a causal network representation of the story was derived. On the
causal chain leading from the beginning to the eventual outcome of
basis of the causal network analysis, each idea unit was coded as to whether
the story (as opposed to  dead end events), and whether events
it was on or off the causal chain. The causal chain is a sequence of events
were high or low in centrality, as determined by college student that are causally linked together and carry the story from the beginning to
the end. All story events are either part of the causal chain (i.e., causally
raters (see Experiment 1 Method section for greater detail). For
connected to prior and subsequent events) or dead-end events (i.e., not
each central or incidental sequence of events, several positions
causally connected to prior and subsequent events on the causal chain).
were identified as potential times for presentation of secondary
Each idea unit also had been rated by 58 college students for its
probes. Watching and understanding the television program was
centrality to the story, on a scale from 1 to 7. The students first watched the
defined as the primary task, and children were told that their
program and then were given scripts representing the individual units in the
knowledge of story events would be tested after viewing. As a
program. In assigning their ratings, they were instructed to consider how
child watched the program, auditory probes were presented at
much plot-relevant information the unit conveyed, how much would be lost
preselected times. The child s secondary task was to press a key as
if the unit were removed from the program, and how much the unit
quickly as possible whenever a probe was presented.
enhanced understanding of the plot. Mean centrality ratings were calcu-
Within the constraints of this task, visual attention was expected lated across all students for each idea unit.
For the purposes of the present study, an individual unit was considered
to be very high, so the cycles of viewing decisions described by
to be central if it both was on the causal chain and had a mean centrality
Huston and Wright (1983) were not expected to occur. Probe RTs,
rating of at least 5.15 out of 7, the upper quartile of centrality ratings. An
however, may capture similar decisions concerning whether to
individual unit was considered to be incidental if it both was off the causal
engage in deeper and more elaborative processing. If children
chain (a dead-end event) and had a mean centrality rating of no more than
build a representation during viewing that reflects the causal
3.30, the lower quartile of ratings. Given these classifications, nine con-
structure of a televised story, they should show increasing cogni-
tinuous sequences of primarily central events and nine continuous se-
tive engagement as a sequence of central story content continues,
quences of primarily incidental events were identified. A description of
because they will encounter events that can be related to the main
each sequence and its order of appearance in the Growing Pains episode
thread of the story. Therefore, the later a probe was presented in a
appears in the Appendix (available on the Web). The length of the se-
sequence of central events, the slower children s responses were quences varied from 15 to 90 s, and sequences contained between 4 and 30
idea units. In central sequences, the mean centrality rating of the units was
expected to be. In contrast, this pattern was not expected to occur
5.53, 77% of the individual units were in the upper quartile of the ratings,
as a sequence of incidental events continued. The lack of connec-
and 72.4% of the individual units were on the causal chain. In incidental
tions to the causal chain of the story should lead to more superfi-
sequences, the mean centrality rating of the units was 2.88, 85% of the
cial processing. Thus, children s RTs might be unrelated to probe
individual units were in the lower quartile of the ratings, and none of the
position or might even become shorter the later a probe was
individual units were on the causal chain. In general, individual units in
presented in a sequence of incidental events.
either type of sequence that did not meet strict criteria for the sequence
tended to be brief and isolated.
Experiment 1 To track online variations in children s cognitive engagement with the
television story, we placed 26 auditory target probes at preselected points
Method
during these sequences of story events. The basic positions for target
probes were approximately 2, 7, or 12 s into the sequence.1 In addition, in
Participants
Participants were 27 boys and 33 girls, ages 9 11 years (M 10.05, SD
1
0.81). Participants were primarily Caucasian (88.33%), with some African The nine central and nine incidental sequences exhausted all possible
American children (8.33%) and children of other ethnic identifications sequences of story events that met criteria for classification as central or
(3.33%). Most parents education included at least some college; the average incidental and continued for a minimum of 15 s. Fifteen seconds was
parent was a college graduate (M 16.98 years of education). Children were chosen as the minimum because this is the asymptote of the attentional
recruited through advertisements in the newspaper and from an existing pool inertia function (Anderson & Lorch, 1983). The basic probe positions of 2,
of experimental volunteers. For Experiment 1, children were screened for 7, and 12 s were chosen to represent early, middle, and later positions
behavioral problems or attentional difficulties, such as ADHD, in a recruitment within the 15-s interval. The number of later probes was constrained by the
phone call and were not included in the study if the parent indicated the child number of sequences longer than 24 s.
COGNITIVE ENGAGEMENT WITH TELEVISION
1209
sequences lasting 24 s or longer, later probe positions were added to the After the child indicated verbally that he or she understood the instructions
basic probes. We created three series of probe assignments to counterbal- and had satisfactorily completed the practice probe program, the experi-
ance probe position across sequences. Each participant was randomly mental session began. The experimenter started the videotape that played
assigned to one of these three series. Each series contained 3 probes in each the television show and synchronized the probe program to the tape before
basic position (i.e., 2, 7, and 12 s), at each centrality level, and 4 probes in leaving the room.
the later positions during sequences longer than 24 s, for a total of 13 When the television program was finished, the experimenter reentered
central and 13 incidental target probes. We inserted 12 filler probes in the the room and began the cued recall questioning session. The testing session
program to prevent participants from detecting a pattern or being able to was videotaped to allow later scoring of cued recall questions. The ques-
predict probes and make anticipatory responses, which resulted in a total of tions were asked in the order in which the content was presented in the
38 probes during the episode. show. If the child did not answer a question correctly, the experimenter
A set of comprehension questions was developed for the television supplied the correct answer and then continued to the next question until all
program. The questions tested factual information that was presented in the questions were asked. The child was then debriefed, paid, and thanked for
program during sequences containing probes. Discrete events that occurred his or her participation.
close in time to a secondary task probe were chosen to be tested, and
straightforward questions were designed to test children s memory for each
Results and Discussion
event. An example question testing memory for a central story event (see
Appendix, available on the Web) is  Ben has to return the camera. What
Dependent Measures
else does he have to do as part of his punishment? (return money and
apologize to neighbors). Twenty-five cued recall questions were devel- The main dependent variable was RT to 26 target probes.
oped; 14 tested central content, and 11 tested incidental content. Every Thirty-eight probes sounded during the program. Thirteen target
sequence was represented by at least 1 question, with two sequences
probes occurred during sequences of central content, and 13 oc-
represented by 2 questions, two represented by 3 questions, and two
curred during sequences of incidental content. The remaining
represented by 4 questions.
probes were fillers and were not analyzed. Two participants RT
data included one impossibly fast RT (32 ms for 1 child, 112 ms
for the other). These responses were not included in the calculation
Procedure
of means for the categories. Two participants failed to respond to
1 probe. We obtained category means for these participants by
The participants were brought to the on-campus television viewing
facility by a parent. Each child participated individually. Before the ex- computing means of the remaining RTs in the category.
periment began, informed consent was obtained from the parent. After a
Performance on the cued recall questions was scored from
brief conversation with the experimenter to help the child feel comfortable
transcripts of the videotape. Each answer was assigned a score of
(e.g., talking to the child about school, activities, pets), the child was shown
1 for correct and 0 for incorrect. An independent rater scored 25%
to his or her seat in the television viewing room. The child was seated at
of participants cued recall responses, with 92% agreement. The
a table, with a computer keyboard on the table in front of him or her. The
proportion correct was computed for each participant separately
probes were generated by an IBM computer that was located on the floor
for central and incidental questions.
next to the child. Responses were made on the keyboard on the table in
Participants visual attention to the television was coded from
front of the child. RTs for each probe were recorded (in milliseconds) from
videotapes. Using a computer program synchronized to the begin-
the beginning of the probe until the child responded by pressing the space
ning of the television program, we obtained a continuous record of
bar. The probe continued until a response was made. If no response was
made, the sound ended after 5 s. the child s looks at and away from the television. This continuous
A television was situated on a desk near one corner of the table, with the
record of looks allowed for identification of looking status at the
television screen about 5 ft (1.5 m) from the child s seat. A video camera
time of every probe as either on or off task. The mean percentage
was mounted on the wall in a position that allowed the image to include
of visual attention was at ceiling (M 0.96, SD 0.04). RT data
both the child and the television. For the child to look at the television, he
were analyzed after removal of all probe RTs that occurred during
or she needed to make a noticeable head movement. This enabled the
a look away from the television (n 51, less than 5% of target
experimenter to record looks toward and away from the television. Each
probes). The pattern of results is identical with these RTs included,
child was videotaped during the entire experimental session, and visual
but analyses reported in this article exclude RTs to probes that
attention to the television was later coded from the videotape.
sounded during looks away from the television (Lorch & Castle,
After the child was seated in the room, the experimenter obtained his or
1997).
her verbal assent to participate in the study. Once the child agreed to
participate, the experimenter explained the procedure to the child, empha-
sizing that watching the television program was the primary task and
RTs
responding to the probe was the secondary task. It was explained to the
participants that they would hear sounds that they should turn off as Mean response times were analyzed in a repeated measures
quickly as possible by pressing the space bar on a computer keyboard
analysis of variance (ANOVA), with centrality (central and inci-
located on the table in front of them. Participants were instructed to keep
dental) and probe position (2, 7, 12 s into sequence, and later
their dominant hand directly in front of the computer keyboard throughout
position) as within-subject variables. Follow-up linear trend anal-
the entire television program. To help the child do this, the experimenter
yses tested the a priori prediction that RTs to probes would
placed a loose Velcro strap across the child s wrist. After the procedure
increase as the time into a sequence of central content increased. In
was explained, the child was given an opportunity to practice. A computer
addition, we tested whether RTs to probes would decrease as time
program presented five probes that were randomly spaced throughout a
into a sequence of incidental content increased. Mean RTs as a
2-min period. Each child pressed the space bar in response to every practice
function of centrality and probe position are depicted in Figure 1.
probe. It was again emphasized to the child that watching and understand-
As hypothesized, a significant interaction was observed between
ing the program was the primary task, and each child was told that after the
program ended there would be some questions about the television show. centrality and probe position, F(3, 171) 3.32, p .022. Linear
LORCH, MILICH, ASTRIN, AND BERTHIAUME
1210
850
800
750
Central
700
Incidental
650
600
550
2 7 12 Later
Probe Position (Seconds into Sequence)
Figure 1. Mean probe response times (RTs) as a function of time into central and incidental sequences for
Experiment 1.
trend analyses indicated that RTs to probes occurring during cally, one might expect that recall performance would be related to
central sequences increased as time into the sequences increased, differing levels of engagement. Each cued recall question tested
F(1, 57) 4.30, p .05 (effect size r .26), and RTs to probes information that was presented during or close to a unit that
occurring during incidental sequences decreased as time into the contained a probe. Mean RTs were computed for probes corre-
sequences increased, F(1, 57) 6.16, p .05 (r .31). sponding to questions that each child answered correctly and
Follow-up analyses controlling family-wise error rate indicated questions answered incorrectly. Regardless of centrality, chil-
that the mean RT in the central, 2-s position was significantly dren s RTs corresponding to questions they answered correctly
shorter than the central, later position RT, t(58) 2.36, p .05 tended to be longer (M 601.77) than their RTs corresponding to
(r .30). The incidental, 2-s position RT was significantly slower questions they answered incorrectly (M 587.83), F(1, 48)
than the incidental, later position RT, t(58) 2.29, p .05 (r 3.71, p .06 (r .27). This suggests an association between
.29). Unexpectedly, children were significantly slower in respond- online cognitive engagement with the televised story and later
ing to probes that sounded during incidental sequences (M recall of specific material.
612.73) than to probes during central sequences (M 572.15), As predicted, 9- to 11-year-old children s cognitive engagement
F(1, 57) 9.84, p .004 (r .38). No main effect was observed was related to the continuity of central or incidental content.
for probe position (F 1, p .05). Children s probe RTs indicated that the longer a sequence of
central story content continued, the more engaged children be-
came, and that as a sequence of incidental content continued, they
Cued Recall Performance
became less engaged. Children s increased engagement was also
We conducted a t test on cued recall scores to evaluate differ-
related to better performance on story comprehension measures.
ences between performance on central and incidental content.
Further discussion of the results from all three experiments is
Performance on the cued recall questions was significantly better
reserved for the general discussion.
for those questions tapping central content (M 0.85, SD 0.11)
than for those tapping incidental content (M 0.61, SD 0.18),
Experiment 2
t(57) 12.05, p .001 (r .84). This result replicates the pattern
observed in numerous previous studies of children s comprehen- Experiment 1 examined cognitive engagement in children ages
sion of both televised and written stories (e.g., Lorch, Bellack, & 9 to 11 years, but an unanswered question is whether younger
Augsbach, 1987; van den Broek, 1989; van den Broek et al., 1996). school-aged children would also use the causal structure of a
A further question concerns the possibility of a relation between televised story to guide their engagement. Memory for story events
cued recall performance and RTs to secondary probes. If children s among preschoolers and young school-age children is influenced
cognitive engagement with the televised story varies systemati- by characteristics of the story s causal structure, but the impact of
RT in milliseconds
COGNITIVE ENGAGEMENT WITH TELEVISION
1211
causal factors increases with age (van den Broek et al., 1996). whether the sequence presented central or incidental content, F(3,
Children also show improvement with age in detecting connec- 306) 9.116, p .001. As shown in Figure 2, linear trend
tions between groups of events (van den Broek, 1989). In addition, analyses revealed that RTs increased the longer a sequence of
young children show differential responses to global story struc- central content continued, F(1, 104) 34.34, p .001 (r .51),
ture, showing increased attentional engagement as normal stories and that RTs decreased the longer a sequence of incidental content
develop but no systematic change in engagement to scrambled continued, F(1, 104) 11.19, p .001 (r .31). Follow-up
stories (Lorch & Castle, 1997). However, it is unknown whether analyses indicated that for central sequences, RTs for each of the
younger children would show online sensitivity to the role differ- first three probe positions differed significantly from RTs for the
ent events play in a coherent story structure. Huston and Wright s later probe position, and RTs for the 2-s position differed from
(1983) theoretical perspective suggests that with development, those for the 7-s position (ts ranged from 2.5 to 5.2, p .05). For
children may become more accomplished at using the story con- the incidental sequences, only the 12-s and later probe positions
tent to determine when deeper and more elaborative processing is differed significantly, t(104) 2.4, p .05 (r .23).
necessary. Thus, Experiment 2 was designed to replicate Experi- Although older children responded more quickly (M 601.52)
ment 1 and to determine whether there are developmental differ- to probes than younger children (M 836.48), F(1, 102) 36.47,
ences in children s cognitive engagement as a function of the p .001 (r .51), there were no significant differences in the
continuity of central and incidental content. On the basis of the patterns of probe RTs as a function of age. As found in Experiment
results of Experiment 1, it was predicted that older school-aged 1, probe RTs were longer during incidental sequences (M
children would show decreasing engagement with the television as 741.99) than during central sequences of content (M 689.58),
incidental content continued and increasing engagement as central F(1, 102) 10.95, p .001 (r .31).
content continued. Children in the younger age group may be more
dependent on formal features of the television program as a guide
Cued Recall Performance
to their attention, so their engagement may not be as sensitive to
the causal structure of the story as that of older children. Children correctly answered a higher proportion of questions
testing central content (M .81) than questions testing inci-
dental content (M .47), F(1, 103) 474.85, p .001 (r
Method
.90). Although older children (M .73) correctly answered a
Participants
higher proportion of questions than younger children (M .58),
F(1, 103) 48.01, p .001 (r .56), the influence of
Participants were 54 younger elementary school-aged children (25
centrality on recall performance did not differ as a function of
boys), ages 6 to 8 (M 7.83, SD 0.64), and 57 older elementary
age, F(1, 103) 1.
school-aged children (31 boys), ages 9 to 11 (M 10.70, SD 0.56), none
We also examined whether children s recall performance was
of whom had participated in Experiment 1. Participants were primarily
Caucasian (89.52%), with some African American children (5.71%) and related to differing levels of engagement. Overall, central content
children of other ethnic identification (4.76%). Most parents education
questions that were answered correctly were associated with
included at least some college; the average parent was a college graduate
longer RTs to probes during presentation of target content (M
(M 16.31 years of education). Children were recruited through adver-
722.8 ms) than RTs to probes associated with central content
tisements in the newspaper and from an existing pool of experimental
questions that were answered incorrectly (M 673.4 ms), F(1,
volunteers. Children were paid $10 for their participation, which lasted
88) 9.63, p .01 (r .31). This difference was significant for
about 1 hr. Data were lost from 6 participants as a result of equipment
older children, t(41) 3.70, p .001 (r .55), but not for
malfunction or failure to follow instructions.
younger children, t(47) 0.90, p .10.
Materials and Procedure
Experiment 3
The materials and procedure were identical to those of Experiment 1.
The findings of Experiments 1 and 2 demonstrate that online
Results and Discussion
variations in children s cognitive engagement with a televised
story are related to the continuity of central or incidental content.
Probe RTs
As indicated by probe RTs, children became increasingly engaged
Mean RTs were analyzed in a mixed ANOVA, with age as a the longer sequences of central events continued but decreased
between-subjects variable and centrality (central and incidental) their cognitive engagement the longer sequences of incidental
and probe position (2, 7, 12 s into sequence, and later position) as events continued. Furthermore, this pattern of results was consis-
within-subject variables. On the basis of the results for Experiment tent from the ages of 6 to 11. These findings indicate that even the
1, follow-up linear trend analyses tested the a priori prediction that younger children were sensitive to the causal structure of the story,
RTs to probes would increase as the time into a sequence of central using it to guide their allocation of cognitive resources. Given the
content increased, whereas RTs to probes would decrease as the reliable findings obtained for comparison children in Experiments
time into a sequence of incidental content increased. More children 1 and 2, Experiment 3 was designed to examine the cognitive
in the younger group (n 11) than in the older group (n 1) engagement of children with ADHD, whose problems in sustain-
failed to respond to at least one probe, 2(1, N 111) 47.34, ing attention and comprehending causal relations have been well
p .01. Mean probe RTs were calculated from the remaining RTs. documented (Lorch, Eastham, et al., 2004).
As hypothesized, children s RTs to secondary probes varied As discussed earlier, two studies have addressed the issue of
jointly as a function of when in a sequence a probe occurred and cognitive engagement among children with ADHD. The results of
LORCH, MILICH, ASTRIN, AND BERTHIAUME
1212
1100
1000
900
800
700
600
500
2 7 12 Later
Probe Position (Seconds into Sequence)
Central Incidental
Figure 2. Mean probe response times (RTs) as a function of time into central and incidental sequences for
Experiment 2.
Lorch, Eastham, et al. (2004) suggest that variations in cognitive importance to the overall story. To accomplish this purpose, we
engagement, as operationalized by time spent in long looks, ac- asked the children in Experiment 3 to sort pictures of story events
count for the problems that children with ADHD have with causal into categories of low, medium, or high importance, and we
relations questions. However, the degree of cognitive engagement compared age and diagnostic group differences in sorting accu-
was inferred from time spent in long looks. By using the secondary racy. In addition, we assessed the role of sorting accuracy in
task implemented in Experiments 1 and 2, Whirley et al. (2003) accounting both for moment-to-moment engagement on the sec-
offered a more direct assessment of cognitive engagement among ondary task and for the postviewing measure of story recall.
9- to 11-year-old boys with ADHD. The boys with ADHD failed
to show the same systematic increase in cognitive engagement
Method
observed in comparison boys as central sequences progressed.
Experiment 3 was designed to replicate and extend the findings
Participants
of Experiments 1 and 2 and from Whirley et al. (2003). In Exper-
This study was part of a larger longitudinal project designed to examine
iment 3, we use a developmental perspective to compare the
story comprehension and its relation to attention among children with
cognitive engagement of children with ADHD and their compar-
ADHD. Although the original sample included 193 children, data for 28
ison peers, using an age range that includes younger children (i.e.,
children had to be eliminated. This resulted in a final sample of 64 children
4 to 9 years) than the previous secondary task studies. Lorch and
(49 boys, 15 girls) with a confirmed diagnosis of ADHD and 101 com-
Castle (1997), using the secondary task, found that 5-year-olds
parison children (61 boys, 40 girls), ranging in age from 4.0 to 9.9 years
demonstrated variations in cognitive engagement as a function of
(M 7.5 years). Each group of children was further classified into two age
dramatic differences in the comprehensibility of television pro-
groups, a group of 64 younger children (between 4 and 7 years 0 months
gramming. However, in Experiment 3, like Experiments 1 and 2,
of age; M 5.75 years), and a group of 101 older children (between 7
the task used investigates whether children make changes in their
years 1 month and 9 years 11 months of age; M 8.5 years).
cognitive engagement as a sequence of central or incidental events To ensure an accurate diagnosis of ADHD, we used a three-step process
develops. Thus, Experiment 3 integrates the developmental per- in the recruitment of children assigned to the ADHD group. First, all
children assigned to this group were referred from medical care settings
spective offered by Experiments 1 and 2 with the questions raised
where, independently of the current study, they had received a diagnosis of
by Whirley et al. concerning children with ADHD.
ADHD/combined type according to criteria in the Diagnostic and Statis-
A second purpose of Experiment 3 was to assess children s
tical Manual of Mental Disorders (4th ed., text rev.; American Psychiatric
metacognitive understanding of the importance of events to the
Association, 2000). Second, referred children s medical records were re-
plot of the story. The secondary task is designed to tap moment-
viewed before they were admitted into the study. This review focused on
to-moment variations in cognitive engagement in relation to the
information regarding the children s behavior, intellectual and academic
centrality of the events. In contrast, a metacognitive understanding
functioning, medical history, age of diagnosis, and other relevant issues.
refers to whether, after viewing the program, children are able to
Children were excluded from the current study if information obtained
make explicit differentiations among events in terms of their during the review described a symptom picture inconsistent with ADHD/
RT in milliseconds
COGNITIVE ENGAGEMENT WITH TELEVISION
1213
Table 1
Comparison of Two Diagnostic Groups in Experiment 3 on Relevant Demographic Variables
ADHD Comparison
(n 64) (n 101)
Factor MSDMSD t(162) p
Age in years
Younger group 5.88 0.85 5.72 0.76 0.82 .417
Older group 8.61 0.96 8.48 0.85 0.72 .471
Mother s education 14.02 2.20 15.63 2.25 4.44 .001
Father s education 14.09 3.38 16.30 3.05 4.18 .001
DSM IV TR
Inattention 6.11 2.21 0.16 0.46 26.23 .001
Hyp/imp 6.19 1.98 0.18 0.48 29.20 .001
Oppositionality 3.49 2.31 0.27 0.71 13.69 .001
Note. ADHD attention-deficit/hyperactivity disorder; DSM IV TR Diagnostic and Statistical Manual of
Mental Disorders (4th ed., text rev.); Hyp/imp hyperactivity/impulsivity.
combined type, if their IQ score was less than 70, or if they were taking in Experiments 1 and 2; the second one was newly prepared for this study,
medications that could not be discontinued for the study. The mere pres- via the same procedures to identify appropriate probe positions in central
ence of comorbid diagnoses (e.g., oppositional defiant disorder or conduct and incidental sequences. Similarly, appropriate cued recall questions were
disorder) was not cause for exclusion from the study. Children whose generated for content that was high or low in importance.
principal ADHD symptoms related to inattention were excluded from this The second subsequent change in the procedures from the prior exper-
study because of growing empirical evidence suggesting that children with iments was that the auditory probes placed 7 s into the sequences were
attentional difficulties in the absence of hyperactivity and impulsivity may deleted, so that auditory probes only were presented at 2, 12, or 24 s or later
best be classified as suffering from a distinct disorder that is not a subtype into a sequence. This change was made to streamline the procedure and
of ADHD (Milich, Balentine, & Lynam, 2001). because in Whirley et al. (2003) the 7-s probe did not supply unique
At a final step, the diagnosis of ADHD/combined type was confirmed by information. For both importance levels (i.e., central and incidental), each
research staff using a semistructured parent interview designed to assess series contained between 4 and 5 probes in each basic position, resulting in
the presence of ADHD or oppositional defiant disorder according to a total of 13 probes for both the central and the incidental sequences. As in
DSM IV criteria. This same interview has been used successfully for the Experiments 1 and 2, each probe continued until a response was made.
classification of children with ADHD in previous studies (Lorch et al., However, because of the younger age of the sample in Experiment 3, the
2000; Lorch, Eastham, et al., 2004; Whirley et al., 2003). An earlier study sound ended after 7 s if no response was made. Twelve filler probes were
found the interrater reliability for number of ADHD symptoms endorsed in inserted in the program, so that each episode of the show consisted of 38
the parent interviews to be 99% (Lorch et al., 1999). Children were not probes, including 13 during presentation of central content, 13 during
assigned to the ADHD group unless this interview confirmed a diagnosis incidental content, and 12 filler probes that were not analyzed.
of ADHD/combined type. Children taking stimulant medication did not The third change in the experimental protocol was to include a sorting
receive their medication on the day of the study. task after completion of the probe task. The sorting task provided a
Children in the comparison group were recruited through newspaper measure of understanding of the importance of story events. The task
advertisements and flyers. Children were excluded from participation in consisted of 12 pictures: 4 depicting very important events from the story,
this group if data gathered through a parent interview and the Child 4 depicting medium important events, and 4 depicting not important
Behavior Checklist (Achenbach, 1991) suggested the presence of any events. Selection of the 12 story events was based on the adult ratings of
behavior or learning disorders. Once admitted into the study, parents had to importance of story events, described earlier. Each event was described in
complete a semistructured parent interview designed to assess the presence a caption at the bottom of the picture. The examiner read each caption
of ADHD or oppositional defiant disorder according to DSM IV criteria. while showing the card to the child. As each card was read, it was placed
Children who met three or more criteria for inattention or hyperactivity on the table in front of the child. Under each of the three category labels
were not included in the final analyses. (i.e., very important, medium important, not important) was placed a small
Demographic characteristics for each group of children can be found in grid, which was segmented into four spaces, each the size of a picture card.
Table 1. Children in the comparison group averaged less than one symptom The three categories were carefully explained to the child. The child was
of ADHD. By comparison, children in the ADHD group averaged 15.84 then asked to place the picture cards under the appropriate category
symptoms, a difference that was statistically significant, t(162) 33.04, heading. Only four events could be placed in each category. The child was
p .001 (r .93). There were no group differences in terms of age, allowed to ask for any event description to be repeated and could make as
t(162) 0.587, p .10, but the groups did differ significantly in terms of many changes to the categorizations as desired.
mothers and fathers average years of education.2
2
Because of the group differences in parent education, analyses were
Procedure
repeated with mothers years of education as a covariate. The pattern of
The materials and procedure were identical to those of Experiments 1 results was unchanged with the covariate included. Similar analyses were
and 2, except for the following three changes. First, because the children not undertaken for fathers years of education because of a higher fre-
were participating in a longitudinal investigation in which they would quency of missing data on this variable. However, the high correlation
repeat the secondary probe task at a later time period, two different between mothers and fathers years of education (r .89) suggests that
Growing Pains episodes were used. One episode was the same as was used the pattern of results would be the same.
LORCH, MILICH, ASTRIN, AND BERTHIAUME
1214
There were three dependent variables concerning importance judgments Position interaction, F(2, 161) 3.5, p .05. Because this
in the sorting task. The measure of gross errors included only those very
interaction was significant and different patterns have been found
important events the child placed in the not important pile or vice versa.
for central and incidental sequences (Experiments 1 and 2; Whir-
The number of seconds to complete the sorting task and the total number
ley et al., 2003), the following results are presented separately for
of moves made in completing the sorting task were measures of impulsivity
the two types of sequences.
and of how effectively the child planned and executed sorting categoriza-
Central sequences. For central sequences, a significant main
tions. These are considered core symptoms of ADHD, so these measures
effect of probe position, F(2, 161) 13.36, p .001, was
were included to determine whether such difficulties explained any group
qualified by a significant Diagnostic Group Probe Position
differences in sorting errors.
interaction, F(2, 161) 11.19, p .001 (see Figure 3). Linear
trend analysis of this interaction indicated that RTs to probes for
Results and Discussion
comparison children increased as central content continued, F(1,
The main dependent variable for the secondary task was RTs to
100) 38.70, p .001 (r .53). Follow-up analyses revealed
26 target probes. Data for children were not included in the final
that RTs to the later probes were significantly longer than both the
analyses unless the children had responded to at least two of the
2-s probes, t(100) 6.22, p .01 (r .53), and the 12-s probes,
probes for any given probe position. Ten children were excluded
t(100) 5.08, p .01 (r .21). In contrast to the pattern for the
because they responded to too few probes. Five of these children
comparison children, linear trend analysis of the RTs of children
were in the ADHD group, whereas 5 were comparison partici-
with ADHD demonstrated no significant change as central se-
pants, 2(1, N 175) 1, p .10. Because of equipment
quences progressed, F(1, 63) 1.
malfunction, data for 18 children (10 comparison, 8 with ADHD)
Younger children were slower to respond to probes than older
were not used. Excluding these 28 children left the final sample of
children, F(1, 161) 10.66, p .001 (r .25), but this main
64 children in the ADHD group and 101 children in the compar-
effect was qualified by a significant Diagnostic Group Age
ison group.
Group interaction, F(1, 161) 4.59, p .05 (r .17). Younger
comparison children were significantly slower than their older
Probe RTs
counterparts, F(1, 99) 24.18, p .001 (r .44), but younger
and older children with ADHD did not differ in overall RT, F(1,
To explore the patterns of engagement for children with ADHD
62) 1.
versus comparison children, we performed a 2 2 2 3 mixed
Incidental sequences. For incidental sequences, there was a
ANOVA. Diagnostic group (comparison and ADHD) and age
main effect of diagnostic group, such that children with ADHD
group (younger and older) were the between-subjects variables,
(M 911 ms) showed significantly longer RTs than comparison
and sequence type (central vs. incidental) and probe position (2,
children (M 794 ms) to probes presented during incidental
12, and 24 s and beyond into sequence) were the within-subject
variables. Given the significant Centrality Probe Position inter- sequences, F(1, 161) 5.75, p .05 (r .19). Younger children
actions found in Experiments 1 and 2, the primary effect of interest (M 948 ms) responded significantly slower than their older
in Experiment 3 was the Diagnostic Group Centrality Probe counterparts (M 757 ms), F(1, 161) 15.03, p .001 (r .29).
1100
1000
900
800
700
600
500
2 12 Later
Probe Position (Seconds into Sequence)
ADHD, Central Comparison, Central
ADHD, Incidental Comparison, Incidental
Figure 3. Mean probe response times (RTs) as a function of time into central and incidental sequences for
children with attention-deficit/hyperactivity disorder (ADHD) and nonreferred children for Experiment 3.
RT in milliseconds
COGNITIVE ENGAGEMENT WITH TELEVISION
1215
There were no significant interactions for incidental sequences moves. There were no main effects or interactions involving di-
(see Figure 3).3 agnostic group for either measure. The only significant effect was
that younger children (M 13.2) made significantly fewer moves
than older children (M 14.2), F(1, 159) 4.56, p .05 (r
Cued Recall Performance
.17).
Percentage correct on cued recall questions was analyzed in a
2 2 2 mixed ANOVA, with diagnostic group and age group
Importance Judgments in Relation to Cognitive
as between-subjects variables and centrality as a within-subject
Engagement and Story Recall
variable. Significant main effects of diagnostic group, F(1, 161)
6.05 p .05 (r .19); age group, F(1, 161) 127.49, p .001
We examined the extent to which importance judgments, as
(r .66); and centrality, F(1, 161) 87.98, p .001 (r .59),
indexed by gross errors, accounted for group differences in the
were qualified by significant Diagnostic Group Age Group, F(1,
pattern of cognitive engagement and in recall performance by
161) 5.18, p .05 (r .18), and Centrality Age Group, F(1,
entering gross errors as a covariate in the original ANOVAs. In
161) 7.48, p .01 (r .21), interactions. Performance of older
terms of the pattern of cognitive engagement, gross errors did not
comparison children (M 55% correct) was significantly better
significantly relate to RT, F(1, 160) 2.05, p .10, nor did any
than that of the older children with ADHD (M 44%), F(1, 99)
of the significant main effects or interactions change. However,
13.58, p .001 (r .35). Younger comparison children (M
gross errors did significantly predict cued recall performance, F(1,
23%) and younger children with ADHD (M 23%) did not differ
160) 17.61, p .001 (r .31), and the main effect of diagnostic
in their cued recall performance, F(1, 62) 1. Children correctly
group was no longer significant, F(1, 160) 2.39, p .10.
answered significantly more questions testing central content
(M 43%) than testing incidental content (M 28%), and this
General Discussion
was true for both the younger, t(63) 4.3, p .001 (r .48), and
the older, t(100) 10.2, p .001 (r .71), children. However,
The findings of these three investigations demonstrate that on-
the difference in recall of central and incidental content was larger
line variations in comparison children s cognitive engagement
for older children (mean difference 20%) than for younger
with a televised story were related to the continuity of central
children (mean difference 11%), t(163) 2.84, p .005 (r
content, and this conclusion held true for children ranging in age
.22).
from 4 to 11. As indicated by probe RTs, comparison children
As in Experiments 1 and 2, we examined whether recall perfor-
became increasingly engaged the longer sequences of central
mance was related to differing levels of engagement. Mean RTs
events continued. In contrast, 4- to 9-year-old children with
were computed for probes corresponding to questions that each
ADHD showed no change in cognitive engagement as central
child answered correctly and questions each child answered incor-
sequences progressed. In the discussion that follows, we first
rectly, and this variable was found to interact with age group, F(1,
examine the general implications of these findings, then address
145) 3.75, p .055 (r .16), but not with diagnostic group,
developmental interpretations and consider specific ways the per-
F(1, 145) 1. Regardless of centrality, older children s RTs
formance of children with ADHD differs from the patterns ob-
corresponding to questions they answered correctly (M 798 ms)
tained for comparison children.
tended to be longer than their RTs corresponding to questions they
The pattern of changes in cognitive engagement is consistent
answered incorrectly (M 754 ms), t(99) 1.86, p .067 (r
with earlier findings suggesting that children s attention becomes
.14), but there was no significant difference for younger children
more engaged as a coherent story segment develops (Hawkins et
(Ms 932 and 954 ms, respectively), t(48) 1. This suggests that
al., 1991; Lorch & Castle, 1997; Meadowcroft & Reeves, 1989). It
for older children, increased online cognitive engagement was
is notable that the earlier studies examined relatively gross differ-
associated with greater recall of story events.
ences in story structure (i.e., events in a meaningful order vs.
events in a scrambled, nonsensical order). The current study,
however, provides evidence of systematic changes in school-age
Importance Judgments
children s cognitive engagement in response to more subtle dif-
Three dependent variables concerning importance judgments in
ferences in the meaning of events. In the stories used in these
the sorting task were analyzed in 2 2 ANOVAs, with diagnostic
experiments, all events occurred in a sensible order. However, as
group and age group as between-subjects variables. The three
dependent variables were (a) gross errors, which involved placing
3
low-importance pictures in the high-importance category, or vice Although children in both diagnostic groups demonstrated high rates of
visual attention to the television, comparison children (M 96%) were
versa; (b) number of seconds to complete the sorting task; and (c)
more likely than children with ADHD (M 90%) to be attending to the
number of moves made in completing the sorting task.
television when target probes were presented, t(163) 4.11, p .001.
For gross errors, there were significant main effects of diagnos-
Therefore, we repeated the analyses, excluding probes when children were
tic group, F(1, 161) 9.01, p .01 (r .23), and age group,
not attending to the television. This required dropping 10 additional chil-
F(1,161) 44.91, p .001 (r .47). Children with ADHD (M
dren (3 comparison, 7 with ADHD) from the analyses, because not all cells
2.5) made more errors than comparison children (M 1.9), and
had a sufficient number of probes. Despite the loss of power, the pattern of
younger children (M 2.9) made more errors than older children
RT results remained the same, with a significant Group Linear Probe
(M 1.5). To determine whether problems in impulsivity or
Position interaction for central sequences, F(1, 151) 11.40, p .001, but
planning accounted for poorer performance among children with
the interaction failed to reach significance for incidental sequences, F(1,
ADHD, we analyzed average sorting time and number of sorting 151) 3.36, p .05.
LORCH, MILICH, ASTRIN, AND BERTHIAUME
1216
central sequences progressed, more content connected to the plot 1997). Similar to the increase in cognitive engagement for very
was provided, eliciting greater cognitive engagement from chil- long looks, in the current study the continuation of central content
dren. In contrast, as incidental sequences progressed, plot devel- also promoted increased cognitive engagement with the story, even
opment was not enhanced, leading either to decreased cognitive though children maintained high levels of visual attention through-
engagement (Experiments 1 and 2) or to no systematic change in out the session. Thus, taken together, these findings highlight the
engagement (Experiment 3; Whirley et al., 2003). importance of long sequences on level of cognitive engagement for
The current findings are consistent with the notion that, during both long looks and long sequences of a particular type of content.
viewing, children engage in the online construction of a story Consistent with other investigations (Lorch et al., 1987; Tra-
representation (Trabasso & Nickels, 1992; Trabasso et al., 1992). basso et al., 1984; van den Broek, 1989; van den Broek et al.,
Sequences of central events consist primarily of events that are 1996), children s memory for central events was superior to their
part of the causal chain defining the plot of the story and are memory for incidental events. Thus, memory for story events
perceived as very important by adult viewers. As such, the events reflects the causal structure of the story. Part of this effect may be
in central sequences figure importantly in a coherent representa- due to postviewing processes of structuring the story as content is
tion of the story. As children view the program and a sequence of retrieved during the recall task. However, the current results sug-
central content continues, they encounter events that can be con- gest that children s systematic increases or decreases in their
nected to the main thread of the story. In contrast, incidental events online allocation of resources also may contribute to recall differ-
are not part of the causal chain and are considered generally ences. Consistent with this interpretation, children tended to show
unimportant by adult viewers. As a sequence of incidental events greater cognitive engagement with events that they later recalled at
continues, children encounter events that cannot be linked to the a higher rate. This was true for central sequences across all three
plot of the story. The lack of connections to the causal chain of the experiments and for incidental sequences in Experiments 1 and 3.
story leads children to stop at more superficial processing; thus, Similar to these findings, Britton, Piha, Davis, and Wehausen
they do not need to increase allocation of attention to events (1978) provided evidence that learning from text was related to
presented later in incidental sequences. These systematic changes adult readers probe RTs.
in children s cognitive engagement indicate that they are building
a coherent representation during viewing that reflects the causal
Developmental Implications
structure of the story.
This online construction of a story representation also is con- One goal of the current investigation was to examine whether
sistent with Huston and Wright s (1983) proposal that children older and younger elementary school children would differ in their
may engage in cycles of decisions concerning whether to perform patterns of cognitive engagement as a function of the centrality and
deeper and more elaborative processing of material. Huston and continuity of story content. Overall, children in both age groups
Wright s theory was designed to predict patterns of visual attention showed similar systematic variations in cognitive engagement.
to television under conditions that enable children to engage in These findings indicate that even the younger children were sen-
alternative activities (e.g., toy play). Under such conditions, visual sitive to the causal structure of the story, using it to guide their
attention is predicted by formal and content characteristics that allocation of cognitive resources.
relate to children s comprehension (Alwitt et al., 1980; Campbell Despite these similar patterns of engagement for the two age
et al., 1987). Thus, visual attention provides one indication of groups, several results suggest that older children were more
children s cognitive processing during viewing. However, the task effective cognitive processors in their story comprehension and
conditions of the current study elicited consistently high levels of allocation of resources. First and not surprising, the older children
visual attention, even from younger children. RTs to secondary were faster to respond to probes and answered more cued recall
probes suggest that even while children maintained visual atten- questions correctly. Second, only older children in Experiments 2
tion, their decisions to engage in deeper processing were influ- and 3 demonstrated the relation between level of engagement and
enced by the centrality and continuity of content. These findings recall of plot content. It may be that younger children can effec-
indicate that the secondary task procedure can serve as a valuable tively vary allocation of attention in response to plot development
technique for revealing subtle variations in children s cognitive but lag behind older children in translating increased attention into
engagement. enhanced story recall. This hypothesized developmental progres-
The clearest indication of systematic change in cognitive en- sion from behavioral response to enhanced understanding is sim-
gagement as a function of time into a sequence is observed when ilar to one observed in online text comprehension. That is, on
RTs to the later probes are considered. These later probes occurred encountering inconsistent information in a text, both younger and
after a sequence of central content had been in progress for at least older children slow their reading, but older children are more likely
24 s. The strength of these effects for later probes suggests a to recall the inconsistency (Harris, Kruithof, Terwogt, & Visser,
conceptual parallel with the phenomenon of attentional inertia 1981; Zabrucky & Ratner, 1986).
(Anderson, Choi, & Lorch, 1987). Attentional inertia refers to the
observation that the longer a look at the television has been in
Cognitive Engagement in Children With ADHD
progress, the higher is the probability that the look will continue,
with this function leveling off at approximately 15 s into a look. Of In dramatic contrast to the systematic changes in cognitive
special relevance to the current study, converging evidence from engagement observed for comparison children, children with
several measures indicates that cognitive engagement is greater if ADHD showed no variation in probe RTs as sequences of central
a look has continued for at least 15 s than it is during a shorter look or incidental material progressed. As such, the group difference in
(Anderson et al., 1987; Burns & Anderson, 1993; Lorch & Castle, the pattern of cognitive engagement was even more marked than
COGNITIVE ENGAGEMENT WITH TELEVISION
1217
that reported by Whirley et al. (2003), who found that only late in sustaining cognitive engagement. In turn, these difficulties impair
central sequences did children with ADHD begin to show in- the story comprehension of children with ADHD.
creased engagement. Given that Whirley et al. s sample was sev-
eral years older than the children in Experiment 3, this suggests
Limitations
that children with ADHD show a pronounced developmental delay
in adjusting their cognitive engagement in response to the impor- One limitation of the current series of experiments is that they
all used one or two episodes of the same situation comedy pro-
tance of current content to the developing story. Even comparison
gram. The degree to which the obtained results would generalize to
children as young as 4 years of age showed the expected linear
other types of programming or even other situation comedies is
increase as central content progressed, whereas it was not until age
unknown. However, both Growing Pains episodes were selected
9 at the earliest (Whirley et al., 2003) that children with ADHD
on the basis of their strong narrative structures. Both episodes
demonstrated an increase in cognitive engagement from the be-
conformed to prototypical story structures (Mandler & Johnson,
ginning to the end of central sequences. However, it should be
1977), in which an initiating event creates a goal that the protag-
noted that the children with ADHD in the Whirley et al. study
onist must reach through a series of attempts to overcome obsta-
showed a decrease in cognitive engagement during central se-
cles to goal attainment. As such, we expect that the current studies
quences prior to the significant increase in RTs at the last probe
consistent finding of increased engagement as central sequences
position. This pattern has not been observed for any age group of
progressed would be replicated in experiments using other pro-
comparison children.
grams with strong narrative structures.
It should be noted that, in general, the children with ADHD had
Although the core finding of increased RTs as central sequences
significantly longer RTs than the comparison children. This might
continued was consistent for comparison children across the three
lead one to the interpretation that children with ADHD are more
studies, there were a few inconsistent or unexpected results, espe-
engaged with the television program than comparison children.
cially concerning the patterns of RTs for incidental sequences. In
However, it is well documented that children with ADHD dem-
Experiments 1 and 2, RTs to probes occurring early in incidental
onstrate slower responses on a variety of reaction time tasks,
sequences were longer than RTs to probes in the same position in
including simple reaction time tasks requiring virtually no cogni-
central sequences. Furthermore, although Experiments 1 and 2
tive processing (Douglas, 1999). Just as we would not argue that
revealed decreases in engagement as incidental sequences pro-
younger children s longer RTs in Experiment 2 indicate they were
gressed, no such decrease was found in Experiment 3 for either
more engaged than older children, we would not conclude pro-
diagnostic group. These inconsistencies for incidental sequences
cessing differences on the basis of overall group RT comparisons.
may reflect the possibility that factors other than the plot relevance
Instead, as we have argued throughout, it is the increase in RTs as
of story content influence children s engagement with these se-
central sequences progress that is indicative of increased
quences. Because the initially long RTs to incidental probes in
engagement.
Experiments 1 and 2 were particularly unexpected, we examined
The present results concerning the children with ADHD are
the content of these sequences to see whether we could identify
conceptually similar to those reported by Meadowcroft and Reeves
any other differences between the central and incidental sequences
(1989). Their finding that only children with well-developed story
to account for these findings. There were no differences between
schema knowledge showed differences in engagement as story
sequence type in the amount of dialogue, the incidence of the laugh
coherence varied is consistent with prior results that children with
track, episode boundaries, or scene changes. Several of the inci-
ADHD are deficient in their appreciation of story structure vari-
dental sequences appear to be for comedic purposes rather than to
ables (Lorch, O Neil, et al., 2004; Renz et al., 2003). The results
provide plot-relevant content. It may be that children s attention
of both Experiment 3 and Meadowcroft and Reeves reinforce the
was drawn to these moments of comic relief, but when children
view that appropriate adjustments in cognitive engagement are
realized that the content was not necessary for understanding the
related to the ability to achieve coherent story representations.
plot, engagement was quickly reduced. The fact that in Experiment
The results for children with ADHD also are conceptually
3 incidental sequences did not provoke initially longer RTs, nor
similar to those of Lorch, Eastham, et al. (2004), who manipulated
was there any change as the sequences progressed, may reflect the
levels of visual attention and operationalized cognitive engage-
addition of a second television program. Perhaps it is not surpris-
ment in terms of long looks (i.e., at least 15 s) at the television.
ing that there is inconsistency in the patterns of responses to
Findings from both investigations support the interpretation that
incidental sequences, because, by definition, these sequences are
comparison children are better able than children with ADHD to
extraneous to the plot and thus are likely to produce unpredictable
alter cognitive engagement in response to changes in story content.
influences on children s engagement. The most critical point is that
In Lorch, Eastham, et al. s study, the amount of time spent in
no group of children ever became more engaged with incidental
deeper cognitive processing during long looks helped to explain
material as the sequences progressed. For future research, the
the differential patterns of recall in children with ADHD and
primary focus should be on children s engagement with central
comparison children. In Experiment 3, the systematic changes in
content, with incidental sequences serving merely as a control to
cognitive engagement demonstrated by comparison children sug- ensure that increases in RTs are specific to central sequences.
gest that these children constructed a more complete story repre-
sentation than children with ADHD. Thus, there is compelling
Future Directions
evidence (Experiment 3; Lorch, Eastham, et al., 2004; Whirley et
al., 2003) across a wide age range and different methodologies that Additional research is needed to better understand and evaluate
children with ADHD have significant difficulties achieving and the development of children s online processes of building story
LORCH, MILICH, ASTRIN, AND BERTHIAUME
1218
representations, both for comparison children and for those diag- Preschool children s visual attention to attributes of television. Human
Communication Research, 7, 52 67.
nosed with ADHD. Many of the conclusions about the develop-
American Psychiatric Association. (2000). Diagnostic and statistical man-
ment of story comprehension have been inferred from studies
ual of mental disorders (4th ed., text rev.). Washington, DC: Author.
examining recall of story events. To create a fuller understanding
Anderson, D. R., Choi, H. P., & Lorch, E. P. (1987). Attentional inertia
of children s comprehension, more research is needed using meth-
reduces distractibility in young children s television viewing. Child
odologies that examine online processes of story comprehension. It
Development, 58, 798 806.
may be possible to use the secondary task methodology with both
Anderson, D. R., & Lorch, E. P. (1983). Looking at television: Action or
televised stories and written stories in a converging operations
reaction? In J. Bryant & D. R. Anderson (Eds.), Children s understand-
approach to gain more specific information about children s online
ing of television: Research on attention and comprehension (pp. 1 33).
processing of stories (Beentjes & van der Voort, 1993). Use of
New York: Academic Press.
televised stories makes it possible to present more lengthy and
Anderson, D. R., Lorch, E. P., Field, D. E., & Sanders, J. (1981). The
complex stories and to study a relatively wide age range of effects of TV program comprehensibility on preschool children s visual
attention to television. Child Development, 52, 151 157.
children. Use of the secondary task methodology during reading of
Anderson, D. R., Lorch, E. P., Smith, R., Bradford, R., & Levin, S. R.
written stories (Britton et al., 1978; Britton & Tesser, 1982),
(1981). The effects of peer presence on preschool children s visual
however, would permit greater control over specific content and
attention to television. Developmental Psychology, 17, 446  453.
variations in story structure. Another methodology that could be
Baer, S. A., & Lorch, E. P. (1990, March). Effects of importance on
used to examine ongoing story processing, particularly in younger
children s attention to television. Paper presented at the meeting of the
children, is online story narration (Renz et al., 2003; Trabasso &
Southeastern Regional Conference on Human Development, Richmond,
Nickels, 1992; Trabasso et al., 1992). This methodology examines
VA.
the extent to which children use a goal-based structure to incor-
Basil, M. U. (1994). Secondary reaction-time measures. In A. Lang (Ed.),
porate new information into their ongoing story representation.
Measuring psychological responses to media messages (pp. 85 98).
Future research also can examine how deficits in cognitive
Hillsdale, NJ: Erlbaum.
engagement for children with ADHD might account for their Beentjes, J. W. J., & van der Voort, T. H. A. (1993). Television viewing
versus reading: Mental effort, retention, and inferential learning. Com-
well-documented academic problems. Results of this and other
munication Education, 42, 191 205.
studies (e.g., Lorch, O Neil, et al., 2004) indicate that the academic
Berthiaume, K. S. (in press). Attentional and story comprehension deficits
deficits associated with ADHD go beyond mere problems in
in children with ADHD: What is the connection? School Psychology
sustaining attention and instead reflect difficulties in sustaining
Review.
cognitive engagement, identifying important content, and making
Black, J. B., & Bower, G. H. (1980). Story understanding as problem
connections among story events (Milich et al., 2005). Future
solving. Poetics, 9, 223 250.
research also can begin to test the efficacy of academic interven-
Britton, B. K., Graesser, A. C., Glynn, S. M., Hamilton, T., & Penland, M.
tions designed to target these specific deficits. Such targeted in-
(1983). Use of cognitive capacity in reading: Effects of some content
terventions may include the mapping of important story events and
features of text. Discourse Processes, 6, 39 57.
their interconnections, training in the focused use of advance- Britton, B. K., Piha, A., Davis, J., & Wehausen, E. (1978). Reading and
organizing techniques, and learning studying strategies that em- cognitive capacity usage: Adjunct question effects. Memory and Cog-
nition, 6, 266 273.
phasize connections among story events (Berthiaume, in press).
Britton, B. K., & Tesser, A. (1982). Effect of prior knowledge on use of
In summary, the current investigation provides evidence that 4-
cognitive capacity in three complex cognitive tasks. Journal of Verbal
to 11-year-old comparison children, but not children with ADHD,
Learning and Verbal Behavior, 21, 421 436.
systematically vary their cognitive engagement while watching a
Burns, J. J. & Anderson, D. R. (1993). Attentional inertia and recognition
televised story in a way that is appropriate for building a coherent
memory in adult television viewing. Communication Research, 20,
story representation. A more thorough understanding of children s
777 799.
story comprehension processes may be helpful for understanding
Calvert, S. L., Huston, A. C., Watkins, B. A., & Wright, J. C. (1982). The
comprehension processes in general. In school, many tasks that
relation between selective attention to television forms and children s
children complete are related to story comprehension, and an
comprehension of content. Child Development, 53, 601 610.
understanding of processing during these tasks may assist and
Campbell, T. A., Wright, J. C., & Huston, A. C. (1987). Form cues and
guide the presentation of material to be maximally beneficial to content difficulty as determinants of children s cognitive processing of
children s learning, especially for children who experience aca- televised educational messages. Journal of Experimental Child Psychol-
ogy, 43, 311 327.
demic difficulties.
Douglas, V. I. (1999). Cognitive control processes in attention-deficit/
hyperactivity disorder. In H. C. Quay & A. E. Hogan (Eds.), Handbook
of disruptive behavior disorders (pp. 105 139). New York: Kluwer.
References
Graesser, A. C., & Clark, L. F. (1985). The structures and procedures of
implicit knowledge. Norwood, NJ: Ablex.
Achenbach, T. M. (1991). Manual for the Child Behavior Checklist.
Harris, P. L., Kruithof, A., Terwogt, M., & Visser, T. (1981). Children s
Burlington: Department of Psychiatry, University of Vermont.
detection and awareness of textual anomaly. Journal of Experimental
Ackerman, B. P. (1986). Referential and causal coherence in the story
comprehension of children and adults. Journal of Experimental Child Child Psychology, 31, 1215 1223.
Psychology, 41, 336 366. Hawkins, R. P., Kim, Y., & Pingree, S. (1991). The ups and downs of
Ackerman, B. P., Paine, J., & Silver, D. (1991). Building a story repre- attention to television. Communication Research, 18, 53 76.
sentation: The effects of early concepts on later causal inferences by Hawkins, R., Tapper, J., Bruce, L., & Pingree, S. (1995). Strategic and
children. Developmental Psychology, 27, 370  380. nonstrategic explanations for attentional inertia. Communication Re-
Alwitt, L. F., Anderson, D. R., Lorch, E. P., & Levin, S. R. (1980). search, 22, 188 206.
COGNITIVE ENGAGEMENT WITH TELEVISION
1219
Huston, A. C., & Wright, J. C. (1983). Children s processing of television: Pingree, S. (1986). Children s activity and television comprehensibility.
The informative functions of formal features. In J. Bryant & D. R. Communication Research, 13, 239 256.
Anderson (Eds.), Children s understanding of television: Research on Renz, K., Lorch, E. P., Milich, R., Lemberger, C., Bodner, A., & Welsh, R.
attention and comprehension (pp. 35 58). New York: Academic Press. (2003). On-line story representation in boys with attention deficit hy-
Huston, A. C., Wright, J. C., Wartella, E., Rice, M. L., Watkins, B. A., peractivity disorder. Journal of Abnormal Child Psychology, 31, 93
Campbell, T., & Potts, R. (1981). Communicating more than content: 104.
Formal features of children s television programs. Journal of Commu- Rolandelli, D. R., Wright, J. C., Huston, A. C., & Eakins, D. (1991).
nication, 31, 32 48. Children s auditory and visual processing of narrated and nonnarrated
Kahneman, D. (1973). Attention and effort. Englewood Cliffs, NJ: television programming. Journal of Experimental Child Psychology, 51,
Prentice-Hall. 90 122.
Lorch, E. P., Anderson, D. R., & Levin, S. R. (1979). The relationship of Salomon, G. (1983). Television watching and mental effort: A social
visual attention to children s comprehension of television. Child Devel- psychological view. In J. Bryant & D. R. Anderson (Eds.), Children s
opment, 50, 722 727. understanding of television: Research on attention and comprehension
Lorch, E. P., & Baer, S. A. (1997, April). Children s strategic adjustments (pp. 181 198). New York: Academic Press.
in attention to and recall from television. Paper presented at the meeting Schmidt, C. R., & Paris, S. G. (1983). Children s use of successive clues
of the Society for Research in Child Development, Washington, DC. to generate and monitor inferences. Child Development, 54, 742 759.
Lorch, E. P., Bellack, D. R., & Augsbach, L. H. (1987). Young children s Thorson, E., Reeves, B., & Schleuder, J. (1985). Message complexity and
memory for televised stories: Effects of importance. Child Development, attention to television. Communication Research, 12, 427 454.
58, 453 463. Trabasso, T., & Nickels, M. (1992). The development of goal plans of
Lorch, E. P., & Castle, V. J. (1997). Preschool children s attention to action in the narration of a picture story. Discourse Processes, 15,
television: Visual attention and probe response times. Journal of Exper- 249 275.
imental Child Psychology, 66, 111 127. Trabasso, T., Secco, T., & van den Broek, P. (1984). Causal cohesion and
Lorch, E. P., Eastham, D., Milich, R., Lemberger, C. C., Sanchez, R. P., story coherence. In H. Mandl, N. L. Stein, & T. Trabasso (Eds.),
Welsh, R., & van den Broek, P. (2004). Difficulties in comprehending Learning and comprehension of text (pp. 83 111). Hillsdale, NJ: Erl-
causal relations among children with ADHD: The role of cognitive baum.
engagement. Journal of Abnormal Psychology, 113, 56  63. Trabasso, T., Stein, N. L., Rodkin, P. C., Munger, M. P., & Baughn, C. R.
Lorch, E. P., Milich, R., Sanchez, R. P., van den Broek, P., Baer, S., Hooks, (1992). Knowledge of goals and plans in the on-line narration of events.
K., et al. (2000). Comprehension of televised stories in boys with Cognitive Development, 7, 133 170.
attention deficit hyperactivity disorder and nonreferred boys. Journal of Trabasso, T., & van den Broek, P. (1985). Causal thinking and the repre-
Abnormal Psychology, 109, 321 330. sentation of narrative events. Journal of Memory and Language, 24,
Lorch, E. P., O Neil, K., Berthiaume, K. S., Milich, R., Eastham, D., & 612 630.
Brooks, T. (2004). Story comprehension and the impact of studying on van den Broek, P. (1989). Causal reasoning and inference making in
recall in children with ADHD. Journal of Clinical Child and Adolescent judging the importance of story statements. Child Development, 60,
Psychology, 33, 506 515. 286 297.
Lorch, E. P., Sanchez, R., van den Broek, P., Milich, R., Murphy, E. L., van den Broek, P. W. (1997). Discovering the cement of the universe: The
Lorch, R., & Welsh, R. (1999). The relation of story structure properties development of event comprehension from childhood to adulthood. In P.
to recall of television stories in young children with attention deficit van den Broek, P. Bauer, & T. Bourg (Eds.), Developmental spans in
hyperactivity disorder and nonreferred peers. Journal of Abnormal Child event comprehension: Bridging fictional and actual events (pp. 321
Psychology, 27, 293 309. 342). Hillsdale, NJ: Erlbaum.
Mandler, J. M., & Johnson, N. S. (1977). Remembrance of things parsed: van den Broek, P. W., Lorch, E. P., & Thurlow, R. (1996). Children s and
Story structure and recall. Cognitive Psychology, 9, 111 151. adults memory for television stories: The role of causal factors, story-
Meadowcroft, J. M., & Reeves, B. (1989). Influence of story schema grammar categories and hierarchical level. Child Development, 67,
development on children s attention to television. Communication Re- 3010 3028.
search, 16, 352 374. Whirley, K. S., Lorch, E. P., Lemberger, C. C., & Milich, R. (2003). Online
Milich, R., Balentine, A., & Lynam, D. R. (2001). ADHD combined type cognitive engagement of boys with ADHD. Journal of Attention Disor-
and ADHD predominantly inattentive type are distinct and unrelated ders, 7, 1 11.
disorders. Clinical Psychology: Science and Practice, 8, 463 488. Zabrucky, K., & Ratner, H. H. (1986). Children s comprehension moni-
Milich, R., Lorch, E. P., & Berthiaume, K. (2005). Story comprehension in toring and recall of inconsistent stories. Child Development, 57, 1401
children with ADHD: Research findings and treatment implications. In 1418.
M. P. Larimer (Ed.), Progress in attention/deficit hyperactivity disorder
research (pp. 111 137). Hauppauge, NY: Nova Science.
Received May 24, 2005
Nezworski, R., Stein, N. L., & Trabasso, T. (1982). Story structure vs.
Revision received December 12, 2005
content in children s recall. Journal of Verbal Learning and Verbal
Behavior, 21, 196 206. Accepted December 12, 2005