J Autism Dev Disord (2006) 36:655–664

DOI 10.1007/s10803-006-0108-z

ORIGINAL PAPER

The Collateral Effects of Joint Attention Training on Social

Initiations, Positive Affect, Imitation, and Spontaneous Speech
for Young Children with Autism
Christina Whalen Æ Laura Schreibman Æ
Brooke Ingersoll

Published online: 30 June 2006

Ó Springer Science+Business Media, Inc. 2006

Abstract Joint attention may be a core deficit in autism Introduction

which underlies the abnormal development of later emerg-
ing social-communication behaviors. Given this theory, Although researchers are still searching for the cause of
researchers have suggested that teaching young children autism, there is no lack of discussion on what might be the
with autism to engage in joint attention may lead to collat- core psychological deficits in this disorder. Since Leo
eral increases in other non-targeted social-communication Kanner’s initial description of autism in 1943, many
behaviors. In this study, children with autism participated in researchers have advocated that one of the core distur-
a 10-week joint attention training program and collateral bances in autism is in social communication (e.g. Baron-
changes in non-targeted behaviors were assessed. Following Cohen, 1995; Kasari, Freeman, & Paparella, 2001). One of
participation in the intervention, positive collateral changes the earliest social impairments noted in children with aut-
were observed in social initiations, positive affect, imitation, ism is their lack of joint attention during infancy and early
play, and spontaneous speech. Results support the hypoth- childhood. Joint attention can be broadly defined as the
esis that teaching joint attention skills leads to improvement ability to coordinate attention between an object and a
in a variety of related skills and have implications for the person in a social context (Adamson & McArthur, 1995).
treatment of young children with autism. Many researchers believe that joint attention may be a core
deficit in children with autism negatively affecting their
Keywords Joint attention Æ Language Æ Social skills Æ development in language, play, and social interactions (e.g.
Play Æ Imitation Charman et al., 1997).
According to Sigman and Capps (1997), three criteria
must be met to identify a core deficit in a particular dis-
C. Whalen
order: (1) Specificity (i.e. the deficit should be specific to
University of Washington Autism Center, Seattle, WA, USA
the disorder and not found in other disorders); (2) univer-
L. Schreibman sality (i.e. the deficit should be present in all children with
Department of Psychology, University of California, San Diego, the disorder); and (3) primacy (i.e. the deficit should
CA, USA
emerge in the early stages of development). In children
B. Ingersoll with autism, all of these criteria appear to be met in terms
Oregon Health & Science University, Portland, OR, USA of joint attention and other early social-communication
skills. First, these deficits have been shown to be specific to
Present Address:
the disorder. Studies that have looked at joint attention
C. Whalen (&)
TeachTown, Inc., 2815 Eastlake Avenue E., Suite 300, Seattle, deficits in other childhood disorders have not found simi-
WA 98102, USA larities to children with autism (e.g. Loveland & Landry,
e-mail: chris@teachtown.com 1986; Ruskin, Kasari, Mundy, & Sigman, 1994). Second,
joint attention deficits appear to be present in most children
Present Address:
B. Ingersoll with autism (Kasari et al., 2001). Finally, joint attention is
Lewis and Clark College, Portland, OR, USA one of the earliest emerging social behaviors and thus

123
656 J Autism Dev Disord (2006) 36:655–664

meets the criteria for primacy. According to the criteria social-communication behaviors (e.g., Mundy, 1995).
proposed by Sigman and Capps (1997), joint attention However, few empirical studies have attempted to directly
appears to be a reasonable candidate for a core deficit in test this assumption.
autism. Some researchers have directly targeted joint attention
Research supports a link between autism and the behaviors in young children with autism with promising
development of other later-emerging social-communica- results. Increases in showing, pointing, and sharing were
tion behaviors. Recent theories of social development have observed in one child with autism after joint attention
paid particular attention to the role of joint attention (e.g. training in an early intervention study by Kasari et al.
Baron-Cohen, 1995; Mundy & Crowson, 1997; Striano & (2001). In addition, significant gains in language were also
Rochat, 1999), most notably its relationship with the observed for this child. This study contributes significantly
development of higher level social behaviors such as to the literature in that it demonstrates that joint attention
intersubjectivity (Mundy & Hogan, 1994) and theory of can be trained and that there might be some collateral
mind (e.g. Baron-Cohen, 1995). In addition to social changes in language.
development, joint attention has also been linked to the In our previous work, we used a multiple-baseline
development of language. Researchers hypothesize that design across five young participants with autism to
young children use joint attention skills to attend to lan- determine the efficacy of a joint attention intervention. This
guage cues in their environment (Bruner, 1974) and that study demonstrated that with developmental maturation
joint attention may act as a precursor to higher social- (2–10 weeks of baseline), children with autism did not
cognitive abilities essential in language development improve in their joint attention skills (Whalen & Schreib-
(Bates, Benigni, Bretherton, Camioni, & Volterra, 1979). man, 2003). Prior to starting treatment, developmental
In addition, protodeclarative joint attention (i.e. purpose of norms in typical children were assessed to establish train-
sharing rather than requesting) may be associated with the ing criteria. Following baseline, all five children were
development of receptive and expressive language (Mundy successfully taught to respond to joint attention initiations
& Gomes, 1998). Joint attention has also been linked to the from the experimenter and four of the children were taught
development of play, although not as strongly (e.g. Mundy to initiate protodeclarative pointing and coordinated joint
& Sigman, 1989). Specifically, affective information in attention. Targeted behaviors successfully generalized to
joint attention allows the child to develop the symbolic unstructured and structured assessments with the experi-
representation needed to acquire symbolic and pretend menters and with each child’s mother.
play. In addition to play, affective information from joint The primary purpose of the present study was to
attention may be important in the acquisition of object examine collateral changes in social initiations, positive
imitation (Meltzoff & Moore, 1994). affect, play, imitation, and language following participa-
Although behavioral interventions for children with tion in a joint attention training program (Whalen &
autism have been effective for language, play, imitation, Schreibman, 2003), to determine whether teaching joint
and social behaviors (e.g. Koegel, Koegel, & Schreibman, attention skills leads to collateral changes in non-targeted
1991; Lovaas, 1987), few studies have attempted to teach social-communication skills.
joint attention behaviors. Pierce and Schreibman (1995)
reported increases in joint attention following a peer-ori-
ented behavioral intervention. In another study, increases Method
in joint attention were observed in children with autism
when their parents imitated them during play but joint Participants
attention initiations were not observed when the environ-
ment was not structured in this manner (Lewy & Dawson, A total of 10 preschool-aged children participated in this
1992). Most studies, however, have not reported significant investigation. Four children with autism participated in an
changes in joint attention following treatment (e.g. Rocha, early intervention program designed to target joint atten-
Sherer, Paredes, & Schreibman, 1999). It is important to tion deficits (Whalen & Schreibman, 2003). The average
note these studies did not directly target joint attention chronological age of the children with autism was
skills. In order to affect increases in spontaneous joint 4 years, 2 months with an average mental age equivalent of
attention, many researchers posit that joint attention 1 year, 5 months using the Bayley Scales of Infant
behaviors should be targeted directly (e.g. Mundy, 1995). Development, Second Edition (Bayley, 1993). The average
Given the association between joint attention skills and language-age equivalent for participants was 1 year,
later emerging social-communication behaviors, research- 5 months using the Bayley and the MacArthur Communi-
ers have suggested that targeting joint attention skills cative Development Inventory (CDI; Fenson et al., 1993).
may also lead to collateral changes in non-targeted All participants met criteria for autism using the Childhood

123
J Autism Dev Disord (2006) 36:655–664 657

Autism Rating Scale (CARS; Schopler, Reichler, DeVellis, sessions took place in rooms in the laboratory that included
& Daly, 1980) (average score: 31) and the Gilliam Autism a small table, two to three small chairs, toys, pictures on the
Rating Scale (GARS; Gilliam, 1995) (average score: 93) walls, and a one-way observation mirror with a viewing
and were required to have a diagnosis of autism or another room on the other side from which sessions were video-
autistic spectrum disorder provided by an outside physician taped for behavioral coding.
or psychologist using DSM-IV criteria (American Psychi-
atric Association, 1994). In addition, because joint atten- Procedure
tion behaviors do not develop in typical children until
about 12–14 months, the participants with autism had non- Each child was administered pre-treatment assessments
verbal mental ages above this level. All children with (See Assessments section and Table 2) and then participated
autism were recruited from a waiting list for participation in baseline for two to ten weeks according to the multiple
in the UCSD Autism Research Laboratory or from referrals baseline design. The joint attention treatment used natural-
from other research and clinical facilities in the San Diego istic behavior modification techniques which incorporated
area on a first come, first serve basis (See Table 1 for components from Discrete Trial Training (e.g., Maurice,
Participant Characteristics). A fifth child with autism par- Green, & Luce, 1996) and Pivotal Response Training (PRT,
ticipated in the original intervention study but was unable Koegel, O’Dell, & Koegel, 1987; Koegel, Schreibman,
to master the targeted joint attention initiations and did not Good, Cerniglia, Murphy & Koegel, 1989). Treatment
complete the intervention study (Whalen & Schreibman, consisted of two phases. In the first phase, Response
2003) and was thus not included in the current study. Training, the child was taught to respond appropriately to
joint attention bids of the experimenter, including placing
Design the child’s hand on an object, tapping an object, showing an
object, following a point, and following gaze. Behaviors
A single subject, multiple baseline design across participants were taught in order and mastery of each target behavior at
was implemented (Kazdin, 1973). This type of design has the 80% correct over four consecutive sessions was required
advantage of controlling for developmental maturation, before teaching the next behavior. This phase took roughly
exposure to the treatment setting, and allows for measure- 3 weeks for each child. In the second phase, Initiation
ment of several concurrent behaviors. In addition, this design Training, the child was taught to initiate joint attention bids
is advantageous for looking at individual differences in the to the experimenter, including coordinated gaze shifting and
effectiveness of an intervention, and focuses on practical protodeclarative pointing. Mastery of each behavior at 30%
significance versus statistical significance. As required for of opportunities for coordinated gaze shifting and 15% of
this design, baselines were staggered across participants and opportunities for protodeclarative pointing was required
ranged from 2 to 10 weeks in length with 10 weeks being before teaching the next behavior. Mastery criteria was
approximately equal to the length of the intervention. For based on behavioral observations of six typical children.
each participant, data were obtained during baseline, treat- This phase took roughly 2 weeks for each child. For a more
ment, post-treatment, and at 3-month follow-up. complete description of the intervention, see Whalen and
Schreibman (2003).
Setting Assessments were administered at post-treatment and
3 months later for follow-up. Language and play probes (see
This research was conducted in the UCSD Autism below) were administered throughout baseline and treat-
Research Laboratory. Baseline, treatment, and assessment ment to assess collateral changes as the child progressed

Table 1 Participant characteristics at intake

Child Chronologicala age Mental agea (Bayley) Language agea (MacArthur CDI) Autism severity Baseline length
(CARS)b (GARS)c

Typical average 2–4 (1–7 to 2–10) 2–5 (1–9 to 3–1) 2–8 (1–8 to 3–4) N/A N/A N/A
Carrie 4–0 1–7 1–4 31.5 90 2 weeks
David 4–3 1–4 1–4 31 90 4 weeks
Alex 4–1 1–4 1–4 32.5 105 6 weeks
Brandon 4–4 1–9 2–1 30 90 10 weeks
a
In years-months
b
Range of autism severity on the CARS: 15–29 = non-autistic, 30–36 = mildly moderately autistic, 37–60 = severely autistic
c
Range probability of autism on the GARS: 90–110 = average probability of autism

123
658 J Autism Dev Disord (2006) 36:655–664

Table 2 Performance on the unstructured Joint Attention Assessment, ECSC, Empathic Response Assessment, and Structured Play Assessment
at pre-treatment, post-treatment, and follow-up
Pre-treatment Post-treatment Follow-up

Social initiations Typical average 15 (0–53)

Unstructured Joint Attention Assessmenta Carrie 0 18 15
David 0 6 5
Alex 0 25 0
Brandon 0 4 4
Positive affect Typical average 22 (0–47)
Unstructured Joint Attention Assessmenta Carrie 0 20 5
David 0 20 10
Alex 0 20 10
Brandon 0 10 0
Social responding Typical average 3 (3–3)
Structured Joint Attention Assessmentb Carrie 3 3 3
David 1 3 3
Alex 2 3 3
Brandon 3 3 3
Social initiations Typical average 2.6 (2–3)
Structured Joint Attention Assessmentb Carrie 2 3 2
David 1 3 3
Alex 1 3 3
Brandon 2 3 2
Response Typical average 2.8 (2–3)
Empathic Response Assessmentc Carrie 3 2 1
David 0 2 3
Alex 1 3 3
Brandon 1 2 2
Reaction Typical average 2.3 (0–3)
Empathic Response Assessmentc Carrie 3 3 3
David 2 3 3
Alex 2 3 3
Brandon 2 3 3
Structured Play Assessmentd Typical average 85 (0–100)
Carrie 40 60 80
David 0 60 60
Alex 40 80 80
Brandon 60 80 80
a
Percent of intervals
b
Scores range from 1 (basic actions directed toward experimenter (e.g. eye gaze, protest, etc.) to 3 (reciprocal exchanges such as turn-taking,
following simple commands, etc.)
c
Scores range from 0 (No Reaction) to 3 (Appropriate Reaction)
d
Percent correct of functional and symbolic play with no prompting from experimenter

through the program. Social behaviors were scored during Collateral changes in behavior were measured at pre-
all sessions (including language and play probes). treatment, post-treatment, and at a 3-month follow-up on a
variety of assessments. In addition, language and play
Dependent measures probes were administered throughout baseline and treat-
ment to assess collateral changes as the child progressed
In our previous paper (Whalen & Schreibman, 2003), we through the program. Six typically developing children
presented data on changes in behaviors that were directly (mean age = 2 years, 4 months) were also measured on all
targeted during the joint attention intervention. These assessments and during a single language and play probe to
behaviors included joint attention responding and protode- establish developmental norms.
clarative initiations. In the current study, we present changes All assessments and language and play probes were
in collateral behaviors which were not directly targeted administered by the first author and trained undergraduate
during the joint attention intervention. These behaviors in- research assistants. Assessments and language and play
clude social initiations, positive affect, empathic response, probes were videotaped and later coded by trained under-
play, imitation, and language. Thus all dependent variables graduate research assistants. Interobserver reliability was
presented in this study are considered collateral behaviors. collected and reported for 33% of all sessions and

123
J Autism Dev Disord (2006) 36:655–664 659

assessments. Kappa statistic was utilized to assess reli- facial expressions and noises to convince the child that they
ability. All kappa statistics were between .85 and 1.0 on all had been injured. Data were recorded on whether or not the
assessments. The advantage of using this statistic is that child discontinued playing with the hammer toy, whether
chance agreement is removed. or not the child looked at the experimenter, and whether or
not the child appeared to be upset about the experimenter’s
Assessments ‘‘injury.’’ One point was given for each of these behaviors
exhibited by the child (range one to three points). In
(1) Unstructured Joint Attention Assessment (adapted addition, the child’s reaction to the ‘‘injury’’ was scored as
from Loveland & Landry, 1986): The purpose of this either concerned (three points), indifferent (two points), or
assessment was to measure the child’s ability to respond positive affect (one point). All scores were based on a
correctly to the protodeclarative joint attention bids of the single trial.
researchers and to measure the child’s unprompted joint (4) Structured Play Assessment (Charman et al., 1997):
attention behaviors (showing, pointing, supported, and Functional and symbolic play were measured in this
coordinated joint attention). Assessment time was assessment. Children were presented with toys that had a
approximately 30-min for each child and involved playing functional purpose (such as a doll and a spoon where the
with the child in a relatively unstructured setting (no table child was expected to feed the doll with the spoon) and toys
and chairs, no demands on the child, and free access to with a symbolic purpose (a doll and a block where the child
toys). Two researchers were present during the assessment. was expected to pretend the block was food and was
The first researcher played with the child and administered expected to feed it to the doll). The children were measured
joint attention probes (e.g. pointed to a picture on the wall) on their ability to engage in functional or symbolic play
while the second researcher recorded the child’s responses either on their own or with prompts from the experimenter.
(e.g. child looked at the picture or did not) and assisted the Percent correct with no prompting, with a verbal prompt,
first researcher. Social initiations of the child and positive and with modeling was scored for all participants. Scores
affect directed toward the experimenter were coded from are reported as an average across four trials (two func-
the videotapes of the assessment to measure collateral tional, two symbolic) for each child.
changes in social behaviors using 10-s interval scoring.
(2) Structured Joint Attention Assessment (adapted from Language and play PRT probes
the Early Social Communication Scales; Mundy, Sigman,
Ungerer, & Sherman, 1996; Siebert & Hogan, 1982): This During baseline and treatment, 10-min probe sessions tar-
measure is prevalent in the joint attention literature and is geting either play or language were administered (weekly
used to assess a variety of joint attention and social inter- during baseline and every other week during treatment) for
action behaviors in a structured, laboratory environment. each child with autism. The purpose of these probes was to
Children were seated at a table facing the experimenter on assess the collateral changes in play and language
the other side of the table in a room with pictures on the wall throughout the child’s participation in the study. PRT was
and toys on a bookshelf behind the experimenter. Each toy implemented during the play and language probe sessions.
was presented one at a time to the child to determine if and PRT is a research-based behavior modification procedure
how the child requested items and to assess social behaviors which enhances motivation by using naturalistic reinforce-
such as joint attention and turn-taking. In this study, social ment, child-choice of tasks, turn-taking, reinforcement of
responding and social initiations were measured to assess attempted responses, and interspersing of maintenance tasks
collateral changes in social behaviors. This assessment was (i.e. mastered tasks) with acquisition tasks (i.e. tasks which
scored using similar criteria to that established in the liter- have not been mastered) (Koegel et al., 1991).
ature, in which a child receives a 0 if he or she did not meet The probe sessions were kept short (i.e. 10 min) in order
the minimum criteria for a 1, 1 for basic actions directed to avoid treatment effects from language and play PRT.
toward experimenter (e.g. eye gaze, protest, etc.), 2 for Although both language and play PRT have been shown to
combining eye contact and gesture, using pointing, reaching, be effective (e.g. Koegel et al., 1987; Stahmer, 1999) in
etc., and 3 for reciprocal exchanges such as turn-taking, increasing language and play behaviors, no changes were
following simple commands, etc. (Mundy et al., 1996; expected with 10-min probes weekly and no child dem-
Siebert & Hogan, 1982). onstrated increases in these behaviors during baseline (2–
(3) Empathic Response (Charman et al., 1997): The 10 weeks). These probes assisted in differentiating devel-
purpose of this assessment was to measure the child’s opmental maturation and treatment effects (if any) in play
empathic response toward an adult who pretended to have and language (during baseline) from changes due to
been injured by a toy hammer. The experimenter was not the acquisition of joint attention behaviors (during and
allowed to use any verbal cues but was required to rely on following joint attention training). Language PRT sessions

123
660 J Autism Dev Disord (2006) 36:655–664

were coded for spontaneous speech and play PRT sessions also showed a positive change in his empathic response
coded for functional and symbolic play and imitation using score and in his reaction to the ‘‘injury.’’ Brandon dem-
10-s interval scoring. All sessions were also coded for onstrated the same pattern as David and Alex by
social initiations and positive affect directed toward the increasing his empathic response score and showing a
experimenter. more appropriate reaction to the ‘‘injury.’’ David, Alex,
and Brandon maintained appropriate empathic responses
at follow-up and all three showed concern at the experi-
menter’s ‘‘injury.’’ Carrie showed no deficit in her
Results
response compared to typical children and appeared
concerned about the experimenter’s ‘‘injury’’ at pre-
Collateral changes in social behaviors
treatment. Although still showing concern (3-point reac-
tion) toward the experimenter, Carrie’s response score
Social behaviors were assessed during the Unstructured
decreased at post-treatment and again at the 3-month
Joint Attention Assessment, Structured Joint Atten-
follow-up.
tion Assessment, and Empathic Response Assessment (see
Table 1), and the language and play PRT probes. On the
Collateral changes in play and imitation
Unstructured Joint Attention Assessment, none of the
participants with autism exhibited any instances of social
Changes in functional and symbolic play were assessed
initiations or positive affect directed toward the experi-
using the Structured Play Assessment and play PRT
menter at pre-treatment. However, all four participants
probes. On the Structured Play Assessment, improvement
showed increases in social initiations and positive affect at
was observed for all participants from pre- to post-treat-
post-treatment with Carrie (18%) and Alex (25%) showing
ment with Alex and Brandon demonstrating scores similar
frequency of social initiations at or above levels of average
to typical children at post-treatment. All four participants
of typical children (15%). Carrie (20%), David (20%), and
maintained their scores and Carrie increased to scores
Alex (20%) demonstrated positive affect at levels similar to
similar to typical children at follow-up.
typical children (average 22%). No dramatic changes in
On the play PRT probes, frequency of spontaneous
social initiations from post-treatment to follow-up were
imitation during PRT probes increased for all participants.
observed for Carrie, David, or Brandon but Alex had a
Baseline averages for imitation ranged from 0% (Carrie
significant decrease. All participants showed a decrease in
and David) to 10% (Brandon and Alex) with an overall
their positive affect at follow-up.
average of 3% for children with autism. At post-treatment,
On the Structured Joint Attention Assessment, all four
imitation increased to an average of 20% with all four
participants with autism received a higher social initiation
children performing at about the same rate. Follow-up
score at post-treatment (similar to average of typical chil-
imitation rates dropped slightly for all four children with an
dren) than pre-treatment. Only two children (David and
average rate of 18%. No changes in the rate of functional or
Alex) received higher social responding scores at post-
symbolic play were observed.
treatment due to ceiling effects for Carrie and Brandon at
pre-treatment. All four participants received the same score
as typical children in social responding at post-treatment. Collateral changes in language
Although Carrie and Brandon received higher scores at
follow-up than baseline in social initiations, they both Language changes were measured during language PRT
received a lower score at follow-up than at post-treatment. probe sessions. Significant changes in language were not
David and Alex showed no change from post-treatment to observed during baseline for any of the participants but all
follow-up in their social responding or social initiations four participants showed increases in spontaneous speech
during the Structured Joint Attention Assessment. by post-treatment and these skills were somewhat main-
On the Empathic Response Assessment, three of the tained at follow-up (See Fig. 1). Rates of spontaneous
children with autism exhibited an improvement on their speech during baseline ranged from near 0% (David) up to
empathic response and emotional reaction from pre- 65% for one session for Brandon. The average rate of
treatment to post-treatment David appeared indifferent (2- spontaneous speech for children with autism during base-
point reaction) to the experimenter at pre-treatment and line was 20%. All four children showed increases in their
showed a severe deficit in his empathic response com- spontaneous speech with a post-treatment average of 55%
pared to typical children. Following treatment, David ranging from 25% (David) to 80% (Brandon). Slight drops
showed improvement in his response and appeared in spontaneous speech were observed at follow-up for all
concerned (3-point reaction) about the ‘‘injury.’’ Alex four children (See Fig. 1).

123
J Autism Dev Disord (2006) 36:655–664 661

Discussion Disturbances in affect have been considered an impor-

tant factor in the study of autism since Kanner (1943) first
Previous research has suggested a correlational link described the disorder. Recent research has paid particular
between joint attention skills and other social-communi- attention to positive affect in social situations (e.g. Bieb-
cation skills in typically developing children and children erich & Morgan, 1998; Joseph & Tager-Flusberg, 1997).
with autism. This study offers additional support for the Smiling and positive affect toward another person may be
relationship between joint attention skills and the devel- interpreted as a type of social sharing (Kasari, Sigman,
opment of other social-communication behaviors in young Mundy, & Yirmiya, 1990) and is considered to be an
children with autism. In addition, it suggests that joint important factor for discriminating joint attention gestures
attention is a particularly important treatment target from requesting behaviors. In the present study, increases
because it can lead to changes in a variety of important in positive affect were observed in all participants and post-
skills without having to target them directly. treatment responses resembled averages of typical children.
These changes were unlikely due to developmental matu-
Collateral changes in social behaviors ration since no positive changes were observed for any of
the participants during baseline. In fact, all participants
Research suggests that joint attention is involved in the showed no positive affect (compared to 20% in typical
acquisition of more advanced social behaviors (Baron- children) prior to beginning the intervention even after 2–
Cohen, 1995). In this study, social initiations, positive 10 weeks in baseline. The collateral changes in positive
affect, and empathic response were measured for the affect suggest support for the theory posited by Kasari et al.
purpose of testing this possibility. Increases in social (1990) that joint attention may be strongly associated with
initiations were observed for all four participants other social behaviors, such as positive affect.
completing the intervention in both structured and Previous research has shown that children with autism
unstructured environments and differences were compa- demonstrate a profound deficit in empathic responses
rable to typical children. These data suggest that joint compared to typical children (Charman et al., 1997). In this
attention may be important for increasing social initia- study, positive changes were observed for all participants
tions. However, several studies have reported changes in in their empathic response to a staged ‘‘injury’’ by the
social initiations by targeting it directly (e.g. Kamps et al., experimenter during toy play. Similar to positive affect,
1992) although few studies have reported collateral empathic responses require the ability to attend to the
changes when targeting other behaviors. social cues of another person and to respond appropriately.

Fig. 1 Changes in spontaneous Average of typical children Baseline average of target children
speech, social initiations,
Post-treatment average of target children Follow-up average of target children
positive affect, and imitation
during baseline, post-treatment,
and 3-month follow-up. 100
Percent of intervals engaged in behavior

Average of six typical children

are shown for each behavior as a 90
reference for developmental
norms 80
70
60
50
40
30 0.5 0 3.0

20
10
0
Spontaneous Social initiations Positive affect Imitation
speech

123
662 J Autism Dev Disord (2006) 36:655–664

Teaching joint attention skills may enhance the compre- development, and autism. First, this study was one of the
hension of such social cues and thus may be important to first to begin to test the hypothesis that teaching joint
the development of social responses such as positive affect attention may lead to gains in other skills. This was done
and empathy. by assessing collateral changes in language, play, and
social behaviors following treatment. By using long base-
Collateral changes in play and imitation lines (2–10 weeks), this study was able to show that sig-
nificant changes were unlikely to be attributed to
There is considerable evidence that children with autism developmental maturation alone and that collateral changes
are severely impaired in their symbolic and pretend play observed were likely to be attributed to the joint attention
skills (e.g. Baron-Cohen, 1987) and many researchers intervention. Although not conclusive, this suggests sup-
believe that this deficit may be closely tied to joint atten- port for the hypothesis that joint attention is linked to the
tion deficits in children with autism (e.g. Mundy, Sigman, development of other behaviors.
& Kasari, 1990). In this study, positive changes were Although this study has some important implications in
observed in play during a structured assessment; however the treatment of young children with autism, there were
changes in spontaneous play skills were not observed limitations. First, although this study offers evidence that
during the play PRT probes. This finding may suggest that teaching joint attention skills can lead to collateral
joint attention and play are not directly linked, or that changes in social-communicative behaviors in children
different mechanisms underlie the other social-communi- with autism, it does not address what mechanism or
cation skills and play skills. For example, it is possible that mechanisms are responsible for these changes. For
changes in the reward value of the social context which example, it is possible that teaching joint attention skills
occurred by teaching joint attention skills increased social, leads to an increase in social motivation, which in turn
affective, and language behaviors, but changes in symbolic affects the development of other social-communication
play skills require advances in representational skills. skills. Or, perhaps joint attention training leads to
Another possibility may be that joint attention and sym- increased attention to social stimuli, making children
bolic play may be related, but that it takes longer for more responsive to their social partner. It is also possible
changes to emerge in spontaneous play skills than in other the joint attention training results in a specific set of
social-communication behaviors. The fact that improve- cognitive skills which also affects the development of
ments were seen on the Structured Play Assessment and in other behaviors. Additional research is needed to discern
imitation on the play PRT probes lends some support to which mechanisms are responsible for the development of
this theory. Perhaps given a longer time, these changes other social-communication behaviors.
would begin to be evident in the children’s spontaneous Second, decreases in many of the collateral behaviors
play. were observed from post-treatment to follow-up. Although
this supports the notion that developmental maturation may
not be sufficient for positive changes in language and
Collateral changes in language certain social behaviors in children with autism, it also
shows that just teaching joint attention may not be enough
Although the relationship between joint attention and play to maintain these positive changes. It will be critical for
is unclear, the significance of joint attention in the devel- future researchers to design interventions in which positive
opment of language is strongly established in the literature. changes are more likely to be maintained at follow-up (e.g.
In this study, substantial gains in spontaneous speech were parent training).
observed for all four participants following the training of Finally, due to the small sample sizes, the observed
joint attention initiations and frequency of spontaneous differences may not be representative of the population.
speech resembled that of typical children. These data Large group designs and longitudinal studies should be
support the hypothesis that joint attention may be devel- implemented which train joint attention and assess collat-
opmentally linked to language. Because baselines were eral changes over time. Future studies should also begin to
administered for 2–10 weeks with no significant changes look at why children with autism are so impaired in joint
during this time, language gains were unlikely to be attention skills. Is this deficit related to an underlying
attributed to developmental maturation alone. attention deficit or social understanding deficit? What are
the neurological correlates? These types of questions
Conclusions and future research would help to unravel the mystery of whether or not there
is a core deficit in autism and whether or not that core
The results from this research may contribute significantly deficit might be joint attention or perhaps, a more general
to the study of joint attention, language, play, social attention deficit.

123
J Autism Dev Disord (2006) 36:655–664 663

Acknowledgements This research was funded in part from support Koegel, R. L., O’Dell, M. C., & Koegel, L. K. (1987). A natural
by U.S. Public Health Grant No. MH39434 from the National Insti- language teaching paradigm for nonverbal autistic children.
tute of Mental Health. This paper was also supported by a center grant Journal of Autism and Developmental Disorders, 17, 187–200.
from the National Institute of Mental Health No. U54MH066399, Koegel, R. L., Schreibman, L., Good, A., Cerniglia, L., Murphy, C., &
which is part of the NIH STAART Centers Program. We are grateful Koegel, L. (1989). How to teach pivotal behaviors to children
to all of the families who participated in this research. We also are with autism: A training manual. Santa Barbara: University of
appreciative of the undergraduate research assistants who helped in California.
the implementation of this study. Lewy, A., & Dawson, G. (1992). Social stimulation and joint atten-
tion in young autisticchildren. Journal of Abnormal Child Psy-
chology, 20, 555–566.
Lovaas, O. I. (1987). Behavioral treatment and normal educational
References and intellectual functioning in young autistic children. Journal of
Consulting and Clinical Psychology, 55, 3–9.
Adamson, L., & McArthur, D. (1995). Joint attention, affect, and cul- Loveland, K., & Landry, S. (1986). Joint attention and language in
ture. In C. Moore, & P. Dunham (Eds.), Joint attention: Its origins autism and developmental language delay. Journal of Autism
and role in development (pp. 205–221). Hillsdale, NJ: Erlbaum. and Developmental Disorders, 16, 335–349.
American Psychiatric Association. (1994). Diagnostic and statistical Maurice, C., Green, G., & Luce, S. C. (Eds.). (1996). Behavioral
manual of mental disorders (4th ed.). Washington, DC: Author. intervention for young children with autism: A manual for par-
Baron-Cohen, S. (1987). Autism and symbolic play. British Journal ents and professionals. Austin, TX, US: PRO-ED, Inc.
of Developmental Psychology, 5, 139–148. Meltzoff, A. N., & Moore, M. (1994). Imitation, memory, and the
Baron-Cohen, S. (1995). Mindblindness: An essay on autism and representation of persons. Infant Behavior and Development, 17,
theory of mind. Cambridge, MA: MIT Press. 83–99.
Bates, E., Benigni, L., Bretherton, I., Camaioni, L., & Volterra, Mundy, P. (1995). Joint attention and social-emotional approach
V. (1979). The emergence of symbols: Cognition and commu- behavior in children with autism. Development and Psychopa-
nication in infancy. New York: Academic Press. thology, 7, 63–82.
Bayley, N. (1993). Bayley infant scales of development (2nd ed.). San Mundy, P., & Crowson, M. (1997). Joint attention and early social
Antonio, TX: Harcourt Brace & Co. communication: Implications for research on intervention with
Bieberich, A. A., & Morgan, S. B. (1998). Affective expression in autism. Journal of Autism and Developmental Disorders, 27,
children with autism or Down syndrome. Journal of Autism and 653–676.
Developmental Disorders, 4, 333–338. Mundy, P., & Gomes, A. (1998). Individual differences in joint
Bruner, J. (1974). The ontogenesis of speech acts. Journal of Child attention skill development in the second year. Infant Behavior
Language, 2, 1–19. and Development, 21, 469–482.
Charman, T., Swettenham, J., Baron-Cohen, S., Cox, A., Baird, G., & Mundy, P., & Hogan, A. (1994). Intersubjectivity, joint attention, and
Drew, A. (1997). Infants with autism: An investigation of autistic developmental pathology. In D. Cicchetti, & S. L. Toth
empathy, pretend play, joint attention, and imitation. Develop- (Eds.), Disorders and dysfunctions of the self. Rochester sym-
mental Psychology, 33, 781–789. posium on developmental psychopathology (Vol. 5, pp. 1–30).
Fenson, L., Dale, P. S., Reznick, J. S., Thal, D., Bates, E., Hartung, J. Rochester, NY: University of Rochester Press.
P., Pethick, S., & Reilly, J. S. (1993). MacArthur communicative Mundy, P., & Sigman, M. (1989). The theoretical implications of
development inventories. San Diego, CA: Singular Publishing joint attention deficits in autism. Development and Psychopa-
Group, Inc. thology, 7, 63–82.
Gilliam, J. E. (1995). Gilliam autism rating scale. Austin, TX: Pro-Ed. Mundy, P., Sigman, M., & Kasari, C. (1990). A longitudinal study of
Joseph, R. M., & Tager-Flusberg, H. (1997). An investigation of joint attention and language development in autistic children.
attention and affect in children with autism and Down syndrome. Journal of Autism and Developmental Disorders, 20, 115–128.
Journal of Autism and Developmental Disorders, 4, 385–396. Mundy, P., Sigman, M., Ungerer, J., & Sherman, T. (1996). Defining
Kamps, D. M., Leonard, B. R., Versnon, S., Dugan, E. P., Delquadri, the social deficits of autism: The contribution of nonverbal
J. C., Gershon, B., Wade, L., & Folk, L. (1992). Teaching social communication measures. Journal of Child Psychology and
skills to students with autism to increase peer interactions in an Psychiatry, 27, 657–669.
integrated first-grade classroom. Journal of Applied Behavior Pierce, K., & Schreibman, L. (1995). Increasing complex social
Analysis, 25(2), 281–288. behaviors in children with autism: Effects of peer-implemented
Kanner, L. (1943). Autistic disturbances of affective content. The pivotal response training. Journal of Applied Behavior Analysis,
Nervous Child, 2, 217–250. 28(3), 285–295.
Kasari, C., Freeman, S., & Paparella, T. (2001). Early intervention in Rocha, M., Sherer, M., Paredes, S., & Schreibman, L. (1999). The
autism: Joint attention and symbolic play. In L. M. Glidden progression of joint attention behaviors in children with autism.
(Ed.), International review of research on mental retardation. Poster Presentation: Chicago, IL: Association for Behavior
New York: Academic Press. Analysis.
Kasari, C., Sigman, M., Mundy, P., & Yirmiya, N. (1990). Affective Ruskin, E., Kasari, C., Mundy, P., & Sigman, M. (1994). Attention to
sharing in the context of joint attention interactions of normal, people and toys during social and object mastery in children with
autistic, and mentally retarded children. Journal of Autism and Down syndrome. American Journal on Mental Retardation,
Developmental Disorders, 20, 87–99. 99(1), 103–111.
Kazdin, A. E. (1973). Methodological and assessment considerations Schopler, E., Reichler, R. J., DeVellis, R. F., & Daly, K. (1980).
in evaluating reinforcement programs in applied settings. Jour- Toward object classification of childhood autism: Childhood
nal of Applied Behavior Analysis, 6, 517–531. Autism Rating Scale (CARS). Journal of Autism and Develop-
Koegel, R. L., Koegel, K. L., & Schreibman, L. (1991). Assessing and mental Disorders, 10, 91–103.
training parents in teaching pivotal behaviors. In R. J. Prinz Seibert, J. M., & Hogan, A. E. (1982). Procedures manual for Early
(Ed.), Advances in behavioral assessment of children and fami- Social Communication Scales (ECSC). University of Miami, FL:
lies (pp. 65–82). Greenwich, CT: JAI Press. Mailman Center for Child Development.

123
664 J Autism Dev Disord (2006) 36:655–664

Sigman, M. (1998). Change and continuity in the development of Striano, T., & Rochat, P. (1999). Developmental link between dyadic
children with autism. Journal of Child Psychology and Psychi- and triadic social competence in infancy. British Journal of
atry & Allied Disciplines, 39, 817–827. Developmental Psychology, 17(4), 551–562.
Sigman, M., & Capps, L. (1997). Children with autism: A develop- Whalen, C., & Schreibman, L. (2003). Joint attention training for
mental perspective. Cambridge, MA; London: Harvard Univer- children with autism using behavior modification procedures.
sity Press. Journal of Child Psychology and Psychiatry and Allied Disci-
Stahmer, A. C. (1999). Using pivotal response training to facilitate plines, 44(3), 456–468.
appropriate play in children with autism spectrum disorders.
Child Language Teaching and Therapy, 15, 29–40.

123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.