You are on page 1of 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/259449716

Emulation and Mimicry in School Students with Typical Development and


with High Functioning Autism

Article in Journal of Autism and Developmental Disorders · July 2014


DOI: 10.1007/s10803-013-2027-0 · Source: PubMed

CITATIONS READS

10 236

3 authors, including:

Luis Jiménez Castor Méndez Paz


University of Santiago de Compostela University of Santiago de Compostela
60 PUBLICATIONS 2,125 CITATIONS 57 PUBLICATIONS 641 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Imitation behaviour in ASD View project

Conflict-triggered learning View project

All content following this page was uploaded by Luis Jiménez on 11 June 2014.

The user has requested enhancement of the downloaded file.


J Autism Dev Disord
DOI 10.1007/s10803-013-2027-0 Pre-print author version of the paper published at http://link.springer.com/article/10.1007/s10803-013-2027-0
ORIGINAL PAPER

Emulation and Mimicry in School Students with Typical


Development and with High Functioning Autism

Luis Jiménez • María José Lorda • Cástor Méndez

Abstract Two samples of participants with typical often ‘‘overimitate’’, that is, they tend to copy even those
development (TD) and high functioning autism performed action patterns which are obviously unrelated to the target
an imitation task where the goal was of high or low salience, outcome (Horner and Whiten 2005; Nielsen 2006).
and where the modeled action complied with or was Automatic imitation and overimitation paradigms have both
contrary to the end-state comfort (ESC) effect. Imitation was proved to be useful to analyze non-explicit, or spontaneous,
affected by the ESC effect in both groups, and participants imitation in autism. However, these two paradigms are
with autism reproduced high salient goals as frequently as affected by somewhat complementary problems. Automatic
did participants with TD, but they reproduced less of the low imitation, on the one hand, has been often confounded with
salient goals. Participants with autism showed a reduced the effects of spatial compatibility, because imitatively
tendency to reproduce those actions which were relatively congruent demonstrations are usually presented at the same
inefficient to reach the goals. The results are discussed in relative location at which the target action should be
terms of either a relative imbalance between emulation and emitted, and vice versa (e.g., Spengler et al. 2010, see Heyes
mimicry in autism, or a reduced tendency to overimitate. 2011, for a systematic description of the issue; Cook and
Keywords Imitation Mimicry Emulation Automatic Bird 2012; Jiménez et al. submitted, for attempts to solve it).
imitation Overimitation On the other hand, studies of overimitation do often focus
on actions which are so obviously irrelevant to the
Introduction demonstrator’s goals that their reproduction seems to
depend more on an explicit decision to join the models in an
Imitation is affected in Autism Spectrum Conditions (ASC). activity acknowledged to be ‘‘irrational’’ (e.g., Nielsen et al.
Despite a wide consensus concerning this general fact, the 2013), rather than on an arguably automatic, or spontaneous,
question arises whether imitation difficulties could be taken facilitation of those actions. Indeed, it is interesting to note
as a core symptom of ASC (Rogers and Pennington 1991; that in one of the few studies of overimitation in which the
Rogers and Williams 2006), or whether they should be emphasis was put on performing the task as fast and
better construed as a consequence of other primary deficits, efficiently as possible, children with TD still copied the
such as the atypical distribution of attention over social and unnecessary actions of the models to a large extent, whereas
nonsocial stimuli (Vivanti et al. 2008), or a specific participants with ASC showed a reduced effect of
difficulty to understand the intentions of others (Cattaneo et overimitation (Marsh et al. 2013). In this context, the aim of
al. 2007). the present study is to further distinguish between goal-
Imitation, at any rate, is not a unitary process, and driven and action-driven imitation effects in conditions
imitation processes are probably not affected unitarily in resembling those induced by the overimitation paradigm but
people with ASC. For example, it has been shown that in which, instead of adding an action which is clearly
individuals with autism differ more from people with typical unrelated with the overall action goal, the model illustrates
development (TD) in the imitation of actions without a clear either an optimal or a suboptimal way to achieve the same
goal, such as gestures or arbitrary movements, than in the goal.
imitation of actions performed upon objects, and with
transparent goals (Wild et al. 2012; Williams et al. 2004). It Emulation and Mimicry
also has been argued that individuals with ASC show lesser
propensity to imitate in conditions of automatic imitation, In addition to the social motivation to copy others’ actions
when imitation is not part of the explicit requirements of the for the sake of sharing their experiences (Nielsen et al.
task (Giganti and Esposito Ziello 2009; Helt et al. 2010; 2013), imitation can be driven by many other processes,
McIntosh et al. 2006; Senju et al. 2007), but that they could including the activation of the observed action goals, and the
be able to imitate at normal rates when they are explicitly automatic facilitation of the motor representations involved
asked to do so (McIntosh et al. 2006), or even when they are in reproducing those actions (Senju 2013). The results of the
simply required to focus their attention on the models studies which have been briefly reviewed above are
(Magne´e et al. 2007; Press et al. 2010; Senju 2013). generally consistent with the existence of a relative
However, even under explicit imitation conditions, it has imbalance between these two latter processes in ASC, and
been reported that participants with ASC tend to imitate they fit nicely with a dual model of imitation proposed by
differently, reproducing the goal rather than the style of the Hamilton (2008, see also Hamilton and Grafton 2007). In
demonstrated actions (Hobson and Lee 1999, see also her EP–M model, the author distinguished between an
Hobson and Meyer 2006), and imitating the instrumental emulation-and-planning (EP) route, which sustains
parts, but not the less functional features of those actions goaloriented and explicit imitation, and a mimicry (M)
(Rogers et al. 2010). In contrast, participants with TD do route, which directly connects visual representations of an

1
J Autism Dev Disord Final version of the paper published on line at http://link.springer.com/article/10.1007/s10803-013-2027-0

observed action with their corresponding motor plans. onstrator who acted in accordance with the ESC effect. The
According to the use made in this model of the terms results indicated that both groups of children were equally
‘‘emulation’’ and ‘‘mimicry’’, we will refer to ‘‘mimicry’’ able to reproduce the goals under both instruction or
as the rote replication of an action as a consequence of demonstration conditions, and that both groups took
having watched it (cf. Tomasello et al. 1993). In contrast, we advantage of the imitation conditions to adopt the most
will use ‘‘emulation’’ to refer to what Whiten and Ham functional grip, which allowed them to finish the action in a
(1992) called ‘‘goal emulation’’, a process of observational comfortable posture.
learning in virtue of which observers who see a model The results of Hamilton et al. (2007) suggested that the
yielding a particular goal will set about achieving the same benefits of imitation were preserved in children with ASC,
goal by their own means (Want and Harris 2002). but they did not make clear whether, in these particular
Results showing that people with ASC imitate less of the conditions, those benefits were caused by either emulation
non-functional features of an observed action (Hobson and or mimicry. Specifically, because in this experiment the
Lee 1999; Rogers et al. 2010), or that they often fail to show model always adopted the most functional grip to achieve
automatic contagion of spontaneous gestures (McIntosh et the goal, it was not possible to assess whether the observers
al. 2006; Senju et al. 2007), may be taken as indicating a reproduced the action just because watching it activated the
specific deficit in ASC involving the mimicry route. corresponding motor program, or because it made the
However, because many of these deficits have been observed goal more salient, and therefore activated the
observed in conditions in which the modeled actions action planning strategies more suitable to achieve that goal.
included facial expressions or emotional reactions, which Moreover, because in this study there were large differences
are known to be impaired in ASC, it is not at all clear in chronological age between TD and ASC groups (their
whether the observed differences should be attributed to a average age were, respectively, 4 and 8 years old), and given
deficit in automatic imitation, or to some other factors that action planning strategies have been reported to develop
pertaining to their specific attentional profile, which tends to shortly after the age of 4 (e.g., Weigelt and Schack 2010), it
prioritize non-social information, and to favor the processing was also possible that the effects observed in this study
of local features over the integrative processing required to could reflect different components of imitation in each
appreciate emotional expressions (Behrmann et al. 2006; group, being mediated by emulation and planning processes
Mundy 1995). Thus, to further analyze the differences in in the older participants with ASC, but relying on mimicry
imitation that may exist between people with TD and ASC for the younger participants with TD.
in relatively simple tasks, not involving face processing or
emotional expressions, we considered a manipulative task The Present Study
which may provide us with a useful tool to dissociate
emulation and mimicry. To sum up, the bar-task experiment of Hamilton et al.
(2007) suggests that imitation may be preserved in children
The Imitative Bar Task and the End-State Comfort Effect with ASC, but this interpretation is problematic because of
the large differences in chronological age between TD and
The original bar task was devised by Rosenbaum et al. ASC groups, and because in their paradigm it was not
(1990) as a tool to demonstrate that, when participants are possible to distinguish between the effects of emulation and
instructed to perform a simple action requiring them to grasp mimicry. As a way to alleviate these concerns, we set up an
and relocate an object, people often plan their actions in imitation bar task similar to that arranged in Hamilton et al.
advance, so that their decision on how to grasp the object (2007), but in which, over different trials, the model
depends crucially on their intended goal. Specifically, the illustrated either the most functional way to achieve the
authors described the end-state comfort (ESC) effect, as a goal, or a less functional one. We did not instruct
general tendency to tolerate adopting a relatively participants on whether the task goal was to imitate the
uncomfortable posture over the first stage of an action, if movements performed by the model, or just to reproduce the
this warrants the agent to finish the action in a comfortable same material outcome (i.e., to put the same extreme of the
end state. In the context of the original bar task, participants bar on the target location). In these conditions, imitation of
were presented with a bicolor bar that stood horizontally on the functional actions could result from any combination of
a central stand, and they were asked to put one its two mimicry and emulation, but the imitation of suboptimal
extremes on a platform. In these conditions, if participants actions could only arise from processes of mimicry or
were told to put the right extreme on the platform, overimitation.
righthanded people did typically grasp the bar with an In addition, we also manipulated the salience of the goal
overhand grip, because in this way they could finish the by alternating the use of mono-color and bicolor bars. We
action with the hand in a comfortable, thumb-up posture. In surmised that the use of a bicolor bar may lead participants
contrast, if their goal was to put the left extreme on the to focus on the material outcome, and therefore on the goal
platform, they would tend to adopt an underhand grip, of putting the same specific extreme of the bar on the
because that initial grasp would also allow agents to finish platform, regardless of the observed hand movements. In
their action in a comfortable thumb-up posture. contrast, the mono-color bar trials would make more
Hamilton et al. (2007; Experiment 3; see also Wo- difficult for the observers to identify which extreme of the
hlschlager et al. 2003 for a related precedent) presented the bar had been actually located on the platform, and therefore
bar task to two groups of children with TD and ASC, and it might arguably lead participants to pay more attention to
compared two conditions in which the actions were either the hand-movements, in an attempt to use them as a proxy to
produced in response to verbal instructions or to a dem- reproduce the same material outcome. Thus, if participants
2
J Autism Dev Disord Final version of the paper published on line at http://link.springer.com/article/10.1007/s10803-013-2027-0

with ASC had a specific difficulty with mimicry, we could participants with TD (45 male), and 17 participants with
expect both groups to reproduce the outcome to a similar ASC (15 male). All participants in the ASC group, and none
extent when the goal was highly salient (i.e., with a bicolor in the group with TD, scored above the cut-off scores of 76
bar), but we would predict that participants with ASC may for the AQ-Child, or of 30 for the AQ-Adolescent tests. To
have more difficulties in those cases in which the two control for the scoring differences between these two
extremes of the bar were not explicitly marked, and thus versions of the AQ tests we normalized the scores according
when they need to rely on the movements of the model to to the means and SD published by the authors of the
ascertain which extreme was located on the target. questionnaires (Auyeung et al. 2008; Baron- Cohen et al.
Finally, because planning, in general, and the ESC effect 2006). Because the distribution of gender was markedly
in particular, have been shown to develop steadily with age uneven in the sample of participants with ASC, each
in healthy children (Thibaut and Toussaint 2010; Weigelt relevant comparison between TD and ASC groups was
and Schack 2010), an additional aim of the present study preceded by an analysis of the effect of gender within the
was to test the expression of the ESC effect in these TD sample, in order to discard that any observed difference
imitation conditions with a relatively large sample of between groups could be accounted for in terms of that
children with TD, with ages between 7 and 16 years old. To gender distribution.
avoid the large discrepancies between participants’ mental The full sample of participants with TD, covering a wide
and chronological ages reported by Hamilton et al. (2007), range of ages between 7 and 16, was entered into a
the ASC group included exclusively participants with a preliminary analysis aimed at exploring whether the ESC
diagnosis of high functioning autism or Asperger syndrome, effect arose in this imitative version of the bar task, that is,
which showed cognitive abilities comparable to those found whether participants imitated comfortable end states more
in participants with TD of a similar age. The comparison often than non-comfortable end states. To gain a better
between groups was conducted on a reduced subsample of understanding of the possible modulatory role played by age
those participants with TD, selected to match the on the expression of this effect, we classified participants
chronological age of the participants with ASC. with TD into four age groups, in terms of their education
cycle. Specifically, we made four intervals, corresponding to
Method Second Cycle of Elementary School (7–9 years old, N = 32),
Third Cycle of Elementary School (10–11 years old, N =
All participants in the experiment completed tests of 25), First Cycle of Secondary School (12–13 years old, N =
vocabulary (Peabody Picture Vocabulary Scale, Dunn et al. 27), and Second Cycle of Secondary School (14–16 years
1997) and fluid intelligence (Raven Progressive Matrices, old, N = 13). We entered Age-Group as a factor in the
SPM, Raven et al. 2003). Their parents were asked to read analysis, to assess whether there was any developmental
and sign an informed consent form, and they were requested tendency in this task, either in the form of an increase in the
to fill the AQ questionnaire for children or adolescents, as proportion of trials in which they responded by adopting the
appropriate for their age (Auyeung et al. 2008; Baron-Cohen most functional action patterns (ESC effect) or, conversely,
et al. 2006). In addition, participants were presented with a as an increase in the proportion of trials in which they
bar task similar to that used by Hamilton et al. (2007; reproduced exactly the observed action patterns, regardless
Experiment 3), and they also completed another automatic of their functional value.
imitation task (analogous to that used by Jiménez et al. Even though participants in both TD and ASC groups
2012), which will be reported elsewhere (Jiménez et al. had been selected to fit within the same age range, the ASC
submitted). The study was conducted in accordance with the sample turned out to be composed of older participants
ethical standards laid down in the 1964 Declaration of (average: 151 months, 95 % CI 137–164), than was the
Helsinki, and was approved by the USC local ethical sample with TD (average: 132 months, 95 % CI 127–137).
committee. To avoid this difference in chronological age to affect the
scores obtained in the Vocabulary and Fluid Intelligence
Participants tests, we relied on raw scores for these two tests. Moreover,
in order to improve comparability between groups, the main
Typically developing participants were recruited from two comparisons between ASC and TD groups were conducted
local primary and secondary schools. Participants with ASC on two age-matched samples obtained by selecting, for each
were contacted through regional parental associations. They participant with ASC, the participant with TD who was
all had received a previous diagnosis of high functioning closer in age. In those cases in which two or more children
autism or Asperger syndrome by external professional with TD were at the same minimum distance (in months)
services, and they were reported to have a level of from a given participant with ASC, all of them were
intelligence within the normal range. We administered the included. With this procedure, we ended up with two
full battery of tasks and tests to a sample of 110 participants comparable samples of 23 participants with TD (11 male)
with TD, and to 22 participants with ASC, with an age range and 17 participants with ASC. Table 1 shows that these two
between 7 and 16 years old. Eleven participants (10 with TD samples were strictly comparable in terms of age and
and 1 with ASC) were excluded from the final sample cognitive abilities, but that there was a highly significant
because they were left-handed. We also excluded the data difference between them in the AQ scores.
from seven participants (3 from the group with TD and 4
from the group with ASC) owing to malfunction of the
experimental apparatus, or to insufficient cooperation from
the participants. The final sample thus consisted of 97
3
J Autism Dev Disord Final version of the paper published on line at http://link.springer.com/article/10.1007/s10803-013-2027-0

Apparatus and Material The wooden bar was either a mono-color (white) or a
bicolor (black-and-white) dowel, 30 cm long 9 3 cm wide,
The Peabody Picture Vocabulary Test was administered which rested on two cradles of 20 cm high, separated from
individually. Raven SPM Test was administered in small each other by a distance of 15 cm. These measures were
class groups for the participants with TD, and individually to designed to allow participants to hold the bar comfortably,
participants with ASC. The Autism-Spectrum Quotient using either an underhand or an overhand grip. The pen
questionnaire was filled in by the parents. The Child version holder was a metal cylindrical container, 17 cm height 9 7
was used for participants below 12 years old, and the cm wide, located at about 20 cm to the left of the cradles.
Adolescent version was administered for older participants. This material was mounted on a rotary ring, with a computer
screen standing on the middle of the ring. The experimenter
remained behind the monitor (see Fig. 1). After each trial,
the experimenter rotated the ring, removed the bar from the
container, and prepared the display for the next trial, while
the computer monitor remained stationary. Participants’
hand movements were continuously video-taped, and their
performance was encoded offline.

Design and Procedure

Participants completed 32 imitation trials, corresponding to


four random repetitions of eight different types of modeling
trials, which combined two modeled end-states
(comfortable, or thumb-up, and uncomfortable, or thumb-
down), two goals (inserting the closer or the farther side of
In the imitation task, participants were instructed to watch a the bar in the container), and two values of goal-salience
series of actions performed by a model, and to ‘‘do as they (depending on whether the bar was mono-color or bicolor).
see’’, right after the demonstration finished. To minimize Thus, the overall design was a mixed 2 × 2 × 2 × 2 design
the impact of social factors, the demonstrations were shown with Group (ASC vs. TD) as a between-participants factor,
as 5-s video-clips which showed the torso, arms, and hands, and with Goal (2), Salience (2) and Modeled End-State (2)
but not the head, of an adult model, who was sitting in front as repeated measures. Participants were verbally reinforced
of a metal pen holder and of a wooden bar resting after every single trial regardless of their performance, and
horizontally on a pair of cradles. During the movie, the the experimenter just asked them to remain with their hands
model grasped the bar from the cradle and inserted it into resting on the table until the clip had finished, and to start
the metal pen holder. Analogous material was located in moving their hands only after the appearance of the message
front of the observers, in a mirror arrangement with respect ‘‘Your Turn.’’ A yellow column displayed on the right side
to the items shown in the movie. Participants were told that of the monitor was gradually filled in green to inform
they should do as they see in the movie, as soon as the video participants about the number of completed and remaining
ended and was replaced by a black screen displaying the trials.
sentence: ‘‘Your Turn.’’ Imitation instructions remained
deliberately vague as to exactly ‘‘what’’ should be imitated,
in order to leave up to participants the final decision on what
to imitate.

Fig. 1 Left panel Illustration of the experimental procedure,


showing a high-salience trial. Right panel Schematic
representation of the eight different types of trials (see the text
for details)

4
J Autism Dev Disord Final version of the paper published on line at http://link.springer.com/article/10.1007/s10803-013-2027-0

Even though participants were not explicitly informed as to mixed Analysis of Variance (ANOVA) conducted with
what to imitate, the task goal was operationally defined as Gender and Age Group as grouping factors, and with
that of inserting the same side of the bar as shown in the Modeled End State as a repeated measure, showed no effect
video clip (i.e., either the extreme located closer or farther to or interaction involving Gender. In contrast, there was a
the container). In those trials displaying a bicolor bar, each significant effect of Age Group, indicating that participants
color was uniformly assigned to a side (i.e., the black side imitated the demonstrated end-states more frequently with
was always located close to the container), and therefore the age (.84, .88, .97, .97), F(3, 89) = 10.21, p<.001, ηp2 = .26,
goal became more salient, as it could be redefined as that of that they imitated comfortable end states more often than
inserting the same color of the bar as shown in the video- non-comfortable end states (.95 vs. 87), F(1, 89) = 14.230,
clip. In contrast, the mono-color trials made more difficult p<.001, ηp2 = .14, and that this latter difference tended to
for participants to ascertain which side of the bar had been disappear with age, as attested by a significant interaction
inserted, and they arguably required participants to pay between both factors, F(3, 89) = 5.67, p = .001, ηp2 = .16.
attention to the specific movements made by the Thus, as observed in Fig. 2a, the end-state effect was clearly
demonstrator, as a way to distinguish which extreme was observed in the youngest participants with TD, but it tended
actually inserted. In that condition, therefore, being able to to disappear with age, as they showed an increased tendency
benefit from mimicry to automatically encode and reproduce to reproduce both comfortable and uncomfortable end states.
the observed movements could be especially helpful for To compare the ESC effect observed in participants with
participants to reproduce the goal. Thus, if participants with TD and ASC, we relied exclusively on the age-matched
ASC had a specific difficulty with mimicry, we could expect samples of participants with TD (N = 23) and ASC (N = 17).
the difference between groups to become larger for these An ANOVA with Group and Modeled End State showed
less salient goals. significant effects of Group F(1, 38) = 4.47, p<.05, ηp2 =
In addition to the manipulation of goal-salience, different .11, and of Modeled End State, F(1, 38) = 17.99, p<.001,
combinations of modeling grips and end-states were ηp2 = .32, as well as a significant interaction between them
arranged to (1) confirm the ESC effect, (2) assess the overall F(1, 38) = 5.50, p<.05, ηp2 = .13. In short, the results
level of goal imitation, and (3) distinguish between the showed that both groups imitated comfortable end states
effects of mimicry and emulation on those trials in which more often than noncomfortable end states (.95 vs. .81), and
participants did accurately reproduce the goal. ESC planning that they imitated the comfortable postures to a similar
was assessed as the difference between the proportions of extent (.95 in both cases), but that participants with TD
trials in which the observers reproduced either comfortable imitated non-comfortable end states more frequently that did
or uncomfortable end states. Overall goal imitation was participants with ASC (.89 vs. .72).
assessed by means of the proportion of trials in which the
observers were able to reproduce the goal. Finally, because Overall Goal Imitation in Participants with TD and ASC
reproducing an action consistent with the ESC effect could
be driven by both imitation and action planning, the To assess the overall level of imitation of the goals in TD
emulation and mimicry components were specifically and ASC groups, we also relied on the comparison between
assessed in those trials modeling uncomfortable end-states. the age-matched samples of participants with TD and ASC,
In those trials, emulation and mimicry were assumed to lead but we first conducted a preliminary analysis to confirm that
to different results: whereas mimicry should increase the the factor of Gender was not affecting the proportion of
probability of reaching the observed goals by reproducing trials in which participants with TD were reproducing the
exactly the demonstrated actions, even when they resulted in modeled goal. A mixed ANOVA conducted for this group,
uncomfortable end states, emulation strategies—in with Gender as a between participants factor, and with Goal
conjunction with the ESC effect—should be expected to Salience (mono-color vs. bicolor bar), and Modeled End
lead observers to disregard those inefficient demonstrations, State (comfortable vs. uncomfortable) as repeated measures,
and to focus on achieving the same goal by means of the showed no effect or interaction involving Gender. Next, as
most efficient action pattern. For instance, if participants for the comparison between ASC and TD groups, the
saw a model who inserted the closer side of the bar using an corresponding ANOVA conducted with Goal Salience and
overhand grip, and thus finishing the action in an awkward Modeled End State, but now with Group (ASC vs. TD) as a
thumb-down posture, emulators would be expected to between participants factor, showed a significant effect of
reproduce the same goal in a more efficient way, by
Goal Salience (.96 vs. .90), F(1, 38) = 18.25, p<.001, ηp2 =
adopting the opposite action pattern.
.32, but not a significant effect of Modeled End State, F(1,
38) = 3.40, p<.08, nor an effect of Group, p[.20. However,
Results
the two-way Group 9 Goal Salience interaction was exactly
on the limit of significance, F(1, 38) = 4.06, p = .05, ηp2 =
The ESC Effect
.10, thus suggesting that goal imitation was more similar
between groups in those trials displaying a bicolor bar (.96
To explore whether the ESC effect arose in this imitative
vs. .96), than in those showing a mono-color bar (.93 vs.
version of the bar task in children with TD, and whether this
86). Putting it differently, even though the effect of goal
effect was modulated by gender or age group, a preliminary
salience was significant in both groups, the effect size was
analysis was conducted by comparing the proportion of
trials in which participants imitated the observed end state larger for the ASC group, F(1, 16) = 14.07, p<.005; ηp2 =
depending on whether the model illustrated a comfortable .47, than for the TD group, F(1, 22) = 4.22, p = .052; ηp2 =
(thumb-up) or an uncomfortable (thumbdown) posture. The .16 (see Fig. 2b).

5
J Autism Dev Disord Final version of the paper published on line at http://link.springer.com/article/10.1007/s10803-013-2027-0

Emulation and Mimicry on Trials not Complying with the


ESC Effect

Finally, to distinguish the effects of emulation and mimicry,


we focused specifically on those trials in which the model
illustrated an action ending in a non-comfortable state, and
analyzed the way in which participants with TD and ASC
achieved those goals. Thus, from all those trials in which the
observers fulfilled the action goal in conditions in which the
model illustrated a suboptimal action, a mimicry score was
computed as the proportion of trials in which the goal was
achieved by reproducing exactly the demonstrated pattern,
instead of recurring to the opposite, but more efficient,
pattern.
Again, the preliminary ANOVA conducted on these
scores for participants with TD, using Gender as a between-
participants factor, and Goal-Salience as a repeated measure,
showed no significant effect or interaction involving
Gender. Next, the ANOVA conducted on the mimicry
scores with Group as a between-participants factor, and with
Goal Salience as a repeated measure, showed only a
significant main effect of Group, F(1, 38) = 5.94, p<.02, η2
= .14, indicating that, for those trials, typical school students
showed a larger tendency to reach the goal by replicating
those patterns than did participants with ASC (.91 vs. .73).
The tendency to reproduce those patterns was numerically
larger when the goal was of low salience than when it was of
high salience (.84 vs. .80), but this difference was not
significant, and neither was significant the interaction
between Goal Salience and Group (F\1). In sum, the results
indicated that, when the model illustrated a suboptimal way
to achieve the goal, participants with TD tended to
systematically reproduce it, whereas participants with ASC
showed a lesser propensity to replicate those relatively
inefficient actions (see Fig. 2c).

Discussion

This study derives from Hamilton’s (2008) proposal on the


existence of two routes for imitation, and from her
suggestion that a deficit in the mimicry route could account
for the imitation pattern that is typically found in children
with autism, who often ‘‘succeed on tasks requiring
emulation of the goal of an action, but show abnormal
performance on those tasks requiring automatic mimicry’’
(p. 109). Taking advantage of the ESC effect (Rosenbaum et
al. 1990), which allowed us to model different actions
fulfilling the same goal with different degrees of efficiency,
we arranged a variant of the imitative bar task reported in
Hamilton et al. (2007, Experiment 3), in which we
manipulated the relative efficiency of the modeled actions
and the salience of the demonstrated goal. Because recent
experiments have shown that the ESC effect does gradually
develop with age (Weigelt and Schack 2010), we analyzed
this effect on a relatively large sample of participants with
typical development, and compared the performance of a
group of participants with high-functioning autism or
Asperger Syndrome with that of a closely age-matched
sample of participants with TD, who showed comparable
levels of cognitive abilities.
The results indicated that both participants with TD and
ASC showed the typical effects of action planning in this
6
J Autism Dev Disord Final version of the paper published on line at http://link.springer.com/article/10.1007/s10803-013-2027-0

imitative version of the bar task, but that older participants that their use of a bicolor bar, together with their modeling
with TD tended to sacrifice efficiency in order to reproduce of actions which invariably led to comfortable end states,
exactly the demonstrated pattern. Results also indicated that may have made their task particularly sensitive to goal-
both groups imitated the observed goals to the same extent driven strategies, hence minimizing the differences in
when the goals were highly salient (i.e., when the task performance between both groups. In contrast, our version
required only to insert the same color of the bar as illustrated of the task modeled different combinations of grips and end-
in the video-clip). However, participants with ASC appeared states, thus allowing us to distinguish between goal-oriented
to have more difficulties than those with TD to reproduce and action-oriented imitation. Moreover, by including
the goal in those trials displaying a monocolor bar, arguably conditions in which the two extremes of the bar were not
because in these conditions participants need to use the marked with a different color, we made the encoding of the
actual movements performed by the model to ascertain specific movements performed by the model particularly
which extreme of the bar was actually inserted in the pen useful for the observers to recognize which extreme of the
holder. Finally, when emulation and mimicry were bar had been inserted in the container, and therefore more
specifically dissociated, in those trials in which the modeled helpful to reproduce low-salience goals. From this
goal could be achieved more easily by adopting an action conceptual standpoint, we hypothesized that, if participants
pattern opposite to that illustrated by the model, the results with ASC showed specific difficulties with mimicry, we
indicated that participants with TD showed a larger tendency should obtain comparable performance between groups in
to sacrifice efficiency and to reproduce exactly the those conditions in which the goals were marked by salient
illustrated action than did participants with ASC. features of the objects (i.e., bar colors), but not so much in
those cases in which the encoding the modeled movements
Action Planning in TD and ASC could be instrumental to identify those goals. According to
the hypothesis, our results indicated that participants’
The results provide evidence on the normal development of performance was affected by the salience of the goal in both
action planning in participants with TD, as well as on the groups, but that the size of the effect was larger for
planning abilities observed in a group of school children participants with ASC.
with ASC. As for the normal development of action
planning, the results indicate that TD children as young as Interactions Between Planning, Mimicry, and Emulation
7–9 years old are already sensitive to the ESC effect, as Constraints
attested by the difference between their tendencies to
reproduce those actions ending either in comfortable or in In addition to showing that action planning effects were
uncomfortable states. This result is consistent with the claim already developed in the youngest participants with TD, our
that such advanced planning processes are already results indicated that these planning effects interacted with
developed by this age (Thibaut and Toussaint 2010). imitation constraints, so that older participants sacrificed
Moreover, the results also show that children with ASC their tendencies to end their actions in comfortable end
were similarly sensitive to the ESC effect as the age- states, and tended to reproduce all actions regardless of the
matched sample of participants with TD. This latter finding efficiency constraints. In the comparison between
is in agreement with the results reported by van Swieten et agematched groups with ASC and TD, the overall tendency
al. (2010), who tested children and adolescents with ASC to imitate comfortable end states did not differ between
and TD of comparable chronological ages, but it differs groups, but the mimicry scores indicated that, in those cases
from those obtained by Hughes (1996), who compared in which the observers reproduced a goal which had been
individuals with ASC and TD matched in terms of their modeled by means of a suboptimal action pattern,
mental age, but differing importantly in chronological age participants with TD tended to reproduce those relatively
(13 vs. 4 years old). The origin of this discrepancy is not inefficient patterns in almost every trial, whereas
completely clear, but it points to the convenience of using participants with ASC showed a lower tendency to adopt
individuals with high functioning autism as the preferred such inefficient actions, and therefore an increased tendency
targets to conduct this kind of comparative study. Even to yield the goal by adopting the opposite, but more
though analyzing imitation in other, non high-functioning, efficient, action.
individuals with autism will be obviously needed to obtain a In general, these results may be taken as consistent with
larger picture of the imitation profile observed over the Hamilton’s dual-route model of imitation, and in particular
whole spectrum of autism, dissociating the impact of with the claim that individuals with ASC are less prone to
imitation deficits from those caused by other cognitive mimic the observed action patterns, and show a larger
factors will surely be more easily attained by using samples tendency to imitate the goals rather than the exact
of participants with ASC and TD as comparable as possible movements performed by a model. However, before
to each other in terms of both chronological and mental endorsing that conclusion, one may need to discuss whether
ages. the tendency observed in participants with TD to reproduce
the suboptimal actions may obey either to mimicry—
Imitation and Goal Salience understood as a spontaneous tendency to reproduce any
observed action, regardless of its goals—or to an explicit
Our results suggest that the manipulation of goal salience decision to set their goal as that of reproducing all the
arranged in this experiment has been useful to bring about movements performed by the models, regardless of their
some differences between participants with TD and ASC material outcome.
which were not found in Hamilton et al. (2007). We surmise
7
J Autism Dev Disord Final version of the paper published on line at http://link.springer.com/article/10.1007/s10803-013-2027-0

From the present results, it is not possible to distinguish this result, Nielsen et al. (2013) reported comparable levels
empirically between these two alternatives, because our of overimitation between a group of typically developing
procedure deliberately aimed at leaving up to the children and a group of participants with high functioning
participants the final decision on what to imitate. Thus, it is autism. They attributed the discrepancy between their own
possible that the difference observed between both groups results and those previously reported in the literature to the
may stem from the fact that participants with autism were fact that they ‘‘assessed children with autism who were
more focused on the material outcome, whereas participants cognitively able.’’ However, our results could be taken as
with TD were more inclined to interpret that they should evidence opposite to this argument, since they show that a
reproduce the full observed movements. This question might deficit in overimitation could be observed by comparing two
have been better assessed by removing the ambiguity, and groups of children with ASC and TD matched in both
by asking participants explicitly to reproduce the same chronological and cognitive age.
material outcome—i.e., to insert the same side of the bar in We contend that the main difference between these two
the container as shown in the video-clip. These conditions studies relies on the subtlety of the manipulations or, more
would have made the demonstrated actions explicitly precisely, on the degree in which the goal to imitate the
irrelevant for the observers’ goal, and therefore they would details of the action was ostensibly conveyed by the
have made the task more strictly comparable to those used in pragmatics of the action. Indeed, in the study by Nielsen et
the standard automatic imitation paradigm (Heyes 2011). In al. (2013), the models did never demonstrate the most
these conditions, if participants were still affected by the functional way to do the task, and they repeated the same
demonstrated actions, one could safely conclude that non-functional action several times before passing the object
imitation effects were independent from the observers’ along to the observer. Moreover, these non-functional
goals. However, because in the present conditions we have actions consisted of repetitive and almost ritualistic patterns,
found a strong default tendency to act in agreement with the which made fairly explicit for the observers that their task
ESC effect, and because asking participants to focus on the was not just to obtain an outcome, but rather to specifically
orientation of the bar could surely have the effect of reproduce the whole action pattern. In contrast, our study
directing their attention away from the moving hand, we was designed to keep the goal more ambiguous, and to use
surmised that it would be unlikely that a non-attended much more plausible candidates as the nonoptimal action
demonstration could affect participants’ default tendencies. patterns, including components (e.g., grips) which were part
Thus, instead of adapting this paradigm to make it more of the optimal actions performed on some other trials. In this
comparable to those of automatic imitation, we decided to way, our procedure aimed at making room for the
test automatic imitation in parallel, presenting the same expression of imitative factors which were not necessarily
participants with another task analogous to that used by driven by the observers’ definition of the imitation goal, but
Jiménez et al. (2012). In this parallel study ( Jiménez et al. which might have been caused by more indirect factors,
submitted), we obtained converging evidence showing that such as the automatic impact of observing a particular
automatic imitation effects were clearly demonstrated in gesture on the observers’ action choice. As pointed out by
participants with TD, whereas the ASC group showed a Marsh et al. (2013), overimitation can be caused by
pattern of performance more consistent with an effect of processes such as the social motivation to conform to the
orthogonal spatial correspondence (Proctor and Cho 2003). perceived norms, but also by a number of other processes,
Taken together, the results of both studies could be more including object learning or priming from the observed
parsimoniously explained in terms of a single difficulty with action patterns. Marsh et al. reported that, when the
mimicry, rather than in terms of the conjunction of a instructions emphasized efficiency, participants with TD
difficulty with mimicry, plus a difference in the still showed a tendency to imitate the non-functional actions
interpretation of the imitation goals. performed by the models, whereas participants with ASC
were less affected by those actions. These results are fully
Overimitation consistent with the pattern obtained in the present study,
which showed that, when the observed action patterns were
The imitation paradigm tested in this study has been used to suboptimal, but still somewhat plausible, participants with
test mimicry, which is more often assessed through a variety TD showed a stronger tendency to replicate those patterns
of paradigms of automatic imitation. However, the present than did participants with ASC.
procedure has much in common with the studies of In sum, this study has shown that participants with ASC
overimitation, in which participants are explicitly asked to are similarly sensitive to action planning constraints as are
imitate a model, but in which the experimenters also include their typically developing counterparts, but that they may
certain actions which are obviously non functional, in an show specific difficulties to imitate action goals when they
attempt to assess whether the imitation profile includes only become less salient, and when the identification of these
the instrumental components of the action, or is extended goals depends on an encoding of the specific movements
also to its less functional parts (Hobson and Lee 1999; performed by the demonstrator. The results also showed that
Hobson and Meyer 2006; Nielsen et al. 2013; Rogers et al. typically developing school students tended to sacrifice
2010). In these tasks, Rogers et al. reported that children action planning constraints in favor of imitating both the
with autism tended to copy fewer of the nonfunctional functional and the less functional features of an observed
components of the observed actions, thus suggesting, action, but that this tendency to imitate unnecessary actions
consistently with the present results, that people with ASC was reduced in participants with ASC. The major limitations
are more focused on reproducing the goal, rather than on of this study, which will need to be addressed in further
replicating the full modeled actions. However, in contrast to research, have to do with the relatively low number of
8
J Autism Dev Disord Final version of the paper published on line at http://link.springer.com/article/10.1007/s10803-013-2027-0

participants with ASC, which did not allow us to explore the identification in autism. Journal of Child Psychology and
existence of a developmental trend in how they imitate in Psychiatry, 40(4), 649–659.
Hobson, J. A., & Meyer, J. (2006). Imitation, identification, and the
these conditions, as we did with the larger group of shaping of mind: Insights from autism. In S. J. Rogers & J. H. G.
participants with TD. In addition, we were not able to give a Williams (Eds.), Imitation and the social mind: Autism and typical
definitive response to the question of whether the observed development (pp. 198–224). New York: Guilford Press.
differences between these two groups depended either on the Horner, V., & Whiten, A. (2005). Causal knowledge and imitation/
impact of mimicry, or rather on the participants’ construal of emulation switching in chimpanzees (Pan troglodytes) and children
the task requirements. Just as it happens in most paradigms (Homo sapiens). Animal cognition, 8(3), 164–181.
Hughes, C. (1996). Planning problems in autism at the level of motor
of overimitation, the observers’ selection of suboptimal control. Journal of Autism and Developmental Disorders, 26, 99–
ways to do the task might be attributed to a host of different 107.
factors, including perceptual-motor priming—or automatic Jiménez, L., Recio, S., Méndez, A., Lorda, M. J., Permuy, B., &
imitation—but also effects related to the social interpretation Méndez, C. (2012). Automatic imitation and spatial compatibility
of the task requirements. The use of relatively plausible in a key-pressing task. Acta Psychologica, 141, 96–103.
Jiménez, L., Lorda, M. J., & Méndez, C. (submitted). Automatic
action patterns was intended to control for some of these
imitation in children and adolescents with typical development and
social factors, and therefore to make more room for the autism spectrum conditions.
expression of mimetic effects. However, more research will Magne´e, M. J., De Gelder, B., Van Engeland, H., & Kemner, C.
be needed to distinguish more clearly between these two (2007). Facial electromyographic responses to emotional
components. information from faces and voices in individuals with pervasive
developmental disorder. Journal of Child Psychology and
Acknowledgments This research was financially supported by the Psychiatry, 48(11), 1122–1130.
Spanish Ministerio de Ciencia e Innovacio´n and by the Xunta de Marsh, L., Pearson, A., Ropar, D., & Hamilton, A. (2013). Children
Galicia with research grants to Luis Jiménez(PSI2009-10823 and with autism do not overimitate. Current Biology, 23(7), R266–
INCITE09211132PR). The authors wish to thank children, parents and R268.
teachers from CEIP Rain˜a Fabiola and IES Rosalı´a de Castro, as well McIntosh, D. N., Reichmann-Decker, A., Winkielman, P., &
as from ASPERGA and BATA Associations, for their collaboration in Wilbarger, J. L. (2006). When the social mirror breaks: Deficits in
this research. We thank Beatriz Permuy, Sergio Recio, for their automatic, but not voluntary, mimicry of emotional facial
assistance in data collection. expressions in autism. Developmental science, 9(3), 295–302.
Mundy, P. (1995). Joint attention and social–emotional approach
behavior in children with autism. Development and
References Psychopathology, 7, 63–82.
Nielsen, M. (2006). Copying actions and copying outcomes: Social
Auyeung, B., Baron-Cohen, S., Wheelwright, S., & Allison, C. (2008). learning through the second year. Developmental Psychology,
The autism spectrum quotient: Children’s version (AQChild). 42(3), 555.
Journal of Autism and Developmental Disorders, 38(7), 1230– Nielsen, M., Slaughter, V., & Dissanayake, C. (2013). Object-directed
1240. imitation in children with high-functioning autism: Testing the
Baron-Cohen, S., Hoekstra, R. A., Knickmeyer, R., & Wheelwright, S. social motivation hypothesis. Autism Research, 6, 23–32.
(2006). The autism-spectrum quotient (AQ)—Adolescent version. Press, C. M., Richardson, D., & Bird, G. (2010). Intact imitation of
Journal of Autism and Developmental Disorders, 36, 343–350. emotional facial actions in autism spectrum conditions.
Behrmann, M., Thomas, C., & Humphreys, K. (2006). Seeing it Neuropsychologia, 48, 3291–3297.
differently: Visual processing in autism. Trends in Cognitive Proctor, R. W., & Cho, Y. S. (2003). Effects of response eccentricity
Sciences, 10(6), 258–264. and relative position on orthogonal stimulus–response
Cattaneo, L., Fabbri-Destro, M., Boria, S., Pieraccini, C., Monti, A., compatibility with joystick and keypress responses. The Quarterly
Cossu, G., et al. (2007). Impairment of actions chains in autism Journal of Experimental Psychology: Section A, 56(2), 309–327.
and its possible role in intention understanding. Proceedings of the Raven, J., Raven, J. C., & Court, J. H. (2003, updated 2004). Manual
National Academy of Sciences, 104(45), 17825–17830. for Raven’s Progressive Matrices and Vocabulary Scales. San
Cook, J. L., & Bird, G. (2012). Atypical social modulation of imitation Antonio, TX: Harcourt Assessment. Rogers, S. J. & Pennington, B.
in autism spectrum conditions. Journal of Autism and F. (1991). A theoretical approach to the deficits in infantile autism.
Developmental Disorders, 42(6), 1045–1051. Development and Psychopathology, 3, 137–162.
Dunn, L. M., Dunn, L. M., Whetton, C., & Burley, J. (1997). British Rogers, S. J., & Williams, J. H. G. (2006). Imitation in autism:
Picture Vocabulary Scale (2nd ed.). Windsor: Nfer-Nelson. Findings and controversies. In S. J. Rogers & J. H. G. Williams
Giganti, F., & Esposito Ziello, M. (2009). Contagious and spontaneous (Eds.), Imitation and the social mind: Autism and typical
yawning in autistic and typically developing children. Current development (pp. 277–310). New York: Guilford Press.
Psychology Letters, 25(1). Retrieved December 21, 2013 from Rogers, S. J., Young, G. S., Cook, I., Giolzetti, A., & Ozonoff, S.
http://cpl.revues.org/4810. (2010). Imitating actions on objects in early-onset and regressive
Hamilton, A. F. (2008). Emulation and mimicry for social interaction: autism: Effects and implications of task characteristics on
A theoretical approach to imitation in autism. The Quarterly performance. Development and Psychopathology, 22(1), 71–85.
Journal of Experimental Psychology, 61, 101–115. Rosenbaum, D. A., Marchak, F., Barnes, H. J., Vaughan, J., Slotta, J.,
Hamilton, A. F., Brindley, R. M., & Frith, U. (2007). Imitation and & Jorgensen, M. (1990). Constraints for action selection: Overhand
action understanding in autistic spectrum disorders: How valid is versus underhand grips. In M. Jeannerod (Ed.), Attention and
the hypothesis of a deficit in the mirror neuron system? performance XIII: Motor representation and control (pp. 321–342).
Neuropsychologia, 45, 1859–1868. Hillsdale, NJ: Lawrence Erlbaum Associates.
Hamilton, A. F., & Grafton, S. T. (2007). The motor hierarchy: From Senju, A. (2013). Atypical development of spontaneous social
kinematics to goals and intentions. In P. Haggard, Y. Rosetti, & M. cognition in autism spectrum disorders. Brain and Development,
Kawato (Eds.), Attention and performance (Vol. 22). Oxford: 35(2), 96–101.
Oxford University Press. Senju, A., Maeda, M., Kikuchi, Y., Hasegawa, T., Tojo, Y., & Osanai,
Helt, M. S., Eigsti, I. M., Snyder, P. J., & Fein, D. A. (2010). H. (2007). Absence of contagious yawning in children with autism
Contagious yawning in autistic and typical development. Child spectrum disorder. Biology Letters, 3, 706–708.
Development, 81(5), 1620–1631. Spengler, S., Bird, G., & Brass, M. (2010). Hyperimitation of actions is
Heyes, C. (2011). Automatic imitation. Psychological Bulletin, 137, related to reduced understanding of others’ minds in autism
463–484. Hobson, R. P., & Lee, A. (1999). Imitation and spectrum conditions. Biological Psychiatry, 68, 1148–1155.
9
J Autism Dev Disord Final version of the paper published on line at http://link.springer.com/article/10.1007/s10803-013-2027-0

Thibaut, J. P., & Toussaint, L. (2010). Developing motor plannings


over ages. Journal of Experimental Child Psychology, 105, 116–
129.
Tomasello, M., Kruger, A. C., & Ratner, H. H. (1993). Cultural
learning. Behavioural and Brain Sciences, 16, 495–552.
van Swieten, L. M., van Bergen, E., Williams, J. H. G., Wilson, A. D.,
Plumb, M., Kent, S. W., et al. (2010). A test of motor (not
executive) planning in developmental coordination disorder and
autism. Journal of Experimental Psychology: Human Perception
and Performance, 36, 493–499.
Vivanti, G., Nadig, A., Ozonoff, S., & Rogers, S. J. (2008). What do
children with autism attend to during imitation tasks? Journal of
Experimental Child Psychology, 101(3), 186–205.
Want, S. C., & Harris, P. L. (2002). How do children ape? Applying
concepts from the study of non-human primates to the
developmental study of ‘imitation’ in children. Developmental
Science, 5(1), 1–14.
Weigelt, M., & Schack, T. (2010). The development of end-state
comfort planning in preschool children. Experimental Psychology,
57, 476–482.
Whiten, A., & Ham, R. (1992). On the nature and evolution of
imitation in the animal kingdom: Reappraisal of a century of
research. Advances in the Study of Behavior, 21, 239–283.
Wild, K. S., Poliakoff, E., Jerrison, A., & Gowen, E. (2012).
Goaldirected and goal-less imitation in autism spectrum disorder.
Journal of Autism and Developmental Disorders, 42(8), 1739–
1749.
Williams, J. H., Whiten, A., & Singh, T. (2004). A systematic review
of action imitation in autistic spectrum disorder. Journal of Autism
and Developmental Disorders, 34(3), 285–299.
Wohlschlager, A., Gattis, M., & Bekkering, H. (2003). Action
generation and action perception in imitation: An instance of the
ideomotor principle. Philosophical Transactions of the Royal
Society of London B: Biological Sciences, 358, 501–515.

10

View publication stats

You might also like