Hindawi

Complexity
Volume 2018, Article ID 4547029, 15 pages
https://doi.org/10.1155/2018/4547029

Research Article
Categorical Cross-Recurrence Quantification Analysis
Applied to Communicative Interaction during Ainsworth’s
Strange Situation

Danitza Lira-Palma, Karolyn González-Rosales, Ramón D. Castillo ,

Rosario Spencer, and Andrés Fresno
Centro de Investigación en Ciencias Cognitivas, Facultad de Psicologı́a, Universidad de Talca, CP 3460000, Chile

Correspondence should be addressed to Ramón D. Castillo; [email protected]

Received 16 May 2018; Revised 21 August 2018; Accepted 26 September 2018; Published 1 November 2018

Guest Editor: Ralf Cox

Copyright © 2018 Danitza Lira-Palma et al. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.

The goal of this study was to characterize the degree of structuring of verbal and motor behaviours, unfolded during the application
of an procedure called the Strange Situation. This procedure is used for assessing children’s attachment quality during early stages
of their development. Many studies have demonstrated that communicative interactions share features with complex dynamic
systems. In such studies, estimations of degree of structure have been used to characterize the system’s synchronization. Thus,
assuming that processes of communicative interaction occur in the Strange Situation procedure, it was expected to find traces
of synchronization. The metrics were estimated through a Categorical Cross-Recurrence Quantification Analysis applied to the
behaviours of individuals and dyads. Two applications of the Strange Situation were implemented and recorded. Verbal and motor
interactions among children, caregivers, and strangers were transcribed, categorized, and organized as time series. From each time
series of original behaviours, randomized time series were created. Measures of recurrence extracted from Recurrent Plots, such
as determinism, entropy, maximum line, laminarity, and trapping time, were calculated. Original and randomized time series
were compared in terms of these measures. Results indicated that communicative interaction during the Strange Situation had
a structure that mimics properties observed in social interactions where synchronization emerges. In our case, verbal behaviours
were more prone to synchronization than motor behaviours, in both individuals and dyads, even though this pattern was more
salient among caregivers and strangers than children. The relevance of having measures that can capture synchronization during
the administration of the Strange Situation is discussed. Our preliminary findings allow us to point out that the application of
RQA and C-RQA to the Strange Situation could not only contribute to methodology, but also contribute to emphasizing the role
of coupling in communicative interaction generated by the application of this procedure to measure attachment patterns.

1. Introduction or more components at the same time or rate. In other words,

this process occurs when two or more systems recurrently
Human communicative interaction is a phenomenon that share a trajectory over a certain period [4].
behaves as a complex dynamic system [1–3]. The reason for Synchronization has shed light on the understanding
the complexity observed in communicative interaction is of the development, permanence, and fluctuation of a
because the number of components and relations between complex system. Research conducted under the dynamic
them increases to such an extent that a new phenomenon system approach proposes that human communicative
emerges, and this phenomenon cannot be explained by behaviour—at the intra- and interindividual levels—
the constitutive components [2]. Complex dynamic systems, produces synchronization and coupling [2]. Concordantly,
whether physical, chemical, biological, or social, share dis- findings indicate that some social-communicative interac-
tinctive properties, among which synchronization is relevant. tions have characteristics of a complex dynamic system,
In broad terms, synchronization refers to the activity of two where synchronization is a pivotal attribute [5].
2 Complexity

Evidence of synchronization of intraindividual behav- presence of synchronization in behaviour is an early marker

iours has been found in a series of perceptual [6–8], motor [9– of healthy development and that synchronization enables
12], and simple decision tasks [13–16]. Signs of synchroniza- predicting the adaptive behaviour of humans in the early
tion between individuals have been observed in cooperative stages of their development.
and noncooperative situations; and traces of coupling have The Theory of Attachment was proposed Bowlby [26] to
been observed in linguistic and nonlinguistic interactions. understand how these early communicative behaviours and
Taking these results into account, it has been proposed that the degree of social adaptive behaviour are configured by
synchronization can be modulated by contextual factors [17]. the caregiver-child interaction. This theory explains how the
For example, in a social context with negative valence, lower first relationships of children with their caregiver are formed,
levels of synchrony were detected among the participants. based on the concept of attachment, which is described as
Vink et al. [18] systematized scientific research and reported “(. . .) the process by means of which children establish and
greater levels of synchronization between dyads when self- maintain a special relationship with another individual who is
reports of rapport were described in terms of more intensity. considered better able to face the world” [27, p. 40]. Patterns
Hove and Risen [19] found levels of sympathy were positively of attachment are the result of instinctive responses for the
associated with levels of synchronization between partici- protection and survival of the child. This process is consid-
pants. They also observed that when a partner (confederate) ered as the bridge of early development and later development
came later to the experiment to perform a task there were low of social relations. Thus, the bond of attachment could be
levels of interpersonal synchrony between participant and considered as the relationship established between babies
confederate. and their caregivers, which influences their development and
Complementarily, other studies have identified that cou- subsequent well-being.
pling dynamics also vary depending on the conversational There is an increasing amount of research about attach-
context. When a person speaks to a listener who is located ment in the early stages of development, personality develop-
somewhere else, better levels of understanding were observed ment [28, 29], social adjustment [23, 24], as well as the devel-
with a delay of two seconds [20]. With this delay, it was also opment of psychopathology [30]. The collected background
found that gazes exhibit synchronization and coupling. On information indicates that the configuration of a particular
the other hand, when two persons speak face to face, in real attachment pattern is the result of an interaction between the
time, alternating the role of speaker and listener, the optimal natural dispositions of the child and the communicative and
coupling time is reduced to zero seconds [21]. bonding patterns that caregivers manifest in a critical period
As a result of communicative interaction, synchroniza- [21]. Disruptions in the period in which this early bond is
tion plays an important role for social development [22]. It configured can have negative effects, of great impact, on the
has been suggested that synchrony in mother-child interac- lives of individuals.
tions is not only significant for language acquisition but also Patterns of attachment are observed under an exper-
significant for the development of social relationships and imental procedure called the Strange Situation (SS) [31].
intersubjectivity. Along the same line, Stern [23] found that The SS, developed by Ainsworth and colleagues, is the gold
a lack of synchronization between parents and their children standard method to assess the quality of infant-caregiver
could affect the latter’s behaviour and affective states. attachment bond [32]. By proposing a mildly to moderately
A longitudinal study developed by Siller and Sigman [24], stressful experience for the infant, this laboratory assess-
with parents of children with autism, found that the commu- ment procedure activates the infant’s attachment behaviour
nication between caregivers and children was predicted by addressed to the attachment figure (the caregiver). The
the degree of synchronization in the interaction. Tunçgenç increase of the infant’s stress activates the infant’s attachment
and Fawcett [25] conducted two studies in which children behavioural system. These attachment behaviours reveal
who were 9 months old and 12 months old were located how the infant organizes his/her expectations regarding the
in social and nonsocial contexts. They found that 9-month- availability of the adult and how he/she can use the caregiver
old babies showed preference for objects in synchronous in order to return to calm. Once the infant is reassured,
movement, regardless of whether the object was in a social the attachment behavioural system is deactivated, and the
context or not. However, 12-month-old infants showed a exploration behavioural system is activated, evidencing the
preference only for stimuli in a social context that moved balance between attachment and exploration [33]. Thus, this
synchronously with respect to them. Thus, synchronization procedure aims to identify patterns of attachment between
of the movements was an important factor to guide the social the child and his or her primary caregiver in a laboratory
preferences of babies [25]. situation where the child’s stress is gradually increased by
In sum, human communicative behaviour is considered a the presence of a stranger and two brief separations from
complex dynamic system, where synchronization is a relevant the caregiver. The procedure is composed of eight episodes
attribute. Signs of synchronization have been detected in with a duration of three minutes each [30]. Episodes 5 and
various types of behaviour and cognitive processes, in both 8 are the moments in which the caregiver meets with the
individuals and groups. There is evidence that levels of child, after having been separated in the previous episodes.
synchronization can affect and be affected by the initial condi- The child’s reactions are scored and categorized according to
tions of the environment or by the affective-emotional state of four criteria: Proximity and contact seeking; contact main-
the individuals. Furthermore, based on the aforementioned taining; avoidance of proximity and contact; and resistance
empirical background, it is possible to conjecture that the to contact and comforting. Finally, according to Ainsworth
Complexity 3

[31] and Ainsworth et al. [32], children’s reactions under Researchers in the field of dynamic systems have devel-
the Strange Situation are classified into four patterns of oped a series of techniques and parameters to study syn-
attachment: (B) Secure: This pattern describes a child who chronization without abandoning its critical aspects, such as
uses the caregiver as a safe base for exploration and can nonstationarity, fluctuations, and transitions. One technique
manifest stress behaviour during separation. During the that has proven to be useful in the analysis of system synchro-
meeting, the child actively seeks caregiver contact through nization is the Recurrence Quantification Analysis (RQA)
behaviours such as smiles, vocalizations, gestures or physical [38]. RQA is a multidimensional nonlinear method used
approach. (A) Insecure Avoidant: This pattern describes a to discover attractors from tenuous correlations and subtle
child that shows exploration behaviours but displays few repetitive patterns in a time series where the data are noisy,
affective behaviours or rarely uses the caregiver as a safe base. irregular, and with many factors or dimensions affecting their
During the separation, the child shows slight or no sign of configuration [8, 39, 40]. RQA does not require additional
stress. At the meeting, the child tends to avoid contact with treatment or assumptions about data distribution or size,
the caregiver, such as avoiding gaze and physical contact and it can be applied to both linear and nonlinear variables
or focusing attention on toys rather than the caregiver. (C) [8, 39, 41]. Measures extracted by mean of RQA are estimated
Resistant Insecure: This pattern depicts a child that during from recurrence plots (RPs). As depicted in Figure 1, the RP
the separation seems extremely stressed. At the meeting, the is a graphical representation of a matrix of recurrence that
child usually alternates contact and seeking of the caregiver highlights aspects that cannot be detected in the original data
with signs of rejection, even tantrums towards the caregiver. set. In formal terms, the RP is an autocorrelation of x(t) with
The child can also be very passive or show behaviours that x(i) through the abscissa and x(j) through the ordinate. Only
denote anger. (D) Disorganized Insecure: In this pattern, points that satisfy the condition x(i) = x(j) are plotted [12,
the child expresses a series of contradictory or incomplete 41, 42]. From a RP, several quantitative and reliable measures
behaviours that would denote a lack of structure, such can be estimated, such as the percentage of recurrence that
as interrupted movements, stereotyped movements, freez- quantifies the proportion of recurrent points that fall within
ing/stilling, indicators of fear/apprehension, disorientation, the recurrent plot with a specified radius. The percentage
and confusion towards the caregiver [21]. of determinism quantifies the degree of randomness based
From a dynamic complex system perspective, these on the proportion of recurrent points that form a diagonal
attachment patterns should interact with other variables to line, called identity line [43]. Determinism allows know-
give rise to a particular type of interaction [2, 8]. Further- ing if future states of the system are determined by their
more, these patterns of attachment would also be an integral previous states. Periodic signals can produce long diagonal
part of the synchronization with other people. Generally, lines; chaotic signals can generate short diagonal lines, and,
children’s attachment would be an important ingredient for finally, stochastic signals cannot generate any diagonal line
synchronization of behaviours observed in social interactions at all. Entropy represents the uncertainty based on Shannon’s
with adults [11, 14, 34]. Thus, as the SS is an experimental entropy, which identifies the degree of disorder expressed by
protocol that promotes social-communicative interactions, a system. This measure is calculated from the lengths of all
and social interactions have shown attributes of complex diagonal lines that are organized in a histogram according
dynamic systems, it was hypothesized that traces of synchro- to their distribution. For simple periodic systems, in which
nization between dyadic interactions of caregivers, strangers all diagonal lines have equal length, the expected entropy is
and children could be found. equal to zero. The maximum line represents the length of the
Research in human communicative interactions has longest diagonal line on the RP when the diagonal line of
shown that synchronization is nonstationary; it experiences identity has been excluded. Hence, it is a measure of system
fluctuations and transitions [35, 36]. From a traditional stability. If the length is shorter, the signal is chaotic, and if
perspective, these aspects are usually controlled or avoided the length is larger, the signal is more stable [8].
because they add error to the results. However, from a Cross-Recurrence Quantification Analysis (C-RQA) is
dynamic system approach these aspects, rather than avoided, used with signals coming from two interacting systems [44–
must be incorporated due to their informative nature. Abrupt 50]. C-RQA, like RQA, quantifies coordinative patterns based
changes in postures, introduction and changes of topics, on an analysis of the sequence of behaviours performed
and breaks in the continuum of the conversation, among in real time [40, 41]. Figure 1 shows RPs with interesting
other factors, could be indicative of qualitative shifts in features, which can be quantified in various ways [38]. One
the mental states of individuals in response to a particular way is to focus on the diagonal line structures, because they
situation [35, 37]. These behaviours are not isolated but depict a sequence of iterations. When the focus is on vertical
rather chained, and they express a pattern that can be lines, two additional measures can be estimated that are
identified when are repeated over time. We hypothesized considered more informative in terms of the structure of two
that Ainsworth’s Strange Situation, even though it is a highly interacting signals: Percentage of laminarity that represents
standardized protocol, has communicative aspects that show the proportion of recurrence points that form vertical lines.
traces of relative synchronization between the actors. These The laminarity percentage is similar to that of determinism,
signs of synchronization should be identified whether the except that it depicts the proportion of recurrent points
behaviour is analysed over time, incorporating all aspects comprising vertical line structures rather than diagonal.
of the dynamics, such as fluctuations, transitions and Finally, there is another measure called trapping time, which
stationarity. represents the mean length of vertical lines.
4 Complexity

600 600

500 500

400 400
i i
300 300

200 200

100 100

100 200 300 400 500 600 100 200 300 400 500 600
j j
(a) (b)

600 600

500 500

400 400
i i
300 300

200 200

100 100

100 200 300 400 500 600 100 200 300 400 500 600
j j
(c) (d)

Figure 1: Recurrence plots (RP) of verbal expressions and grammar coming from an individual (Panels (a) and (b)) and dyads (Panels (c)
and (d)). Panel (a) shows a RP with a delay of 1 and an embedded dimension equal to 1, in which it is possible to observe a diagonal line
and coloured squares that show repetitions—in speech—of certain grammatical structures (e.g., verb, pronoun, adjective, article, adverb,
among others) while the individual was interacting communicatively. Panel (b) shows a RP in which the original verbal expressions were
randomized (reshuffled). Randomization destroyed the sequences and, therefore, the coherence with which the behaviours was appearing
while the interaction was taking place. Thus, the RP represented by Panel (b) shows a random pattern of behaviours, in which predictability
is very low, there is maximum entropy, and there is no system stability at all. Panel (c) shows a RP generated from a dyadic interaction in
which C-RQA was applied. This RP and its randomized version (Panel (d)) have a characteristic that differentiates them from the RPs in
panels (a) and (b). This is because the diagonal identity line is not present. Being two systems that interact, the diagonal line of identity tends
to disappear.

RQA and C-RQA have been applied to detect recurring by two persons during a communicative interaction. In fact,
features and patterns of complex dynamical systems, which there is a large corpus of evidence in which these nonlinear
result from one or two signals interacting over time [38, techniques have been used to analyse postural fluctuations
43, 48]. In psychology, such signals can be fluctuations in [42], conversational interaction between caregiver and child
gait, postural changes, eye movements, or informational [34, 51, 52], coupling of time series of verbalizations and
patterns, such as syntactic structures or words exchanged gestures [50], and reading comprehension [53]. Additionally,
Complexity 5

through these analyses, interpersonal coordination has been infant and the caregiver. The episodes lasted three minutes
characterized in terms of coupling of eye movements [42], each. In the first episode, the caregiver enters the room with
body movements [41, 54], child behaviour and sleep [55], the child. In the second episode, the caregiver takes a seat
patterns of aggression [29], minimal and effective forms of while the child can interact with the toys. In this phase, the
coordination during the dialogue [4], and early language caregiver can interact with the child if the child requests
development [46, 47]. it. In the third episode, the stranger enters the room, and
Even when RQA and C-RQA have been used in diverse takes a seat without interacting with the child or caregiver
contexts of communicative interaction, until now no research for a minute. The stranger then talks to the caregiver for a
has been aimed at studying the communicative interaction minute and then plays with the child for one minute. Episode
of standardized routines with children, such as the Strange four begins when the caregiver leaves the room, while the
Situation. The implementation of Ainsworth’s Strange Sit- stranger stays with the child in the room. If the child is
uation is organized in a very systematic and interesting distressed, the stranger can comfort the child. In the fifth
way, and new techniques can help researchers to examine episode, the caregiver returns to the room and the stranger
its temporal organization. Even though the main goal is leaves. The caregiver knocks on the door before entering
to define the kind of attachment pattern of the infant, it and says the name of the child. After waiting for a moment,
is equally important to know how verbal expressions and she is free to respond to the child. She must then make the
motor behaviours are unfolded, and whether the structure child interested in the toys and sit down again. In the sixth
observed in such variables can be used to estimate how traces episode, the child is left alone in the room. In episode seven,
of synchronization emerge among the individuals present the stranger enters the room. If the child is in distress, the
during the evaluation. stranger can comfort the child. In episode eight, the mother
Based on this background, our research aimed at enters and the stranger leaves the room. The caregiver can
analysing the unfolded verbal and motor behaviours among behave in the same way as in the previous meeting. If the
the participants of the Strange Situation (caregiver, child, and child is very stressed by the separations, these periods can
stranger), and to estimate, in reliable terms, the structure of have a shorter duration (30 seconds). The behaviours of the
these behaviours. The questions that we intend to answer child observed during the two meetings (episodes 5 and 8)
are: What are the structural indicators of these behaviours are coded in four scales (Proximity and contact seeking;
that allow establishing traces of synchronization between the contact maintaining; avoidance of proximity and contact; and
actors of the Strange Situation? What values do recurrence resistance to contact and comforting). Based on these scores
measures adopt in the behaviours of each individual and dyad and taking into account the behaviour of the child throughout
(caregiver-child, stranger-child, and caregiver-stranger)? the procedure, the child is classified in the category of secure
attachment (B), insecure avoidant (A), or resistant insecure
(C). If, during the episodes that provided contact with the
2. Materials and Methods mother, the child presented disorganized behaviours that
Before implementing this research, the protocol was reviewed disrupted the organization of their attachment relationship,
and approved by the University’s Institutional Review Board the classification applied is disorganized insecure attachment
(IRB # 1161533 and IRB # 1130773). (D).

(2) Analysis and Categorization of Verbal and Motor

2.1. Participants. Two female infants, 14- and-18-month old, Behaviours
with their respective biological mothers, were recruited for
this research. Both mothers were married and had completed (2.1) Words Were Labelled with a Number by Using a Text
undergraduate studies, and, according to their income, they Converter. Regardless of the language, this text converter
were classified as middle class. The “strangers” were two assigns a number that is characteristic and unique to each
clinical psychologists, 25 and 30 years old each. Both spe- word (http://cognaction.org/rick/ati/). The procedure was
cialists were trained and certified in the use of the protocol repeated twice, comparing the numbers assigned to the word
of the Strange Situation. Before carrying out the study, each sequences. The reliability in assigning numbers to each word
mother-caregiver read and signed the informed consent that was perfect, with a Kappa value equal to one.
was explained in detail by the researchers in charge of
the study. Clinical psychologists were randomly assigned as (2.2) The Text Converter Does Not Analyse Verbal Expressions,
evaluators when the procedure of the Strange Situation was Grammar in Spanish, or Any Other Type of Communicative
administered. The sessions were recorded in a Gesell room Behaviour. The coding of these variables was implemented
equipped with high-resolution cameras and microphones. manually. Verbal expressions and grammar were grouped in
43 descriptors. Body movements were grouped in 59 descrip-
2.2. Procedure tors. The labels and their descriptions are provided in Tables
1 and 2. Two trained researchers, who did not participate in
(1) Evaluation with Strange Situation Protocol. Following the the application of the Strange Situation procedure, analysed
protocol proposed by Ainsworth and collaborators [32], the the videos and categorized the behaviours of caregivers,
assessment took about twenty minutes, organized in eight strangers and children. These researchers were blind to the
episodes that combined the separation and the reunion of the children’s attachment pattern. Verbal expressions, babblings
6 Complexity

Table 1 Spectral Analysis [56–58], and also applied with complexity

Code Verbal expressions and Grammar
measures such as RQA and C-RQA [13, 41]. Even when there
are opinions contrary to this strategy and its assumptions
1 Silence
[59, 60], some researchers had shown that this strategy is
2 Inarticulate language (Crying)
convergent with other procedures that have proven to be
3 Inarticulate language (babbling) robust in demonstrating the dynamic and complex nature of
4 Inarticulate language (Scream) cognitive and behavioural processes [61].
5 Article Figure 2 depicts an example of this manipulation in which
6 Pronoun verbal expressions were categorized. For the original series
7 Noun (Panel (a)), 45 consecutive events were selected in which
8 Verb caregiver and stranger were interacting (between event 205
9 Adjective and event 251). The numbers represent certain types of words
10 Adverb (verbs, nouns, articles, pronouns, adverbs, among others).
11 Conjunction For the randomized series (Panel (b)), the words that were
12 Preposition located between events 205 and 251 were selected, Differences
between the original and randomized series can be visually
13 Interjection
detected. However, it is difficult to detect differences in series
14 Contraction
of 676 or 400 events. The same procedure was implemented
15 Adverbial phrase with child-caregiver and child-stranger dyads. The complete
16 Own name string of verbal expressions and grammar was made up of 676
21 Crying events for Video 1 and 400 events for Video 2. For individuals
22 Crying, complaint and dyads, original and randomized series were analysed by
23 Much stronger crying means of categorical RQA and C-RQA, respectively. From
31 Babbles sound “Aa” these analyses, measures of synchronization, such as deter-
32 Babbles sound “Ee” minism, maximum line, entropy, laminarity, and trapping
33 Babbles sound “Ii” time were estimated. Analyses were applied on original and
randomized series, in order to be compared. Assuming that
34 Babbles sound “Oo”
all communicative interactions have a dynamic structure in
41 Shouts “Aa”
natural conditions, randomization should annihilate such
42 Shouts “Ee” structure. If the original and randomized series have the same
43 Shout s“Ii” value of determinism, entropy, laminarity, maximum line,
and trapping time, it can be concluded that such a system has
no coupling or synchronization.
and periods of silence were transcribed and coded. A similar Following the guidance of Dale and colleagues [51] for the
transcription and coding was done for body movements and implementation of RQA and C-RQA with categorical data,
gestures, where each behaviour was registered according to recurrence measures were estimated with one embedded
the moment it appeared. The interrater reliability between dimension, and with delays of one lag. Unlike the analysis
these two researchers was estimated with a Kappa Coefficient with continuous variables, in this case radius and delay were
(Table 3) during three segments. The Kappa values were not estimated.
highly reliable, fluctuating between 0.71 and 1.0.
Each video was divided into a discrete number of events.
There were 676 events for Video 1 and 400 for Video 2. For 3. Results and Discussion
each participant in the Strange Situation (child, caregiver,
and stranger) verbal expressions and grammar, as well as 3.1. Categorical Recurrence Quantification Analysis for Indi-
body movements were maintained in the order in which viduals in Videos 1 and 2. The results for children, caregivers,
they appeared. For each participant, three files were extracted and strangers are summarized in Table 4. It is possible
with the original time series of the three types of variables. to observe that the randomized and original series had
For each variable, randomized time series were generated. the same level of recurrence. These results were expected,
Thus, for each child, caregiver, and stranger there were six considering that randomized series were generated from the
files containing the three original series and their respective same data than the original data. Thus, the events that are
randomized series. part of the recurrence were organized in different order,
From a dynamic perspective it has been proposed that but they are the same. Randomization breaks down the
original series of behaviour form regular patterns, which original structure, mainly affecting measures of determinism,
have certain properties different from the same randomized entropy, maximum line, and laminarity. When the original
series [6–10, 12–16]. A strategy to account for the dynamic structure of words, verbal expressions and grammar, and
character of the behaviour has been to compare the original body movement was shuffled, a decrease in determinism and
series with the same randomized series [53]. This strategy has laminarity (lower predictability) and an increase in entropy
been implemented with fractal techniques, such as Detrended (lower values of entropy in this case) were expected. These
Fluctuation Analysis, Standardized Dispersion Analysis, and results, as predicted, were clearly observed in the words,
Complexity 7

Table 2
Code Body Movements
1 Silence
2 Takes toy 1
3 Takes toy 2
4 Takes toy 3
5 Takes toy 4
6 Takes the paper
7 Gets up off the floor (or the chair)
8 Lightens up
9 Moves hands
10 Moves arms
11 Moves head towards the girl
12 Smiles
13 Moves head towards the stranger
14 Bends
15 Moves head affirmatively (yes)
16 Takes the girl’s hand
17 Takes the girl
18 Walks
19 Exits the room
20 Enters the room
21 Kisses the girl
22 Sits the girl on the floor
23 Touches the girl’s head
24 Hugs the girl
25 Takes toys 1 and 3
26 Plays with toys 1 and 3
27 Drops toy or paper
28 Points forward
29 Aims towards toy 3
30 Moves the girl
31 Sings
32 Touches the girl
33 Aims for the chair
34 Moves toy 1
35 Moves toy 2
36 Moves toy 3
37 Bites toy 1
38 Bites toy 3
39 Hits toy 1
40 Hits toy 3
41 Drops toy 1
42 Drops toy 2
43 Drops toy 3
44 Picks up toy 3
45 Moves torso forward
46 Moves torso towards the mother
47 Crawls towards the mother
48 Raises hands to the mother
49 Moves torso towards stranger
50 Hugs the mother
51 Takes the mother’s hand
8 Complexity

Table 2: Continued.
Code Body Movements
52 Touches toy 2
53 Moves leg
54 Crawls
55 Touches toy 4
56 Plays with toys 1 and 4
57 Plays with toy 4
58 Aims for toy 3
59 Takes toys 1 and 3

Table 3: Kappa values of interrater reliability.

Ca Ch St
Segment 1 Segment 2 Segment 3 Segment 1 Segment 2 Segment 3 Segment 1 Segment 2 Segment 3
Verbal Expressions and Grammar
Video 1 1∗ 1∗ .99∗ 1∗ 1∗ 1∗ .98∗ 1∗ 1∗
Video 2 1∗ 1∗ 1∗ 1∗ 1∗ .94∗ .98∗ 1∗ 1∗
Body Movements
Video 1 .91∗ 1∗ 1∗ .91∗ 1∗ 1∗ 1∗ 1∗ 1∗
Video 2 1∗ 1∗ .97∗ 1∗ 1∗ .87∗ .71∗ 1∗ 1∗
Note: Ch = child; Ca = caregiver; St = stranger; ∗ p < .0001.

Table 4: RQA measures for individuals in Videos 1 and 2.

Recurrence Determinism Entropy Maximum Line Laminarity

Variables Unit Orig Ran Orig Ran Orig Ran Orig Ran Orig Ran
Ch 75.51 75.51 92.13 92.2 3.49 3.12 130 60 0.98 0.99
String of Words Ca 51.41 51.41 95.64 77.43 3.96 1.9 130 6 0.92 0.72
St 19.51 19.51 88.64 33.72 4.02 0.71 153 11 0.97 0.82
Ch 75.22 75.22 92.61 94.44 3.39 3.19 69 43 0.98 0.99
Video 1 Verbal Expressions and Grammar Ca 53.78 53.78 96.53 80.12 3.73 2.24 126 10 0.82 0.63
St 24.65 24.65 80.6 38.23 3.25 0.88 155 13 0.93 0.84
Ch 53.41 53.41 80.2 78.32 2.22 1.9 10 10 0.86 0.86
Body Movements Ca 43.39 43.39 73.39 68.79 2.71 1.55 37 8 0.75 0.81
St 57.44 57.44 84.41 83.95 3.42 2.33 57 26 0.94 0.96
Ch 80.84 80.84 95.31 96.04 4.23 3.58 173 25 0.96 0.97
String of Words Ca 23.31 23.31 85.96 40.95 3.48 0.89 63 11 0.98 0.93
St 40.18 40.18 94.64 59.67 3.95 1.7 60 6 0.93 0.66
Ch 82.6 82.6 95.63 97.01 4.4 3.6 172 32 0.96 0.98
Video 2 Verbal Expressions and Grammar Ca 26.83 26.83 78.18 47.23 3.09 1.2 64 13 0.98 0.91
St 43.64 43.64 91.82 69.3 3.95 1.76 62 7 0.81 0.54
Ch 37.71 37.71 68.81 66.52 1.74 2.68 12 12 0.92 0.93
Body Movements Ca 32.67 32.67 58.71 56.79 2.64 1.34 33 15 0.85 0.88
St 61.42 61.42 87.52 86.58 3.37 2.44 38 14 0.92 0.94
Note: Ch = child; Ca = caregiver; St = stranger.

verbal expressions, and grammar of caregivers and strangers, the percentage of determinism in children tended to remain
while in children the levels of determinism, laminarity, stable, and entropy, especially in Video 1, tended to maintain
and entropy tended to remain constant. In relation to the its values, while in Video 2, a decrease was not as noticeable
body movements displayed during the Strange Situation, as in the case of adults.
adults showed slight increases in determinism levels and Based on determinism and laminarity, it is possible to
decreases in entropy and maximum line levels. In contrast, establish that communicative behaviours of caregivers and
Complexity 9

Table 5: Categorical CRQA measures for dyads in Videos 1 and 2.

Recurrence Determinism Entropy Maximum Line Laminarity Trapping Time

Variables Dyad Orig Ran Orig Ran Orig Ran Orig Ran Orig Ran Orig Ran
Ch-Ca 62.07 62.07 93.81 85.51 3.92 2.37 131 6 97.8 98.8 18.13 10.53
String of Words Ch-St 37.89 37.89 90.78 59.74 3.82 1.32 131 12 97.4 82.7 35.33 3.87
Ca-St 31.39 31.39 92.46 52 4.22 1.12 131 8 96.8 82.2 35.19 3.87
Ch-Ca 62.49 62.49 95.31 87.22 3.72 2.6 70 11 98 99.3 18.81 10.7
Video 1 Verbal Expressions and Grammar Ch-St 38.06 38.06 91.66 55.53 3.75 1.12 70 13 97.4 88 38.43 3.82
Ca-St 34.46 34.46 90.06 52.38 3.83 1.16 128 11 90.3 80.9 26.52 3.76
Ch-Ca 47.87 47.87 76.83 73.24 2.5 1.78 9 8 87.5 86.9 3.89 3.66
Body Movements Ch-St 54.89 54.89 83.46 81.07 2.68 2.12 9 10 93.4 95.3 12 5.6
Ca-St 49.45 49.45 78.85 76.29 3.07 1.87 39 10 93.2 95.1 12 5.6
Ch-Ca 42.95 42.95 90.96 67.75 3.98 1.57 64 12 96.2 97.3 16.6 8
String of Words Ch-St 56.79 56.79 95.12 77.8 4.31 2.33 61 6 94.5 67.6 23.57 2.62
Ca-St 30.43 30.43 90.42 49.83 3.85 1.2 61 7 94.8 67.8 23.57 2.62
Ch-Ca 42.74 42.74 91.16 67.24 3.96 1.85 65 14 96.5 98.5 15.9 8.3
Video 2 Verbal Expressions and Grammar Ch-St 58.32 58.32 95.28 83.17 4.47 2.31 63 7 95.4 62.5 24 2.44
Ca-St 33.5 33.5 85.63 56.7 3.5 1.45 64 5 90.4 59.5 19.56 2.44
Ch 34.52 34.52 64.77 61.26 2.2 1.44 13 13 91.7 92.6 5.13 4.24
Body Movements Ca 47.67 47.67 79.92 76.45 2.34 1.92 13 13 92.1 94.4 13.56 5.43
St 44.45 44.45 71.43 71.39 3 1.78 40 14 91.9 94 13.49 5.43
Note: Ch-Ca = child-caregiver dyad; Ch-St = child-stranger dyad; Ca-St = caregiver-stranger dyad.

strangers had higher levels of synchronization with them- involved showed diverse degrees of synchronization with
selves compared with communicative behaviours of children. themselves. Both adults, caregiver and stranger, expressed
That is to say, the initial structure of verbal and motor better indicators of such internal coupling than infants.
behaviours was a strong predictor of subsequent behaviours. Finally, the synchronization indicators appeared clearly in
On the other hand, the decrease in entropy and maximum all dyads, even when the caregiver-stranger dyad presented
line indicated that there was a structure or pattern in the way better indicators of coupling than the dyads where the
behaviours were organized in natural conditions. And this infants were involved. The communicative interaction is a
pattern was different from a pattern of random organization. multidimensional phenomenon, in which a series of variables
operating at different time scales are intertwined [44]. The
analysis techniques used by us reduced the multidimension-
3.2. Categorical Cross-Recurrence Quantification Analysis for ality to the two dimensions present in the recurrence plot
Dyads in Videos 1 and 2. The results for dyads in Videos 1 [39, 41, 48]. Verbal behaviours were expressed on a different
and 2 (see Table 5) partially replicated the values observed time scale than motor behaviours. These verbal behaviours
with individuals. For string of words, verbal expressions, and were a better example of the dynamic present in the Strange
grammar, the most notorious changes were observed at the Situation. The dynamic was observed more clearly in the
level of determinism, entropy, maximum line, and trapping verbal behaviour of caregivers and strangers, especially when
time, with the original time series having higher values than they were together in a dyad.
the randomized series. In contrast, for body movements,
changes were detected in entropy and, to a lesser extent,
in maximum line and laminarity. In the three dyads, the 3.3. Comparison of Means between Groups Segmented by
entropy levels were higher in the original series than in the Communicative Behaviours, Individuals, and Dyads. Based
randomized series, while only in the caregiver-stranger dyad on the recurrence measures obtained from both videos,
the maximum line of the original series was greater than we proceeded—in heuristic terms—to compare the original
that of the randomized series. In verbal expressions, the and randomized series with a Wilcoxon signed-rank test, a
dyads presented values that allow us to assume a degree of nonparametric test for related samples. The first comparison
synchronization. However, in terms of motor behaviour, the was segmenting by type of communicative behaviour (string
caregiver-stranger dyad was the only one that showed slight of words, grammar, and body movements). In this case,
signs of synchronization. the recurrence measures from individuals and dyads were
Our findings indicate at least three relevant aspects. grouped to estimate an average. The second comparison was
Verbal and motor behaviours revealed different degrees of among individuals (child, caregiver, and stranger). Finally,
synchronization [5, 11]. Words, verbal expressions, and gram- the third comparison was segmenting by dyads (child-
mar had more clear-cut indicators of synchronization and caregiver, child-stranger, and caregiver-stranger). In both
structure than body movements [43]. Likewise, individuals case, for individual and dyads, recurrence measures from
10 Complexity

Original Series Ca-St Dyad

16
15 15
14
13 13 13 13
12 12 12 12 12 12 12 12
11 11 11 11
Grammar Code 10 10 10 10 10 10 10 10 10 10 10 10 10
9 9
8 8 8 8 8 8 8 8
7 7 7 7 7
6 6 6
5 5 5 5 5
4

2
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47

Stranger
Caregiver

(a)
Randomized Series Ca-St Dyad
18

16 16
15 15
14

12 12 12 12 12 12
Grammar Code

10 10 10 10 10 10 10
9
8 8 8 8 8 8 8 8 8 8 8 8 8 8 8
7 7 7 7
6 6 6 6

2
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47

Stranger
Caregiver

(b)

Figure 2: Panels (a) and (b) show the sequence of words that have been classified according to their grammatical function. In this segment of
47 events, only 12 categories appear. 1 = silence, 5 = article, 6 = pronoun, 7 = noun, 8 = verb, 9 = adjective, 10 = adverb, 11 = conjunction, 12 =
preposition, 13 = interjection, 15 = adverbial phrase, and 16 = own name. Grammar observed in original series (Panel (a)) shows typical patterns
of communicative interaction between caregiver and stranger. This typical pattern is cancelled when original time series are randomized (Panel
(b)).

string of words, grammar and verbal expressions, and body (Z = -3.06, p = .002), entropy (Z = -2.51, p = .012), maximum
movements were clustered to estimate their respective aver- line (Z = -2.32, p = .021), and laminarity (Z = -2.43, p = .016)
age. than the randomized series.
As described in Figure 3 (Panels (a), (b), (c), and (d)), In Figure 4 (Panels (a), (b), (c), and (d)), segmenting
for the string of words the original series had a significantly by individuals, the children in the original series presented
higher percentage of determinism (Z = -2.82, p = .005), marginally higher levels of maximum line than in the ran-
entropy (Z = -3.06, p = .002) and maximum line (Z = -2.20, domized series (Z = -1.83, p = .06); however no differences
p = .028), and a marginally higher laminarity (Z = -1.65, p were detected in terms of determinism, entropy, and laminar-
= .09) than in the randomized series. The same trend was ity (Zs ≤ - .11, ps ≥ .91). For caregivers (Panels (e), (f), (g), and
observed with verbal expressions and grammar (Panels (e), (h)) and strangers (Panels (i), (j), (k), and (l)), the original
(f), (g), and (f)), where the original series had significantly series showed higher levels of determinism, entropy, and
more determinism (Z = -2.82, p = .005), entropy (Z = -3.06, maximum line (Zs ≤ -2.20, ps ≤ 0.028) than the randomized
p = .002), and the maximum line (Z = -3.06, p = .002) had series. However, no differences were observed in terms of
a marginally higher laminarity (Z = -1.65, p = .099) than the laminarity (Zs ≤ -1.58, ps ≥ .11).
randomized series. For body movements (Panels (i), (j), (k), When focusing on dyads (Figure 5), it is possible to
and (l)), the original series had higher levels of determinism observe that the child-caregiver dyad (Panels (a), (b), (c),
Complexity 11

5.0 120.0 70.0

100.0
100.0 60.0

Maximum Line
%Determinism 4.0
80.0

Laminarity
50.0

Entropy
3.0 80.0
60.0 40.0
60.0 30.0
40.0 2.0
40.0 20.0
20.0 1.0 20.0 10.0
0.0 0.0 0.0 0.0
Original Randomized Original Randomized Original Randomized Original Randomized

(a) (b) (c) (d)

5.0 120.0 70.0
100.0 60.0
100.0

Maximum Line
4.0

Laminarity
%Determinism

80.0 50.0

Entropy
3.0 80.0 40.0
60.0
60.0 30.0
40.0 2.0
40.0 20.0
20.0 1.0 10.0
20.0
0.0 0.0 0.0
0.0
Original Randomized Original Randomized Original Randomized Original Randomized

(e) (f) (g) (h)

100.0 120.0 70.0
5.0
60.0
100.0

Maximum Line
%Determinism

80.0

Laminarity
4.0 50.0
80.0
Entropy

60.0 3.0 40.0

60.0 30.0
40.0 2.0 40.0 20.0
20.0 1.0 20.0 10.0
0.0 0.0 0.0 0.0
Original Randomized Original Randomized Original Randomized Original Randomized

(i) (j) (k) (l)

Figure 3: Mean and standard errors of determinism (Panels (a), (e), and (i)), entropy (Panels (b), (f), and (j)), maximum line (Panels (c), (g),
and (k)), and laminarity (Panels (d), (h), and (l)), segmenting by words, verbal expressions and grammar, and body movements.

100.0 5.0 120.0 1.0

Maximum Line

4.0 100.0 0.8

%Determinism

Laminarity
80.0
80.0
Entropy

60.0 3.0 0.6

60.0
40.0 2.0 0.4
40.0
20.0 1.0 20.0 0.2
0.0 0.0 0.0 0.0
Original Randomized Original Randomized Original Randomized Original Randomized

(a) (b) (c) (d)

5.0 120.0 1.0
100.0
0.8
Maximum Line

4.0 100.0
Laminarity
%Determinism

80.0
80.0
Entropy

3.0 0.6
60.0
60.0 0.4
40.0 2.0
40.0
1.0 0.2
20.0 20.0
0.0 0.0 0.0
0.0
Original Randomized Original Randomized Original Randomized Original Randomized

(e) (f) (g) (h)

100.0 120.0 1.0
5.0
Maximum Line

100.0 0.8
%Determinism

Laminarity

80.0 4.0
80.0 0.6
Entropy

60.0 3.0
60.0 0.4
40.0 2.0 40.0
20.0 1.0 0.2
20.0
0.0 0.0 0.0 0.0
Original Randomized Original Randomized Original Randomized Original Randomized

(i) (j) (k) (l)

Figure 4: Mean and standard errors of determinism (Panels (a), (e), and (i)), entropy (Panels (b), (f), and (j)), maximum line (Panels (c), (g),
and (k)), and laminarity (Panels (d), (h), and (l)) when segmenting by individuals (child, caregiver, and stranger).

(d), and (e)) in the original series had significantly more determinism, entropy, and trapping time (Zs = -2.21, ps =
determinism, entropy, maximum line, and trapping time 0.03), as well as a marginally higher maximum line (Z = -1.75,
than in the randomized series (Zs ≥ -2.21, ps ≤ .03), but p = 0.08) than the randomized series. However, no differences
no difference was observed in laminarity. The caregiver- were detected in laminarity (Z = -1.51, p = .12).
stranger dyad (Panels (l), (m), (n), (o), and (p)) and child- With this nonparametric analysis, we corroborate what
stranger dyad (Panels (f), (g), (h), (i), and (j)) showed similar was previously reported from the visual inspection sum-
trends with the original series expressing higher levels of marized in Tables 4 and 5. Words and verbal expressions
12 Complexity

100.0 5.0 100.0 100.0 30.0

Maximum Line
%Determinism

Trapping Time
25.0

Laminarity
80.0 4.0 80.0 80.0

Entropy
60.0 3.0 60.0 60.0 20.0
15.0
40.0 2.0 40.0 40.0 10.0
20.0 1.0 20.0 20.0 5.0
0.0 0.0 0.0 0.0 0.0
Original Randomized Original Randomized Original Randomized Original Randomized Original Randomized

(a) (b) (c) (d) (e)

100.0 5.0 100.0 120.0 30.0

Maximum Line
%Determinism

Trapping Time
100.0 25.0

Laminarity
80.0 4.0 80.0

Entropy
60.0 3.0 60.0 80.0 20.0
60.0 15.0
40.0 2.0 40.0 40.0 10.0
20.0 1.0 20.0 20.0 5.0
0.0 0.0 0.0 0.0 0.0
Original Randomized Original Randomized Original Randomized Original Randomized Original Randomized

(f) (g) (h) (i) (j)

100.0 5.0 100.0 120.0 30.0

Maximum Line

Trapping Time
%Determinism

100.0 25.0

Laminarity
80.0 4.0 80.0
Entropy

60.0 3.0 60.0 80.0 20.0

60.0 15.0
40.0 2.0 40.0 40.0 10.0
20.0 1.0 20.0 20.0 5.0
0.0 0.0 0.0 0.0 0.0
Original Randomized Original Randomized Original Randomized Original Randomized Original Randomized

(k) (l) (m) (n) (o)

Figure 5: Mean and standard errors of determinism (Panels (a), (f), and (l)), entropy (Panels (b), (g), and (m)), maximum line (Panels (c),
(h), and (n)), laminarity (Panels (d), (i), and (o)), and trapping time (Panels (e), (j), and (p)) when segmenting by dyads (child-caregiver,
child-stranger, and caregiver-stranger).

and grammar showed a structure that resembles a typical type of synchronization with a recurrent structure, where the
coupling pattern. This structure was more defined and initial states enabled predicting the final states of the system,
clearer than the one observed in body movements. The but whose stability was not different from a series where the
actors involved in the Strange Situation showed noteworthy motor behaviours appear in a random manner [40, 43–48,
differences among them. While children clearly showed no 51]. Thus, our results indicate that verbal behaviours—in the
traces of a structure, other than that observed by chance; Strange Situation—are part of a communicative phenomenon
the behaviours of caregivers and strangers had clear traces that expresses higher levels of synchronization than motor
of dynamic patterns, typical of coupled systems. Finally, all behaviours [45, 46]. This preliminary finding emphasizes that
dyads presented clear synchronization indicators, especially the communicative interaction has synchronization features,
caregiver-child dyads. It is important to note that verbal but these attributes are not homogeneous. If, until now,
and motor behaviours expressed by children did not have we assumed that communicative interaction—among three
elements that indicate a coupling pattern. However, when people interacting during a protocolized evaluation—was
these children interacted with their respective caregivers, the globally synchronized, our results suggest that some aspects
communicative behaviour between them clearly expressed have more dynamic characteristics than others.
signs of synchronization. For children, the values of determinism, entropy, max-
imum line, and laminarity remained constant between
4. Conclusions the original and randomized series. Thus, the structure
of verbal and motor behaviours expressed for children
The purpose of this research was to characterize the degree was not different from what was to be expected if these
of structuring of behaviours in order to identify the param- behaviours appeared in a random manner. In contrast, for
eters of synchronization in a protocolized communicative adults—caregivers or strangers—the values of determinism,
interaction, Ainsworth’s Strange Situation [21, 27, 31–33], by entropy, laminarity, and maximum line were significantly
analysing the unfolded verbal and motor behaviours among reduced when their original series were randomized, sug-
caregivers, children, and strangers [31–33]. These behaviours gesting that the original series of communicative behaviours
were scrutinised using a nonlinear technique named Cate- had a synchronization pattern that was far from a random
gorical Cross-Recurrence Quantification Analysis [5, 38, 39, organization. This made us aware that, in a communica-
41–43]. From these analyses, it was expected to estimate tive interaction, not all actors involved have synchronized
measures that have been used to characterize degrees of behaviours. However, when analysing the recurrence of
coupling between systems [44–48]. two people interacting, the system itself shows traces of
The findings showed that words and verbal expressions synchronization, even when one of the actors (in our case the
and grammar had clear parameters of synchronization, children) does not show synchronization traits.
taking into account the fact that determinism, entropy, We are still blind to the attachment pattern of these two
maximum line, and laminarity were higher in the original girls who participated in the Strange Situation. However,
series in comparison to the randomized series [16]. In the there are two possible scenarios that we conjecture. In the
case of body movements, communicative patterns showed a first one it can assumed that both infants have the same
Complexity 13

attachment pattern—regardless of whether this pattern is A, Data Availability

B, C, or D—and therefore the observed values indicate a
similar pattern of recurrence among them as observed in our The data used to support the findings of this study are
study. Another possibility is that the attachments of these available from the corresponding author upon request.
infants are different (generating the possible combinations
of A and B, A and C, A and D, B and C, B and D, or C Conflicts of Interest
and D) [31–33]. Under this scenario, the observed recurrence
values (entropy, laminarity, determinism, among others) did The authors declare that there are no conflicts of interest
not detect the differences that the infants manifested in their regarding the publication of this paper.
attachment behaviours or how this pattern of attachment
was unfolded in the interaction with adults. So, in order Acknowledgments
to disambiguate this problem our current work is aimed
at analysing more Strange Situation videos, where each of The authors thank Dr. José Luis Ulloa Fulgeri (Universidad
the four types of attachment patterns can be represented de Talca, Chile), for his collaboration with the R program,
proportionally. Thus, this preliminary investigation can be and Dr. David López (University of Warsaw, Poland), for
improved with the incorporation of more children classi- his suggestions and comments about preliminary version
fied according the four types of attachment [21], in such of this manuscript. This research has been funded by the
a quantity that comparisons can be made in terms more Comisión Nacional de Investigación Cientifica y Tecnológica
robust [44]. Attachment patterns are discrete categories (CONICYT) under Research Grants REDES-170155 and PCI-
defined by certain behaviours. Some of them promote com- PAI80160101 to Ramón D. Castillo and by the Fondo Nacional
municative interactions with adults, while others restrict de Desarrollo Cientı́fico y Tecnológico de Chile (FONDE-
them [30]. Therefore, it is expected that recurrence param- CYT) under Research Grant no. 1161533 to Ramón D. Castillo
eters tend to vary from one type of attachment pattern to and Research Grant no. 1130773 to Rosario Spencer.
another.
Our study resembles a study with small samples or
two unique cases study. The nonparametric contrasts that Supplementary Materials
we conducted were purely heuristic and corroborated what The databases, coding matrices, and transcriptions of the
the tables expressed. However, the possibility of comparing interactions can be requested from Ramón D. Castillo
means depends on having an adequate sample size. Despite ([email protected]) and ([email protected]).
this limitation, the direct observation of the recurrence (Supplementary Materials)
parameters in the recurrence plots is an extended practice,
because the means and deviations are deceptive, insofar as
they can hide the temporal structure of the behaviours [2, References
8, 40, 43]. Two groups could have the same mean and the
[1] A. L. Goldberger, D. R. Rigney, and B. J. West, “Chaos and
same standard deviation and not show significant differences. fractals in human physiology,” Scientific American, vol. 262, no.
However, they could have a different structure of variability 2, pp. 42–49, 1990.
over time. This last aspect is what the recurrence plots and [2] J. A. S. Kelso, Dynamic Patterns: The Self-Organization of Brain
RQA identify and that we have applied to Ainsworth’s Strange and Behavior, The MIT Press, Cambridge, MA, 1995.
Situation.
[3] K. De Bot, W. Lowie, and M. Verspoor, “A dynamic systems
The Ainsworth’s Strange Situation protocol is intended theory approach to second language acquisition,” Bilingualism:
to study the infant’s attachment pattern. In this protocol, Language and Cognition, vol. 10, no. 1, pp. 7–21, 2007.
two adults, the caregiver and the stranger, interact with [4] M. M. Louwerse, R. Dale, E. G. Bard, and P. Jeuniaux, “Behavior
the infant. However, it is also possible to observe that matching in multimodal communication Is synchronized,”
these adults interact with each other for a few minutes. In Cognitive Science, vol. 36, no. 8, pp. 1404–1426, 2012.
that sense, the interaction between the caregiver and the [5] K. Shockley and M. Riley, “Interpersonal couplings in human
stranger is susceptible to be analysed in terms of verbal and interactions,” in Recurrence quantifications analysis: Theory and
motor behaviour. Considering that the central objective is best practices, C. Webber and N. Marwan, Eds., pp. 399–421,
to establish the attachment pattern of the infant, a rational Springer, Heidelberg, Germany, 2015.
and feasible decision could be to omit information regarding [6] D. L. Gilden, T. Thornton, and M. W. Mallon, “1/F Noise in
the few minutes of interaction between the caregiver and human cognition,” Science, vol. 267, no. 5205, pp. 1837–1839,
the stranger; however, we believe that it is early to make 1995.
such a decision, considering that we are still estimating the [7] D. L. Gilden, “Cognitive emissions of 1/f noise,” Psychological
weight that each actor has in the interactions that the protocol Review, vol. 108, no. 1, pp. 33–56, 2001.
promotes. Social interactions have a dynamic character; [8] M. A. Riley and M. T. Turvey, “Variability and determinism in
therefore what will happen in the future is a function of what motor behavior,” Journal of Motor Behavior, vol. 34, no. 2, pp.
has happened in previous phases. Therefore, we suppose the 99–125, 2002.
interaction of these two adults in a previous phase could be [9] G. C. Van Orden, J. G. Holden, and M. T. Turvey, “Self-
an important ingredient that affects their interaction with the organization of cognitive performance,” Journal of Experimental
infant in the following phases. Psychology: General, vol. 132, no. 3, pp. 331–350, 2003.
14 Complexity

[10] C. T. Kello, B. C. Beltz, J. G. Holden, and G. C. Van Orden, [28] C. López and M. Ramı́rez, “Apego,” Revista Chilena de Medicina
“The Emergent Coordination of Cognitive Function,” Journal Familiar, vol. 6, no. 1, pp. 20–24, 2005.
of Experimental Psychology: General, vol. 136, no. 4, pp. 551–568, [29] A. Lichtwarck-Aschoff, F. Hasselman, R. Cox, D. Pepler, and I.
2007. Granic, “A characteristic destabilization profile in parent-child
[11] M. T. Turvey, “Action and perception at the level of synergies,” interactions associated with treatment efficacy for aggressive
Human Movement Science, vol. 26, no. 4, pp. 657–697, 2007. children,” Nonlinear Dynamics, Psychology, and Life Sciences,
[12] D. G. Stephen and D. Mirman, “Interactions dominate the vol. 16, no. 3, pp. 353–379, 2012.
dynamics of visual cognition,” Cognition, vol. 115, no. 1, pp. 154– [30] B. E. Vaughn, S. Goldberg, L. Atkinson, S. Marcovitch, D. Mac-
165, 2010. Gregor, and R. Seifer, “Quality of Toddler-Mother Attachment
[13] N. A. Kuznetsov and S. Wallot, “Effects of accuracy feedback on in Children with Down Syndrome: Limits to Interpretation of
fractal characteristics of time estimation,” Frontiers in Integra- Strange Situation Behavior,” Child Development, vol. 65, no. 1,
tive Neuroscience, vol. 5, no. 62, pp. 1–12, 2011. pp. 95–108, 1994.
[14] C. A. Coey, M. Varlet, and M. J. Richardson, “Coordination [31] M. D. S. Ainsworth, Infancy in Uganda: infant care and growth
dynamics in a socially situated nervous system,” Frontiers in of attachment, University Press, Baltimore, Maryland: Johns
Human Neuroscience, vol. 6, no. 2012, pp. 1–12, 2012. Hopkins, 1967.
[15] M. Malone, R. D. Castillo, H. Kloos, J. G. Holden, M. J. [32] M. D. S. Ainsworth, M. C. Blehar, E. Waters, and S. Wall, Pat-
Richardson, and R. Balasubramaniam, “Dynamic Structure of terns of attachment: A Psychological Study of Strange Situation,
Joint-Action Stimulus-Response Activity,” PLoS ONE, vol. 9, no. Lawrence Erlbaum, Hillsdale, NJ, USA, 1978.
2, p. e89032, 2014. [33] M. D. Ainsworth and S. M. Bell, “Attachment, exploration, and
separation: illustrated by the behavior of one-year-olds in a
[16] R. D. Castillo, H. Kloos, J. G. Holden, and M. J. Richardson,
strange situation.,” Child Development, vol. 41, no. 1, pp. 49–67,
“Long-range correlations and patterns of recurrence in children
1970.
and adults’ attention to hierarchical displays,” Frontiers in
Physiology, vol. 6, 2015. [34] C. Beckner, R. Blythe, J. Bybee et al., “Language is a complex
adaptive system: Position paper,” Language Learning, vol. 59, no.
[17] E. Delaherche, M. Chetouani, A. Mahdhaoui, C. Saint-Georges,
1, pp. 1–26, 2009.
S. Viaux, and D. Cohen, “Interpersonal synchrony: A survey
of evaluation methods across disciplines,” IEEE Transactions on [35] S. Wallot, “Recurrence Quantification Analysis of Processes and
Affective Computing, vol. 3, no. 3, pp. 349–365, 2012. Products of Discourse: A Tutorial in R,” Discourse Processes, vol.
54, no. 5-6, pp. 382–405, 2017.
[18] R. Vink, M. L. Wijnants, A. H. N. Cillessen, and A. M. T.
Bosman, “Cooperative learning and interpersonal synchrony,” [36] D. H. Abney, A. S. Warlaumont, A. Haussman, J. M. Ross, and S.
Nonlinear Dynamics, Psychology, and Life Sciences, vol. 21, no. 2, Wallot, “Using nonlinear methods to quantify changes in infant
pp. 189–215, 2017. limb movements and vocalizations,” Frontiers in Psychology, vol.
5, p. 771, 2014.
[19] M. J. Hove and J. L. Risen, “It’s all in the timing: Interpersonal
synchrony increases affiliation,” Social Cognition, vol. 27, no. 6, [37] F. Ramseyer and W. Tschacher, “Nonverbal synchrony of head-
pp. 949–960, 2009. and body-movement in psychotherapy: Different signals have
different associations with outcome,” Frontiers in Psychology,
[20] M. I. Coco and R. Dale, “Cross-recurrence quantification anal- vol. 979, no. 5, pp. 1–9, 2014.
ysis of categorical and continuous time series: An R package,”
[38] J. P. Zbilut and C. L. Webber Jr., “Embeddings and delays as
Frontiers in Psychology, vol. 5, article 510, 2014.
derived from quantification of recurrence plots,” Physics Letters
[21] M. Main and J. Solomon, “Procedures for identifying infants as A, vol. 171, no. 3-4, pp. 199–203, 1992.
disorganized/disoriented during the Ainsworth Strange Situa-
[39] N. Marwan, M. Carmen Romano, M. Thiel, and J. Kurths,
tion,” in Attachment in the preschool years, M. T. Greenberg, D.
“Recurrence plots for the analysis of complex systems,” Physics
Cicchetti, and E. M. Cummings, Eds., pp. 121–160, University of
Reports, vol. 438, no. 5-6, pp. 237–329, 2007.
Chicago Press, Chicago, 1990.
[40] F. Orsucci, A. Giuliani, C. Webber Jr., J. Zbilut, P. Fonagy,
[22] P. Fitzpatrick, V. Romero, J. L. Amaral et al., “Evaluating the
and M. Mazza, “Combinatorics and synchronization in natural
importance of social motor synchronization and motor skill
semiotics,” Physica A: Statistical Mechanics and its Applications,
for understanding autism,” Autism Research, vol. 10, no. 10, pp.
vol. 361, no. 2, pp. 665–676, 2006.
1687–1699, 2017.
[41] K. Shockley, “Cross recurrence quantification of
[23] D. Stern, The interpersonal world of the infant, Basic Books, New interpersonal postural activity,” in Tutorials in contemporary
York, NY, USA, 1985. nonlinear methods for the behavioral sciences, M. A.
[24] M. Siller and M. Sigman, “The behaviors of parents of children Riley and G. C. Van Orden, Eds., pp. 142–177, 2005,
with autism predict the subsequent development of their chil- https://www.nsf.gov/sbe/bcs/pac/nmbs/chap4.psd.
dren’s communication,” Journal of Autism and Developmental [42] M. A. Riley, R. Balasubramaniam, and M. T. Turvey, “Recur-
Disorders, vol. 32, no. 2, pp. 77–89, 2002. rence quantification analysis of postural fluctuations,” Gait &
[25] B. Tunçgenç, E. Cohen, and C. Fawcett, “Rock With Me: The Posture, vol. 9, no. 1, pp. 65–78, 1999.
Role of Movement Synchrony in Infants’ Social and Nonsocial [43] C. L. Webber and J. P. Zbilut, “Recurrence Quantification
Choices,” Child Development, vol. 86, no. 3, pp. 976–984, 2015. Analysis of Nonlinear Dynamical Systems,” in Tutorials in
[26] J. Bowlby, “The nature of the child’s tie to his mother,” The contemporary nonlinear methods for the behavioral sciences, M.
International journal of psycho-analysis, vol. 39, no. 5, pp. 350– A. Riley and G. C. Van Orden, Eds., vol. 94, pp. 26–94, 2005,
373, 1958. https://www.nsf.gov/sbe/bcs/pac/nmbs/chap4.psd.
[27] J. Bowlby, A secure base. Clinical applications of Attachment [44] D. C. Richardson and R. Dale, “Looking To Understand: The
Theory, Buenos Aires: Editorial Paidós, 1989. Coupling Between Speakers’ and Listeners’ Eye Movements
Complexity 15

and Its Relationship to Discourse Comprehension,” Cognitive [59] E.-J. Wagenmakers, H. L. J. Van Der Maas, and S. Farrell,
Science, vol. 29, no. 6, pp. 1045–1060, 2005. “Abstract Concepts Require Concrete Models: Why Cogni-
[45] M. J. Spivey and R. Dale, “Continuous dynamics in real-time tive Scientists Have Not Yet Embraced Nonlinearly Coupled,
cognition,” Current Directions in Psychological Science, vol. 15, Dynamical, Self-Organized Critical, Synergistic, Scale-Free,
no. 5, pp. 207–211, 2006. Exquisitely Context-Sensitive, Interaction-Dominant, Multi-
fractal, Interdependent Brain-Body-Niche Systems,” Topics in
[46] D. C. Richardson, R. Dale, and N. Z. Kirkham, “The Art of Cognitive Science, vol. 4, no. 1, pp. 87–93, 2012.
Conversation Is Coordination,” Psychological Science, vol. 18, no.
5, pp. 407–413, 2007. [60] E.-J. Wagenmakers, S. Farrell, and R. Ratcliff, “Human cognition
and a pile of sand: A discussion on serial correlations and
[47] R. F. A. Cox and M. van Dijk, “Microdevelopment in Parent- self-organized criticality,” Journal of Experimental Psychology:
Child Conversations: From Global Changes to Flexibility,” General, vol. 134, no. 1, pp. 108–116, 2005.
Ecological Psychology Journal, vol. 25, no. 3, pp. 304–315, 2013.
[61] D. L. Gilden, “Global model analysis of cognitive variability,”
[48] J. P. Zbilut, A. Giuliani, and C. L. Webber Jr., “Detecting Cognitive Science, vol. 33, no. 8, pp. 1441–1467, 2009.
deterministic signals in exceptionally noisy environments using
cross-recurrence quantification,” Physics Letters A, vol. 246, no.
1-2, pp. 122–128, 1998.
[49] R. Fusaroli, I. Konvalinka, and S. Wallot, “Analyzing Social
Interactions: The Promises and Challenges of Using Cross
Recurrence Quantification Analysis,” in Translational Recur-
rences: From Mathematical Theory to Real-World Applications,
N. Marwan, M. Riley, A. Giuliani, and C. Webber, Eds., vol. 103,
pp. 137–155, Springer International Publishing, London, UK,
2014.
[50] L. De Jonge-Hoekstra, S. Van der Steen, P. Van Geert, and
R. F. Cox, “Asymmetric Dynamic Attunement of Speech and
Gestures in the Construction of Children’s Understanding,”
Frontiers in Psychology, vol. 7, 2016.
[51] R. Dale and M. J. Spivey, “Unraveling the dyad: Using recurrence
analysis to explore patterns of syntactic coordination between
children and caregivers in conversation,” Language Learning,
vol. 56, no. 3, pp. 391–430, 2006.
[52] A. Warlaumont, D. Oller, and R. Dale, “Vocal interaction
dynamics of children with and without autism,” in Proceedings
of the 32nd Annual Meeting of the Cognitive Science Society,
S. Ohlsson and R. Catrambone, Eds., pp. 121–126, Cognitive
Science Society, Austin, TX, 2010.
[53] S. Wallot, B. A. O’Brien, A. Haussmann, H. Kloos, and M. S.
Lyby, “The role of reading time complexity and reading speed
in text comprehension,” Journal of Experimental Psychology:
Learning, Memory, and Cognition, vol. 40, no. 6, pp. 1745–1765,
2014.
[54] R. Schmidt and M. Richardson, “Dynamics of interpersonal
coordination,” in Coordination: Neural, Behavioral and Social
Dynamics, Fuch and V. Jirsa, Eds., pp. 281–308, Springer-Verlag,
Heidelberg, Germany, 2008.
[55] J. A. De Graag, R. F. A. Cox, F. Hasselman, J. Jansen, and C.
De Weerth, “Functioning within a relationship: Mother-infant
synchrony and infant sleep,” Infant Behavior & Development,
vol. 35, no. 2, pp. 252–263, 2012.
[56] J. G. Holden, I. Choi, P. G. Amazeen, and G. Van Orden,
“Fractal 1/ Dynamics Suggest Entanglement of Measurement
and Human Performance,” Journal of Experimental Psychology:
Human Perception and Performance, vol. 37, no. 3, pp. 935–948,
2011.
[57] J. G. Holden, G. C. Van Orden, and M. T. Turvey, “Dispersion
of Response Times Reveals Cognitive Dynamics,” Psychological
Review, vol. 116, no. 2, pp. 318–342, 2009.
[58] C. T. Kello, G. G. Anderson, J. G. Holden, and G. C. Van
Orden, “The pervasiveness of 1/f scaling in speech reflects the
metastable basis of cognition,” Cognitive Science, vol. 32, no. 7,
pp. 1217–1231, 2008.