Compared to other mammalian species, human newborns present larger and more adaptable brains that are also particularly immature and dependent on caregiving behaviors and environment (Trevathan, 2015). Not only is human infants' survival and safety granted by the caregiving environment, but the developmental outcomes themselves are shaped by the early context of care and caregiver-infant interaction (O'Connor, 2003; Swain et al., 2007). Thus, human caregivers—more often the mothers—have to respond to infant basic needs (such as feeding, sleep, and temperature regulation), and also play a crucial role in fostering regulative (Conradt and Ablow, 2010), emotional (Provenzi et al., 2015a), cognitive (Malmberg et al., 2016) and social abilities (Kivijärvi et al., 2001), that form developmental key competencies for later life adaptation and both physical and psychological health. Consequently, the multilayered and complex interactional processes occurring between infants and their mothers is of vital importance for healthy developmental trajectories.

The scientific interest in the form, quality and developmental relevance of early caregiver-infant relationship is as old as Bowlby's trilogy on attachment (Bowlby, 1969/1982, 1973, 1980). Based on these works, Ainsworth and colleagues (Ainsworth et al., 1978) considered maternal sensitivity, i.e., the ability to perceive infant signals, to interpret them correctly and to promptly and appropriately respond to them, as the key component in mother-infant relationship. While this concept has been critical to understand infant attachment (Behrens et al., 2016), it was recognized as a global, unidirectional scale that does not help in depicting the underlying mechanisms which occur moment-by-moment and require bidirectional or reciprocal contributions of both the mother and the infant behavioral and emotional states (Mesman, 2010).

Stern (1971) was among those who pioneered the idea that mother and infant engage in a bidirectional moment-to-moment process, in which, together, both partners form repeated dyadic patterns of body-movements and gaze-behaviors—a view that emphasized infant readiness for social encounters and social-based meaning making. Consequently, the interest in infants' communicative abilities, dyadic exchanges and early forms of intersubjectivity rapidly grew during the 70s and 80s (e.g., Sander, 1977; Brazelton, 1979; Trevarthen, 1979, 1980; Stern, 1985). Bateson (1979) described the interaction between caregiver and infant as eye-contacts and vocalizations forming a “proto-conversation” primarily characterized by mutuality. Others highlighted the active role played by infants in early face-to-face interactions proposing that the trend and drive to intersubjectivity is something innate at birth in human infants (e.g., Murray and Trevarthen, 1986). Infants were increasingly viewed as competent social partners, who anticipate the caregivers' responses to their own communicative signals and react with, for example, withdrawal or protest to the caregiver's social disengagement (e.g., the Still-Face paradigm; Brazelton et al., 1975; Papoušek and Papoušek, 1977; Tronick et al., 1978).

Tronick was among the first to provide a formalized and consistent theoretical model of the reciprocal nature of caregiver-infant-interactions called the Mutual-Regulation-Model (Gianino and Tronick, 1988). This model was mainly grounded in the growing assumption supported by the infant research field, which highlighted that the caregiver-infant-dyad forms a mutually coordinated, communicative unit that quickly oscillates between synchronous states of affective-behavioral matches and asynchronous states of affective-behavioral mismatches in a continuous moment-to-moment process of mutual behavioral adaption, emotional exchange and affect regulation.

To date, the view on the caregiver-infant dyad has been expanded by non-linear dynamic system approaches (Hollenstein, 2007; DiCorcia and Tronick, 2011; Sravish et al., 2013; Beebe et al., 2016). Consistent with this framework, the caregiver-infant-dyad is viewed as an interactional system, in which both partners hierarchically organize levels of functioning (behavioral, affective, physiological, etc.) by reciprocally and mutually coordinating their behaviors, communicative signals and emotional states in various domains, in which changes on one level affect the functioning (and development) on others.

The nature of the reciprocal interconnections and interactions between a mother and her infant is complex and multifaceted and what takes place between the two interactive partners is made up of multiple processes. To understand these dyadic mechanisms, research started to analyze the micro-temporal features of these social exchange processes. In their review, Leclère et al. (2014) focused on the dyadic concept of “synchrony” and defined it as the intermodal temporal coordination of verbal and non-verbal communicative and emotional behaviors between interactive partners. They confirmed that synchrony is higher between familiar, healthy partners and is associated with more positive child outcomes. However, they recognized that various terms (mutuality, reciprocity, rhythmicity, etc.) and assessment methods (global scales, specific synchrony scales, micro-temporal time series analyses) have been used to describe and measure dyadic synchrony.

This paucity of conceptual clarity doesn't only apply to the concept of synchrony, but also to other dyadic concepts that have been proposed by diverse authors and throughout scientific progress in the infant research field. Stern (2010) dedicated more attention to affective attunement between mother and infant, i.e., the maternal matching and modulation of the infant affective tone. Sander (1988) addressed the importance of mutual event-structure in moment-to-moment interactions between caregivers and infants on neonatal state organization and biological regulation. Beebe and Lachmann (1998) captured Fogel's (1992) concept of co-regulation (i.e., a concept denoting individuals' behaviors which continuously modify and shape each other) and later understood self- and interactive contingency as the temporal relation between the occurrence of affective-behavioral events, that involves sequential coordination (Beebe et al., 2016). Fonagy and Target (2006) see the caregivers' attuned, contingent and empathic mirroring responses to infant's emotional signals as the key element in the acquisition of emotion regulation, whereas they consider “marked” mirroring (i.e., exaggerated modulations of infant signals) as the optimal mode for infants to internalize the expression of their own feelings. Again, Tronick stated that most dyadic interaction processes are “messy” in the sense that they normally feature naturally occurring mismatched states, but the successful reparation of these dyadic mismatches is highly important for the development of infant regulation, resilience and of other domains (DiCorcia and Tronick, 2011). Thus, perfect synchrony between the temporal sequences of each partners' behaviors might be neither desirable nor possible, nor can the process of interactive reparation be sufficiently measurable by assessments of synchrony.

First, it is not clear if these theoretical constructs are meant to be actual interactive processes describing what happens in the dyad in a moment-by-moment fashion or rather if they are broader meta-theoretical accounts of mother-infant dyads. Notably, there is a lack of systematization regarding which theoretical items are concepts (i.e., meta-theoretical views of mother-infant dyadic interactions) and which of them are processes (i.e., detailed descriptions of specific joint actions observable within the mother-infant dyad). Moreover, the relationships among these dyadic concepts has not been previously accounted for. It is our belief, that all these dyadic constructs reflect specific facets of the complex micro-temporal dyadic nature of the mother-infant system. Additionally, there is also a lack of agreement and systematization at the methodological level. For instance, interactive reparation might be alternatively measured as the frequency of transitions from mismatched to matched states (e.g., Provenzi et al., 2015b) or as the average mismatch duration (e.g., Müller et al., 2015), while synchrony is usually assessed by time-series-analyses and lead-lag-relationships (e.g., Feldman, 2003).

In the previous paragraphs a very brief sketch on previous infant research decades has been laid out in order to highlight that when it comes to measuring and understanding what happens within the mother-infant dyad in research, different terms are often used interchangeably. This leads to an overall confusion in both theoretical and methodological level. Consistently, it is plausible that this theoretical and methodological disarray of infant research might have led to mixed findings in previous research. After more than four decades of studies in the field, we believe it is time to provide a first attempt to systemize not only the evidence, but also the theoretical underpinnings and the methodologies used in this area of investigation. On the one hand, it might be useful to develop a unified dyadic dictionary in which each term could have a theoretical definition based on previous contributions in literature and in which the actual and potential relationships among different dyadic processes are highlighted or, at least, hypothesized. On the other hand, a methodological roadmap is also needed to navigate the multifaceted landscape of methods that have been previously used to quantify mother-infant interactive states and behaviors. Finally, this comprehensive systemization also holds the potentials of highlighting what previous decades of infant research tradition has found about the protective and risk factors inherent to adaptive and maladaptive caregiver-infant environments.

In the present work, we aimed to disentangle the different dyadic constructs (i.e., both concepts and processes) which can be depicted and measured when observing mother-infant interacting dyads. Differences and similarities among different dyadic constructs will be highlighted through a three-step systematic review. First, we will review the theoretical definition of dyadic constructs, according to a qualitative approach plus a computer-aided text-analysis of the included papers (section: Theoretical Review). Second, methodological aspects will be abstracted and reviewed, in order to provide a systematic overview of the procedures adopted in previous literature to measure specific dyadic processes in the mother-infant interaction (section: Methodological Review). Third, findings of previous studies will be reviewed describing specific characteristics of dyadic processes, their predictors and effects on infants and children development, as well as differences among low- and high-risk mother-infant dyads (section: Systematic Review of Outcomes).

The procedures of the present study are consistent with the guidelines for systematic review included in the Referred Reporting items for Systematic Review and Meta-Analysis (PRISMA) guidelines (Liberati et al., 2009). The literature search was performed on three different databases: Scopus, PubMed and Web of Science. The search string included: “mother” AND “infant” AND (“attunement” OR “contingency” OR “coordination” OR “matching” OR “mirroring” OR “mutuality” OR “reciprocity” OR “reparation” OR “synchrony”). The records were checked for duplicates using Mendeley 1.17.6 (© 2008–2016 Mendeley Ltd.). The remaining papers were than screened independently by three authors (LP, MM, and LG) by reading titles, abstracts and full-texts. Disagreements were solved in conference. Exclusion criteria and the whole selection process are reported in the flow chart in Figure 1.

The records were reviewed and the following data was extracted: authors, year, journal source, dyadic construct(s) included, theoretical definition, operation definition, measurement and instrument/tool, sample characteristics, sample size, setting, observational procedure, findings (separated in four categories: characterization, antecedents, consequences and between-group differences for the included dyadic process). According to the PICOS questions for systematic review, population was mother-infant dyads, comparators were sample characteristics, setting, type of observational procedure, outcomes were different effects of dyadic processes on maternal and infants' domains, whereas no exclusion criterion was applied to study designs (but see quality appraisal below). Interventions were not included in the present review. The final pool of included studies is reported in Table 1.

The methodological quality of the included papers was assessed according to the Quality Assessment Tool for Quantitative Studies (Jackson et al., 2005). Sections A-F (A, selection bias; B, study design; C, confounders; D, blinding; E, data collection methods; F, withdrawal and dropouts) were coded by three independent researchers (LP, GSM, EG) as 3 (weak), 2 (moderate) or 1 (strong) according to the component rating scale criteria. A summary 1-to-3 score was assigned to each paper according to the presence of 2 or more weak scores (3, weak), only 1 weak score (2, moderate), no weak scores (1, strong). A 93% agreement was reached for the A-F components. Disagreement was solved in conference through the supervision of the third author.

We reported the definition of the different dyadic constructs in Table 3. First, it is immediately clear that two (i.e., mutuality and reciprocity) among the nine constructs abstracted here seem to be much better framed as wide, global and meta-theoretical concepts rather than as processes. Both these concepts provide a specific point of view on mother-infant interaction, assuming that there are reciprocal or mutual contributions within the “dyadic dance.” Nonetheless, they are different in the way they underline the relative contributions of each of the interactive patterns. Indeed, reciprocity assumes that the contribution of the mother and the infant are in some way equal in terms of relevance, frequency and intensity (Trevarthen, 1980), whereas mutuality may represent a more cautious and conservative approach in which both partners contribute to the dyadic system, but with different quantities and qualities (Beebe et al., 2010). The first approach appears to be more conducible to the first contributions from Trevarthen and Stern, whereas the mutual model is more consistent with the proposals of Beebe and Tronick.

The other seven constructs appear to be real dyadic processes, with both theoretical and methodological definitions. These processes appear to be in specific relationships among themselves. First, contingency emerges as the low-level process, in which the dyadic encounter is expressed by the concomitant expression of specific gestures. When these contingent moments repeat in time they generate more prolonged states of coordination, which require not only an in-moment contingency but a sequence of non-necessarily continuous contingent moments organized in a stable pattern in space and time. Coordination appears to be a critical node that leads on the one hand to attunement and mirroring and on the other to the matching-reparation cycle and to synchrony.

This latter cycle is well known and represent the core theoretical proposal made by Tronick (e.g., Tronick and Beeghly, 2011). It means that the matching state is not meant to be the stable nor the all-time desirable state of the dyad; instead, continuous interactive ruptures and re-negotiations occur within the dyad which alternatively experience states of matching, ruptures, reparation, and matching again. Tronick has greatly described how the ability to repair interactive ruptures is one of the central dyadic processes that associates with adequate and protective caregiving (DiCorcia and Tronick, 2011) as well as effective psychotherapeutic work (Harrison and Tronick, 2011; Fonagy, 2015). Based on the present work, synchrony appears to be a higher level process in which not only the dyad navigates the process of moment-by-moment negotiation of dyadic states, but also does that in a common and joint way. Again, synchrony is also meant to be a higher level process as it is critically trans-modal: whereas matching is generally described as the simultaneous display of similar affective states (e.g. mother positive and infant positive), synchrony is not limited to perfect matches, but rather to smooth on-line changes in the agenda of each of the interactive partners.

The other cycle regards the vertical emanation of coordination reported in Figure 4. When coordination is achieved in the dyad, it is also possible to coordinate not only behavioral outputs, but also intentions and affects. This is what Stern meant by the use of the word “attunement,” referring both to the mother-infant dyad and to adult lovers. The attunement of intentions and affects also makes it easier for the member of the dyad—more often the mother—to understand what is going on between them, to anticipate the other subject's act on the basis of this intention forecasting and to mirror the reciprocal expression with exaggerations that hold the meaning: “I understand exactly how you feel.” Finally, the ability of being in “togetherness” states characterized by high levels of attunement and mirroring is also a facilitating dyadic context for the emergence of new contingent states, which makes this set of relationship a cycle which repeats in time and space.

Even though mother-infant interaction is one of the most natural processes of our life, the majority (approximately 2 out of 3) of the included studies reported lab-based researches. The most used settings include free face-to-face interactions and the stress-inducing FFSF procedure. Importantly, these processes have been observed ubiquitously at different ages, ranging from the very first week to the third year of life. Consistently with the model reported above in which attunement is meant to be one of the highest level process, this has been more studied in older children. Moreover, as mirroring is more often displayed by the mother who anticipate infants' intention by providing a meaning-making context for his/her communicative signals this has been observed even in younger infants.

The great majority of this research also focused on healthy mother-infant dyads, in which neither the infant nor the mother presented clinical complications or risk factors. Nonetheless, there is a notable proportion of studies which reported how dyadic processes occur when the infants' present specific clinical conditions such as preterm birth or risk-factors such as being a sibling of a child diagnosed with autism spectrum disorder or when the mothers present signs of psychological disturbances, including depression, anxiety, and stress.

As expected, different methods emerged for each of the seven dyadic processes abstracted here. For some of the processes (i.e., attunement and mirroring) only one measure has been reported, i.e., frequency of the specific dyadic behavioral coding. Reparation only reported two methods as well, which included the number of repairs made by the dyad and the latency to the first repair, especially when it was measured in association with the FFSF procedure. Other processes featured three or more measures, such as synchrony (correlational indexes, frequency, proportions of time, shared variance, odds ratio and conditional probability), contingency (qualitative global rates, frequencies, time series analysis, odds ratio and conditional probability), coordination (correlational indexes, proportion of time, frequencies), and matching (raw amount of time, frequencies, proportion of time, qualitative global rates, latency to match, duration of behavioral state, and odds ratio and conditional probability). Despite the majority of these measures were based on the fine-tuned and frame-by-frame or second-by-second micro-analytical coding of videotaped mother-infant interactions, qualitative/macro-analytical account of specific processes (e.g., matching) also exist. Nonetheless, as previously suggested (Cerezo et al., 2012), the micro-analytical account appears to be more consistent with the theoretical and meta-theoretical framework of the non-linear dynamic systems, despite it is more expensive in terms of human resources and time.

Research on the antecedents of dyadic processes have poorly focused on contingency and reparation. Whereas this is less surprising for contingency—as it is a very low complexity-level dyadic process—it is really unfortunate that previous studies did not provide specific information for what pertains which are the main factors associated with more effective and protective reparation processes within the mother-infant dyad during normal (Riva Crugnola et al., 2013) and stressful (Tronick, 2007) face-to-face interactions. As Tronick (2007) intended dyadic reparation as a key element of the dyadic functioning which has direct implications for behavioral and emotional health and well-being, future research is meant to dedicate specific efforts in this direction. As for the other processes, generally there is evidence that both infants' and maternal behavioral characteristics are factors associated with the display of more adaptive dyadic processes. For example, infants at different ages who play more, display more positive emotionality and freely explore the environment also experience dyads in which the amount of attunement, coordination, matching and synchrony is higher. The mother affects the quality of the dyadic encounter, as those more securely attached show more attunement and those with higher prenatal anxiety are less able to engage in synchronous interactions.

Research on the consequences of dyadic processes for the behavioral, cognitive and emotional development of infants have provided relevant information. First, the majority of the processes investigated here have multiple effects on infants' growth and development, including both cognitive (e.g., IQ is higher in infants who experienced more attuned and synchronous interactions with the mother), behavioral (e.g., greater matching is predictive of more positive and stable behavior in infants), affective (e.g., more contingency is associated with greater probability to develop a secure attachment), and biological markers of stress regulation (e.g., better reparation rates are associated with less pronounced cortisol reactivity to the FFSF procedure). In sum, all the processes have been shown to have protective and beneficial effects on both healthy and at-risk infants. Unfortunately, very little is known about the effects of engaging in dyadic interactions on maternal health and well-being. In the light of the mutuality and reciprocity features of the different theoretical models of mother-infant dyadic functioning, the lack of studying the consequences of mother-infant dyadic processes on maternal behavioral and emotional states is surprising and surely deserves more attention in future research.

This review aimed at a first attempt to organize terms, methods and knowledge on caregiver-infant research. Beyond this initial systemization, the research area is still confronted with several uncertainties due to gaps in the empirical level. First, the majority of studies on caregiver-infant systems were implemented in western societies (see Figure 3). Cross-cultural comparisons to uncover society-specific characteristics in caregiver-infant processes, thus, are impossible and consequently one major challenge of future research. Moreover and not less vital, some interactional formats have widely been neglected in caregiver-infant research. On the one hand, most of our knowledge on caregiver-infant dyads derives from studies that observe interaction processes between mothers and their infants. There are only a few studies on father-infant dyads: For example, Feldman (2003) found indications that gender differences regarding synchrony may refer to gender-specific patterns of arousal and depend on the matching between the infant and parental gender in the observed dyads. Thus, more comprehensive studies on paternal engagement and the dyadic processes in father-infant dyads would extend existing knowledge on caregiver-infant systems. On the other hand, there are very few studies on interaction processes in triadic settings.

Second, the model reported in Figure 4 is the product of a theoretical attempt of integration. Future research may use this concept as a starting point for targeted and direct examinations of reciprocal interconnections between the dyadic processes that were the focus of this framework. In other words, this model might be considered as an a priori model to be tested with original data in different populations of infants and mothers. Additionally, the suggested existence of developmental steps from contingency to mirroring warrants further investigation in infants with different ages or risk-status. In this context, little is known about the developmental antecedents of dyadic reparation processes. Thus, parental and infant characteristics that determine the reparation process need to be addressed in future investigations, especially as they represent a main feature for healthy infant development (DiCorcia and Tronick, 2011).

Third, a comprehensive systemization of the complex processes involved in caregiver-infant-dyads might lead to new insights for early therapeutic interventions that aim at transforming maladaptive interaction patterns into adaptive ones (Reck et al., 2004; Downing et al., 2014). Moreover, very little information was gathered about the effects of adaptive dyadic engagement patterns on the behavioral and emotional states as well as on long-term well-being of the caregivers. For the purpose of therapeutic approaches regarding parental mental health, this lack of knowledge has to be compensated.

Fourth, the great majority of research on dyadic processes have been conducted in laboratory setting (70% of studies). Despite many of these studies reported on well-validated and standardized procedures (e.g., the FFSF), interactions occurring in the laboratory environment may partially overlap with observations in more naturalistic contexts. More research is needed on dyadic processes in daily and home environments and to provide methodological relevant insights on the comparison of dyadic functioning in daily vs. novel environments. This kind of research would add to generalizability of the findings.

Fifth, the study of the contribution of individual dimensions (e.g. infant characteristics, such as gender: Weinberg et al., 2006; maternal characteristics, such as attachment classification: Riva Crugnola et al., 2014) appear to be under-represented among the pool of papers included in the present review. Nonetheless, the investigation of how individual dimensions contribute to the emerging dyadic processes that characterize the dyadic system is warranted to be further pursued in future research, as this field of studies may provide relevant insights on the protective factors that parents and infants bring into the dyadic functioning.

Last but not least, more research on the interconnections between macro-analytical concepts in caregiver-infant research, such as sensitivity and attachment, and micro-analytical processes is desirable. For instance, it was revealed that secure attachment may lead to more adaptive dyadic processes (Leyendecker et al., 1997) and that more adaptive dyadic processes are key to develop secure attachment (Beebe et al., 2010). Thus, the evolvement of secure attachment and adaptive interaction patterns may represent a dynamic auto-enforcing process in itself. Consistently, future investigations on the relations between macro- and micro-analytical concepts would not only connect different methodological approaches but also enhance our understanding of the dynamics in developmental trajectories.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.