Conceptual and methodological approaches to the study of Self-Regulation: An inquiry within Developmental Science

Abstract


Introduction
At the beginning of the present millennium, Posner & Rothbart (2000;2007) identified the comprehensive study of self-regulation (SR) as a mayor objective for advancing an understanding of development and psychopathology.Nearly two decades later, efforts made towards this goal by researchers stepping from multiple disciplinary backgrounds have condensed into a considerable body of published works, which grows more voluminous with each passing year.
In a broad sense, SR stands for the ability to direct one's thoughts, attention, behavior and emotions in order to achieve goals (Vohs & Baumeister, 2016).The capacity to exercise control over motor, emotional and other cognitive processes develops as a result of complex interactions between an individual's genetic endowment and its life experiences (Bell & Deater-Deckard, 2007;Blair & Raver, 2012).Developmental science focuses in describing, explaining and optimizing intraindividual changes in adaptive developmental regulations, as well as the interindividual differences in such relations, across life (Lerner, 2012).Since developmental science is especially compelled to understand phenomena characterized by adaptive coregulations among individuals and their environments, it follows that interest in processrelational constructs such as SR should be significant.The use of the term process (i.e., the ordered succession of acts), in the context of developmental science reflects the adoption of an epistemological stance, in which the fundamental categories employed for the analysis of developmental phenomena are change and process, as opposed to stasis and substance which derive from classic Cartesian reductionism (Overton, 2015).In this process-relational view, the historical and physiological events that constitute developmental change for either a single organism or a whole community are conceived as a a set of complex transformational processes through which fundamental features of the dynamic systems relate, differentiate and move towards dialectic integration (Sameroff, 2010).Therefore, from this perspective, SR (in all of its conceptual variants) will be referred to as a process or as series of interdependent processes.
SR has been pointed out as an important aspect of people´s behavioral adaptation to the demands of their social and physical environments, by means of different emotional skills, the ability to delay gratification and to comply with social norms, goal-directed cognitive control, linguistic and mathematical reasoning, among other competences that appear to be relevant for academic performance and optimization in multiple areas of functioning, across cultures (Eisenberg & Sulik, 2012;Lerner, 2018).Still, there remains much debate on how to define and measure SR.Some authors consider that scientific literature in this area lacks theoretical order, and that disagreements in how to define self-regulatory constructs generates confusion that hinders the coherent integration of this body of knowledge (Cervone, Shadel, Smith, & Fiori, 2006).In this sense, it is important to mention that systematic review-based efforts have already been put into action by developmental psychopathologists in order to assess conceptual relations between different theoretical constructs (Nigg, 2017).In addition, an interesting metaanalysis on the convergent validity of heterogenous self-control measures was published almost ten years ago (Duckworth & Kern, 2011).
What exactly is SR it and how do we measure it?The need to tackle these issues constitute the rationale of the present work.Discerning the latest conceptual and methodological advancements in this matter would enable the reassessment of how developmental scientists study SR, and thus achieve a better understanding of how societies eventually could benefit from their insights.Contemporary research paradigms in developmental science have been shaped extensively by philosophical debates on the concept of agency, systems theory and by technical breakthroughs in the field of neurobiology (Lerner, 2018).The identification of several mechanisms involved in neural plasticity, stress physiology and the development of attentional networks has sparked support for the integration of cognitive and temperamental psychobiological models (Rueda, Posner, & Rothbart, 2005;Williams et al., 2008).While perspectives on temperament and personality tend to emphasize the relative rank-order stability of self-regulatory traits over time and across contexts, cognitive frameworks usually underscore the idea that SR encompasses adaptive processes and that the executive function components can be trained and influenced by contextual factors (Bornstein et al., 2015).During infancy and early childhood, children gradually acquire the SR skills and strategies necessary to cope with a variety of developmental challenges, thus gaining autonomy, control and separate identities from their caregivers (Calkins & Fox, 2002;Posner & Rothbart, 2007).
Further on, researchers usually employ external behavioral measures as proxies to study internal regulatory processes.Coding systems for direct observation, controlled performance tasks, self-reports, as well as caregiver and teacher questionnaires are the most common tools found in basic and applied research.However, each type of instrument has often been designed for specific purposes and may carry inherent strengths and limitations (Duckworth & Yeager, 2015;McCoy, 2019).For instance, scales designed for clinical or experimental settings are often considered limited in validity for studying how behavioral and emotional aspects of SR unfold in ecological contexts like school, work and other "real-world" social environments (McClelland, Ponitz, Messersmith, & Tominey 2010;McCoy, 2019).In addition, many questionnaires are vulnerable to reference and social desirability biases and thus are not recommended for program evaluation, inter-school accountability or individual diagnosis (Duckworth & Yeager, 2015).Although we do not intend to judge on the appropriateness of such methods, we consider it relevant to contextualize the goal of our work in the light of current debates that permeate this area of study.

Objectives
The purpose of this review was to delve into contemporary developmental research and discern the latest theorical, technical and methodological advancements in the conceptualization and assessment of SR.In order to do so in an exhaustive and replicable manner, empiric research and review articles published in the current century were considered.We sought answers to the following questions: 1.

Study design
Scientific articles on the typical development of SR in human populations, published either in English or Spanish by peer reviewed journals during the 2000-2018 period, constitute the main focus of this review.This decision is grounded in the exponential growth in the number of publications that has taken place in the fields of psychology and developmental cognitive neuroscience in the last two decades, which is related to technical and methodological advances in both disciplines and to the emerging theorical perspectives for the integrated study of cognition and emotion.

Search strategy, data sources and extraction
Advanced search engine parameters were adjusted using keywords: self-regulation OR selfcontrol; and incorporated filters: peer reviewed journals AND in humans.This choice of keywords is based in the interchangeable use that is often made out of these terms (Maranges & Baumeister, 2017;Vohs & Baumeister, 2016).Although a wide variety of terms associated to executive functioning, temperament and personality could have been chosen for this query, the lack of recursivity among these theorical frameworks would have proven an obstacle for the straight forward identification of relevant literature, as authors do not always align these constructs as stemming from the same group of processes, nor make reference to over-arching concepts like SR or self-control.Data bases included in the EBSCOhost Psychology and Behavioral Sciences Collection and PubMed Biomed Central search engines were consulted, as these contain articles published in journals from the humanities, health and social sciences.
Data extraction and organization into spreadsheets was undertaken for each article that met all criteria to be selected for the review.

Data analysis
Data analysis consisted in the following mixedmethod procedures: (1) Frequency analysis of previously enumerated variables using R, SPSS 15.0 and MS Excel 2016; and (2) Qualitative analysis of semantic categories employed in the conceptualization of SR utilizing Atlas.ti7.5.18software and following Grounded Theory guidelines (Justicia & Padilla, 2011;Strauss & Corbin, 1998).Atlas.tiaided in (a) data storing and organization by means of a manageable digital file (i.e., hermeneutic unit); (b) segmenting text fragments in which conceptual definitions or descriptions of SR were listed; (c) coding information; (d) establishing relations between assigned labels; and (e) generating graphical models for the representation of terms that were found in greater association with conceptual definitions of SR.
For this last step, the Word Count tool from Atlas.ti was used over collected text fragments.After filtering personal names, articles, pronouns, prepositions, interjections and redundant words, the most frequent key terms employed in the conceptual definition of SR were identified.A separate analysis was carried out to identify top quoted authors in these fragments.The outcome of the first word count was also used to generate categories that correspond to different aspects of self-regulatory processes.They were then applied to inquire further into the heterogeneity of conceptual definitions in the article sample, by relating one or more category to each corresponding paper and then making comparisons between them based on other variables that distinguish sub-groups of articles (discipline, subdiscipline and age range of the samples).
The purpose of carrying on an open coding process, also known as microanalysis (Strauss & Corbin, 1998) was to complement the word count approach.Different codes were chosen to represent relevant conceptual features and theoretical frameworks that are usually encountered in self-regulation researc, guided by the word count results and the authors' understanding of different research lines that permeate the field (Vohs & Baumeister, 2016), These codes were then used to label all the definitions in the sample, thus making it possible to pinpoint which theoretical features had a stronger presence in them.For more information on grounded theory and qualitative research methods, see Strauss & Corbin, 1998).

Study selection and characteristics
Search results in February 2018 allowed the initial identification of 5296 articles, , which were then screened for the following exclusion criteria: (a) duplicate results; (b) unavailable articles; (c) unrelated articles; (d) studies using sampling approaches that correspond to clinical, psychopathological, deficit or at risk populations (i.e.; socioeconomic risk, poverty, ADHD, cerebral paralysis, dyslexia, physical abuse, cognitive disfunction, autism, perinatal risk, mood disorders, premature birth, low weight at birth, language development delay, and drug abuse); (e) studies that do not focus on domain general aspects of SR, but on domain specific ones (i.e., selfregulated learning, SR of feeding behavior; SR in sexual relationships; SR of health behavior; neurofeedback, SR in sports, SR in meditation; SR in consumer behavior, SR in driving; racism and stigmatization, quality control, substance abuse, SR of wake; SR of work behavior, etc.); and (f) interventions.The final sample was composed by a selection of 142 reviews and 326 original research articles retrieved from 170 different journals.Figure 1 outlines these results by showing the flow of information throughout the different phases implemented in this review, based on the PRISMA guidelines (Moher et al., 2009).
When considering the claim that SR is currently an expanding area of study, this review found in confirmation that the number of published articles since 2010 doubles the amount corresponding to the 2000-2009 period (Figure 2).Populations studied in original research articles cover 40 different countries (Figure 3).Despite this great diversity, studies from North America hold a majority over the rest of the world (57%) and transcultural studies are relatively few (7%).
In terms of researched age groups, it was found that the whole life course is contemplated in these empirical articles, but that studies involving only adult participants are the most numerous (50%) (Figure 4).Research articles reviewed here belong to four different disciplinary backgrounds: psychology, cognitive neuroscience, health sciences and education sciences (Table 1).Psychology is the discipline from which more studies were originated (~81%).

Defining self-regulation from a developmental perspective
The present work's proposal is to generate a conceptual synthesis from the articles reviewed and to analyze the variability in the definitions offered by different article subgroups.Figure 5 shows the semantic composition of SR, in relation to the terms that are more frequently used to define it.Following this result, eight different categories were generated to qualitatively compare definitions across articles.Although they represent diverse features of self-regulatory processes, because they derive from different lines of research, this does not imply that they are mutually exclusive or that they cannot conceptually overlap: Cognition.It refers to the facet of SR related to cognitive control and those functions involved in information processing and instrumental reasoning, such as working memory, inhibitory control and attentional flexibility (Miyake & Friedman, 2012).Theories on Executive Functions are the main framework for this category.
Emotion.Emotional SR stands for the ability to initiate, inhibit, maintain, modulate and change the occurrence of internal emotional processes and their concomitant behaviors, like facial expressions and gestures (Eisenberg et al., 2000;Gross & Thompson, 2007).
Temperament.Such studies focus on constitutionally based individual differences in reactivity and SR, in the domains of affect, activity, and attention (Rothbart & Bates, 2006).
Goal-oriented activity.It refers to the motivational dimension of SR included in the management of behavior according to goals or objectives.This approach is distinctive of cybernetic models of SR, where feedback circuits play an essential role in the regulation of behavior (Carver & Scheier, 2000).
Interpersonal factors.It corresponds with the view that links self-regulated processes to psychosocial phenomena, as social interactions both shape the development of SR competences and become targets for self-regulated action (McClelland et al., 2010).
Conscious effort.This outlook distinguishes the volitional side of self-regulated processes from other automatic or more reactive forms of regulation, as intentional and conscious effort are displayed in order to resolve control demands that arise from the internal and external environments (Gestsdóttir & Lerner, 2007;Kaplan, 2017).
Gene-environment interaction.It deals with views that highlight bidirectional influences among individuals and their environment (Geldhof & Little, 2011).This category considers how genetic variability and environmental factors interact to shape a phenotype; as for example, the influence that certain parenting practices have over the expression of different alleles corresponding to genes that regulate the development of temperament and attention (Sheese, Voelker, Rothbart, & Posner, 2007).

Resource allocation
This standpoint relates SR with the ability to distribute finite resources for the attainment of multiple tasks.Studies in this category utilize willpower or cognitive load models to explain ego depletion phenomena and individual differences in task performance requiring sustained self-control (Kelley, Wagner, & Heatherton, 2015;Vohs & Baumeister, 2016).
Figure 6 shows how these categories are averagely represented in percentage values by definitions in the sample of articles.It could be synthesized that most articles reviewed here define SR in terms of conscious, goal-oriented control of cognitive and emotional aspects of behavior.There seems to be a consensus that SR refers to the ability to activate, monitor and inhibit behavior, attention, emotion and other cognitive processes in a flexible and adaptive manner, in response to internal or external stimuli and for the achievement of desired purposes (Dias, del Castillo, & Moilanen, 2014).Note.This figure shows the percentual distribution of conceptual labels which were used to tag each of the 468 studies included in the review.Each paper could be tagged with up to nine different labels.For example, Emotion, the most frequently assigned label, was present in over 78.8% of the studies; and Resource allocation, the least frequently assigned label, only showed up in 25% of the articles reviewed.
When different groups of articles are observed separately, it is possible to appreciate that conceptual definitions of SR can vary according to an article's disciplinary background or to the age of the sample which was studied (in the case of empirical studies).Definitions from cognitive neuroscience and psychophysiological studies more frequently mention the incidence of physiological factors in SR phenomena.In turn, developmental psychology, health and education sciences studies more often name emotional and interpersonal factors.Temperament is a more commonly used category in infant and toddler studies, while mentions of goal-oriented activity and cognition increase relatively with the age range of the sample of articles included in the review.
As a final note to this section, the most influential authors in the conceptualization of SR were determined by the number of citations each author received in text fragments containing definitions.Top cited researchers include Mary Rothbart (380), Roy Baumeister (350), Michael Posner (185), Nancy Eisenberg (161) and Grazyna Kochanska (141).All of them are psychologists based in the United States.

Instruments, indicators, temporal design, levels of organization and contexts of evaluation included in SR research
Instruments used to collect indicators at the behavioral level of organization, like report measures, performance tasks and structured observations, constitute around 92 percent of the total assessments registered.The remaining 8% is almost evenly distributed between measurements made at the physiological, neuroanatomical and molecular-genetic levels.
It was found that 14% of original research articles included in this review contemplate multiple contexts in their assessments (i.e., school, household, community centers).However, evaluations that take place only in laboratoy settings represent the most frequent scenario in SR research (54%).
In addition, longitudinal research (i.e., studies where measurements are made at different time intervals) accounts for less than a third part of all the original research articles reviewed.In other words, cross-sectional design is still prevalent in this field of study based on the reviewed literature.
Also, the distribution of the type of instuments used for data collection was found to be the following: Self-report and other report measures together comprise 40% of total assessments, followed by performance tasks (25%), other laboratory tasks (21%), structured observations (6%), physiological measurements (6%), molecular-genetic analyses (1%) and neuroimaging techniques (1%).In Table 2, the names of the most frequently used measures in empiric research articles are shown.

Summary of main findings
Since the number of published articles on SR is rising, and since studies usually approach the subject from different theoretical and disciplinary frameworks, defining SR can often result problematic (Bridgett, Oddi, Laake, Murdock, & Bachmann, 2013).Some authors consider that scientific literature in this area lacks theoretical order, and that disagreements in how to define self-regulatory constructs generates confusion that hinders the coherent integration of this body of knowledge (Cervone et al., 2006).It has been noted that some works offer vague conceptualizations, with insufficient clarity or tautological statements (Burman, Green, & Shanker, 2015).In fact, the present study found that over a 30% of the articles reviewed did not define SR explicitly, but that readers are often left to make their own inferences or forced to interpret the meaning of the concept from the context in which it is used or from the presence of associated constructs.This finding is not incidental and has been reported previously (Barkley, 2001).
The results from the semantic analysis of text fragments that conceptualize SR can be summarized by stating that aspects which generally have a stronger presence in the conceptual definition of the construct are emotionality, cognition, goal-oriented activity, conscious effort, interpersonal and physiological factors.There seems to be a consensus that SR refers to the ability to activate, monitor and inhibit behavior, attention, emotion and other cognitive processes in a flexible and adaptive manner, in response to internal or external stimuli and for the achievement of desired purposes (Dias et al., 2014).It was found that the way in which researchers operationalize SR can vary according to their disciplinary background, theoretical framework, or even in relation to the characteristics of the population they study.In other words, the description of self-regulatory processes that a researcher chooses can be biased by grounding disciplinary practices and also by the developmental stage which is being studied.Factors from the socio-cultural context can also shape research practices, acting like ideological frameworks (Kuhn, 1962;Latour & Woolgar, 1979;Lerner, 2018).
Growing interdisciplinary collaboration in the study of SR holds a hopeful prospect for unified definitions.
Developmental scientists (i.e., psychologists, cognitive neuroscientists, health and educational practitioners) tend to utilize SR to designate processes involved in the development of conscious self-control over intentional activity, which includes cognitive, emotional, physiological and social phenomena that can be influenced across the life span by different biological and contextual factors.
Furthermore, original research articles contemplate SR phenomena at different levels of organization, including physiological, attentional, emotional, other cognitive and interpersonal functioning domains (Calkins & Fox, 2002).Many of these studies have paid attention to the everyday settings in which people develop and interact with each other, in their own houses or at educational institutions, for instance.However, this conception is not always followed by the use of appropriate instruments, methods and design to capture SR phenomena in a naturalistic or ecological manner.Despite this diversity in assessment contexts, the laboratory setting continues to be the most frequent scenario for SR research.
In the last decades, developmental theories have strived to leave behind simplistic approaches and conceptual reductionism, by taking into account the interactions that occur over time among different systems involved in human development (Lerner, 2006;2018;Posner & Rothbart, 2007).A shift into interdisciplinary perspectives is necessary to achieve real progress, with the incorporation of methods emerging from computing sciences, economic sciences, neuroscience, molecular biology, sociology, statistics and psychology.
Another of our objectives was to understand how researchers operationalize SR.This crucial phase in research design, in which variables or indicators are defined for their effective observation and measurement, allows phenomena to be approached empirically and quantitively (Hernández Sampieri, Fernández Collado, & Baptista Lucio, 1991).The detailed inspection of methodological sections in empirical studies offers a general overview on how SR has been operationalized in recent years.While it is possible to appreciate the inclusion of genome sequencing techniques and physiological measures, like vagal autonomic nervous system activity, salivary cortisol concentrations and neuroimaging (i.e., FMRI, EEG, PET), their use is not widespread (Bell & Deater-Deckard, 2007;Sheese et al., 2007).The behavioral level of organization continues to be the focus of descriptions on cognitive and emotional control.Report scales, cognitive performance tasks and other laboratorybased procedures are among the main instruments implemented during empiric assessments of SR.Physiological and structured observational measurements in other than laboratory settings are scarse.
Both experimental and correlational methods are used in the study of SR (Eisenberg, Duckworth, Spinrad & Valiente, 2014).The present work focused in the temporal aspect of research design, because different reviews highlight the importance of chronologic sensitivity in the study of development and SR (Lerner, 2006;McClelland et al., 2010).Findings show that twothirds of original research articles employed crosssectional designs, but an increase in the number of longitudinal studies could be expected in the upcoming years.Many authors that make use of cross-sectional designs remark the need to adopt longitudinal approaches as an imortant future direction.In addition, while the study of how SR develop takes into account the whole life cycle, our results show that adult samples are overrepresented.The study of SR during infancy, childhood and adolescence constitutes a crucial target for future research.
Despite SR has been studied in different cultures, the great majority of articles analyzed here have worked with North American samples.This means that not every conclusion to which these studies arrive to are generalizable to other cultural contexts, because variability in sociocultural patterns, like socialization practices, condition the development of SR processes (Liu et al., 2017).Efforts to tackle the study of SR and its development in local cultures through original research proposals, scale adaptations and replication studies, continues to be an imperant necesity for public policy planning.
Concerning the debate on whether SR should be considered a number of fixed, contextindependent traits or a group of skills that can be improved, the present scientific consensus holds that despite the construct appears to show relatively rank-order stable features over time in the absence of exogenous forces (i.e., intentional intervention, life events, changes in social roles), evidence is at odds with the connotation of immutability, because it has been demostrated that such personal qualities are potentially responsive to intervention and dependent on situational factors for their expression (Duckworth & Yeager, 2015).As for the problem of increasing reliability and validity in the assessment of SR, perfectly unbiased, unfakeable, and error-free measures do not exist at the present time, so multimethod, application-dependent approaches to measurement are desireable in order to transcend the limitations and exploit the strenghts of individual tools (Duckworth & Kern, 2011;Duckworth & Yeager, 2015;McCoy, 2019).

Limitations
The purpose of this review was to identify contemporary conceptualizations of SR in normative developmental research, including theoretical definitions and methods used to assess it.This reduction constitutes the scope of our work and it necessarily leaves out contributions form clinical research, interventions programs and domain specific studies on SR, like self-regulated learning or SR of health behavior.Although such studies definitely provide different and valuable views on SR research, the amount of published works on these issues alone would justify that they are reviewed separately.
It is possible that the databases that were consulted did not index all existing articles on the normal development of SR.In such case, relevant studies that could modify the results of our analysis may have been left out, unintentionally.Because of this, the conclusions to which this review has arrived must be contemplated within the limits of our sample.Furthermore, the restrained or simplified use of keywords in our search strategy can be considered another important limitation, because the number of articles associated to constructs like executive functions, temperament and personality (which are intertwined in the theorization of SR) greatly exceed those that are properly labeled or addressed as part of the SR literature.
Another limitation is that despite abidance to PRISMA criteria and the sole consideration of peer reviewed journals, in order to guarantee some methodological rigor, the authors cannot completely control the publication bias and therefore cannot guarantee full access to the data in the realm of this review.
Comparing the strength and suitability of different instruments for the assessment of SR requires a meta-analytical approach that is beyond the scope of this review and it is therefore an issue that should be addressed by future studies (readers interested in this discussion are advised to look into the following essays: Duckworth & Kern, 2011;Duckworth & Yeager, 2015;McCoy, 2019).

Conclusions
Developmental science has seen profuse conceptual, methodological and technical advancement in the study of SR over past decades.Current efforts in this interdisciplinary field seek to implement research designs that reflect a longitudinal, multidimensional and ecological view of self-regulatory processes, through diverse, age-appropriate, reliable and valid measures, that are sensitive to cultural variability and contextual influences (Bell & Deater-Deckard, 2007;Bridgett, Burt, Edwards, & Deater-Deckard, 2015;McClelland et al., 2010).The main aspect of SR that is usually emphazised in its definition is that multiple general domain processes are involved in the development of conscious self-control over intentional activity.SR processes include interlocking cognitive, emotional, physiological and social phenomena that can be influenced across the life span by different biological and contextual factors.
However, there is still not sufficient agreement on the specific pathways that coact in the development of SR, nor on the exact degree in which SR processes can be altered for the benefit of individuals and their communities.These will probably continue to be important isssues during the next decades of research.

Figure 1 .
Figure 1.Flow diagram of the studies retrieved for the review.

Figure 2 .
Figure 2. Number of articles published yearly

Figure 4 .
Figure 4. Distribution of empirical studies in the sample according to different age groups.

Figure 5 .
Figure 5. Cloud Word corresponding to the most frequent terms used in the definition of SR.Note.Words that appear more frequently in the definitions look relatively bigger and darker.The location of words in the cloud is random, except for the central word (emotion) which has the highest number of occurrences.

Figure 6 .
Figure 6.Conceptual profile of an average SR definition in the reviewed articles.Note.This figure shows the percentual distribution of conceptual labels which were used to tag each of the 468 studies included in the review.Each paper could be tagged with up to nine different labels.For example, Emotion, the most frequently assigned label, was present in over 78.8% of the studies; and Resource allocation, the least frequently assigned label, only showed up in 25% of the articles reviewed.

Table 1 .
Percentage distribution of empirical studies in the sample according to different disciplines.

Table 2 .
List of the top 30 instruments utilized in empirical studies.

Table 3 .
List with indicators associated to the operational definition of SR in empirical research articles