We consider the models included in this section as classic, as they have served to forge a body of assumptions shared by different psychological perspectives, which constituted an important stimulus for the research on SRL (Zeidner et al., 2000; Usher and Schunk, 2018).

The proposals of Zimmerman (1989, 1990, 1994, 1998a,b, 2000) are some of the first and most widely recognised in relation to SRL. Specifically, in the Triadic Analysis of SRL, Zimmerman (1989, 1990) adopts the assumptions of the social cognitive perspective of human self-regulation and proposes multidimensionality as its essential characteristic. This model considers three sources of reciprocal influence involved in self-regulation and that should be considered in the field of education: personal (covert beliefs, such as self-perception and knowledge of one’s own regulatory processes, and affective processes), behavioural (covert and overt conduct) and environmental (physical and social context). Projection of these sources of influence on the agentic functioning of students led (Zimmerman, 1994, 1998a) to distinguish different dimensions of self-regulation related to fundamental research questions: motivational (why is the individual taking part in the learning process), methodological (how the individual approaches the learning process), temporal (when the different steps of the personal action plan are applied), behavioural (what overt conduct is initiated/modified), contextual (where the learning takes place, in terms of the physical environment) and social (who the student can and would like to count on for support during the learning process).

Finally, the cyclical phase model (Zimmerman, 1998b; Zimmerman and Moylan, 2009) delimits three recursive stages of SRL (see Figure 1):

The models developed by Boekaerts (1991, 1995, 2006, 2007) have also had an important influence in the field of educational psychology. This author focuses on the role of the motivational dynamics that drive the individual within the SRL cycle. Her adaptable learning model (see Figure 2) considers two alternative processing modes (mastery and coping), which correspond to the preponderance of one or other type of the principal motives of the student when confronted with a learning task.

The mastery mode originates in the aspiration to expand the personal repertoire of knowledge and skills and entails activation of learning strategies. On the other hand, the coping mode is brought about by the desire to preserve well-being and prevent any possible loss, damage or distortion of this state (Boekaerts, 1991, 1995) and involves activation of self-defence strategies, which may hamper learning. The balance between these modes depends on the appraisal based on an internal model of the learning situation (working model, WM), conformed according to three sources of information: (1) the characteristics of the task in question (demands and conditions in which it is presented); (2) the domain, declarative and procedural information that the student possess relative to the task, and (3) the contents of the self-system which are activated by the task (motivational values and beliefs). The appraisal may involve the perceived congruence between the value attributed to and the resources available for conducting the task, which will produce positive affective reactions and direct the student towards the mastery mode. On the other hand, appreciation of incongruencies, which may threaten personal well-being, will generate negative affective states and direct the individual towards the coping mode. The individual’s actions linked to a task may be initiated by either of these routes and then change depending on successive appraisals of the task being undertaken.

After the initial formulation of her model, Boekaerts showed increasing interest in the circumstances that determine the transition between the two alternative routes of processing and the role of the satisfaction of basic psychological needs and also in volitional processes that maintain or, where applicable, return the student to the mastery mode (see, e.g., Boekaerts and Niemivirta, 2000). The author further explored theoretical arguments in support of her view of SRL, compiled empirical evidence and finally presented her a dual-processing model (Boekaerts, 2006, 2007) (see Figure 3). This model includes a volitional self-regulation path, which ranges from the coping pathway (now denominated well-being) to the mastery pathway (now growth) and which reflects the student’s attempts to remain focused on the task, despite any obstacles or distractions that may arise. This would involve the use of different strategies aimed at controlling affective and motivational reactions associated with the difficulties found in undertaking the task.

Coinciding with publication of the previous models, Winne and Hadwin (1998) also made an influential proposal, elaborated from the perspective of information processing. In the proposal, the student’s monitoring of their own cognitive activity acquires a central role, as a key process that guides control of the activity in each of the phases of SRL (see Figure 4). Illustrative examples of the way in which monitoring is manifested in the different phases of SRL are given in the original release of the model and in later publications by Winne and colleagues (Winne and Perry, 2000; Winne, 2001, 2011; Winne and Hadwin, 2008).

The planning phase identified in Zimmermann’s model (Zimmerman, 1998b, 2000) is divided into (1) task definition and (2) goal setting and planning. The available information about the task conditions, including the specifications provided for execution (e.g., materials and knowledge required, individual or group learning method, etc.), is processed in the first phase. These appear in Figure 4 as standard (s). The student can decide to monitor this initial representation of the task, ensuring correct understanding of the demands. In the second phase, the student generates a personal profile of selected standards, i.e., a set of objectives in terms of ideal, optimal or satisfactory states to achieve, regarding behaviour, cognition or motivation. Once this profile has been established, the operations (tactics and strategies) that constitute the plan of action are activated. The plan can be monitored metacognitively (mentally tested), which may lead to redefinition of the task or of the plan of action. In the third phase (enacting tactics), the student carries out the plan, the results of which are also monitored and, in this case, checked against the personalized standards. This internal feedback will be complemented by any external feedback provided and can also lead to changes in the profile of standards and the plan of action. Optionally, on completing the task, in a fourth (adaptation) phase, the student monitors the overall way that the procedure has occurred, generating an improved representation of the task features and the best way of acting, which can later be applied to similar tasks.

The phases are implicitly included in the diagram of the model (Figure 4), in which a series of boxes that include essential components of SRL stand out. We will consider these components in the following section.

Finally, any review of the classic models must include mention of those developed by Pintrich (2000, 2004), who, also inspired by a social cognitive view, integrated both phases and areas of self-regulation, which are, respectively, placed in the columns and rows of a representative table (see Table 1). The model subdivides the performance phase into two (monitoring and control) and subsumes the 6 dimensions of self-regulation differentiated by Zimmerman in four areas (cognition, motivation/affect, behaviour and context). Pintrich also adds specifications about the personal processes and dimensions involved, based on metacognitive and motivational notions.

Regarding the metacognitive facet, in the first phase of SRL (forethought, planning and activation) the model of Pintrich (2000) includes efficacy judgements and ease of learning judgements, derived from the knowledge about the task and its context and self-knowledge in relation to the task (in the table, metacognitive knowledge). The monitoring phase includes feelings of knowing and judgements of learning, the latter being related to the fluidity of task processing and impediments that arise during its execution.

Regarding the motivational facet, the model includes self-efficacy, attributions, task value and affective reactions; the activation, monitoring and control of these are explicitly contemplated in the different phases of the model. The inclusion of self-regulation of motivational dimensions is, in fact, one of the distinctive features of the model. The so-called goal orientations deserve special mention for the special role that the author confers them in SRL. Thus, the author conceives these as general motives explaining why the student engages in academic tasks, which originate from the representations of desired results and/or states, and which condition the monitoring and control processes used during task execution (Pintrich, 2000).

Indeed, Pintrich’s interest in both the (meta) cognitive and motivational facets of learning is patent in studies published prior to the model, and it is projected in the Motivated Strategies for Learning Questionnaire (MSLQ; Pintrich et al., 1991), which is probably the questionnaire most widely used internationally to evaluate SRL. Pintrich’s model has been given visibility by its integrative nature as it explicitly includes cognitive, motivational, behavioural, and contextual dimensions and thus highlights the multidimensionality of SRL (Limón et al., 2004).

In general, the models presented below can be characterized as derivations of the classic models in four directions: from a macro (generic) to a micro (detailed/situated) focus; from cold to hot self-regulation (by the gradual weighting of affective-motivational states and their self-regulation); from conscious to implicit activity; and from individual to interindividual functioning.

The Metacognitive and Affective Model of Self-Regulated Learning (MASRL) developed by Efklides (2011, 2018) is a clear example of the first three tendencies. This model distinguishes two levels of processing, which include three types of metacognitive phenomena differentiated by Flavell (1979) (see Figure 5). Thus, the macro or Person level, comprising stable characteristics of the individual that transcend specific learning episodes, includes metacognitive knowledge. This refers to declarative information about oneself, academic tasks and learning strategies stored in memory and that form the basis of performance of academic tasks. The micro or Task x Person level includes online processing of the task, in which metacognitive experiences become important. These experiences include judgements and feelings generated during task monitoring, in the three phases of SRL contemplated in the model (task representation, cognitive processing and task performance) and the active knowledge related to the task. Finally, metacognitive skills condense procedural knowledge (represented in the macro level) and the practical application (represented at the micro level) of strategies for controlling cognition (executive processes), understood as conscious and intentionally displayed procedures.

The author considers in detail the role of metacognitive experiences, regarded as manifestation of metacognition in everyday situations, and she emphasizes the role of metacognitive feelings (e.g., confidence in carrying out a task correctly or satisfaction with having achieved an established objective). These feelings emerge unconsciously and transmit the personal relevance attributed to a particular learning task, endowing the cognitive act with affective load (pleasant or unpleasant emotions) associated with the cognitive act. The author of the model also contemplates the possibility that unconscious heuristic processes, i.e., routines established by experience with other similar tasks, also participate in SRL processes. On the other hand, Efklides (2011) draws attention to the role of metacognitive experiences in the social shaping of cognition, as well as in teaching SRL and in collaborative learning dynamics, aspects which this author studied prior to publication of the MASRL model (Salonen et al., 2005; Efklides, 2008).

The model developed by Hadwin et al. (2011), Järvelä and Hadwin (2013), and Hadwin et al. (2018) illustrates precisely the necessary social nature of SRL and differentiates three modes in which this can be manifested, in an interactive and collaborative learning environment (see Figure 6).

First, self-regulated learning, which refers to the functioning of each student separately, regarding the same task. The authors emphasize that, even in this case the SRL process is a socio-historic and environmentally situated process, in the sense that it is shaped by personal and group beliefs and experiences, by the context of the task and by the involvement, along with others, in its execution. Second, co-regulated learning, consisting of the stimulation produced by the self-regulated learning experience of another, giving rise to exchange or internalization of self-regulation processes. Third, socially shared regulation, produced when the self-regulation processes are interdependent and/or jointly constructed during episodes of cooperative learning.

We consider the motivational regulation model of Schwinger and Stiensmeier-Pelster (2012) a clear example of the gradual incorporation of hot aspects of self-regulation. This model directs our attention towards the role of the strategies that the students use to regulate their motivation, as an essential factor determining their performance. Such strategies are activated in response to the realization that the motivational state is insufficient for continuing with a task once initiated (see Figure 7).

Once the deficit has been perceived, the student deduces the cause, which may be situational (transitory) or fundamental (stable). Taking these aspects into account, the student will select which strategies of motivational self-regulation to apply and the way of doing so (either maintaining the activity or elevating the basic motivation). The efficacy of the process is assumed to depend on the student’s skill in detecting a possible deficit in motivation and adjusting the strategy accordingly. Regarding the latter, the contribution made by Schwinger’s group can be considered essential, i.e., design of the Motivational Regulation Questionnaire (MRQ; Schwinger et al., 2009), a tool that has been well received by the scientific community.

Miele and Scholer (2018) provide a more recent, detailed conceptualization of motivational self-regulation in their Metamotivational Model (see Figure 8).

In this model, the initial motivation to engage in a task is understood to be formed as a function of a specifically established objective (oval G in Figure 8; e.g., getting a good mark in an exam), which in turn depends on some type of aspiration of a higher order (oval F; e.g., performing well throughout the course). The self-efficacy and the task value are specified as motivational dimensions to be monitored and controlled (box B₁ in Figure 8), at the start and throughout execution of the task. The state of these dimensions is assumed to be bidirectionally associated with the processing mode used in the task (box B₂); this association is modulated by the predicted cost and obstacles that will occur during execution (box A). Monitoring the motivational state (metamotivational monitoring; routes between C and D) can occur in a downwards direction, when it is controlled by executive processes, such as, e.g., when the student evaluates whether their motivation is sufficient to allow a plan of action to be carried out. On the other hand, it can occur in an upwards direction when it is guided by metamotivational feelings (phenomenological experiences such as pleasure or frustration), indicative of the state of the motivational components and, when applicable, of the possible risk of abandonment or change in the initial objective. This process is sustained in metamotivational knowledge (oval H), i.e., that related to the motivational requisites of the task, the motivational self-regulation strategies and the personal ability to execute the strategies. Finally, the motivational self-regulation strategies play a key role in the metamotivational control (route from box E to B), i.e., in maintaining or increasing the level of motivation for carrying out a specific task, with an established objective. The model authors point out that metamotivational monitoring and control can proceed in a conscious or automatic manner.

Both of these modes of processing are specifically represented in the interactive layers model recently proposed by Wirth et al. (2020) (Figure 9). The authors include sensorial memory as a necessary explanatory structure, through which information from the environment and that activated by the task in the individual student enters the cognitive system. The information may be of three types (learning content, cognitive procedures and metacognitive procedures) corresponding to three simultaneous layers of processing. The authors provide a representative figure for each of the layers. The figure including the learning content layer is shown below by way of example.

If the information that the sensory memory accesses coincides with that stored in the long-term memory, resonance occurs, i.e., the coincidental information is reinforced and acquires prominence in the learning process, which can proceed unconsciously. However, if the resonance is sufficiently intense or lasting, a conscious process may occur, whereby the resonant information is intentionally processed in the short-term memory. On the other hand, the non-resonant information is discarded, except when it is sufficiently strong, in which case a search is initiated for concordant information in the long-term memory.

A common feature of all of the models described is that they all have a dynamic view of SRL, i.e., they propose a cyclical sequence of events that form a prototypical generic learning episode (Zeidner and Stoeger, 2019). Different interdependent components come into play throughout the sequence (processes, dispositions, states and environmental conditions). As an alternative mode of analyzing the nature of SRL, some authors have considered differentiating and classifying these components, proposing what have come to be known as static or componential models (Wirth and Leutner, 2008; Sitzmann and Ely, 2011), in contrast to the models described so far, distinguished as dynamic. Below we present an analysis of the treatment of SRL components in both types of proposals.

We have compiled five published classifications of SRL components, which we present in Table 2, with the aim of facilitating comparison of the similarities and differences. As can be observed, the first two classifications subdivide the components depending on whether they correspond to motivation or cognition, understood as domains of self-regulation in the approach used by Garcia and Pintrich (1994) and as regulatory systems in that used by Boekaerts (1996b). A coincident transverse organization can also be noted, which corresponds to the facets of metacognition differentiated by Flavell (1979) and specifically recognised in the dynamic models of Pintrich (2000, 2004) and Efklides (2011, 2018): metacognitive knowledge, skills and experiences. However, the last type of component is only included in the last three classifications, giving these a more situated nature than the first two.

The proposal of Winne and Hadwin (1998) deserves special mention. This model includes five componential categories (represented by the acronym COPES), which we understand are also similar to the metacognitive facets. Thus, the category conditions groups components that determine the personal representation of the task, including the available resources and restrictions, derived from the external context (e.g., instructional clues and social dynamics of the classroom) and internal conditions (e.g., prior knowledge of the learning strategies and styles); this is therefore the equivalent of metacognitive knowledge. The operations category combines the different modes of cognitive manipulation of the information (tactics and strategies) and thus corresponds to metacognitive skills. These generate products, a third category that includes the cognitive, motivational, affective and behavioural results of the operations. These products constitute new conditions for successive phases of SRL. In fact, in the dynamic model of Winne and Hadwin (1998), the phases take the names of the characteristic products generated: (1) task definition, (2) goals and plan (s), (3) study tactics and (4) adaptations. The category evaluations includes the judgements and feelings generated during execution of the task, thus coinciding with the metacognitive experiences. Finally, the category standards combines attributes, in terms of ideal, optimal and satisfactory states, which the student aspires to in the task being executed. These constitute the task objectives and serve as reference points for successive evaluations. In this respect, the standards can be considered part of metacognitive knowledge.

With the aim of evaluating the importance attributed to the SRL components and localizing their position in the cyclical sequence reflected in the dynamic models, we have elaborated a comparative table (Table 3), in which we list the components that explicitly appear in the models.

In the dynamic models, the components corresponding to metacognitive knowledge are located systematically in the phase prior to the start of task execution, except in the model of Miele and Scholer (2018). This model emphasizes that self-efficacy and the task value are modulated by metamotivational monitoring and control throughout the whole SRL cycle. The authors thus adopt a componential approach that we can qualify as “state-based,” complementary to the view of the components analysis in terms of traits, which prevail in the models.

The components related to metacognitive skills are linked to execution of the task, although in the model of Miele and Scholer (2018) motivation is controlled from the start to the end of the learning episode.

The components related to metacognitive experiences reflect the monitoring process that takes place during task execution. However, in the model of Miele and Scholer (2018), this process extends to the phase(s) prior to task execution, while those of Zimmerman and Moylan (2009) and Winne and Hadwin (1998) also explicitly includes self-evaluation after finalization of the task.

Finally, we can see that the dynamic models delimit SRL processes and states. However, the degree of detail varies depending on the metacognitive facets. Thus, metacognitive and metamotivational knowledge and beliefs are very detailed (particularly the latter). The learning strategies are generally referred to in a global way in the dynamic models, with the exceptions of the detailed cognitive strategies included in the models of Zimmerman and Moylan (2009) and Wirth et al. (2020), of resource management strategies in the model of Zimmerman (1998a) and of the motivational strategies in the model of Miele and Scholer (2018). Metacognitive experiences are also not detailed, although the model of Pintrich (2000) refers to metocognitive judgements and feelings, and the model of Miele and Scholer (2018) includes metamotivational feelings.