Keywords

1 Introduction

This paper urges an integration of findings from the field of Cognitive Science with the rich literature on self-regulated learning from the fields of Educational and Industrial/Organizational Psychology, respectively. Specifically, we will examine the implications of dual modes of cognitive processing for two important components of self-regulated learning: metacognition and motivation. We believe that these three domains of study will mutually benefit from cross-fertilization of this sort, and hope that other scholars will join us in this endeavor. The authors are two Industrial/Organizational Psychologists (JB and TB), a field from which much research on self-regulated learning has emerged, and a Cognitive Scientist (WS), the field which spawned the dual-system theory of cognition. As we hope the reader of this article will discover, we have identified several interesting and important points of intersection among these disciplines.

2 Four Kinds of Metacognition

A key finding from Cognitive Science is that the human brain processes information in two ways. In [1] they are called Type 1 processes—fast, effortless, non-conscious, and often automatic—and Type 2 processes—slower, requiring cognitive effort, and available for conscious introspection.Footnote 1 Type 1 processes are sometimes described as pattern recognition, intuition, or “gut feel”, and Type 2 processes are sometimes known as analytic or critical thinking. [2] provides an explanation of a variety of cognitive heuristics and biases in terms of dual processes. [3] shows that dual processes can help reconcile the apparent conflict between naturalistic decision making (where Type 1 processes are strongly valued because of their role in expert performance) and probabilistic thinking (where Type 1 processes are viewed with skepticism because they sometimes result in logical errors.)

A second finding from Cognitive Science is that expertise develops as a function of deliberate practice [4, 5]. This is true for learners on their way to being experts, as well. The “deliberate” part of deliberate practice, of course implies that learners – often working with a coach or mentor – choose the nature of the practice they engage in, and therefore it implies a degree of self-regulation. This kind of self-regulated practice is key to the development of expertise across a range of disciplines including music, dance, sports, martial arts, and chess among others.

Deliberate practice has interesting and differential effects on these two cognitive processes. Knowledge, skills, and aptitudes (KSAs) start out as declarative information, typically learned in didactic environments such as a classroom, and are learned via Type 2 processes. As learners gain domain-specific experience, though, the KSAs become “chunked” and “compiled” into more efficient Type 1 processes [6,7,8].

Expert-like performance requires several years of sustained deliberate practice, which involves: skills practice that is specifically targeted toward one’s greatest areas of weakness, diagnostic performance feedback, deliberate self-reflection on that feedback, and strategically-timed periods of recuperation [5]. During this extended period of training, learners set specific, difficult goals for improving their future performance, such as running a mile in under 8 min, or completing a half-marathon within 2 h. Goals are critical to any theory of self-regulated learning because they direct the learner’s attention, increase task-related effort, and help the learner to persist in the face of adversity [10]. Along with increased skills, the learner’s self-confidence invariably increases with expertise. This further helps to motivate the learner during the long, difficult, and sometimes painful process of deliberate practice [11]. Therefore, any discussion of self-regulated learning must also include elements of motivation as well.

Type 1 and Type 2 cognitive processes are not mutually exclusive; they often work together. For example, Type 2 processes often prime the Type 1 processes by providing critical contextual information that is needed for successful task performance. For example, baseball players have very limited time to decide where and how to swing at a pitched ball [12], so they rely on Type 1 processing to make that decision. But that decision is heavily influenced by game context—the score, the stage of the game, the known style of the pitcher—which is provided by Type 2 processes.

Figure 1 shows a notional overview of the development of Type 1 and Type 2 cognitive processing with increased domain experience.

Fig. 1.
figure 1

Notional overview of the development of Type 1 and Type 2 processing as a function of the accumulated amount of deliberate practice.

Full size image

The process of thinking about one’s thinking is generally referred to as metacognition [9], and is a critical hallmark of domain expertise. The challenge is that Type 1 processes are generally not available for introspection. They occur without the learner’s conscious knowledge and often involve automatic pattern recognition. This gives rise to four distinct kinds of metacognition, as shown in Table 1. In the upper left hand quadrant (Type 1/Type 1) might be a fire ground commander who immediately recognizes that the fire has spread into the structure’s walls and knows instinctively that he can recognize this category of patterns, and orders all personnel out of the building. In the lower left quadrant (Type 2/Type 1), the fire ground commander realizes that he has never seen this type of fire behavior before, and begins sequentially reviewing alternatives to find one that fits the current circumstances. In the upper right hand quadrant (Type 1/Type 2), a student realizes that he has been reading the same sentence repeatedly, but has not been able to comprehend the content. He therefore decides to put the book down and return to it after a break. In the lower right hand quadrant (Type 2/Type 2), a student identifies a gap in his understanding and deliberately seeks out the instructor’s assistance to help resolve the gap.

Table 1 Four kinds of metacognition. Metacognition has a process type (how is the person thinking?) and an object (what is the person thinking about?), either of which could be Type 1 or Type 2 thinking.
Full size table

Metacognition using a Type 1 process, that is, automatically recognizing patterns in either other pattern recognition domains or in declarative knowledge and critical thinking domains, is currently a kind of black box, not only because Type 1 processes are not available for introspection, but also because they have not been widely discussed. [13] discuss metacognition in the context of dual-process cognition, but they focused only on the second column in Table 1 (that is, they considered Type 1 and Type 2 thinking about Type 2 thinking only). They made the astute observation (in terms of Table 1) that only Type 2 thoughts can be communicated to others, and therefore they viewed Type 1 metacognition as useful for intra-personal cognitive control, and Type 2 metacognition as useful for shared supra-personal cognitive control. The analysis in this paper extends their work by explicitly acknowledging that accumulated amount of deliberate practice leads to an increased role for Type 1 processing in the domain being learned.

In any case, both cells in Column 1 provide good fodder for future research, because there are well-developed methodologies for investigating rapid pattern recognition (for example, temporal occlusion [14,15,16] and drift diffusion modeling [17,18,19] to name two of many.)

Metacognition involving Type 1 processes is thus necessarily indirect, more so as the learner gains proficiency and more Type 1 processes come into play. Since metacognition is one component of self-regulated learning (along with motivation, as in [20, 21]), the nature of the metacognitive component of self-regulated learning must therefore evolve with increasing amounts of domain-specific deliberate practice.

2.1 Guide to the Rest of the Paper

In the remainder of this paper, we will discuss self-regulated learning from a dual-process perspective in three contexts: in the classroom, in the workplace, and in learning an advanced perceptual-motor-cognitive skill, namely landing a jet on an aircraft carrier. We believe those environments provide a good sample of real-world human learning situations in which self-regulated learning comes into play. After those discussions, we conclude with recommendations for future research.

3 Self-regulated Learning in the Classroom from a Dual-Process Perspective

In the field of Educational Psychology, much of the research on self-regulated learning emanates from two distinct schools of thought. The first focuses on the learning of specific domain knowledge, such as learning specific math or science concepts. The second focuses on generalized principles of “learning to learn”. The discussion that appears below draws heavily from this second school of thought.

3.1 Metacognition in the Classroom

Because learning new declarative knowledge is a Type 2 cognitive process, metacognitive research in the classroom has focused entirely on the right-hand column of Table 1. The upper right-hand quadrant depicts what happens when a learner uses pattern-recognition (Type 1) processes to dynamically monitor and regulate the learning process in real time. For example, while studying for a test, the learner may recognize that she has just re-read the same sentence 3 times, but still does not understand the concept. As a result, the learner puts the book down, takes a break, and returns to the material with a fresh perspective. Similarly, while taking an exam, the learner may come to a test question for which they do not know the answer. Recognizing that the exam is timed, the learner skips that question in the hopes that answering the subsequent test questions will cue recall of the forgotten information.

The lower right-hand quadrant depicts what happens when a learner uses deliberate or analytic (Type 2) processes to engage in metacognitive planning. For example, before reading a textbook chapter, the learner may: actively seek out a quiet place to study; skim the chapter headings beforehand to identify key concepts, and; review the end-of-chapter self-assessment questions before reading the chapter text. Similarly, while reading the chapter text, the learner may identify a critical gap in her understanding of the material and then actively seek out the instructor’s assistance to help resolve that gap.

Research on metacognition has identified three inter-related components: metacognitive planning, self-monitoring, and self-regulation [22, 23]. The first component, metacognitive planning, appears to be a Type 2 process. It includes behaviors such as setting goals for studying, skimming the chapter headings to identify key concepts, and reviewing the end-of-chapter assessment questions before reading the chapter text. It is proactive and future-oriented, rather than reactive and present-oriented. The last two components, self-monitoring and self-regulation, appear to be Type 1 processes. Examples of self-monitoring include: monitoring one’s attention in real time, self-testing to assess one’s comprehension of the material, and using “smart” test-taking strategies such as process-of-elimination to rule out potential incorrect responses. Examples of self-regulation include: modulating one’s reading speed as a function of the material difficulty, re-reading key text passages to improve comprehension, and resolving perceived inconstancies by systematically comparing one’s class notes to the chapter text [22, 23].

While they are theoretically distinct, in practice these three metacognitive components are highly inter-correlated [24, 25]. As a result, they are often measured together.

3.2 Motivation in the Classroom

The field of Educational Psychology has also made a number of important contributions to the study of motivation. Perhaps one of the most critical is recognizing the dynamic interplay between “cold” cognition and “hot” motivation [24]. Simply put, learners are not machines. Their feelings and emotions influence their ability to self-regulate the learning process. Another is the field’s conceptualization of the learner as an active information processor whose feelings and beliefs can be contextually-activated [24]. While the student may have learned though extended experience certain automatic dispositions that cause them to respond in a particular manner in ambiguous situations (Type 1 processes), those characteristics can be over-ruled by the classroom “culture” (Type 2 processes). As a result, highly-skilled instructors systematically modify the classroom environment to their advantage [26], especially when teaching difficult courses like mathematics. In the following paragraphs, we describe two motivational constructs—goal orientation and self-efficacy— both of which appear to have both automatic (Type 1) and situationally malleable (Type 2) components.

The term “goal orientation” refers to one’s goal-related preferences in achievement-related situations. Research suggests that there are three different, although non-exclusive, types of goal orientation [27]. Individuals with a strong “Learning Goal Orientation” (LGO) approach new situations as an opportunity to learn, grow, or develop themselves. As a result, they interpret failure as a learning opportunity, and welcome it. Individuals with a strong “Performance Goal Orientation” (PGO) approach new situations as an opportunity to demonstrate their skills and abilities to others. In essence, it is an opportunity to “show off” in front of one’s peers. Such individuals tend to interpret the course grade or exam score as an end in itself; they tend to be ambivalent to performance feedback. Finally, individuals with a strong “Fear of Failure” (FOF) approach new situations as potential opportunities to demonstrate their lack of knowledge or skills to their peers. As a result, they try to avoid receiving negative performance feedback. They often do this by choosing easy tasks (because doing so guarantees success) and choosing tried-and-true methods (which have worked well in the past) rather than trying new methods. Previous research suggests that goal orientation has some automatic (Type 1) characteristics; however, it can be situationally induced (Type 2). For example, an instructor can deliberately induce a LGO classroom culture by: teaching the learners effective goal-setting strategies; involving students in classroom decision making process; making grades contingent on improvements vis-à-vis one’s prior performance, or; rewarding learners when they try new approaches regardless of their performance outcomes [26, 28].

The term “self-efficacy” refers to a learner’s self-confidence in their ability to successfully perform a specific learning-related task [11]. Previous research shows that self-efficacy influences the goals that learners set for themselves, how much effort they mobilize toward their goals, how long they persevere in the face of difficulty, and whether their thought processes are self-aiding or self-hindering [11]. Although self-efficacy was originally conceptualized as a situationally-specific belief [11, 29], there is some evidence to suggest that it may also have stable components. In essence, some learners have a generalized belief in their ability to be successful across a wide range of learning-related tasks and situations [30]. Also, like goal orientation, instructors can deliberately induce high levels of self-efficacy among their students, for example by systematically engineering the learning experience by starting with relatively simple tasks (to ensure initial success) and then slowly building up to more complex ones, by actively modeling the correct way to perform the task, by verbally persuading learners, as needed, when they are struggling to perform the task, and by helping the learner to correctly attribute their physiological states (e.g., anxiety, stress, fatigue) as being both normal and common even among high-performers [29].

Previous research suggests that regardless of how learning is measured (e.g., tests of basic skills, classroom performance, scores on standardized tests), self-efficacy exhibits strong positive correlations with learning-related outcomes. Moreover, the results hold for learners in elementary school, high school, and college [31].

4 Self-regulated Learning in the Workplace from a Dual-Process Perspective

Historically, researchers in the field of Industrial/Organizational Psychology have focused on how to maximize the effectiveness of formal, employer-provided training courses. Over the years, the field has amassed an impressive body of literature on how to accomplish this [32]. However, there is an emerging body of research which suggests that for many employees, a substantial majority (70–90%) of their on-the-job learning experiences occur via informal, semi-structured, and self-directed means [33].

While this shift in emphasis from organizationally-provided training to self-directed learning places greater control in the hands of the learner, it requires a significant amount of self-regulation by the learner, both in terms of metacognition and motivation. However, there is a wealth of evidence that learners have difficulty when it comes to self-regulated learning. For example, they often make poor decisions about what to focus on and how much effort to invest in the learning process [34,35,36]. Therefore, it is critical to understand the factors that might influence employees’ ability to effectively self-regulate, as well as what interventions can support them.

As noted previously, motivation and metacognition are central to self-regulated learning in general, and to the deliberate practice approach, in particular. To highlight the key points of intersection, we first need to consider the conditions under which effective self-regulation takes place, including goal setting, task selection, feedback, concentrated periods of effort, and motivation [37]. Inherently, the first three conditions align more with metacognition. That is, to set effective goals and identify appropriate tasks to focus on, learners need sufficient domain expertise to know what they know, to know what they don’t know, and to prioritize what they should focus on next [38]. Without some baseline level of expertise, learners can easily set goals and identify tasks that lead to shallower learning curves or that result in bad habits.

The latter two conditions focus more on the motivation, or effort regulation, aspect of self-regulated learning. That is, learners need to have the discipline and motivation to engage in concentrated, effortful practice and to choose to invest effort into developing new skills, rather than practicing what they already know or enjoy doing. In general, this suggests that employees who are at an expert-level of learning might be better equipped and more capable of engaging in self-directed learning, while novice employees might need more support and guidance to engage in productive self-regulation. However, regardless of where a learner falls on the expertise continuum, there are challenges to engaging in effective self-regulatory learning as seen in Table 2.

Table 2. Implications for self-regulatory learning based on the current state of the learner.
Full size table

As depicted in the table, understanding the intersection of self-regulatory learning and Type 1/Type 2 processes provides more focused insight as to the likely strengths and limitations of different types of learners for engaging in self-regulation, as well as what types of strategies or interventions might be needed to support their ability to engage in self-regulated learning. However, while [39] provides a rich framework of strategies for promoting more effective self-directed learning, noticeably missing from the discussion is how these strategies may be more or less effective for individuals at different levels of expertise. [39] explicitly notes that research is lacking on how the effectiveness of different self-regulatory strategies (or the need for them) varies based on individual differences, with the seeming assumption that the strategies hold across all individuals. However, the limited research on other individual differences, including self-efficacy [35], goal orientation [40], and personality [41] suggests that this assumption is likely not true. As such, the intersection highlighted in Table 2 above is believed to be an important extension of this work, as it can help inform how organizations utilize and support self-directed learning of their employees, when considering their level of expertise, to ensure maximum benefits, while minimizing the potential negative side-effects (e.g., withdrawal, attrition).

5 Self-regulated Learning of an Advanced Perceptual-Motor-Cognitive Skill from a Dual Process Perspective

An extension to self-regulated learning in the workplace is self-regulated learning that involves difficult perceptual-motor-cognitive skills. A good example of such learning is learning to land a jet on an aircraft carrier. In addition to the normal challenges associated with landing an aircraft, this task requires landing a fast-moving aircraft on a very short runway that moves, both because the carrier is in motion and because the carrier is affected by the motion of the ocean. Pilots have considerable training in the classroom and landing on a fixed airfield before they attempt a carrier landing, and the initial attempts at landings of the typical pilot learning to land on the carrier can often be described as “colorful”.Footnote 2

Because of the extensive amount of training involved, many of the skills are Type 1 skills: they are practiced enough that they do not require cognitive resources and they are more-or-less automatic. Yet there are substantial Type 2 components, as well. For example, expert pilots cue on the carrier’s wake to infer the wind conditions over the deck: if the wake is frothy, the carrier must be moving fast enough to generate its own wind, meaning that the wind will be in a known direction at a known range of speeds. Normally this tells the pilot that they will have to keep the left wing down on landing. On the other hand, if the wake is minimal, then environmental wind will prevail, which could mean that the carrier has turned into the wind to aid with the recovery, or could mean something else—the pilot will need to stay alert to the possibilities during landing.

Pilot motivation is strong and straightforward. They need to land safely on the carrier to ensure their own safety, the safety of other personnel on the carrier, and the aircraft itself. Further, there is a spirit of sporting competition among pilots. The quality of every landing is evaluated by a Landing Signals Officer (LSO), and average scores are posted in the ward room on a “greenie board”, providing a way for pilots to compare their performance with one another, and offering motivation to achieve high scores. This incentivizes expert pilot effort (as discussed in the Type 1/motivation cell in Table 2) and suggests that expert pilots have a Performance Goal Orientation [27], as described in Sect. 3.2. On the other hand, LSOs provide explicit structure and guidance to less experienced pilots (as discussed in the Type 2/motivation cell in Table 2), suggesting that novice pilots have a Learning Goal Orientation.

[14] conducted an experiment in a simulator-based training study using a temporal occlusion paradigm [42] looking at several issues surrounding optimal training environments for carrier landing. One such issue was potential differences between the cognitive processes used by expert and by novice pilots. Pilots were shown an 8-second video of a portion of a carrier landing and were asked effectively if they were where they should be or if they anticipated needing to make an aggressive correction to ensure a safe landing. Correct responses were identified by a panel of subject matter experts, and the speed and accuracy of the pilot participants was assessed via computer. This temporal occlusion measurement was made both before and after a simulation-based carrier landing training session in order to assess the amount of learning.

One of the hypotheses under test was that experts use better-developed Type 1 responses because of their greater experience and pattern recognition capabilities, and that this should show up as faster and more correct responses compared to the novice pilots. Figure 2 shows the results. As predicted, experts were faster and more accurateFootnote 3, more so after the training session.Footnote 4 This is consistent with the idea that experts engage in Type 1 pattern recognition more than novices, who must primarily reason using slower Type 2 cognitive processes.

Fig. 2.
figure 2

Results of temporal occlusion experiment for expert and novice pilots, before and after a simulation-based training session.

Full size image

What does this say about the kinds of metacognition that pilots engage in? Definitive answers await future research, be we can see evidence of all three of the kinds of metacognition described in [21] and Sect. 3.1 at different levels of experience.

Metacognitive planning is likely a Type 2 endeavor, and likely of central importance for novices. For example, one pilot described a time when he was a novice and had just completed three landings in a row where he was much higher than the normal glideslope during landing. That pilot recalled deliberately thinking through what he needed to do in order to correct the problem before his next attempted carrier landing. On the other hand, experts are more likely to engage in self-monitoring and, should they detect unwelcome patterns, dynamically self-regulate their behavior. For one thing, Type 1 processing is prominent during their landings, so automatically recognizing and correcting problematic patterns in their own performance is also a task at which they have had considerable practice. For another, experts—unlike novices—do not generally get extensive guidance from the LSOs about ways they could improve their landing skills. For experienced pilots, LSOs generally limit their debriefs to a simple description of what happened during the landing. It is expected that the expert pilots will figure out for themselves how they need to improve.

6 Conclusions and Future Research

Dual-processing theory makes interesting contributions to the understanding of metacognition. There are certain predictable changes—in cognitive processing, metacognition, and motivation—that occur as learners accumulate increasingly greater amounts of domain expertise. Having little or no direct hands-on experience, novices generally understand the domain based on what they have read, heard in a lecture, or learned vicariously. Their knowledge is abstract, poorly organized, and decays quickly. During skills practice, their performance is slow and effortful. They perform the task in discrete steps following the textbook description, their movements are “jerky” rather than “smooth”, and they make common errors such as omitting critical steps or performing steps in the wrong order. Because they must consciously monitor their performance to avoid making errors, novices experience high levels of workload thereby making it difficult to self-monitor or self-self-regulate. Finally, given that they are also prone to failure, their self-efficacy may be fragile, thereby leading to intrusive thoughts during task performance.

With increased domain-specific deliberate practice, the learners’ performance becomes more expert-like, their behavior becomes goal-directed and contextualized. Depending on their particular goal (or mission), certain situational cues now take on greater relevance, thereby priming their Type 1 pattern recognition processes. Moreover, as the task becomes increasingly “compiled”, additional cognitive resources are freed up—these resources can now be used for self-monitoring and self-regulation purposes. Finally, given their relatively high levels of self-efficacy, they are less likely to experience intrusive thoughts during task performance.

We believe that the four distinct “types” of metacognition that have described in this paper represent a critical extension of prior research from the fields of Cognitive Science, Educational Psychology, and Industrial/Organizational Psychology—all of which have looked at very narrow “slices” of this complex phenomenon.

We fully recognize that, despite being consistent with the literature, some our ideas about metacognition are both speculative and anecdotal. That being said, we believe this kind of theoretic exploration is healthy, and that it can lead to useful hypothesis testing and modeling. Fortunately, since Type 1 processing requires little in the way of cognitive resources and is more-or-less automatic, there are many experimental paradigms that can be brought to bear on metacognition that involves Type 1 processing as method or object. This includes dual-task protocols, temporal and spatial occlusion protocols, two-alternative forced choice protocols, a large fraction of the techniques that have been applied to perceptual pattern recognition, and many others, including neuroscientific techniques. In addition, Table 2 provides the beginnings of the guidance for self-regulated learning that will result a deeper understanding of the role of dual cognitive processes in metacognition.

The future of the dual-process perspective in self-regulated learning is bright. We look forward to the rich discoveries that await.