Strategies to read and learn: overcoming learning by consumption


Danielle S McNamara, Department of Psychology, University of Memphis, 202 Psychology Boulevard, Memphis, Tennessee 38152, USA. Tel: 00 1 901 212 5829; Fax: 00 1 901 678 1336; E-mail:


Medical Education 2010: 44: 340–346

Objectives  This article discusses the need for, and value of, providing students with instruction in how to use comprehension strategies as well as the effectiveness of inducing strategy use through cognitive disequilibrium. The leading assumption that guides this article is that learning facts and figures is not enough. Students need to build deep knowledge that is interconnected, coherent and includes understanding of potential causal mechanisms. Doing so requires going beyond the printed page by generating inferences and developing coherent explanations. Inferences and explanations allow the student to make links between concepts in the material and, importantly, to make connections to prior knowledge. These connections render students’ understanding of new material more coherent and, in consequence, deeper and more stable.

Discussion  This article describes two means of inducing students to construct a deeper understanding of new material. One means of challenging students is through cohesion gaps in a text (or a lecture) that require the student to generate inferences to understand the relationships between concepts. Although low-knowledge readers are not able to generate these inferences, relatively high-knowledge readers (e.g. medical students) are more likely to successfully generate inferences to bridge conceptual gaps, and doing so results in a deeper understanding of the material. A second means of inducing active processing is to provide students with instruction and practice in how to use comprehension strategies. This article describes methods of providing such instruction, including the intelligent tutoring system, iSTART.

Conclusions  The overarching goal of the research described in this article is to scaffold students towards ideal learning strategies. This cannot happen simply by telling students about good strategies. It is ineffective to inform a student that the content will be better understood if it is explained or evaluated. Such an approach is a victim of learning by consumption attitudes towards education.


The most common approach to education is to provide content to students. Teachers cover the content in class and students are instructed to read and study it. In science courses, students read about science; in English courses, students read literature, and so on. There is an overwhelming assumption in our educational system that the most important thing to deliver to students is content. Teachers assume that when they cover something in a course, it should be absorbed by the student. Teachers and students assume that the act of reading information is sufficient for learning. These assumptions of learning by consumption weaken our educational system.

Research that focuses on how students comprehend and learn from text highlights the importance of overcoming assumptions that learning can occur by shallow processing and mere exposure to information (i.e. through a lecture or text). Comprehension requires active processing of information by making connections between incoming and familiar information and making connections between concepts in content to be learned. This does not naturally occur through passive listening or reading. This article makes the case for teaching students to use comprehension and learning strategies as well as inducing active processing of information through cognitive disequilibrium.

The importance of strategies

Before content, should come strategies. Students need to learn how to learn. Although expecting students to read and learn from texts is certainly a necessary part of education, students also need to acquire skills that facilitate and enhance how they read, understand, study and learn. Reading and learning strategies help students to understand and learn content across a wider variety of situations. Strategy instruction ensures that students can perform well both with and without a teacher’s direct instruction.

Strategy instruction across a variety of domains builds on the notion that less skilled students should learn strategies that mimic those exhibited by skilled students.1 Likewise, strategies can help less skilled students compensate for their weaknesses.2 Providing strategy instruction has been found to be beneficial to both comprehension and learning.3–8 Further, strategy instruction is particularly needed and effective for those students who are struggling most, namely those with less knowledge and lower reading skills.9–12

However, strategy instruction can also be important for skilled, successful students. Strategy use is often crucial for developing an understanding of challenging content at deeper levels. Researchers and educators commonly distinguish between shallow and deep knowledge.4,13–16 Although there is no ironclad agreement on what differentiates deep from shallow knowledge, many scientists agree that shallow knowledge consists of simple ideas explicitly mentioned in the learning material, such as facts, properties or definitions of key terms. By contrast, deeper knowledge requires inferences, an understanding of complex mechanisms, and the ability to apply knowledge to new situations.4,13–16 Deep knowledge normally depends on an understanding of the simple ideas, but goes beyond this basic level of understanding to embrace an understanding of causal mechanisms, coherent mental models, complex relations, logical reasoning, translation of knowledge into action, and application of existing solutions to dissimilar situations.

Developing a deep understanding of material requires using a variety of strategies that afford making connections between ideas expressed in the content, and between those ideas and what the student already knows. When students struggle with challenging content, it is often the case that they have no idea how to take action to rectify the problem. This may pose a problem for successful students who enter medical school without ever having had to struggle to obtain good grades. Although shallow strategies such as memorisation may have been sufficient until that point, students may suddenly be faced with learning and comprehension challenges they have never before encountered.

Deep understanding of content emerges from the process of generating inferences as learners attempt to connect explicit ideas in the learning materials. They generate inferences to connect new information, recently encountered content and prior knowledge. Active learners who seek deep understanding are more likely to engage in learning activities or strategies that promote inference generation. Successful students ask deep how and why questions.17,18 They engage in active inquiry14,19–21 by searching for the answers to challenging questions and critically evaluating the quality of those answers. Active learners construct explanations and apply those explanations to difficult problems.22–24 Further, they consciously reflect on these cognitive activities.25,26 Successful students engage in self-regulated learning by generating their own learning goals and then attempting to achieve these goals by monitoring, regulating and controlling their thoughts and behaviour.3,27,28 Likewise, metacognition and metacomprehension are critical components because successful learning and comprehension are often accompanied by conscious and deliberate reflection about behaviour, emotions and thoughts.14,29,30 As depicted in Figure 1, these strategic activities come together to effectuate deep understanding of challenging material, as opposed to the shallow understanding that results from passive reading and memorisation.

Figure 1.

 Active reading and learning processes produce deeper understanding of challenging material

Although such qualities are desirable, students who spontaneously and skilfully engage in inquiry, explanation, self-regulation and metacognitive strategies are rare.25 Moreover, these processes are seldom exhibited in normal classrooms or in typical one-on-one sessions with human tutors.31–33 Unfortunately, these ideal learning processes also fail to emerge by simply exposing students to rich learning environments.34

Inducing active reading strategies

In sum, there are few students who frequently exhibit the strategic processes depicted in Figure 1, and surrounding students with a stimulating environment may do little to induce them to engage in such active learning processes. Indeed, much more is needed to equip an ideal learner. Most learners need to be challenged with questions, problems and tasks that place them in cognitive disequilibrium or present desirable difficulties. These kinds of situations encourage higher standards of comprehension and learning.14,35–41

Characteristics of the text can also induce readers to engage more or less in active reading processes. For example, McNamara et al.42–44 demonstrated that text cohesion can play an important role in inducing or inhibiting active reading processes. Cohesion refers to the overlap in ideas between sentences and paragraphs in a text, as well as the degree to which the text contains explicit cues to guide understanding (e.g. connectives, headers, topic sentences). When a reader has less background knowledge about a text, he or she can be stumped by the lack of cohesive cues in a text. The low-knowledge reader cannot make the inferences necessary to understand a low-cohesion text. By contrast, low-cohesion text induces the high-knowledge reader to generate inferences that connect the ideas in the text with prior knowledge. Making such connections is crucial to learning. Indeed, that’s what learning is all about: understanding new information by making connections to what is already known. However, when a more knowledgeable reader is faced with a high-cohesion text, in which these connections between ideas are explicitly expressed, the reader can process the text more superficially. The reader is not induced to generate inferences and, in consequence, forms only a shallow interpretation of the text.

Notably, further research on this phenomenon has shown that high-knowledge readers are not induced into this false sense of understanding if they are also skilled, active readers.2,45 Skilled, strategic readers approach text comprehension in a different way from less skilled readers. Skilled readers process text actively. They are more likely to engage in inference generation as well as the other active processing strategies listed in Figure 1. Thus, when they are faced with text that contains many cohesive cues and explicit conceptual connections, they are just as likely to generate the inferences necessary to construct a deep understanding of the text.

Teaching students to use reading strategies

These findings on the effects of text cohesion underscore the importance of teaching readers to use strategies such as those in Figure 1. For this purpose, McNamara23 developed and tested Self-Explanation Reading Training (SERT), which teaches students to use a combination of self-explanation22 and reading comprehension strategies4,8 while reading challenging science text. Students learn to self-explain text using five reading strategies: monitoring comprehension; paraphrasing; making bridging inferences between the current sentence and prior text; making predictions, and elaborating the text with links to what the reader already knows. SERT was motivated by empirical findings showing that students who self-explain text are more successful at solving problems, more likely to generate inferences, construct more coherent mental models, and develop a deeper understanding of the concepts covered in the text.22,46 However, without training, only high-ability students gain from self-explanation. SERT was designed to help less skilled students learn how to benefit from self-explanation and, in turn, to learn effective reading strategies.

McNamara23 examined the effectiveness of SERT by having college students self-explain a challenging text about cell mitosis. Half the students were provided with SERT before they attempted to self-explain the cell mitosis text and half were not. A median split based on a test for prior knowledge about cells provided an indication of high and low knowledge. Approximately half of the participants who received SERT were low knowledge and vice versa, half of those who did not receive training were low knowledge. The results showed that SERT was most effective for readers with less prior knowledge about the science domain. Students who had received SERT understood the text as well as the high-knowledge students did. By contrast, self-explaining the text was ineffective for those who had not received focused practice in using the reading strategies.

Further research9,47 has shown that reading strategy training is also helpful for students who are less skilled comprehenders. However, for both low-knowledge and less skilled readers, the benefits of reading strategy training are limited to enhancing comprehension of the explicit contents of the text (i.e. the textbase). These readers need to acquire more knowledge and become more proficient readers before crossing over to understanding challenging science text at deeper levels.

The story is quite different for more skilled readers. Among both adolescent students47 and college students9 research has shown that skilled readers benefit from reading strategy training by learning to build a more coherent situation model, or deeper-level representation of the text content. Whereas prior to training these students may be more content to settle for a superficial understanding of the text, after training these students show significant gains in terms of making more connections between ideas in the text and understanding the text at deeper levels.

Thus, it appears that many students benefit from reading strategy training, but in different ways, and according to their zone of proximal development (e.g.48). Readers need to first learn to form an adequate representation of the text-based information (i.e. essentially, the information presented in each individual sentence). Then, readers can learn how to understand the text at a deeper level by processing the relationships between the ideas conveyed across sentences and making links to world knowledge. There are two important consequences of such strategies. First, they allow readers with less knowledge about a text to compensate for those knowledge gaps. Second, such training induces more skilled readers to generate more inferences and construct a deeper understanding of the material. However, students need to be provided with instruction in how and when to use such strategies. A crucial part of such instruction is practice. Students need to be provided with opportunities to engage in deliberate practice in using such strategies.

Using technologies to scaffold strategy instruction

Unfortunately, logistical constraints can make it impossible for students to effectively learn strategies that support deep comprehension. For example, many classrooms do not provide sufficient opportunities for students to observe ideal learning strategies in action, nor the time necessary to practise them. Strategy instruction in the classroom rarely includes practice combined with immediate feedback, primarily because it is too time-consuming to achieve in a typical classroom setting. This practical constraint in classrooms motivates the need for automated tools that provide strategy instruction and practice with feedback.

Computer-based technologies are a means of relieving these problems. Automated tutoring technologies can increase the availability of needed instruction, reach more students and allow more opportunities for students to practise the strategies. Moreover, computational linguistic technologies fortify our ability to provide adaptive instruction and to give students immediate, accurate feedback on their performance. Tutoring technologies afford interaction with students using natural language via an avatar, which approaches real-life interactions with human tutors. This approach helps to engage and motivate the student. These technologies thus afford the provision of instruction that is adaptive and responsive to the student’s needs.

iSTART (Interactive Strategy Training for Active Reading and Thinking49) is an example of an intelligent tutoring system that helps high school and college students learn and practise reading comprehension strategies. Its design is based on SERT. However, in iSTART, avatars interact with one another and with the student to provide training and to scaffold the practice of the strategies. The avatars provide feedback to students based on computational algorithms that evaluate the quality of student input.49–51 The agents instruct trainees in the use of self-explanation and other active reading strategies to explain the meaning of science text while they read.

Empirical studies on the effectiveness of iSTART have been positive. Studies at both college52 and high school levels53,54 have indicated that iSTART improves text comprehension and strategy use over control groups. Two studies have further confirmed that iSTART training is as good as and can exceed the effectiveness of SERT, the live, classroom-based version of the training.9,55

Importantly, reading strategy training has been found to improve not only comprehension of science texts, but also grades in science courses. Three experiments were conducted with students in college-level science courses in which subsets of students were provided with SERT training. These studies showed reliable advantages on examination scores for students who received SERT training in comparison with control students, who did not receive training, which ranged between 5% and 14%. In addition, prior knowledge of scientific facts generally showed the strongest correlations with examination performance, whereas prior reading skill showed the lowest correlations (which were generally non-significant). Most importantly, training generally had the greatest benefits for those students with less prior knowledge about science. Unfortunately, within all of these experiments, low-knowledge students who did not receive training often left the science course without a passing grade.56

iSTART is just one of many technologies being developed at the Institute for Intelligent Systems (IIS) at the University of Memphis. All of the tutoring systems at the IIS include interactive dialogue between the user and the pedagogical agents which guide the learning process. These tutoring technologies rest on the principles that active processing is critical to learning and that interactive dialogue helps to support and enhance that level of processing. Hence, developing technologies to support natural language processing is also a major focus of projects at the IIS. These latter projects support the tutoring projects by providing a means to hold natural language dialogues with the students who interact with the tutoring systems. Two of the systems at IIS, AutoTutor and MetaTutor, are particularly relevant to medical education because they focus, like iSTART, on students’ learning of science content.


An overarching goal of many of the technologies being developed by researchers at the IIS is to help students to understand and learn challenging content at deeper levels. This objective follows from the recognition that an emphasis on shallow memory and recognition rather than deeper understanding of the relationships between ideas and underlying constructs represents a significant problem in our education system. Our approach to this problem is to build technologies that are based on scientific principles established in cognitive science. These principles include those relevant to knowledge and skill acquisition, reading comprehension and motivation. One guiding principle is to engage the learner through interactive dialogue and question answering. Importantly, the learner must generate answers to questions and practise the relevant skills with adaptive feedback. To meet this objective, pedagogical agents and natural language processing play integral roles in our tutoring systems.

Our overarching goal is to scaffold students towards the ideal learning strategies depicted in Figure 1. This will not be induced simply by telling students that these are good strategies. It is ineffective to inform a student that the content will be better understood if it is explained or evaluated. Such an approach is a victim of learning by consumption attitudes towards education. Just as all content needs to be taught, metacognitive strategies need to be taught in a manner that includes the provision of examples, as well as deliberate focused practice with feedback. Moreover, these strategies need to be practised in concert with the materials the students are expected to learn. Such instruction and practice can occur in the classroom. However, intelligent tutoring systems also offer a means of providing such training. Such systems afford engaging and adaptive training that scaffolds the student towards becoming a more ideal learner who seeks deep and coherent understandings of content, rather than a learner who passively consumes facts and figures.


Acknowledgements:  the opinions, findings, conclusions and recommendations expressed in this material are those of the author and do not necessarily reflect the views of supporting agencies.

Funding:  this research was supported by multiple granting agencies, including the Institute for Education Sciences (IES R305G020018-02, R305G040046, R305A080589) and the National Science Foundation (NSF REC0241144, IIS-0735682).

Conflicts of interest:  none.

Ethical approval:  not applicable.