Much is known from decades of research with children, college students, and older adults about the conditions that affect cognition and learning and how cognition and learning change across the life span. In this chapter, we describe principles of learning that have sufficiently strong and broad support to warrant their application to the design of instruction for adolescents and adults. We draw on and update several recent efforts to distill principles of learning from research for educators that include
• Organizing Instruction and Study to Improve Student Learning (Pashler et al., 2007), an initiative of the Institute of Education Sciences (IES) in the U.S. Department of Education.
• Lifelong Learning at Work and at Home (Graesser, Halpern, and Hakel, 2007), an initiative of the Association of Psychological Sciences (APS) and the American Psychological Association.
• How People Learn (National Research Council, 2000).
• What Works in Distance Learning Guidelines (O’Neil, 2005).
• e-Learning and the Science of Instruction and Multimedia Learning (Clark and Mayer, 2003; Mayer, 2009).
These reports and hundreds of published studies inform the committee’s conclusions about the elements of instruction with potential to support adult learning and the research that is needed to discover how to apply these principles most effectively to improve the literacy skills of diverse populations of adult learners.
There is substantial convergence between the conditions that facilitate
learning in general and the principles of effective literacy instruction for typical and struggling learners presented in Chapter 2. This convergence leads to having greater confidence in the findings and further indicates the value of incorporating them into the design of instruction for other populations, such as adult learners. How to use the principles of learning and effective literacy instruction presented in this report to substantially enhance the literacy of diverse populations outside school is an important question for future research.
The ideal culmination of successful learning is the development of expertise. Learners who achieve expertise tend to be self-regulated (Azevedo and Cromley, 2004; Pintrich, 2000b; Schunk and Zimmerman, 2008; Winne, 2001). They formulate learning goals, track progress on these goals, identify their own knowledge deficits, detect contradictions, ask good questions, search relevant information sources for answers, make inferences when answers are not directly available, and initiate steps to build knowledge at deep levels of mastery. The “meta” knowledge of language, cognition, emotions, motivation, communication, and social interactions that is part of self-regulated learning is well developed. The expert learner forms conceptually rich and organized representations of knowledge that resist forgetting, can be retrieved automatically, and can be applied flexibly across tasks and situations. The development of expertise has specific features:
1. Experts acquire and maintain skill through consistent and long-term engagement with domain-relevant activities, deliberate practice, and corrective feedback (Ericsson, 2006).
2. Experts notice features and meaningful patterns in situations and tasks that are not noticed by novices (Chase and Simon, 1973; Chi, Glaser, and Rees, 1982; Rawson and van Overschelde, 2008).
3. Experts have content knowledge that is organized around core mental models and concepts that reflect deep understanding (Mosenthal, 1996; Vitale, Romance, and Dolan, 2006).
4. Experts have the metacognitive skills to think about and apply strategies (Hacker, Dunlosky, and Graesser, 2009).
5. Expert knowledge is tuned and conditionalized, so it includes representing the contexts in which particular knowledge, skills, and strategies apply (Anderson et al., 1995).
6. Experts retrieve and execute relevant knowledge and skills automatically, which enables them to perform well on complex tasks and to free cognitive resources for more attention-demanding activities (Ackerman, 1988).
7. Experts approach tasks flexibly, so they recognize when more knowledge is needed and take steps to acquire it while monitoring progress (Bilalić, McLeod, and Gobet, 2008; Metcalfe and Kornell, 2005; Spiro et al., 1991).
8. Within certain physical limits of speed and endurance associated with aging and health status, experts retain domain-related skills through adulthood as long as they are practiced (Krampe and Charness, 2006).
Expertise is usually difficult to achieve—and for a complex skill such as literacy requires many hours of practice over many years—experts tend to have 1,000-10,000 hours of experience in their field of expertise (Chi, Glaser, and Farr, 1988). With respect to literacy expertise taught in schools, an hour per day from kindergarten through twelfth grade amounts to about 2,000 hours in total, after taking out the inevitable days when no real instruction occurs, which is at the low end of the range needed to gain expertise. Adult literacy learners can be assumed to have missed out on many of these hours or to need substantially more practice. Adults bring varied goals to adult literacy education, but it is clear that given the hours of practice needed to develop literacy skills for functioning well in the realms of work, family, education, civic engagement, and so on, instruction needs to be designed to ensure that learning proceeds as efficiently as possible. Efficiency is especially important considering that adolescents and adults live in complex worlds with many competing demands (Riediger, Li, and Lindenberger, 2006).
Learning involves being proficient with the tools needed to complete the tasks to be mastered and so requires practice with using tools. Tools can be anything from a physical tool (pen, computer, textbook, or graphic organizer) to more abstract tools—such as the appropriate lexicon of a particular domain or knowledge of how people in a domain construct written arguments or literature. Tools can contribute to the development of deeper understandings of a concept or idea by presenting learners with varied ways of representing the idea (Eisner, 1994; Paivio, 1986; Siegel, 1995).
The learning principles described in this chapter vary in their attention to explicit and implicit teaching and learning. Both explicit and implicit learning contribute to the development of expertise in complex skills, such as reading and writing, as illustrated in previous chapters. The principles also vary in their emphasis on promoting initial acquisition of knowledge and skills over transfer and generalization of acquired knowledge and skills to new situations. Initial acquisition involves attention to and encoding of relevant material, so that it can be retrieved from memory or applied to problems within short retention intervals. Transfer and generalization are
maximized when acquired knowledge and skills are successfully applied to relevant new situations that differ from the initial context of acquisition (Banich and Caccamise, 2010). It has been widely acknowledged in the cognitive sciences for decades that transfer and generalization can be very difficult or nearly impossible when the surface characteristics of the material and context differ between training and transfer problems and when the correspondences are not highlighted or recognized (Forbus, Gentner, and Law, 1995; Gick and Holyoak, 1980; Hayes and Simon, 1977). For example, Hayes and Simon’s classic study shows that college students experienced zero transfer between successive problems that were solved when the problems were structurally identical at a deep level but had different surface features (e.g., missionaries and cannibals versus monsters and globes). Each of the learning principles can be analyzed from the standpoint of ease of initial acquisition versus successful transfer and generalization. However, the principles that favor the latter are far from settled (Banich and Caccamise, 2010).
Researchers have identified a number of factors that improve retention of information and transfer of acquired knowledge to new situations. These factors are important for educators and product developers to consider when designing curricula, texts, materials, and technologies and selecting or creating lesson plans for use in adult education programs. For adult learners who have underdeveloped literacy skills, following these guidelines is especially important for ensuring that new concepts are absorbed, even though literacy skill is, to some extent, the ability to overcome the less-than-optimal designs of information sources.
Novices or those working to further develop their knowledge and skills often need help in attending to the parts of a task that are most relevant to their learning goal. Adults of all ages benefit from a clear (Dickinson and Rabbitt, 1991; Gao et al., 2011; Wingfield, Tun, and McCoy, 2005) and organized presentation that helps them to learn and remember new information (Craik and Jennings, 1992; Hess and Slaughter, 1990; Morrow et al., 1996; Smith et al., 1983). It is important to remove any irrelevant information, even if interesting, that could detract from learning to minimize cognitive load and competing demands on attention (Kalyuga, Chandler, and Sweller, 1999; Moreno, 2007; Van Merrienboer et al., 2006). Seductive details that do not address the main points to be conveyed also
risk consuming the learner’s attention and effort so that they miss the main points. Visual displays that are hard to read or spoken presentations that are presented in noisy environments can compromise learning because they distract attention away from deeper semantic processing (Dickinson and Rabbitt, 1991; Heinrich, Schneider, and Craik, 2008).
According to the coherence principle, learners need to get a coherent, well-connected representation of the main ideas to be learned. Providing structure and organization is important to help them understand concepts and how they relate to one another. The particular method used to organize ideas depends on the relations to be depicted. Outlines can be used to show structural hierarchies (Ausubel, 1968). Graphic organizers show the structure of interrelated ideas pictorially, with ideas represented as concepts in circles and relationships as lines that connect the circles (Vitale and Romance, 2007). Tables can be used to organize ideas in two or three dimensions, and diagrams can help to convey more complex relationships.
According to the contiguity principle, materials and lesson plans should be organized so that the elements and ideas to be related are presented near each other in space and time (Clark and Mayer, 2003; Mayer, 2005; Mayer and Moreno, 2003). For example, the verbal label for a picture needs to be placed spatially near the picture on the display, not on the other side of the screen. An explanation should be given at the time a concept is depicted rather than many minutes, hours, or days later. According to the segmentation principle, new material should be presented in discrete units so that new learners are not overwhelmed with too much new information at once.
There is substantial evidence that knowledge, skills, and strategies acquired across multiple and varied contexts are better generalized and applied flexibly across a range of tasks and situations (Atkinson, 2002; Catrambone, 1996; Paas and van Merrienboer, 1994; Schmidt and Bjork, 1992; Spiro et al., 1991). Memories are triggered by multiple cues so knowledge is available when needed. Acquisition can be slower, but learners retain and transfer their knowledge and skills better than if learned only in one context (Swezy and Llaneras, 1997).
Information is encoded and remembered better when it is delivered in multiple modes (verbal and pictorial), sensory modalities (auditory and visual), or media (computers and lectures) than when delivered in only a single mode, modality, or medium (Clark and Mayer, 2003; Kalchman
and Koedinger, 2005; Kozma, 2000; Mayer, 2009; Mayer and Moreno, 2003; Moreno and Mayer, 2007; Paivio, 1986). For example, it is effective to combine graphics with text, graphics with spoken descriptions, speech sounds with printed words, and other combinations of modalities. Graphic depictions with spoken descriptions are particularly effective for subject matter in science and technology (Mayer, 2009). Multiple codes provide richer and more varied representations that allow more memory retrieval routes.
However, implementation of this principle must be balanced against Principle 1: the amount of information should not overwhelm the learner to the point of attention being split or cognitive capacities being overloaded (Kalyuga, Chandler, and Sweller, 1999; Mayer and Moreno, 2003; Moreno, 2007; Sweller and Chandler, 1991). There needs to be a careful selection of the pictures, graphs, or other visual representations in order to be relevant to the material being taught. Graphics do not have to be completely realistic to be useful; sometimes a more abstract or schematic picture will best illustrate a key idea, whereas a more photorealistic graphic may actually distract the learner with details that are irrelevant to the main point. There is also substantial evidence that memory retention increases when a person studies the material at deeper, semantic levels of processing than exclusively at the surface levels of processing (Craik and Lockhart, 1972; Kintsch et al., 1990).
There is some evidence that, with aging, learners can increasingly benefit from the environmental support provided by augmenting the material to be learned with multimodal presentations (Craik and Jennings, 1992; Luo et al., 2007). However, multimodal presentations can be relatively less effective for older adults if the information across modalities is difficult to integrate (Luo et al., 2007; Stine, Wingfield, and Myers, 1990).
According to Vygotsky’s (1986) concept of the zone of proximal development (ZPD), the effectiveness of a text, technology, tutor, or instructional approach in promoting learning can be assessed by comparing performance with and without the supports provided in the intervention. Does the intervention allow the person to perform better than they would have been able to without the particular material, tool, or approach to instruction? There is moderate evidence that the answer depends partly on the selection of learning goals, materials, and tasks, which should be sensitive to what the student has mastered and be appropriately challenging—not too easy or too difficult, but just right (Metcalfe and Kornell, 2005; VanLehn et al., 2007; Wolfe et al., 1998).
Consider a text used to help an adult learn about a medical procedure: if the text is extremely easy and overlaps perfectly with what readers already know, then the text will not stretch their knowledge beyond what they already knew without the text. Similarly, the adult will not gain much medical knowledge by reading a text that is too complex and riddled with technical jargon far beyond what he or she can handle. People will learn most from a text that appeals to some of what they already know and expands knowledge in a way that is neither too challenging nor redundant.
Individualized student instruction is expected to be more effective when it takes into account the ZPD of individual learners. The U.S Common Core Standards for reading and writing have adopted the ZPD principle by proposing that text assignments push the envelope on text difficulty, as reflected in Lexile scores and other text characteristics, but not too much beyond what the student can handle. Evidence is accumulating that reading skills are acquired better when interventions consider the characteristics of individual learners. This has been demonstrated for beginning reading in children, in that some types of readers benefit from one instructional method and other types of readers benefit from another (Connor et al., 2007). In that research, Assessment to Instruction (A2i) web-based software was used to compare students’ lexical decoding skills (i.e., letter and word reading skills) and vocabulary. Instruction methods were differentially effective depending on the readers’ starting skill levels on these dimensions. Readers with low lexical decoding benefited most from explicit teacher-managed code-focused instruction; this instruction was not helpful to readers with higher lexical decoding skills but low vocabulary. Readers with low vocabulary needed a combination of explicit teacher-managed code-focused instruction and explicit meaning-focused instruction. Those with high vocabulary benefited from explicit meaning-focused instruction or independent reading. Indeed, students with high lexical decoding skills and vocabulary would best be left alone to conduct independent reading on topics they are interested in.
Several factors affect growth experienced in the ZPD. First, having more knowledge about the domain to be learned can increase the efficiency of learning (Beier and Ackerman, 2005; Miller, 2009; Miller, Cohen, and Wingfield, 2006; O’Reilly and McNamara, 2007). During adulthood (in contrast to childhood) knowledge is highly individualized (Ackerman, 2008), so instruction should first assess and then build on the knowledge the learner already has. Finally, gradual age-related declines in speed of processing, attentional control, associative binding, and working memory may decrease learning efficiency (Hertzog et al., 2008; Myerson et al., 2003; Park et al., 2002; Waszak, Li, and Hommel, 2010), so slower pacing or more practice or both may be required to reach a given level of performance.
It is better to distribute the presentation of materials and tests over time than to concentrate the learning experiences within a short time span (Bahrick et al., 1993; Bloom and Shuell, 1981; Cepeda et al., 2006; Cull, 2000; Rohrer and Taylor, 2006). When studying for an exam, it is better to space the same amount of study over days and weeks than to cram it into a single study session the night before the test. Spaced practice has been shown to be advantageous for adults of a variety of ages (Kausler, Wiley, and Philips, 1990; Kornell et al., 2010; Logan and Balota, 2009).
Reexposure to course material after a delay often markedly increases the amount of information that students remember. Delayed reexposure can be promoted through homework assignments, in-class reviews, quizzes, and other instructional exercises (Pashler et al., 2007). Evidence for this principle is primarily based on memory for isolated information units (such as facts or vocabulary definitions). However, there is evidence that rereading can enhance metacomprehension skills and long-term retention of text material, especially if it is spaced and especially for low-ability students (Griffin, Wiley, and Thiede, 2008; Rawson and Kintsch, 2005; Rawson, Dunlosky, and Thiede, 2000).
There is substantial evidence that periodic testing helps learning and slows down forgetting (Bangert-Drowns et al., 1991; Bjork, 1988; Butler and Roediger, 2007; Dempster, 1997; Karpicke and Roediger, 2007; McDaniel, Roediger, and McDermott, 2007; McDaniel et al., 2007; Roediger and Karpicke, 2006). One indirect benefit is that regular testing, which can be quite brief and embedded in instructional materials, keeps students constantly engaged in the material and guides instructors or computers in making decisions about what to teach (Shute, 2008). The precise frequency of testing presumably depends on the nature of materials to be learned. Students benefit more from repeated testing when they expect a final exam than when they do not expect one (Szupnar, McDermott, and Roediger, 2007). Spacing retrieval has been shown to improve performance for adults from a wide age range (Bishara and Jacoby, 2008).
There is substantial evidence that it is important to link concepts to be read or learned to concrete perceptions and actions (Glenberg and Kaschak, 2002; Glenberg and Robertson, 1999; Glenberg et al., 2004; Piaget, 1952). For example, when reading instructions on assembling a piece of furniture,
it helps to be able to view and hold the parts while reading the instructions. Perceptual-motor experience is particularly important when there is a need for precision of ideas and communication and when a concept is first introduced. Some cognitive frameworks have emphasized the importance of grounding comprehension and learning in perceptual-motor experience (called embodied cognition), but there is a debate on the role of abstract representations and symbols in comprehension in addition to the embodied perceptual-motor representations (de Vega, Glenberg, and Graesser, 2008; Glenberg, 1997). As noted below, there is some evidence that it is effective to integrate abstract with concrete representations of concepts. It is critical to keep in mind that new knowledge is built on and interpreted in light of existing knowledge, and much knowledge comes from everyday activities. Building atop barely learned and abstract ideas is much more difficult and error-prone than building atop well-learned concepts that are experienced daily.
Stories and other types of narrative are usually about everyday experiences and create perceptual-motor memories similar to daily experience. There is substantial evidence that stories are easier to read, comprehend, and remember than other types of learning materials (Bower and Clark, 1969; Casey et al., 2008; Graesser and Ottati, 1996; Rubin, 1995). For many millennia, the primary way of passing wisdom down from generation to generation was through stories. Stories have concrete characters, objects, locations, plots, themes, emotions, and actions that bear some similarity to everyday experiences and are natural packages of knowledge (Bower, Black, and Turner, 1979; Graesser, Olde, and Klettke, 2002).
It is interesting to note that active experiencing, a theatrical technique in which dialogue is learned by acting out scenes with physical and emotional expression, facilitates learning large passages of dialogue without explicit memorization (Noice and Noice, 2006, 2008; Noice et al., 1999; Noice, Noice, and Kennedy, 2000). This finding is consistent with the notion that stories are easier to understand and remember partly because of the generation of perceptual-motor memories similar to the memories of everyday experience. Perceptual-motor memory is well preserved, if not enhanced, in adulthood (Dijkstra et al., 2004; Radvansky and Djikstra, 2007; Radvansky et al., 2001) and performing actions related to material to be remembered enhances memory for adults in a wide age range (Bäckman and Nilsson, 1985; Feyereisen, 2009). Thus, stories may be powerful tools for practicing and building comprehension skills and developing and reinforcing background knowledge across the life span.
At the same time, there also is a tendency for other genres than narratives to be underused in literacy instruction, and literacy does require the ability to handle a number of forms other than stories. In order to acquire
the ability to read and write in other forms, practice on those forms will be required.
Interventions are needed that encourage the learner to actively generate language, content, and patterns of reasoning rather than passively processing the material delivered by the learning environment. Learning is enhanced when learners have to organize the information themselves and exert cognitive effort during acquisition or retrieval. Simply put, it is the student who should be doing the acting, thinking, talking, reading, and writing for learning. Encouraging learners to engage in deeper levels of thinking and reasoning is especially helpful to adults needing to develop these skills for education, work, and other purposes involving complex materials and tasks.
Learning is enhanced when learners produce answers themselves instead of reading or recognizing them (Chi, Roy, and Hausmann, 2008; National Research Council, 2000; Tulving, 1967). This fact explains why free recall or essay tests that require the test-taker to generate answers with minimal cues often produce better retention than recognition tests and multiple-choice tests in which the learner only needs to be able to recognize correct answers. It also explains why tutors learn more than tutees in peer tutoring when students start out on an even playing field (Fuchs et al., 1994; Mathes and Fuchs, 1994; Topping, 1996). Learner-generated content can lack detail and contain misconceptions, however, that need to be monitored to ensure adequate learning and to prevent learning incorrect information.
Strategies that require learners to be actively engaged with reading material also produce better retention over the long term (McNamara, 2007a, 2007b; Pressley et al., 1998). Learners can, for example, develop their own mini-testing situations as they review material, such as stating the information in their own words (without viewing the text) and synthesizing information from multiple sources, such as from class and textbooks (Bjork, 1994). Programs exist to help students learn to do this (Beck and McKeown, 2006). Although the strategies require cognitive effort, their use is important to encourage since they improve learning and are underdeveloped in many children and adults (Pearson and Duke, 2002; Pressley, 2002; Snow, 2002). For complex and coherent bodies of material, outlining, integrating, and synthesizing information produce better learning than rereading materials or other more passive strategies.
There is evidence that adults from a wide age range benefit from con-
tent generation to improve learning (Johnson, Schmitt, and Pietrukowicz, 1989; Mitchell et al., 1986; Taconnat et al., 2008). Past their 20s, learners slowly may become less likely to spontaneously generate content that is rich, elaborative, and distinctive if they are learning in a domain outside their previous knowledge and experience; consequently, more contextual support may be needed as the learner generates content to optimize the benefits of generation (Dunlosky, Hertzog, and Powell-Moman, 2005; Luo, Hendricks, and Craik, 2007).
There is substantial evidence that learning is facilitated by constructing explanations and arguments (Ainsworth and Loizou, 2003; Anderson et al., 2001; Chi et al., 1994; Magliano, Trabasso, and Graesser, 1999; McNamara, 2004; McNamara and Magliano, 2008; Reznitskaya et al., 2008; VanLehn et al., 2007). Explanations consist of causal analyses of events, logical justifications of claims, and functional rationales for actions. Explanations provide coherence to the material and justify why information is relevant and important. Students may be prompted to give self-explanations of material by thinking aloud or answering questions that elicit explanations connecting the material to what they already know. The self-explanations of students can be improved by explicit instruction on self-explanations and by setting up collaborations with a student or tutor to help with the process of constructing useful explanation. Studying good explanations facilitates deeper comprehension, learning, memory, and transfer.
Explanations of material and reasoning are elicited by deep questions, such as why, how, what-if, and what-if not, as opposed to shallow questions that require the learner to simply fill in missing words, such as who, what, where, and when (Graesser and Person, 1994). There is substantial evidence that training students to ask deep questions facilitates comprehension of material from text, classroom lectures, and electronic media (Beck et al., 1997; Craig et al., 2006; Dillon, 1988; King, 1994; Pressley et al., 1992; Rosenshine, Meister, and Chapman, 1996). The learner gets into the mindset of having deeper standards of comprehension (Baker, 1985), and the resulting representations are more elaborate.
One method of stimulating thought, content generation, and reasoning is to present some challenges, obstacles, or contradictions that place the learner in “cognitive disequilibrium.” The occurrence of cognitive disequilibrium is anticipated by instructors who purposefully select topics, texts, and questions that clash with the students’ knowledge, beliefs, or attitudes. Cognitive disequilibrium is confirmed when students ask relevant questions
(Graesser and McMahen, 1993; Rosenshine, Meister, and Chapman, 1996), when a classroom launches into a spirited discussion addressing the challenge (Nystrand, 2006), and when students exhibit facial expressions of confusion (D’Mello and Graesser, 2010). Such “desirable difficulties” slow down initial learning but promote long-term retention and transfer (Bereiter and Scardamalia, 1985; Bjork, 1988, 1999; Bjork and Linn, 2006). Presenting a challenging problem before students read a text can stimulate inquiry, curiosity, thinking, deep questions, and deeper learning during text comprehension (Schwartz and Bransford, 1998). Cognitive disequilibrium and questions occur when there are obstacles to goals, contradictions, conflicts, anomalous events, failures of the text to satisfy a task need, salient gaps in knowledge, uncertainty, equally attractive alternatives, and other types of impasses (Chinn and Brewer, 1993; Graesser and McMahen, 1993; Graesser and Olde, 2003). When these impasses occur, adaptive learners engage in reasoning, thought, problem solving, and planning en route to restoring cognitive equilibrium. Adaptive readers slow down and construct elaborations or explanations while reading misconceptions, contradictions, and false information (Kendeou and Van den Broek, 2007; O’Brien et al., 1998; Rapp, 2008).
However, it is noteworthy that readers often do not notice blatant contradictions (e.g., burying survivors, tranquilizing stimulants) that on second glance appear to be quite obvious (Daneman, Lennertz, and Hannon, 2006; Hannon and Daneman, 2004). Less skilled readers are more vulnerable to such shallow processing, so that explicit instruction and practice in monitoring coherence and self-explanation (McNamara and Magliano, 2009) may be useful.
There is moderate evidence that opportunities to consider multiple viewpoints and perspectives about a phenomenon contribute to understanding a concept and to greater cognitive flexibility in accessing and using the concept in a range of contexts. If a concept is understood in only a specific and rigid manner, it will be encoded, accessed, and used in a more restricted way. Cognitive flexibility increases when interventions support multiple layers of knowledge that interconnect facts, rules, skills, procedures, plans, and deep conceptual principles (Spiro et al., 1991). The cognitive complexity and multiple viewpoints are believed to be helpful when learners need to transfer knowledge and skills to tasks that have unique complexities that cannot be anticipated.
Two examples illustrate attempts to promote multiple points of view and perspectives. Kozma (2000) developed a computerized learning envi-
ronment for chemistry that shows four different viewpoints simultaneously during the course of a chemical reaction, such as the action of a person mixing chemicals in beakers, the action of molecules, mathematical formulae, and graphs that plot measures over time. The extent to which a student views the different perspectives depends on their preferences and prior training, so their mental models do not necessarily converge on a single correct understanding. As another example, readers who comprehend stories can be instructed to adopt the perspectives of different characters and their resulting recall protocols and story representations end up being quite different (Anderson and Pichert, 1978). Readers eventually can be trained to adopt multiple character viewpoints while reading stories and thereby achieve greater cognitive flexibility.
Laboratory experiments and classroom studies have shown the benefits of connecting and interleaving both abstract and concrete representations of problems at the K-12 and college levels, particularly in the domains of mathematics, science, and technology (Bottge et al., 2007; Goldstone and Sakamoto, 2003; Goldstone and Son, 2005; Sloutsky, Kaminisky, and Heckler, 2005). Students have an easier time acquiring an initial understanding of a concept presented in a concrete form, but they also need a more abstract symbolic representation to apply that knowledge in a different context. So, for example, when most college students read texts on physics and technology, they do not acquire a deep enough representation or understanding to support inferences and the building of situation models without some pedagogical activities that encourage multiple representations and cognitive flexibility (VanLehn et al., 2007; Wiley et al., 2009).
Students often lack the knowledge, skills, and meta-awareness needed to focus attention on content relevant to a task or goal, to comprehend text, to study material sufficiently, or to perform effectively on complex cognitive tasks. In particular, reading strategies at deeper levels are underdeveloped in many children and adults, especially for expository text, so they would benefit from comprehension strategy training (McNamara, 2007a; Pearson and Duke, 2007; Pressley, 1998; Snow, 2002).
There is moderate evidence that complex strategies can be acquired by well-engineered instruction that is structured, explicit, scaffolded, and intensive. Scaffolded instruction is the systematic selection and sequencing
of content, materials, and tasks that both prompt the student to provide information and deliver relevant information to achieve learning. This is well documented for comprehension strategy instruction (McNamara, 2007b; Pressley, 2000; Williams et al., 2009; Williams, Hall, and Lauer, 2004). The instruction typically goes from simple to complex, with substantial practice at each step. It incorporates meaningful and interactive tasks, as well as clear templates that exhibit instruction points.
Strategies of solving mathematical problems can also be acquired by observing experts solving example problems step by step or by interleaving worked example solutions with problem-solving exercises. That is, students learn more by alternating between studying examples of worked-out problem solutions and solving similar problems on their own than they do when just given problems to solve on their own (Catrambone, 1996; Cooper and Sweller, 1987; Kalyuga et al., 2001; Pashler et al., 2007). Procedural skills can be modeled effectively through modeling-scaffolding-fading (McNamara, 2007a; Renkl, Atkinson, and Grosse, 2004; Renkl et al., 2002; Rogoff, 1990; Rogoff and Gardner, 1984): the expert first models the solution, then the student tries with periodic feedback and scaffolding from the expert, and then the expert assistance eventually fades. Strategies of argumentation can be developed from structured practice with argument stratagems in collaborative reasoning that transfer to writing (Anderson et al., 2001; Reznitskaya et al., 2008). One central question is how much learning of knowledge, strategies, and skills can be acquired through information delivery and scripted exercises without the more flexible and interactive scaffolding (Connor et al., 2007; McNamara, 2007b).
There is some evidence that adults from a wide age range can benefit from instruction in memory monitoring strategies to improve memory performance (Dunlosky, Kubat-Silman, and Hertzog, 2003). Mnemonic training, especially if embedded in otherwise valued classroom literacy activities, may be more effective in augmenting the repertoire of memory skills of adolescents and young adults than of children (Brehmer et al., 2007, 2008). Although even older adults benefit, it is possible that age-related decreases in fluid abilities may slow the acquisition of new strategies in later life (Brehmer et al., 2007, 2008; Hertzog et al., 2008).
It is well documented that both children and adults can experience serious limitations in metacognition (Hacker, Dunlosky, and Graesser, 2009)—their ability to understand, assess, and act on the adequacy of their memory, comprehension, learning, planning, problem-solving, and decision processes. One would expect children to have limited metacognitive knowledge, but it is somewhat remarkable that adults also have limited metacognitive proficiency after their years of experience. More specifically, the vast majority of adults are not good at judging their own comprehension of text (Dunlosky and Lipko, 2007; Maki, 1998). They also are not good at plan-
ning, selecting, monitoring, or evaluating their strategies for self-regulated learning (Azevedo and Cromley, 2004; Azevedo and Witherspoon, 2009; Winne, 2001), inquiry learning (Graesser, McNamara, and VanLehn, 2005; White and Frederiksen, 2005), or discovery learning (Kirschner, Sweller, and Clark, 2006; Klahr, 2002). Therefore, explicit training, modeling, and guided practice are needed before students acquire adequate strategies of comprehension, critical thinking, metacomprehension, self-regulated learning, and discovery learning (Dunlosky and Hertzog, 1998). Domain knowledge can also enhance self-regulated learning (Griffin, Jee, and Wiley, 2009).
There is moderate evidence that strategy instruction should be deeply integrated with subject-matter content rather than being lists of abstract rules or scripted procedures that ignore the content (National Research Council, 2000). For example, it is a good strategy for readers to be asking the question “why” when reading texts because it encourages the student to build explanations of the content. This strategy is ideally implemented across the curriculum, so students ask such questions as why catalysts are important when reading a chemistry text, why the Spanish-American War was important in U.S. history, why an action of a character in a novel has a particular motive, and why an author bothers to describe the layout of a city. Substantial subject-matter knowledge is needed to effectively apply many reading strategies because comprehension involves the integration of prior knowledge and text.
Many reading researchers claim that comprehension skills and strategies are facilitated when they are embedded in content areas (e.g., science, history, social studies) (Duke and Pearson, 2002; Guthrie et al., 2004; Moje, 2008b; Neufeld, 2005; Pearson and Duke, 2002; Pressley, 2002; Williams et al., 2009), although some claim that more evidence should be amassed to have greater certainty (Lee et al., 2006). Comprehension can improve after instruction on the structure of expository text, such as compare-contrast, problem-solution, cause-effect, description, sequence, and other rhetorical frames (Chambliss, 1995; Meyer and Poon, 2001; Williams, Hall, and Lauer, 2004; Williams et al., 2005, 2009). Such structure training, which is often contextualized in subject matter, can improve comprehension for adults from a wide age range (Meyer and Poon, 2001; Meyer, Young, and Bartlett, 1989).
Feedback affects learning in a number of ways that are well documented (Azevedo and Bernard, 1995; Kluger and DiNisi, 1996; Shute, 2008). Adults from young to old can take advantage of feedback to acquire new skills (Hertzog et al., 2007; Stine-Morrow, Miller, and Nevin, 1999; West, Bagwell, and Dark-Freudeman, 2005). Feedback helps learners fine-tune their knowledge, skills, and strategies. It can be explicitly delivered by people or computers (supervised learning), or it can be implicitly provided in situations that are engineered to make knowledge and skill gaps evident to the learner (unsupervised learning). The feedback may identify and possibly correct inaccurate skills (bugs) and misconceptions (errors of commission) or may identify missing information (errors of omission).
There is substantial evidence that students benefit from feedback on their performance in a learning task, but the optimal timing of the feedback depends on the task (Pashler et al., 2005; Shute, 2008). Immediate feedback has the advantage of maximizing contiguity of correct information and of preventing elaboration of incorrect information. Just as people learn correct information from accurate feedback, they also can learn incorrect information. For example, when incorrect alternatives on multiple-choice tests are presented, the wrong answers can be learned instead of the correct answers (Butler and Roediger, 2007; Roediger and Marsh, 2005; Toppino and Luipersbeck, 1993), and accuracy may be compromised as a function of the number of distracters (Roediger and Marsh, 2005). This may also occur for true-false tests (Toppino and Brochin, 1989) and when misconceptions are planted in texts (Kendeou and van den Broek, 2005). These effects can be reduced when learners receive feedback immediately after a test (Butler, Karpicke, and Roediger, 2008; Kang, McDermott, and Roediger, 2007; Metcalfe and Kornell, 2007; Roediger and Marsh, 2005) or while performing an action in a procedure (Anderson et al., 1995; Ritter et al., 2007) or completing a task. They can also be reduced by a rhetorical structure (Kendeou and van den Broek, 2007) or critical stance (Wiley et al., 2009) that encourages the learner to be skeptical or to refute the presented information.
Immediate feedback can be useful under many conditions, but it does have potential liabilities. A learner’s motivation can be threatened when there is a barrage of corrections and negative feedback. Frequent interruptions of organized action sequences (such as reading a text aloud) can be not only irritating but also counterproductive in the acquisition of complex motor skills. Immediate feedback blocks the possibility of the students’
correcting their own reading errors and regulating their own learning more generally. The impact and timing of feedback differ for tasks that involve memory, simple procedural skills, reasoning, problem solving, and complex domains of knowledge that have entrenched misconceptions.
The optimal administration of feedback is a complex mechanism that depends on timing, the nature of the knowledge or skill to be developed, and characteristics of the student. It is unlikely that an instructor can track all of these levels for 30 students in a class—or even a single student for a tutor. As discussed further in Chapter 6, technologies can keep track of the details that are beyond the horizon of human capacities. Computerized learning environments are poised to provide adaptive feedback that is sensitive to all of these constraints.
There is moderate evidence that feedback should both point out errors to the learner and explain why the information is incorrect instead of merely flagging that an answer is incorrect or giving a student an overall score that does not provide information about the nature of the needed improvements (Aleven et al., 2003; Ritter et al., 2007; Roscoe and Chi, 2007; Shute, 2008). Much of this research is on subject-matter content rather than literacy per se, but the principles are expected to apply universally. That said, more research is needed on the type of qualitative feedback that is optimal for different types of material and different types of learners (Shute, 2008). How specific should the feedback be (Ritter et al., 2007)? At what point will a negative feedback frustrate or dispirit students, especially those with low self-efficacy (Graesser, D’Mello, and Person, 2009; Lepper and Woolverton, 2002)? How can task-specific feedback productively guide subsequent learning (Hunt and Pellegrino, 2005; Shute, 2008)? When should students have control over the nature and extent of feedback they receive (Aleven et al., 2003)? Under what conditions is it appropriate to have an open learning environment, in which the students have full knowledge of their extent of mastering knowledge, skills, and strategies at a fine-grained level (Bull and Kay, 2007)? As in much of instructional design, there are a number of trade-offs and sensitivities to the nature of the knowledge and skills being trained. Instructionally perfect feedback may be expensive to provide, but to the extent that technology can be recruited, costs can decrease. Fine-grained feedback is best for specific well-defined skills, but some modicum of feedback is also appropriate for general, ill-defined skills. Excessive feedback also runs the risk of preventing the development of self-regulated learning, and so a fading process is needed to gradually shift control to the student.
There is some evidence among older learners that quantitative feedback on skill acquisition is more effective if it is framed in terms of positive feedback (what is good about one’s performance) relative to goal attainment, compared with a raw score (West et al., 2005). Also, adult learners with initially high levels of perceived control may benefit more from feedback than those with lower levels of perceived control (Miller and West, 2010). (This may be true for younger populations, although further data are needed.)
As previously discussed, training in complex strategies, metacognition, and self-regulated learning may to some extent be accomplished by well-engineered training materials that guide all learners through the same regimen in a scripted fashion. However, some researchers think that students need to be guided by knowledgeable tutors, mentors, and computer learning environments that adaptively interact in a fashion that is sensitive to the characteristics of the learner, called the learner profile (Conley, Kerner, and Reynolds, 2005; Connor et al., 2007; Graesser, D’Mello, and Person, 2009; McNamara, 2007b; Woolf, 2009). In essence, human or machine intelligence facilitates learning when it fits the needs of the particular student in a context-sensitive fashion, particularly in the case of complex skills and knowledge (see Chapter 6 for more on technology). This research is consistent with sociocultural theories of learning positing that learning depends on interaction with a more knowledgeable other (Lave and Wenger, 1991, 1998; Rogoff, 1990, 1993, 1995; Rogoff and Lave, 1984; Rogoff and Wertsch, 1984; Scribner and Cole, 1981; Vygotsky, 1986; Wertsch, 1991).
There is moderate evidence that learning of complex material requires adaptive learning environments that are sensitive to the learner’s general profile and to the level of his or her mastery at any given point in time. Indeed, this assumption underlies research showing learning gains through intelligent tutoring systems and learning environments (Anderson et al., 1995; Dodds and Fletcher, 2004; Doignon and Falmagne, 1999; Koedinger et al., 1997; Lesgold et al., 1992; Ritter et al., 2007; VanLehn, 2006; Woolf, 2009) and other reading systems that adapt to the learner, either computer systems (Connor et al., 2007; McNamara, 2007a; Meyer and Wijekumar, 2007) or human tutors (Palincsar and Brown, 1984; Rosenshine and Meister, 1994). When the knowledge conveyed by a text is complex, fine-tuned diagnosis and remediation may need to be sensitive to a large spectrum of learners’ states of knowledge, skills, and strategies, as well as how the presence or
absence of various supporting knowledge and skills impacts other components of effective performance (Connor et al., 2009).
At this point, researchers have not differentiated the contributions of context-sensitive adaptive strategies from the content in the learning experience. Simply put, is it adaptive instruction or the content of the instruction that matters? Individualized adaptive training has been used successfully to build cognitive skills among older learners (Erickson et al., 2007; Jaeggi et al., 2008; Kramer et al., 1999; Kramer, Larish, and Strayer, 1995). However, as for younger populations, there is a lack of experimentation that isolates the adaptive nature of the instruction as a cause of learning gains. Differentiating the two requires a precise mathematical treatment of the information delivered by the interventions. Such control over content is rarely imposed in research investigations (although see VanLehn et al., 2007).
Computer environments, rather than human instructors, may have the most promise in manipulating and controlling these complex interventions because of the complexity of diagnoses and remediation mechanisms. For example, accomplished human tutors have a difficult time being adaptive to many aspects of the learner (Chi, Roy, and Hausmann, 2008; Chi, Siler, and Jeong, 2004; Graesser, D’Mello, and Person, 2009). Examples of the kinds of computer interventions that can be achieved include analysis of reading times for segments of a text against models of the strategies that would distribute time in a given way, followed by coaching of specific ways to read more effectively. Even without any machine intelligence, it is possible to mark text segments according to the amount of time past readers have spent on them and thus guide students to consider their efforts more carefully.
There is moderate evidence that learners benefit from instructional interactions in which they receive fine-grained feedback (i.e., feedback specific to the immediate momentary task at hand) with hints that prompt them to generate knowledge (Ainsworth, 2008; Chi, Roy, and Hausmann, 2008; Graesser, D’Mello, and Person, 2009; Graesser, Person, and Magliano, 1995; VanLehn et al., 2007). Various teaching methods include such interactions: reciprocal teaching method, modeling-scaffolding-fading, the Socratic method, refutation, and others. Efficacy studies are needed, however, to determine the effects on learning and if the effects vary for different learners (see McNamara, 2007b).
Learning is enhanced by opportunities to practice and use skills for a purpose (Ford and Forman, 2006; Forman, Minick, and Stone, 1993; Lave and Wenger, 1991; Rogoff, 1990; Street, 1984). There is some evidence that anchored learning practices help learning (Bottge et al., 2007; Collins, Brown, and Newman, 1989; Dede and Grotzer, 2009; National Research Council, 2000). Anchored learning refers to developing knowledge and skill while working on problems encountered in the real world. Students often work in teams for several hours or days trying to solve a practical problem that matters to them and that connects to their knowledge. The problem is also challenging, so learners need to engage in problem solving and recruit multiple levels of knowledge and skills. With coaching, these activities can be organized coherently around solving the practical problem.
Examples of anchored learning are problem-based curricula in medical schools, in which students work on genuine medical cases, and communities of practice, in which students try to solve problems of pollution in their city. The students may spend 2 weeks learning about ecology to explain why fish are dying in a pond or how to save an eagle in a forest. Medical students may spend days analyzing the cases of patients in a hospital for diagnosis and treatment (Vernon and Blake, 1993).
Anchored learning has features that are likely to motivate struggling adult learners who are sensitive to the value of their learning experience. Yet much needs to be understood about how to design effective anchored learning experiences to achieve goals related to literacy and learning. For instance, for any particular topic, what learning goals do students pursue and what material should be read to achieve the learning goals? When an article is accessed, what do they read, how much do they read, and when do they give up? How much of the information in an article gets incorporated in messages to peers, documents they write, and behavior? What deficits in reading components present barriers to effective participation in a community of learners? There is little or no empirical evidence on answers to these fundamental questions about goal-based reading (McCrudden and Schraw, 2007). More research is needed on the principles and dynamics of how adults sift through and select material for focused study (Pirolli, 2005, 2007; Pirolli and Card, 1999).
Motivation is inextricably bound to learning, and decades of research have attempted to explain the relationship (Deci and Ryan, 2002a; Dweck, 2002; Lepper and Henderlong, 2000; Linnenbrink and Pintrich, 2002;
Meyer and Turner, 2006). Chapter 5, on supporting persistence, reviews in detail research findings related to motivation and distills principles for creating learning environments to inspire and support persistence and engagement. We note two important points here. First, the affective response that learners have to the learning experience influences not only engagement and persistence in a task, but also their capacity for cognitive processing. It is well known that adults are more motivated when the learning experience and materials are consonant with existing interests and dispositions (Ackerman and Rolfhus, 1999; Beier and Ackerman, 2001, 2003, 2005), and when engaged in reading or writing for a real purpose. Engaging narrative, expository, or procedural texts on topics that interest the learner and deliver knowledge the learner values are more likely to sustain the attention needed for learning (Hultsch and Dixon, 1983; Morrow et al., 2009; Stine-Morrow et al., 2004).
Second, motivation among adults is also more likely to be enhanced when instruction helps to build self-confidence and self-efficacy and develops the student’s identity as a person who reads. Adults with literacy problems often have experienced being stigmatized or marginalized, which makes enhancing self-confidence especially important. Because past experiences may have been very painful, interventions need to accommodate the occurrence of negative emotions, such as frustration, anger, boredom, and disengagement. Social support from peers, family members, tutors, and mentors facilitate motivation and mitigate their dropping out of adult literacy programs.
Research on cognition and learning shows elements to include in the design of instruction (see Box 4-1). Some of these findings have emerged from research on literacy. The principles are expected to generalize across populations, but how to apply them to the development of effective literacy instruction for diverse adult learners in various forms of adult education and developmental instruction in college must be determined in future research. Given the findings from research on learning, three questions should guide this research.
1. There is a high level of complexity involved in the design of learning environments consistent with principles of learning (e.g., ideal levels of information delivery, task difficulty, and feedback tailored to the individual learner). This complexity must be considered in the development of hypotheses and research designs. The research must also determine the expertise required to flexibly deliver instruction consistent with the principles once developed. To what
extent can technology leverage and augment the literacy instructor’s expertise to provide the adaptive learning environments that are optimal for the learner?
2. For adolescents and adults to invest the time required to develop their literacy, the instruction they receive must provide valued content knowledge and literacy skills (see Chapter 5 on motivation, engagement, and persistence). Thus, a promising direction for practice and research that is consistent with principles of learning and motivation is to discover how to build effective literacy instruction (curricula, practices, texts, and tools) that connects with the personal interests of learners and delivers the knowledge they need in content domains (e.g., electronics). To what degree is it possible for reading and writing instruction to piggyback onto instruction to develop content knowledge, instead of content knowledge being secondary to the acquisition of reading and writing skills? In other words, to what extent can content drive the development of adults’ literacy?
3. Similarly, certain skills are in demand in the 21st century for social interaction and for success in college and in the workplace. To what extent can reading and writing skills be developed as part of developing these forms of literate practice? Given that most literate practice in today’s world involves technologies, a goal for research is to determine how to effectively integrate important technologies into literacy instruction and practice to enable adults to function effectively in their educational, work, and social environments.
Summary of Principles of Learning for Instructional Design
Attention, Retention, and Transfer
• Present material in a clear and organized format. To facilitate learning, remove irrelevant information, even if interesting, to minimize distraction, provide structure and organization (coherence principle), present related elements to be learned near each other in space and time (continuity principle), and present new material in units that do not overwhelm with information (segmentation principle).
• Use multiple and varied examples. Knowledge, skills, and strategies acquired across multiple and varied contexts are better generalized and applied flexibly across a range of tasks and situations,
• Present material in multiple modalities and formats. Information is encoded and remembered better when it is delivered in multiple modes (verbal and pictorial), sensory modalities (auditory and visual), or media (computers and lectures) than when delivered in only a single mode, modality, or medium.
• Teach in the zone of proximal development. Select learning goals, materials, and tasks that are sensitive to what the student has mastered and that are appropriately challenging. Scaffold learning with instructional interactions and systematic selection and sequencing of content, materials, and tasks that are both at the appropriate level of difficulty and provide prompts and information needed to learn.
• Space presentations of new material. Learning is facilitated by the temporally distributed presentation of materials and tests instead of concentrated learning experiences within a short time span. Reexposure to course material after an optimal amount of delay often markedly increases the amount of information that students remember.
• Test on multiple occasions, preferably with spacing. Periodic testing helps learning and slows down forgetting. Regular testing, which can be quite brief and embedded in instructional materials, keeps students constantly engaged in the material and guides instructors or computers in making decisions about what to teach.
• Ground concepts in perceptual-motor experiences. Learning of concepts is facilitated with instruction that employs or evokes concrete perceptions and actions. Stories, for example, which generate perceptual-motor memories similar to the memories of everyday experience, may be powerful tools for practicing and building comprehension skills and developing and reinforcing background knowledge. Consider using content presented in stories to scaffold learning from other genres.
Generation of Content and Reasoning
• Encourage the generation of explanations, substantive questions, and the resolution of contradictions. These active learning processes impart coherence and meaning to the material to be learned, facilitates habitual generation of complex representations of information, and result in deeper understanding. Learner-generated content can lack detail and contain misconceptions that must be monitored and corrected.
• Construct ideas from multiple points of view and different perspectives. Considering multiple viewpoints and perspectives contributes to understanding a concept and to greater cognitive flexibility in accessing and using the concept in a range of contexts.
Complex Strategies, Critical Thinking, Inquiry, and Self-Regulated Learning
• Structure instruction to develop the effective use of complex strategies. Explicit training, modeling, and guided practice in the use of complex strategies is especially important for those with serious limitations in metacognition (the ability to understand, assess, and act on the adequacy of one’s memory, comprehension, learning, planning, problem-solving, and decision processes) and difficulties with regulating their own strategy use.
• Combine complex strategy instruction with the learning of content. To facilitate learning and application of new knowledge in a subject domain, strategy instruction should be integrated with subject-matter content.
• Effective feedback is immediate, accurate, and timely. Feedback should not contain too many corrections, too much negative feedback, or frequent interruptions of organized action sequences (such as reading a text aloud) because these can be demotivating and counterproductive in the acquisition of complex skills.
• Qualitative feedback is better for learning than test scores and error flagging. Feedback is more effective if it points out errors and explains why the response is incorrect. The type of qualitative feedback that is optimal for different types of material and different types of learners requires further study.
Adaptive and Interactive Learning Environments
• Adaptive learning environments foster understanding in complex domains. Adaptive learning environments are sensitive to the learner’s general profile, and level of mastery at any given point in time can facilitate the learning of complex material. The degree to which adaptive instruction from human instructors and computerized learning environments can facilitate and accelerate learning requires further study.
• Interactive learning environments facilitate learning. Fine-grained feedback provided while learners engage in a task with hints that prompt generation of knowledge facilitates learning. Research is needed to evaluate the effectiveness of specific interactive instructional approaches (e.g., reciprocal teaching method, modeling-scaffolding-fading, the Socratic method, refutation).
• Learning is facilitated in genuine and coherent learning environments. Learning is enhanced by opportunities to practice and use skills for a purpose, although the effectiveness of specific approaches consistent with this principle remains to be tested.
Motivation and Emotion
• Motivation is essential for learning. A learner's affective response to the learning experience influences not only engagement and persistence in a task but also the capacity for cognitive processing.