Learning

Imagine that every time that people entered a new environment they had to learn how to behave without the guidance of prior experiences. Slightly novel tasks, like shopping online, would be disorienting and dependant on trial-and-error tactics. Fortunately, people use aspects of their prior experiences, such as the selection of goods and subsequent payment, to guide their behavior in new settings. The ability to use learning gained in one situation to help with another is called transfer.

Transfer has a direct bearing on education. Educators hope that students transfer what they learn from one class to another–and to the outside world. Educators also hope students transfer experiences from home to help make sense of lessons at school. There are two major approaches to the study of transfer. One approach characterizes the knowledge and conditions of acquisition that optimize the chances of transfer. The other approach inquires into the nature of individuals and the cultural contexts that transform them into more adaptive participants.

Knowledge-Based Approaches to Transfer

There are several knowledge-based approaches to transfer.

Transferring out from instruction. Ideally, the knowledge students learn in school will be applied outside of school. For some topics, it is possible to train students for the specific situations they will subsequently encounter, such as typing at a keyboard. For other topics, educators cannot anticipate all the out-of-school applications. When school-based lessons do not have a direct mapping to out-of-school contexts, memorization without understanding can lead to inert knowledge. Inert knowledge occurs when people acquire an idea without also learning the conditions of its subsequent application, and thus they fail to apply that idea appropriately. Memorizing the Pythagorean formula, for example, does not guarantee students know to use the formula to find the distance of a shortcut.

Knowing when to use an idea depends on knowing the contexts in which the idea is useful. The ideas that people learn are always parts of a larger context, and people must determine which aspects of that context are relevant. Imagine, for example, a young child who is learning to use the hook of a candy cane to pull a toy closer. As the child learns the action, there are a number of contextual features she might also learn. There are incidental features–it is Christmas; there are surface features–the candy is small and striped; and there are deep features–the candy cane is rigid and hooked. Instruction for transfer must help the child discern the deep features. This way the child might subsequently use an umbrella handle to gather a stuffed animal instead of trying a candy-striped rope.

When people learn, they not only encode the target idea, they also encode the context in which it occurs, even if that context is incidental. For a study published in 1975, Gooden and Baddeley asked adults to learn a list of words on land or underwater (while scuba diving). Afterwards, the adults were subdivided; half tried to remember the words underwater and half on land. Those people who learned the words underwater remembered them better underwater than on land, and those people who learned the words on land remembered them better on land than underwater. This result reveals the context dependency of memory. Context dependency is useful because it constrains ideas to appear in appropriate contexts, rather than cluttering people's thoughts at odd times. But context dependency can be a problem for transfer, because transfer, by definition, has to occur when the original context of learning is not reinstated–when one is no longer in school, for example.

Surface features, which are readily apparent to the learner, differ from incidental features, because surface features are attached to the idea rather than the context in which the idea occurs. Surface features can be useful. A child might learn that fish have fins and lay eggs. When he sees a new creature with fins, he may decide it is a fish and infer that it too lays eggs. Surface features, however, can be imperfect cues. People may overgeneralize and exhibit negative transfer. For example, the child may have seen a dolphin instead of a fish. People may also undergeneralize and fail to transfer. A child might see an eel and assume it does not lay eggs. Good instruction helps students see beneath the surface to find the deep features of an idea.

Deep features are based on structures integral to an idea, which may not be readily apparent. To a physicist, an inclined plane and scissors share the same deep structure of leverage, but novices cannot see this similarity and they fail to use a formula learned for inclined planes to reason about scissors.

Analogies are built on deep features. For example, color is to picture as sound is to song. On the surface, color and sound differ, as do pictures and song. Nonetheless, the relation of used to create makes it possible to compare the common structure between the two. Analogy is an important way people discover deep features. In the 1990s, Kevin Dunbar studied the laboratory meetings of cell biologists. He found that the scientists often used analogies to understand a new discovery. They typically made transfers of near analogies rather than far ones. A far analogy transfers an idea from a remote body of knowledge that shares few surface features, as might be the case when using the structure of the solar system to explain the structure of an atom. A near analogy draws on a structure that comes from a similar body of knowledge. The scientists in Dunbar's study used near analogies from biology because they had precise knowledge of biology, which made for a more productive transfer.

Instruction can help students determine deep features by using analogous examples rather than single examples. In a 1983 study, Mary Gick and Keith Holyoak asked students how to kill a tumor with a burst of radiation, given that a strong burst kills nearby tissue and a weak burst does not kill the tumor. Students learned that the solution uses multiple weak radiation beams that converge on the tumor. Sometime later, the students tried to solve the problem of how a general could attack a fortress: If the general brought enough troops to attack the fortress, they would collapse the main bridge. Students did not propose that the general could split his forces over multiple bridges and then converge on the fortress. The students' knowledge of the convergence solution was inert, because it was only associated with the radiation problem. Gick and Holyoak found they could improve transfer by providing two analogous examples instead of one. For example, students worked with the radiation problem and an analogous traffic congestion problem. This helped students abstract the convergence schema from the radiation context, and they were able to transfer their knowledge to the fortress problem.

Transferring in to instruction. In school, transfer can help students learn. If students can transfer in prior knowledge, it will help them understand the content of a new lesson. A lesson on the Pythagorean theorem becomes more comprehensible if students can transfer in prior knowledge of right triangles. Otherwise, the lesson simply involves pushing algebraic symbols.

Unlike transfer to out-of-school settings, which depends on the spontaneous retrieval of relevant prior knowledge, transfer to in-school settings can be directly supported by teachers. A common approach to help students recruit prior knowledge uses cover stories that help students see the relevance of what they are about to learn. A teacher might discuss the challenge of finding the distance of the moon from the earth to motivate a lesson on trigonometry. This example includes two ways that transferring in prior knowledge can support learning. Prior knowledge helps students understand the problems that a particular body of knowledge is intended to solve–in this case, problems about distance. Prior knowledge also enables learners to construct a mental model of the situation that helps them understand what the components of the trigonometric formulas refer to.

Sometimes students cannot transfer knowledge to school settings because they do not have the relevant knowledge. One way to help overcome a lack of prior knowledge is to use contrasting cases. Whereas pairs of analogies help students abstract deep features from surface features, pairs of contrasting cases help students notice deep features in the first place. Contrasting cases juxtapose examples that only differ by one or two features. For example, a teacher might ask students to compare examples of acute, right, and obtuse triangles. Given the contrasts, students can notice what makes a right triangle distinctive, which in turn, helps them construct precise mental models to understand a lesson on the Pythagorean theorem.

Person-Based Approaches to Transfer

The second approach to transfer asks whether person-level variables affect transfer. For example, do IQ tests or persistence predict the ability to transfer? Person-based research relevant to instruction asks whether some experiences can transform people in general ways.

Transferring out from instruction. An enduring issue has been whether instruction can transform people into better thinkers. People often believe that mastering a formal discipline, like Latin or programming, improves the rigor of thought. Research has shown that it is very difficult to improve people's reasoning, with instruction in logical reasoning being notoriously difficult. Although people may learn to reason appropriately for one situation, they do not necessarily apply that reasoning to novel situations. More protracted experiences, however, may broadly transform individuals to the extent that they apply a certain method of reasoning in general, regardless of situational context. For example, the cultural experiences of American and Chinese adults lead them to approach contradictions differently.

There have also been attempts to improve learning abilities by improving people's ability to transfer. Ann Brown and Mary Jo Kane showed young children how to use a sample solution to help solve an analogous problem. After several lessons on transferring knowledge from samples to problems, the children spontaneously began to transfer knowledge from one example to another. Whether this type of instruction has broad effects–for example, when the child leaves the psychologist's laboratory–remains an open question. Most likely, it is the accumulation of many experiences, not isolated, short-term lessons, that has broad implications for personal development.

Transferring in to instruction. When children enter school, they come with identities and dispositions that have been informed by the practices and roles available in their homes and neighborhoods. Schools also have practices and roles, but these can seem foreign and inhospitable to out-of-school identities. Na'ilah Nasir, for example, found that students did not transfer their basketball "street statistics" to make sense of statistics lessons in their classrooms (nor did they use school-learned procedures to solve statistics problems in basketball). From a knowledge approach to transfer, one might argue that the school and basketball statistics were analogous, and that the children failed to see the common deep features. From a person approach to transfer, the cultural contexts of the two settings were so different that they supported different identities, roles, and interpretations of social demands. People can view and express themselves quite differently in school and nonschool contexts, and there will therefore be little transfer.

One way to bridge home and school is to alter instructional contexts so children can build identities and practices that are consistent with their out-of-school personae. Educators, for example, can bring elements of surrounding cultures into the classroom. In one intervention, African-American students learned literary analysis by building on their linguistic practice of signifying. These children brought their cultural heritage to bear on school subjects, and this fostered a school-based identity in which students viewed themselves as competent and engaged in school.

Conclusion

The frequent disconnect between in-school and out-of-school contexts has led some researchers to argue that transfer is unimportant. In 1988, Jean Lave compared how people solved school math problems and best-buy shopping problems. The adults rarely used their school algorithms when shopping. Because they were competent shoppers and viewed themselves as such, one might conclude that school-based learning does not need to transfer. This conclusion, however, is predicated on a narrow view of transfer that is limited to identical uses of what one has learned or to identical expressions of identity.

From an educational perspective, the primary function of transfer should be to prepare people to learn something new. So, even though shoppers did not use the exact algorithms they had learned in school, the school-based instruction prepared them to learn to solve best-buy problems when they did not have paper and pencil at hand. This is the central relevance of transfer for education. Educators cannot create experts who spontaneously transfer their knowledge or identities to handle every problem or context that might arise. Instead, educators can only put students on a trajectory to expertise by preparing them to transfer for future learning.