LEARNING

ANALOGICAL REASONINGDedre GentnerJeffrey Loewenstein

CAUSAL REASONINGJoseph P. MaglianoBradford H. Pillow

CONCEPTUAL CHANGECarol L. Smith

KNOWLEDGE ACQUISITION, REPRESENTATION, ANDORGANIZATIONDanielle S. McNamaraTenaha O’Reilly

NEUROLOGICAL FOUNDATIONHoward Eichenbaum

PERCEPTUAL PROCESSESJohn J. Rieser

PROBLEM SOLVINGRichard E. Mayer

REASONINGThomas D. Griffin

TRANSFER OF LEARNINGDaniel L. SchwartzNa’ilah Nasir

ANALOGICAL REASONING

Analogy plays an important role in learning and in-struction. As John Bransford, Jeffrey Franks, NancyVye, and Robert Sherwood noted in 1989, analogiescan help students make connections between differ-ent concepts and transfer knowledge from a well-understood domain to one that is unfamiliar or notdirectly perceptual. For example, the circulatory sys-tem is often explained as being like a plumbing sys-tem, with the heart as pump.

The Analogical Reasoning Process

Analogical reasoning involves several sub-processes:(1) retrieval of one case given another; (2) mappingbetween two cases in working memory; (3) evaluat-ing the analogy and its inferences; and, sometimes,(4) abstracting the common structure. The core pro-cess in analogical reasoning is mapping. Accordingto structure-mapping theory, developed by DedreGentner in 1982, an analogy is a mapping of knowl-edge from one domain (the base or source) into an-other (the target) such that a system of relations thatholds among the base objects also holds among thetarget objects. In interpreting an analogy, peopleseek to put the objects of the base in one-to-one cor-respondence with the objects of the target so as toobtain the maximal structural match. The corre-

sponding objects in the base and target need not re-semble each other; what is important is that theyhold like roles in the matching relational structures.Thus, analogy provides a way to focus on relationalcommonalities independently of the objects inwhich those relations are embedded.

In explanatory analogy, a well-understood baseor source situation is mapped to a target situationthat is less familiar and/or less concrete. Once thetwo situations are aligned—that is, once the learnerhas established correspondences between them—then new inferences are derived by importing con-nected information from the base to the target. Forexample, in the analogy between blood circulationand plumbing, students might first align the knownfacts that the pump causes water to flow through thepipes with the fact that the heart causes blood to flowthrough the veins. Given this alignment of structure,the learner can carry over additional inferences: forexample, that plaque in the veins forces the heart towork harder, just as narrow pipes require a pump towork harder.

Gentner and Phillip Wolff in 2000 set forth fourways in which comparing two analogs fosters learn-ing. First, it can highlight common relations. For ex-ample, in processing the circulation/plumbinganalogy, the focus is on the dynamics of circulation,and other normally salient knowledge—such as thered color of arteries and the blue color of veins—issuppressed. Second, it can lead to new inferences, asnoted above. Third, comparing two analogs can re-veal meaningful differences. For example, the circu-lation/plumbing analogy can bring out thedifference that veins are flexible whereas pipes arerigid. In teaching by analogy, it is important to bringout such differences; otherwise students may missthem, leading them to make inappropriate infer-ences. Fourth, comparing two analogs can leadlearners to form abstractions, as amplified below.

What Makes a Good Analogy

As Gentner suggested in 1982, to facilitate makingclear alignments and reasonable inferences, an anal-ogy must be structurally consistent—that is, itshould have one-to-one correspondences, and therelations in the two domains should have a parallelstructure. For example, in the circulation/plumbingsystem analogy, the pump cannot correspond toboth the veins and the heart. Another factor influ-encing the quality of an analogy is systematicity:Analogies that convey an interconnected system of

LEARNING: ANALOGICAL REASONING 1421

relations, such as the circulation/pumping analogy,are more useful than those that convey only a singleisolated fact, such as ‘‘The brain looks like a walnut.’’Further, as Keith Holyoak and Paul Thagard arguedin 1995, an analogy should be goal-relevant in thecurrent context.

In addition to the above general qualities, sever-al further factors influence the success of an ex-planatory analogy, including base specificity, trans-parency, and scope. Base specificity is the degree towhich the structure of the base domain is clearly un-derstood. Transparency is the ease with which thecorrespondences can be seen. Transparency is in-creased by similarities between corresponding ob-jects and is decreased by similarities betweennoncorresponding objects. For example, in 1986Gentner and Cecile Toupin found that four- to six-year-old children succeeded in transferring a story tonew characters when similar characters occupiedsimilar roles (e.g., squirrel → chipmunk; trout →salmon), but they failed when the match was cross-mapped, with similar characters in different roles(e.g., squirrel → salmon; trout → chipmunk). Thesame pattern has been found with adults. Transpar-ency also applies to relations. In 2001 Miriam Bassokfound that students more easily aligned instances of‘‘increase’’ when both were continuous (e.g., speedof a car and growth of a population) than when onewas discrete (e.g., attendance at an annual event). Fi-nally, scope refers to how widely applicable the anal-ogy is.

Methods Used to Investigate Analogical Learning

Much research on analogy in learning has been de-voted to the effects of analogies on domain under-standing. For example, in 1987 Brian Ross foundthat giving learners analogical examples to illustratea probability principle facilitated their later use ofthe probability formula to solve other problems. Inclassroom studies from 1998, Daniel Schwartz andJohn Bransford found that generating distinctionsbetween contrasting cases improved students’ subse-quent learning. As reported in 1993, John Clementused a technique of bridging analogies to induce re-vision of faulty mental models. Learners were givena series of analogs, beginning with a very close matchand moving gradually to a situation that exemplifiedthe desired new model.

Another line of inquiry focuses on the sponta-neous analogies people use as mental models of theworld. This research generally begins with a ques-

tionnaire or interview to elicit the person’s own ana-logical models. For example, Willet Kempton in1986 used interviews to uncover two common ana-logical models of home heating systems. In the (in-correct) valve model, the thermostat is like a faucet:It controls the rate at which the furnace producesheat. In the (correct) threshold model, the thermo-stat is like an oven: It simply controls the goal tem-perature, and the furnace runs at a constant rate.Kempton then examined household thermostat re-cords and found patterns of thermostat settings cor-responding to the two analogies. Some familiesconstantly adjusted their thermostats from high tolow temperatures, an expensive strategy that followsfrom the valve model. Others simply set their ther-mostat twice a day—low at night, higher by day,consistent with the threshold model.

Analogy in Children

Research on the development of analogy shows a re-lational shift in focus from object commonalities torelational commonalities. This shift appears to resultfrom gains in domain knowledge, as Gentner andMary Jo Rattermann suggested in 1991, and perhapsfrom gains in processing capacity as suggested byGraeme Halford in 1993. In 1989 Ann Brownshowed that young children’s success in analogicaltransfer tasks increased when the domains were fa-miliar to them and they were given training in therelevant relations. For example, three-year-olds cantransfer solutions across simple tasks involving fa-miliar relations such as stacking and pulling, and six-year-olds can transfer more complex solutions. In1987 Kayoko Inagaki and Giyoo Hatano studiedspontaneous analogies in five- to six-year-old chil-dren by asking questions such as whether they couldkeep a baby rabbit small and cute forever. The chil-dren often made analogies to humans, such as ‘‘Wecannot keep the baby the same size forever becausehe takes food. If he eats, he will become bigger andbigger and be an adult.’’ Children were more oftencorrect when they used these personification analo-gies than when they did not. This suggests that chil-dren were using humans—a familiar, well-understood domain—as a base domain forreasoning about similar creatures.

Retrieval of Analogs: The Inert KnowledgeProblem

Learning from cases is often easier than learningprinciples directly. Despite its usefulness, however,

1422 LEARNING: ANALOGICAL REASONING

training with examples and cases often fails to leadto transfer, because people fail to retrieve potentiallyuseful analogs. For example, Mary Gick andHolyoak found in 1980 that participants given an in-sight problem typically failed to solve it, even whenthey had just read a story with an analogous solu-tion. Yet, when they were told to use the prior exam-ple, they were able to do so. This shows that the priorknowledge was not lost from memory; this failure toaccess prior structurally similar cases is, rather, aninstance of ‘‘inert knowledge’’—knowledge that isnot accessed when needed.

One explanation for this failure of transfer isthat people often encode cases in a situation-specificmanner, so that later remindings occur only forhighly similar cases. For example, in 1984 Ross gavepeople mathematical problems to study and latergave them new problems. Most of their later re-mindings were to examples that were similar only onthe surface, irrespective of whether the principlesmatched. Experts in a domain are more likely thannovices to retrieve structurally similar examples, buteven experts retrieve some examples that are similaronly on the surface. However, as demonstrated byLaura Novick in 1988, experts reject spurious re-mindings more quickly than do novices. Thus, espe-cially for novices, there is an unfortunatedissociation: While accuracy of transfer dependscritically on the degree of structural match, memoryretrieval depends largely on surface similarity be-tween objects and contexts.

Analogical Encoding in Learning

In the late twentieth century, researchers began ex-ploring a new technique, called analogical encoding,that can help overcome the inert knowledge prob-lem. Instead of studying cases separately, learners areasked to compare analogous cases and describe theirsimilarities. This fosters the formation of a commonschema, which in turn facilitates transfer to a furtherproblem. For example, in 1999 Jeffrey Loewenstein,Leigh Thompson, and Gentner found that graduatemanagement students who compared two analogicalcases were nearly three times more likely to transferthe common strategy into a subsequent negotiationtask than were students who analyzed the same twocases separately.

Implications for Education

Analogies can be of immense educational value.They permit rapid learning of a new domain by

transferring knowledge from a known domain, andthey promote noticing and abstracting principlesacross domains. Analogies are most successful, how-ever, if their pitfalls are understood. In analogicalmapping, it is important to ensure that the base do-main is understood well, that the correspondencesare clear, and that differences and potentially incor-rect inferences are clearly flagged. When teaching fortransfer, it is important to recognize that learnerstend to rely on surface features. One solution is tominimize surface features by using simple objects.Another is to induce analogical encoding by askinglearners to explicitly compare cases. The better edu-cators understand analogical processes, the betterthey can harness them for education.

See also: Learning, subentry on Transfer ofLearning; Learning Theory, subentry on Histor-ical Overview.

BIBLIOGRAPHY

Bassok, Miriam. 2001. ‘‘Semantic Alignments inMathematical Word Problems.’’ In The Analogi-cal Mind: Perspectives from Cognitive Science, ed.Dedre Gentner, Keith J. Holyoak, and Biocho N.Kokinov. Cambridge, MA: MIT Press.

Bransford, John D.; Franks, Jeffrey J.; Vye,Nancy J.; and Sherwood, Robert D. 1989.‘‘New Approaches to Instruction: Because Wis-dom Can’t Be Told.’’ In Similarity and Analogi-cal Reasoning, ed. Stella Vosniadou and AndrewOrtony. New York: Cambridge University Press.

Brown, Ann L. 1989. ‘‘Analogical Learning andTransfer: What Develops?’’ In Similarity andAnalogical Reasoning, ed. Stella Vosniadou andAndrew Ortony. New York: Cambridge Univer-sity Press.

Brown, Ann L., and Kane, Mary Jo. 1988. ‘‘Pre-school Children Can Learn to Transfer: Learn-ing to Learn and Learning from Example.’’Cognitive Psychology 20:493–523.

Chen, Zhe, and Daehler, Marvin W. 1989. ‘‘Posi-tive and Negative Transfer in Analogical Prob-lem Solving by Six-Year-Old Children.’’Cognitive Development 4:327–344.

Clement, John. 1993. ‘‘Using Bridging Analogiesand Anchoring Intuitions to Deal with Students’Preconceptions in Physics.’’ Journal of Researchin Science Teaching 30:1241–1257.

Gentner, Dedre. 1982. ‘‘Are Scientific AnalogiesMetaphors?’’ In Metaphor: Problems and Per-

LEARNING: ANALOGICAL REASONING 1423

spectives, ed. David S. Miall. Brighton, Eng.:Harvester Press.

Gentner, Dedre. 1983. ‘‘Structure-Mapping: ATheoretical Framework for Analogy.’’ CognitiveScience 7:155–170.

Gentner, Dedre, and Rattermann, Mary Jo.1991. ‘‘Language and the Career of Similarity.’’In Perspectives on Thought and Language: Inter-relations in Development, ed. Susan A. Gelmanand James P. Brynes. London: Cambridge Uni-versity Press.

Gentner, Dedre; Rattermann, Mary Jo; andForbus, Kenneth D. 1993. ‘‘The Roles of Simi-larity in Transfer: Separating Retrievability fromInferential Soundness.’’ Cognitive Psychology25:524–575.

Gentner, Dedre, and Toupin, Cecile. 1986. ‘‘Sys-tematicity and Surface Similarity in the Devel-opment of Analogy.’’ Cognitive Science 10:277–300.

Gentner, Dedre, and Wolff, Phillip. 2000.‘‘Metaphor and Knowledge Change.’’ In Cogni-tive Dynamics: Conceptual Change in Humansand Machines, ed. Eric Dietrich and Arthur B.Markman. Mahwah, NJ: Erlbaum.

Gick, Mary L., and Holyoak, Keith J. 1980. ‘‘Ana-logical Problem Solving.’’ Cognitive Psychology12:306–355.

Gick, Mary L., and Holyoak, Keith J. 1983.‘‘Schema Induction and Analogical Transfer.’’Cognitive Psychology 15:1–38.

Goswami, Usha. 1992. Analogical Reasoning inChildren. Hillsdale, NJ: Erlbaum.

Halford, Graeme S. 1993. Children’s Understand-ing: The Development of Mental Models. Hills-dale, NJ: Erlbaum.

Holyoak, Keith J., and Koh, K. 1987. ‘‘Surface andStructural Similarity in Analogical Transfer.’’Memory and Cognition 15:332–340.

Holyoak, Keith J., and Thagard, Paul R. 1995.Mental Leaps: Analogy in Creative Thought.Cambridge, MA: MIT Press.

Inagaki, Kayoko, and Hatano, Giyoo. 1987.‘‘Young Children’s Spontaneous Personificationas Analogy.’’ Child Development 58:1013–1020.

Kempton, Willet. 1986. ‘‘Two Theories of HomeHeat Control.’’ Cognitive Science 10:75–90.

Kolodner, Janet L. 1997. ‘‘Educational Implica-tions of Analogy: A View from Case-Based Rea-soning.’’ American Psychologist 52:(1)57–66.

Loewenstein, Jeffrey; Thompson, Leigh; andGentner, Dedre. 1999. ‘‘Analogical EncodingFacilitates Knowledge Transfer in Negotiation.’’Psychonomic Bulletin and Review 6:586–597.

Markman, Arthur B., and Gentner, Dedre.2000. ‘‘Structure Mapping in the ComparisonProcess.’’ American Journal of Psychology113:501–538.

Novick, Laura R. 1988. ‘‘Analogical Transfer,Problem Similarity, and Expertise.’’ Journal ofExperimental Psychology: Learning, Memory, andCognition 14:510–520.

Perfetto, Greg A.; Bransford, John D.; andFranks, Jeffrey J. 1983. ‘‘Constraints on Ac-cess in a Problem Solving Context.’’ Memoryand Cognition 11:24–31.

Reed, Steve K. 1987. ‘‘A Structure-Mapping Modelfor Word Problems.’’ Journal of ExperimentalPsychology: Learning, Memory, and Cognition13:124–139.

Ross, Brian H. 1984. ‘‘Remindings and Their Ef-fects in Learning a Cognitive Skill.’’ CognitivePsychology 16:371–416.

Ross, Brian H. 1987. ‘‘This Is Like That: The Useof Earlier Problems and the Separation of Simi-larity Effects.’’ Journal of Experimental Psycholo-gy: Learning, Memory, and Cognition 13:629–639.

Ross, Brian H. 1989. ‘‘Distinguishing Types of Su-perficial Similarities: Different Effects on theAccess and Use of Earlier Problems.’’ Journal ofExperimental Psychology: Learning, Memory, andCognition 15:456–468.

Schank, Roger C.; Kass, Alex; and Riesbeck,Christopher K., eds. 1994. Inside Case-BasedExplanation. Hillsdale, NJ: Erlbaum.

Schwartz, Daniel L., and Bransford, John D.1998. ‘‘A Time for Telling.’’ Cognition and In-struction 16:475–522.

Spiro, Rand J.; Feltovich, Paul J.; Coulson,Richard L.; and Anderson, Daniel K. 1989.‘‘Multiple Analogies for Complex Concepts:Antidotes for Analogy-Induced Misconceptionin Advanced Knowledge Acquisition.’’ In Simi-larity and Analogical Reasoning, ed. Stella Vos-niadou and Andrew Ortony. New York:Cambridge University Press.

Dedre GentnerJeffrey Loewenstein

1424 LEARNING: ANALOGICAL REASONING

CAUSAL REASONING

A doorbell rings. A dog runs through a room. A seat-ed man rises to his feet. A vase falls from a table andbreaks. Why did the vase break? To answer this ques-tion, one must perceive and infer the causal relation-ships between the breaking of the vase and otherevents. Sometimes, the event most directly causallyrelated to an effect is not immediately apparent (e.g.,the dog hit the table), and conscious and effortfulthought may be required to identify it. People rou-tinely make such efforts because detecting causalconnections among events helps them to make senseof the constantly changing flow of events. Causalreasoning enables people to find meaningful orderin events that might otherwise appear random andchaotic, and causal understanding helps people toplan and predict the future. Thus, in 1980 the phi-losopher John Mackie described causal reasoning as‘‘the cement of the universe.’’ How, then, does onedecide which events are causally related? When doesone engage in causal reasoning? How does the abilityto think about cause–effect relations originate anddevelop during infancy and childhood? How cancausal reasoning skills be promoted in educationalsettings, and does this promote learning? Thesequestions represent important issues in research oncausal reasoning

Causal Perceptions and Causal Reasoning

An important distinction exists between causal per-ceptions and causal reasoning. Causal perceptionsrefer to one’s ability to sense a causal relationshipwithout conscious and effortful thought. Accordingto the philosopher David Hume (1711–1776), per-ceptual information regarding contiguity, prece-dence, and covariation underlies the understandingof causality. First, events that are temporally andspatially contiguous are perceived as causally related.Second, the causal precedes the effect. Third, eventsthat regularly co-occur are seen as causally related.In contrast, causal reasoning requires a person toreason through a chain of events to infer the causeof that event. People most often engage in causal rea-soning when they experience an event that is out ofthe ordinary. Thus, in some situations a person maynot know the cause of an unusual event and mustsearch for it, and in other situations must evaluatewhether one known event was the cause of another.The first situation may present difficulty because thecausal event may not be immediately apparent. Phi-losophers have argued that causal reasoning is based

on an assessment of criteria of necessity and suffi-ciency in these circumstances. A necessary cause isone that must be present for the effect to occur.Event A is necessary for event B if event B will notoccur without event A. For example, the vase wouldnot have broken if the dog had not hit the table. Acause is sufficient if its occurrence can by itself bringabout the effect (i.e., whenever event A occurs, eventB always follows). Often, more than one causal fac-tor is present. In the case of multiple necessarycauses, a set of causal factors taken together jointlyproduces an effect. In the case of multiple sufficientcauses, multiple factors are present, any one ofwhich by itself is sufficient to produce an effect.

The Development of Causal Perception andCausal Reasoning Skills

Causal perception appears to begin during infancy.Between three and six months of age, infants re-spond differently to temporally and spatially contig-uous events (e.g., one billiard ball contacting asecond that begins to roll immediately) compared toevents that lack contiguity (e.g., the second ball be-gins to roll without collision or does not start tomove until half a second after collision). Thus, thepsychologist Alan Leslie proposed in 1986 that in-fants begin life with an innate perceptual mechanismspecialized to automatically detect cause–effect rela-tions based on contiguity. However, psychologistsLeslie Cohen and Lisa Oakes reported in 1993 thatfamiliarity with role of a particular object in a causalsequence influence ten-month-old infants’ percep-tion of causality. Therefore, they suggest that infantsdo not automatically perceive a causal connectionwhen viewing contiguous events. The question ofwhether infants begin with an innate ability to auto-matically detect causality, or instead gradually devel-op casual perception through general learningprocesses remains a central controversy concerningthe origins of causal thought.

Although infants perceive causal relationships,complex causal reasoning emerges during earlychildhood and grows in sophistication thereafter.Thus, information about precedence influencescausal reasoning during childhood. When asked todetermine what caused an event to occur, three-year-olds often choose an event that preceded it,rather than one that came later, but understandingof precedence becomes more consistent and generalbeginning at five years of age. Unlike contiguity andprecedence, information about covariation is not

LEARNING: CAUSAL REASONING 1425

available from a single casual sequence, but requiresrepeated experience with the co-occurrence of acause and effect. Children do not begin to use co-variation information consistently in their casualthinking before eight years of age. Because the vari-ous types of information relevant to causality do notalways suggest the same causal relation, children andadults must decide which type of information ismost important in a particular situation.

In addition to the perceptual cues identified byHume, knowledge of specific causal mechanismsplays a central role in causal reasoning. By threeyears of age, children expect there to be some mech-anism of transmission between cause and effect, andknowledge of possible mechanisms influences bothchildren’s and adults’ interpretation of perceptualcues. For instance, when a possible causal mecha-nism requires time to produce an effect (e.g., a mar-ble rolling down a lengthy tube before contactinganother object), or transmits quickly across a dis-tance (e.g., electrical wiring), children as young asfive years of age are more likely to select causes thatlack temporal spatial contiguity than would other-wise be the case. Because causal mechanisms differfor physical, social, and biological events, childrenmust acquire distinct conceptual knowledge to un-derstand causality in each of these domains. By threeto four years of age, children recognize that whereasphysical effects are caused by physical transmission,human action is motivated internally by mentalstates such as desires, beliefs, and intentions, andthey begin to understand some properties of biologi-cal processes such as growth and heredity. Further-more, conceptual understanding of specific causalmechanisms may vary across cultures and may belearned through social discourse as well as throughdirect experience.

A fundamental understanding of causality ispresent during early childhood; however, prior toadolescence children have difficulty searching forcausal relations through systematic scientific experi-mentation. Preadolescents may generate a singlecausal hypothesis and seek confirmatory evidence,misinterpret contradictory evidence, or design ex-perimental tests that do not provide informative evi-dence. In contrast, adolescents and adults maygenerate several alternative hypotheses and test themby systematically controlling variables and seekingboth disconfirmatory and confirmatory evidence.Nevertheless, even adults often have difficulty de-signing valid scientific experiments. More generally,

both children and adults often have difficulty identi-fying multiple necessary or sufficient causes.

Teaching Causal Reasoning Skills

The psychologist Diane Halpern argued in 1998 thatcritical thinking skills should be taught in primary,secondary, and higher educational settings. Casualreasoning is an important part of critical thinkingbecause it enables one to explain and predict events,and thus potentially to control one’s environmentand achieve desired outcomes.

Three approaches to teaching causal reasoningskills may be efficacious. First, causal reasoning skillscan be promoted by teaching students logical deduc-tion. For example, teaching students to use counter-factual reasoning may help them assess whetherthere is a necessary relationship between a potentialcause and an effect. Counterfactual reasoning re-quires student to imagine that a potential cause didnot occur and to infer whether the effect would haveoccurred in its absence. If it would occur, then thereis no causal relationship between the two events.

Second, causal reasoning skills can be promotedby teaching students to generate informal explana-tions for anomalous events or difficult material. Forinstance, learning from scientific texts can be partic-ularly challenging to students, and often studentshave the misconception that they do not have ade-quate knowledge to understand texts. The psycholo-gist Michelene Chi demonstrated in 1989 thatstudents who use their general world knowledge toengage in causal, explanatory reasoning while read-ing difficult physics texts understand what they readconsiderably better than do students who do notdraw upon general knowledge in this way. Further-more, in 1999 the psychologist Danielle McNamaradeveloped a reading training intervention that pro-motes explanatory reasoning during reading. In thisprogram, students were taught a number of strate-gies to help them to use both information in the textand general knowledge to generate explanations fordifficult material. Training improved both compre-hension of scientific texts and overall class perfor-mance, and was particularly beneficial to at-riskstudents.

Third, the psychologist Leona Schauble demon-strated in 1990 that causal reasoning skills can bepromoted by teaching students the principles of sci-entific experimentation. A primary goal of experi-mentation is to determine causal relationships

1426 LEARNING: CAUSAL REASONING

among a set of events. Students may be taught toidentify a potential cause of an effect, manipulate thepresence of the cause in a controlled setting, and as-sesses whether or not the effect occurs. Thus, stu-dents learn to use the scientific method to determinewhether there are necessary and sufficient relation-ships between a potential cause and an effect. Be-cause the principles of science are often difficult forstudents to grasp, teaching these principles wouldprovide students with formal procedures for evalu-ating causal relationships in the world around them.

See also: Learning, subentry on Reasoning;Learning Theory, subentry on Historical Over-view; Literacy, subentry on Narrative Compre-hension and Production; Reading, subentries onComprehension, Content Areas.

BIBLIOGRAPHY

Bullock, Merry; Gelman, Rochel; and Baillar-geon, Renee. 1982. ‘‘The Development ofCausal Reasoning.’’ In The Developmental Psy-chology of Time, ed. William J. Friedman. NewYork: Academic Press.

Chi, Michelene T. H., et al. 1989. ‘‘Self-Explanation: How Students Study and Use Ex-amples in Learning to Solve Problems.’’ Cogni-tive Science 13:145–182.

Cohen, Leslie B., and Oakes, Lisa M. 1993. ‘‘HowInfants Perceive a Simple Causal Event.’’ Devel-opmental Psychology 29:421–433.

Epstein, Richard L. 2002. Critical Thinking, 2ndedition. Belmont, CA: Wadsworth.

Halpern, Diane F. 1998. ‘‘Teaching Critical Think-ing for Transfer across Domains.’’ AmericanPsychologist 53:449–455.

Hume, David. 1960. A Treatise on Human Nature(1739). Oxford: Clarendon Press.

Kuhn, D.; Amsel, Eric; and O’Loughlin, Mi-chael. 1988. The Development of ScientificThinking Skills. San Diego, CA: Academic Press.

Leslie, Alan M. 1986. ‘‘Getting Development offthe Ground: Modularity and the Infant’s Per-ception of Causality.’’ In Theory Building in De-velopmental Psychology, ed. Paul Van Geert.Amsterdam: North-Holland.

Mackie, John L. 1980. The Cement of the Universe.Oxford: Clarendon Press.

McNamara, Danielle S., and Scott, Jeremy L.1999. Training Reading Strategies. Hillsdale, NJ:Erlbaum.

Schauble, Leona. 1990. ‘‘Belief Revision in Chil-dren: The Role of Prior Knowledge and Strate-gies for Generating Evidence.’’ Journal ofExperimental Child Psychology 49:31–57.

Sedlak, Andrea J., and Kurtz, Susan T. 1981. ‘‘AReview of Children’s Use of Causal InferencePrinciples.’’ Child Development 52:759–784.

Wellman, Henry M., and Gelman, Susan A. 1998.‘‘Knowledge Acquisition in Foundational Do-mains.’’ In Handbook of Child Psychology: Cog-nition, Perception, and Language, 5th edition, ed.Deanna Kuhn and Robert Siegler. New York:Wiley.

White, Peter A. 1988. ‘‘Causal Processing: Originsand Development.’’ Psychological Bulletin104:36–52.

Joseph P. MaglianoBradford H. Pillow

CONCEPTUAL CHANGE

The term conceptual change refers to the develop-ment of fundamentally new concepts, through re-structuring elements of existing concepts, in thecourse of knowledge acquisition. Conceptual changeis a particularly profound kind of learning—it goesbeyond revising one’s specific beliefs and involvesrestructuring the very concepts used to formulatethose beliefs. Explaining how this kind of learningoccurs is central to understanding the tremendouspower and creativity of human thought.

The emergence of fundamentally new ideas isstriking in the history of human thought, particular-ly in science and mathematics. Examples include theemergence of Darwin’s concept of evolution by nat-ural selection, Newton’s concepts of gravity and in-ertia, and the mathematical concepts of zero,negative, and rational numbers. One of the chal-lenges of education is how to transmit these complexproducts of human intellectual history to the nextgeneration of students.

Although there are many unresolved issuesabout how concepts are mentally represented, con-ceptual-change researchers generally assume that ex-planatory concepts are defined and articulatedwithin theory-like structures, and that conceptual

LEARNING: CONCEPTUAL CHANGE 1427

change requires coordinated changes in multipleconcepts within these structures. New concepts thathave arisen in the history of science are clearly partof larger, explicit theories. Making an analogy be-tween the organization of concepts in scientists andchildren, researchers have proposed that childrenmay have ‘‘commonsense’’ theories in which theireveryday explanatory concepts are embedded andplay a role. These theories, although not self-consciously held, are assumed to be like scientifictheories in that they consist of a set of interrelatedconcepts that resist change and that support infer-ence making, problem solving, belief formation, andexplanation in a given domain. The power and use-fulness of this analogy is being explored in the earlytwenty-first century.

A challenge for conceptual-change researchers isto provide a typology of important forms of concep-tual change. For example, conceptual differentiationis a form of conceptual change in which a newer (de-scendant) theory uses two distinct concepts wherethe initial (parent) theory used only one, and the un-differentiated parent concept unites elements thatwill subsequently be kept distinct. Examples of con-ceptual differentiation include: Galileo’s differentia-tion of average and instantaneous velocity in histheory of motion, Black’s differentiation of heat andtemperature in his theory of thermal phenomena,and children’s differentiation of weight and densityin their matter theory. Conceptual differentiation isnot the same as adding new subcategories to an ex-isting category, which involves the elaboration of aconceptual structure rather than its transformation.In that case, the new subcategories fit into an exist-ing structure, and the initial general category is stillmaintained. In differentiation, the parent concept isseen as incoherent from the perspective of the subse-quent theory and plays no role in it. For example, anundifferentiated weight/density concept that unitesthe elements heavy and heavy-for-size combines twofundamentally different kinds of quantities: an ex-tensive (total amount) quantity and an intensive (re-lationally defined) quantity.

Another form of conceptual change is coales-cence, in which the descendant theory introduces anew concept that unites concepts previously seen tobe of fundamentally different types in the parenttheory. For example, Aristotle saw circular planetaryand free-fall motions as natural motions that werefundamentally different from violent projectile mo-tions. Newton coalesced circular, planetary, free-fall,

and projectile motions under a new category, accel-erated motion. Similarly, children initially see plantsand animals as fundamentally different: animals arebehaving beings that engage in self-generated move-ment, while plants are not. Later they come to seethem as two forms of ‘‘living things’’ that share im-portant biological properties. Conceptual coales-cence is not the same as simply adding a moregeneral category by abstracting properties commonto more specific categories. In conceptual coales-cence the initial concepts are thought to be funda-mentally different, and the properties that will becentral to defining the new category are not repre-sented as essential properties of the initial concepts.

Different forms of conceptual change mutuallysupport each other. For example, conceptual coales-cences (such as uniting free-fall and projectile mo-tion in a new concept of accelerated motion, orplants and animals in a new concept of living things)are accompanied by conceptual differentiations(such as distinguishing uniform from acceleratedmotion, or distinguishing dead from inanimate).These changes are also supported by additionalforms of conceptual change, such as re-analysis ofthe core properties or underlying structure of theconcept, as well as the acquisition of new specific be-liefs about the relations among concepts.

Mechanisms of Conceptual Change

One reason for distinguishing conceptual changefrom belief revision and conceptual elaboration isthat different learning mechanisms may be required.Everyday learning involves knowledge enrichmentand rests on an assumed set of concepts. For exam-ple, people use existing concepts to represent newfacts, formulate new beliefs, make inductive or de-ductive inferences, and solve problems.

What makes conceptual change so challengingto understand is that it cannot occur in this way. Theconcepts of a new theory are ultimately organizedand stated in terms of each other, rather than theconcepts of the old theory, and there is no simpleone-to-one correspondence between some conceptsof the old and new theories. By what learning mech-anisms, then, can scientists invent, and studentscomprehend, a genuinely new set of concepts andcome to prefer them to their initial set of concepts?

Most theorists agree that one step in conceptualchange for both students and scientists is experienc-ing some form of cognitive dissonance—an internal

1428 LEARNING: CONCEPTUAL CHANGE

state of tension that arises when an existing concep-tual system fails to handle important data and prob-lems in a satisfactory manner. Such dissonance canbe created by a series of unexpected results that can-not be explained by an existing theory, by the pressto solve a problem that is beyond the scope of one’scurrent theory, or by the detection of internal incon-sistencies in one’s thinking. This dissonance can sig-nal the need to step outside the normal mode ofapplying one’s conceptual framework to a moremeta-conceptual mode of questioning, examining,and evaluating one’s conceptual framework.

Although experiencing dissonance can signalthat there is a conceptual problem to be solved, itdoes not solve that problem. Another step involvesactive attempts to invent or construct an under-standing of alternative conceptual systems by usinga variety of heuristic procedures and symbolic tools.Heuristic procedures, such as analogical reasoning,imagistic reasoning, and thought experiments, maybe particularly important because they allow bothstudents and scientists to creatively extend, combine,and modify existing conceptual resources via theconstruction of new models. Symbolic tools, such asnatural language, the algebraic and graphical repre-sentations of mathematics, and other invented nota-tional systems, allow the explicit representation ofkey relations in the new system of concepts.

In analogical reasoning, knowledge of concep-tual relations in better-understood domains arepowerful sources of new ideas about the less-understood domain. Analogical reasoning is oftensupported by imagistic reasoning, wherein onecreates visual depictions of core ideas using visualanalogs with the same underlying relational struc-ture. These depictions allow the visualization of un-seen theoretical entities, connect the problem to thewell-developed human visual-spatial inferencingsystem, and, because much mathematical informa-tion is implicit in such depictions, facilitate the con-struction of appropriate mathematical descriptionsof a given domain. Thought experiments use initialknowledge of a domain to run simulations of whatshould happen in various idealized situations, in-cluding imagining what happens as the effects of agiven variable are entirely eliminated, thus facilitat-ing the identification of basic principles not self-evident from everyday observation.

Case studies of conceptual change in the historyof science and science education reveal that new in-tellectual constructions develop over an extended

period of time and include intermediate, bridgingconstructions. For example, Darwin’s starting ideaof evolution via directed, adaptative variation initial-ly prevented his making an analogy between thisprocess and artificial selection. He transformed hisunderstanding of this process using multiple analo-gies (first with wedging and Malthusian populationpressure, and later with artificial selection), imagisticreasoning (e.g., visualizing the jostling effects of100,000 wedges being driven into the same spot ofground to understand the tremendous power of theunseen force in nature and its ability to produce spe-cies change in a mechanistic manner), and thoughtexperiments (e.g., imagining how many small effectsmight build up over multiple generations to yield alarger effect). Each contributed different elements tohis final concept of natural selection, with his initialanalogies leading to the bridging idea of selectionacting in concert with the process of directed adap-tive variation, rather than supplanting it.

Constructing a new conceptual system is alsoaccompanied by a process of evaluating its adequacyagainst known alternatives using some set of criteria.These criteria can include: the new system’s abilityto explain the core problematic phenomena as wellas other known phenomena in the domain, its inter-nal consistency and fit with other relevant knowl-edge, the extent to which it meets certainexplanatory ideals, and its capacity to suggest newfruitful lines of research.

Finally, researchers have examined the personal,motivational, and social processes that support con-ceptual change. Personal factors include courage,confidence in one’s abilities, openness to alterna-tives, willingness to take risks, and deep commit-ment to an intellectual problem. Social factorsinclude working in groups that combine differentkinds of expertise and that encourage considerationof inconsistencies in data and relevant analogies. In-deed, many science educators believe a key to pro-moting conceptual change in the classroom isthrough creating a more reflective classroom dis-course. Such discourse probes for alternative studentviews, encourages the clarification, negotiation, andelaboration of meanings, the detection of inconsis-tencies, and the use of evidence and argument in de-ciding among or integrating alternative views.

Educational Implications

Conceptual change is difficult under any circum-stances, as it requires breaking out of the self-

LEARNING: CONCEPTUAL CHANGE 1429

perpetuating circle of theory-based reasoning, mak-ing coordinated changes in a number of concepts,and actively constructing an understanding of new(more abstract) conceptual systems. Students needsignals that conceptual change is needed, as well asgood reasons to change their current conceptions,guidance about how to integrate existing conceptualresources in order to construct new conceptions,and the motivation and time needed to make thoseconstructions. Traditional education practice oftenfails to provide students with the appropriate signals,guidance, motivation, and time.

Conceptual change is a protracted process call-ing for a number of coordinated changes in instruc-tional practice. First, instruction needs to begrounded in the consideration of important phe-nomena or problems that are central to the experts’framework—and that challenge students’ initialcommonsense framework. These phenomena notonly motivate conceptual change, but also constrainthe search for, and evaluation of, viable alternatives.Second, instruction needs to guide students in theconstruction of new systems of concepts for under-standing these phenomena. Teachers must knowwhat heuristic techniques, representational tools,and conceptual resources to draw upon to make newconcepts intelligible to students, and also how tobuild these constructions in a sequenced manner.

Third, instruction needs to be supported by aclassroom discourse that encourages students toidentify, represent, contrast, and debate the adequa-cy of competing explanatory frameworks in terms ofemerging classroom epistemological standards. Suchdiscourse supports many aspects of the conceptual-change process, including making students aware oftheir initial conceptions, helping students constructan understanding of alternative frameworks, moti-vating students to examine their conceptions morecritically (in part through awareness of alternatives),and promoting their ability to evaluate, and at timesintegrate, competing frameworks.

Finally, instruction needs to provide studentswith extended opportunities for applying new sys-tems of concepts to a wide variety of problems. Re-peated applications develop students’ skill atapplying a new framework, refine their understand-ing of the framework, and help students appreciateits greater power and scope.

See also: Categorization and Concept Learn-ing; Learning, subentry on Knowledge Acquisi-tion, Representation, and Organization.

BIBLIOGRAPHY

Carey, Susan. 1999. ‘‘Sources of ConceptualChange.’’ In Conceptual Development: Piaget’sLegacy, ed. Ellin K. Scholnick, Katherine Nelson,Susan A. Gelman, and Patricia H. Miller. Mah-wah, NJ: Erlbaum.

Chi, Micheline. 1992. ‘‘Conceptual Change withinand across Ontological Categories: Examplesfrom Learning and Discovery in Science.’’ InCognitive Models of Science, ed. Ronald N. Giere.Minnesota Studies in the Philosophy of Science,Vol. 15. Minneapolis: University of MinnesotaPress.

Chinn, Clark A., and Brewer, William F. 1993.‘‘The Role of Anomalous Data in KnowledgeAcquisition: A Theoretical Framework and Im-plications for Science Instruction.’’ Review ofEducational Research 63(1):1–49.

Clement, John. 1993. ‘‘Using Bridging Analogiesand Anchoring Intuitions to Deal with Students’Preconceptions in Physics.’’ Journal of Researchin Science Teaching 30(10):1241–1257.

Dunbar, Kenneth. 1995. ‘‘How Scientists ReallyReason: Scientific Reasoning in Real-WorldLaboratories.’’ In The Nature of Insight, ed. Rob-ert J. Sternberg and Janet E. Davidson. Cam-bridge, MA: MIT Press.

Gentner, Dedre; Brem, Sarah; Ferguson, Ron-ald; Markman, Arthur; Levidow, Bjorn;Wolff, Phillip; and Forbus, Kenneth. 1997.‘‘Analogical Reasoning and Conceptual Change:A Case Study of Johannes Kepler.’’ Journal of theLearning Sciences 6(1):3–40.

Laurence, Stephen, and Margolis, Eric. 1999.‘‘Concepts and Cognitive Science.’’ In Concepts:Core Readings, ed. Eric Margolis and StephenLaurence. Cambridge, MA: MIT Press.

Lehrer, Richard; Schauble, Leona; Carpenter,Susan; and Penner, David. 2000. ‘‘The Inter-related Development of Inscriptions and Con-ceptual Understanding.’’ In Symbolizing andCommunicating in Mathematics Classrooms: Per-spectives on Discourse, Tools, and InstructionalDesign, ed. Paul Cobb, Erna Yackel, and KayMcClain. Mahwah, NJ: Erlbaum.

1430 LEARNING: CONCEPTUAL CHANGE

Millman, Arthur B., and Smith, Carol L. 1997.‘‘Darwin’s Use of Analogical Reasoning in The-ory Construction.’’ Metaphor and Symbol12(3):159–187.

Nercessian, Nancy J. 1992. ‘‘How Do ScientistsThink? Capturing the Dynamics of ConceptualChange in Science.’’ In Cognitive Models of Sci-ence, ed. Ronald N. Giere. Minnesota Studies inthe Philosophy of Science, Vol. 15. Minneapolis:University of Minnesota Press.

Pintrich, Paul R.; Marx, Ronald W.; and Boyle,Robert A. 1993. ‘‘Beyond Cold ConceptualChange: The Role of Motivational Beliefs andClassroom Contextual Factors in the Process ofConceptual Change.’’ Review of Educational Re-search 63(2):167–199.

Posner, Gerald; Strike, Kenneth; Hewson,Peter; and Gertzog, W. A. 1982. ‘‘Accommo-dation of a Scientific Conception: Toward aTheory of Conceptual Change.’’ Science Educa-tion 66:211–227.

Smith, Carol; Maclin, Deborah; Grosslight,Lorraine; and Davis, Helen. 1997. ‘‘Teachingfor Understanding: A Study of Students’ Pre-instruction Theories of Matter and a Compari-son of the Effectiveness of Two Approaches toTeaching about Matter and Density.’’ Cognitionand Instruction 15(3):317–393.

van Zee, Emily, and Minstrell, Jim. 1997. ‘‘UsingQuestioning to Guide Student Thinking.’’ Jour-nal of the Learning Sciences 6:227–269.

Vosniadou, Stella, and Brewer, William F.1987. ‘‘Theories of Knowledge Restructuring inDevelopment.’’ Review of Educational Research57:51–67.

White, Barbara. 1993. ‘‘ThinkerTools: CausalModels, Conceptual Change, and Science In-struction.’’ Cognition and Instruction 10:1–100.

Wiser, Marianne, and Amin, Tamir. 2001. ‘‘‘IsHeat Hot?’ Inducing Conceptual Change by In-tegrating Everyday and Scientific Perspectiveson Thermal Phenomena.’’ Learning and Instruc-tion 11(4–5):331–355.

Carol L. Smith

KNOWLEDGE ACQUISITION,REPRESENTATION, AND

ORGANIZATION

Knowledge acquisition is the process of absorbingand storing new information in memory, the successof which is often gauged by how well the informa-tion can later be remembered (retrieved from mem-ory). The process of storing and retrievinginformation depends heavily on the representationand organization of the information. Moreover, theutility of knowledge can also be influenced by howthe information is structured. For example, a busschedule can be represented in the form of a map ora timetable. On the one hand, a timetable providesquick and easy access to the arrival time for each bus,but does little for finding where a particular stop issituated. On the other hand, a map provides a de-tailed picture of each bus stop’s location, but cannotefficiently communicate bus schedules. Both formsof representation are useful, but it is important to se-lect the representation most appropriate for the taskat hand. Similarly, knowledge acquisition can be im-proved by considering the purpose and function ofthe desired information.

Knowledge Representation and Organization

There are numerous theories of how knowledge isrepresented and organized in the mind, includingrule-based production models, distributed net-works, and propositional models. However, thesetheories are all fundamentally based on the conceptof semantic networks. A semantic network is a meth-od of representing knowledge as a system of connec-tions between concepts in memory.

Semantic Networks

According to semantic network models, knowledgeis organized based on meaning, such that semanti-cally related concepts are interconnected. Knowl-edge networks are typically represented as diagramsof nodes (i.e., concepts) and links (i.e., relations).The nodes and links are given numerical weights torepresent their strengths in memory. In Figure 1, thenode representing DOCTOR is strongly related toSCALPEL, whereas NURSE is weakly related toSCALPEL. These link strengths are represented herein terms of line width. Similarly, some nodes in Fig-ure 1 are printed in bold type to represent theirstrength in memory. Concepts such as DOCTORand BREAD are more memorable because they are

LEARNING: KNOWLEDGE ACQUISITION, REPRESENTATION, AND ORGANIZATION 1431

FIGURE 1

more frequently encountered than concepts such as

SCALPEL and CRUST.

Mental excitation, or activation, spreads auto-

matically from one concept to another related con-

cept. For example, thinking of BREAD spreads

activation to related concepts, such as BUTTER and

CRUST. These concepts are primed, and thus more

easily recognized or retrieved from memory. For ex-

ample, in David Meyer and Roger Schvaneveldt’s

1976 study (a typical semantic priming study), a se-

ries of words (e.g., BUTTER) and nonwords (e.g.,

BOTTOR) are presented, and participants deter-

mine whether each item is a word. A word is more

quickly recognized if it follows a semantically related

word. For example, BUTTER is more quickly recog-

nized as a word if BREAD precedes it, rather than

NURSE. This result supports the assumption that se-

mantically related concepts are more strongly con-

nected than unrelated concepts.

Network models represent more than simple as-

sociations. They must represent the ideas and com-

plex relationships that comprise knowledge and

comprehension. For example, the idea ‘‘The doctor

uses a scalpel’’ can be represented as the proposition

USE (DOCTOR, SCALPEL), which consists of the

nodes DOCTOR and SCALPEL and the link USE

(see Figure 2). Educators have successfully used sim-

ilar diagrams, called concept maps, to communicate

important relations and attributes among the key

concepts of a lesson.

Types of Knowledge

There are numerous types of knowledge, but themost important distinction is between declarativeand procedural knowledge. Declarative knowledgerefers to one’s memory for concepts, facts, or epi-sodes, whereas procedural knowledge refers to theability to perform various tasks. Knowledge of howto drive a car, solve a multiplication problem, orthrow a football are all forms of procedural knowl-edge, called procedures or productions. Proceduralknowledge may begin as declarative knowledge, butis proceduralized with practice. For example, whenfirst learning to drive a car, you may be told to ‘‘putthe key in the ignition to start the car,’’ which is adeclarative statement. However, after starting the carnumerous times, this act becomes automatic and iscompleted with little thought. Indeed, proceduralknowledge tends to be accessed automatically andrequire little attention. It also tends to be more dura-ble (less susceptible to forgetting) than declarativeknowledge.

Knowledge Acquisition

Listed below are five guidelines for knowledge acqui-sition that emerge from how knowledge is represent-ed and organized.

Process the material semantically. Knowledge is or-ganized semantically; therefore, knowledge acquisi-tion is optimized when the learner focuses on themeaning of the new material. Fergus Craik andEndel Tulving were among the first to provide evi-dence for the importance of semantic processing. Intheir studies, participants answered questions con-

1432 LEARNING: KNOWLEDGE ACQUISITION, REPRESENTATION, AND ORGANIZATION

FIGURE 2

cerning target words that varied according to thedepth of processing involved. For example, semanticquestions (e.g., Which word, friend or tree, fits ap-propriately in the following sentence: ‘‘He met a____ on the street’’?) involve a greater depth of pro-cessing than phonemic questions (e.g., Which word,crate or tree, rhymes with the word late?), which inturn have a greater depth than questions concerningthe structure of a word (e.g., Which word is in capi-tal letters: TREE or tree?). Craik and colleaguesfound that words processed semantically were betterlearned than words processed phonemically orstructurally. Further studies have confirmed thatlearning benefits from greater semantic processingof the material.

Process and retrieve information frequently. Asecond learning principle is to test and retrieve theinformation numerous times. Retrieving, or self-producing, information can be contrasted with sim-ply reading or copying it. Decades of research on aphenomenon called the generation effect have shownthat passively studying items by copying or readingthem does little for memory in comparison to self-producing, or generating, an item. Moreover, learn-ing improves as a function of the number of timesinformation is retrieved. Within an academic situa-tion, this principle points to the need for frequentpractice tests, worksheets, or quizzes. In terms ofstudying, it is also important to break up, or distrib-ute retrieval attempts. Distributed retrieval can in-clude studying or testing items in a random order,

with breaks, or on different days. In contrast, repeat-ing information numerous times sequentially in-volves only a single retrieval from long-termmemory, which does little to improve memory forthe information.

Learning and retrieval conditions should be simi-lar. How knowledge is represented is determined bythe conditions and context (internal and external) inwhich it is learned, and this in turn determines howit is retrieved: Information is best retrieved when theconditions of learning and retrieval are the same.This principle has been referred to as encoding speci-ficity. For example, in one experiment, participantswere shown sentences with an adjective and a nounprinted in capital letters (e.g. The CHIP DIP tasteddelicious.) and told that their memory for the nounswould be tested afterward. In the recognition test,participants were shown the noun either with theoriginal adjective (CHIP DIP), with a different ad-jective (SKINNY DIP), or without an adjective(DIP). Noun recognition was better when the origi-nal adjective (CHIP) was presented than when noadjective was presented. Moreover, presenting a dif-ferent adjective (SKINNY) yielded the lowest recog-nition. This finding underscores the importance ofmatching learning and testing conditions.

Encoding specificity is also important in termsof the questions used to test memory or comprehen-sion. Different types of questions tap into differentlevels of understanding. For example, recalling in-

LEARNING: KNOWLEDGE ACQUISITION, REPRESENTATION, AND ORGANIZATION 1433

formation involves a different level of understand-ing, and different mental processes, than recognizinginformation. Likewise, essay and open-ended ques-tions assess a different level of understanding thanmultiple-choice questions. Essay and open-endedquestions generally tap into a conceptual or situa-tional understanding of the material, which resultsfrom an integration of text-based information andthe reader’s prior knowledge. In contrast, multiple-choice questions involve recognition processes, andtypically assess a shallow or text-based understand-ing. A text-based representation can be impover-ished and incomplete because it consists only ofconcepts and relations within the text. This level ofunderstanding, likely developed by a student prepar-ing for a multiple-choice exam, would be inappro-priate preparation for an exam with open-ended oressay questions. Thus, students should benefit by ad-justing their study practices according to the expect-ed type of questions.

Alternatively, students may benefit from review-ing the material in many different ways, such as rec-ognizing the information, recalling the information,and interpreting the information. These latter pro-cesses improve understanding and maximize theprobability that the various ways the material isstudied will match the way it is tested. From a teach-er’s point of view, including different types of ques-tions on worksheets or exams ensures that eachstudent will have an opportunity to convey their un-derstanding of the material.

Connect new information to prior knowledge.Knowledge is interconnected; therefore, new materi-al that is linked to prior knowledge will be better re-tained. A driving factor in text and discoursecomprehension is prior knowledge. Skilled readersactively use their prior knowledge during compre-hension. Prior knowledge helps the reader to fill incontextual gaps within the text and develop a betterglobal understanding or situation model of the text.Given that texts rarely (if ever) spell out everythingneeded for successful comprehension, using priorknowledge to understand text and discourse is criti-cal. Moreover, thinking about what one alreadyknows about a topic provides connections in memo-ry to the new information—the more connectionsthat are formed, the more likely the information willbe retrievable from memory.

Create cognitive procedures. Procedural knowledgeis better retained and more easily accessed. There-fore, one should develop and use cognitive proce-

dures when learning information. Procedures caninclude shortcuts for completing a task (e.g., usingfast 10s to solve multiplication problems), as well asmemory strategies that increase the distinctivemeaning of information. Cognitive research has re-peatedly demonstrated the benefits of memory strat-egies, or mnemonics, for enhancing the recall ofinformation. There are numerous types of mnemon-ics, but one well-known mnemonic is the method ofloci. This technique was invented originally for thepurpose of memorizing long speeches in the timesbefore luxuries such as paper and pencil were readilyavailable. The first task is to imagine and memorizea series of distinct locations along a familiar route,such as a pathway from one campus building to an-other. Each topic of a speech (or word in a word list)can then be pictured in a location along the route.When it comes time to recall the speech or word list,the items are simply found by mentally traveling thepathway.

Mnemonics are generally effective because theyincrease semantic processing of the words (orphrases) and render them more meaningful by link-ing them to familiar concepts in memory. Mnemon-ics also provide ready-made, effective cues forretrieving information. Another important aspect ofmnemonics is that mental imaging is often involved.Images not only render information more meaning-ful, but they provide an additional route for findinginformation in memory. As mentioned earlier, in-creasing the number of meaningful links to informa-tion in memory increases the likelihood it can beretrieved.

Strategies are also an important component ofmeta-cognition, which is the ability to think about,understand, and manage one’s learning. First, onemust develop an awareness of one’s own thoughtprocesses. Simply being aware of thought processesincreases the likelihood of more effective knowledgeconstruction. Second, the learner must be aware ofwhether or not comprehension has been successful.Realizing when comprehension has failed is crucialto learning. The final, and most important stage ofmeta-cognitive processing is fixing the comprehen-sion problem. The individual must be aware of, anduse, strategies to remedy comprehension and learn-ing difficulties. For successful knowledge acquisitionto occur, all three of these processes must occur.Without thinking or worrying about learning, thestudent cannot realize whether the concepts havebeen successfully grasped. Without realizing that in-

1434 LEARNING: KNOWLEDGE ACQUISITION, REPRESENTATION, AND ORGANIZATION

formation has not been understood, the studentcannot engage in strategies to remedy the situation.If nothing is done about a comprehension failure,awareness is futile.

Conclusion

Knowledge acquisition is integrally tied to how themind organizes and represents information. Learn-ing can be enhanced by considering the fundamentalproperties of human knowledge, as well as by the ul-timate function of the desired information. Themost important property of knowledge is that it isorganized semantically; therefore, learning methodsshould enhance meaningful study of new informa-tion. Learners should also create as many links to theinformation as possible. In addition, learning meth-ods should be matched to the desired outcome. Justas using a bus timetable to find a bus-stop locationis ineffective, learning to recognize information willdo little good on an essay exam.

See also: Learning, subentry on ConceptualChange; Reading, subentry on Content Areas.

BIBLIOGRAPHY

Anderson, John R. 1982. ‘‘Acquisition of a Cogni-tive Skill.’’ Psychological Review 89:369–406.

Anderson, John R., and Lebière, Christian. 1998.The Atomic Components of Thought. Mahwah,NJ: Erlbaum.

Bransford, John, and Johnson, Marcia K. 1972.‘‘Contextual Prerequisites for UnderstandingSome Investigations of Comprehension and Re-call.’’ Journal of Verbal Learning and Verbal Be-havior 11:717–726.

Craik, Fergus I. M., and Tulving, Endel. 1975.‘‘Depth of Processing and the Retention ofWords in Episodic Memory.’’ Journal of Experi-mental Psychology: General. 194:268–294.

Crovitz, Herbert F. 1971. ‘‘The Capacity of Mem-ory Loci in Artificial Memory.’’ PsychonomicScience 24:187–188.

Glenberg, Arthur M. 1979. ‘‘Component-LevelsTheory of the Effects of Spacing of Repetitionson Recall and Recognition.’’ Memory and Cog-nition 7:95–112.

Guastello, Francine; Beasley, Mark; andSinatra, Richard. 2000. ‘‘Concept MappingEffects on Science Content Comprehension of

Low-Achieving Inner-City Seventh Graders.’’RASE: Remedial and Special Education 21:356–365.

Hacker, Douglas J.; Dunlosky, John; and Gr-aesser, Arthur C. 1998. Metacognition in Edu-cational Theory and Practice. Mahwah, NJ:Lawrence Erlbaum.

Jensen, Mary Beth, and Healy, Alice F. 1998.‘‘Retention of Procedural and Declarative Infor-mation from the Colorado Drivers’ Manual.’’ InMemory Distortions and their Prevention, ed.Margaret Jean Intons-Peterson and Deborah L.Best. Mahwah, NJ: Lawrence Erlbaum.

Kintsch, Walter. 1998. Comprehension: A Para-digm for Cognition. Cambridge, Eng.: Cam-bridge University Press.

Light, Leah L., and Carter-Sobell, Linda. 1970.‘‘Effects of Changed Semantic Context on Rec-ognition Memory.’’ Journal of Verbal Learningand Verbal Behavior 9:1–11.

McNamara, Danielle S., and Kintsch, Walter.1996. ‘‘Learning from Text: Effects of PriorKnowledge and Text Coherence.’’ DiscourseProcesses 22:247–287.

Melton, Arthur W. 1967. ‘‘Repetition and Re-trieval from Memory.’’ Science 158:532.

Meyer, David E., and Schvaneveldt, Roger W.1976. ‘‘Meaning, Memory Structure, and Men-tal Processes.’’ Science 192:27–33.

Paivio, Allen. 1990. Mental Representations: ADual Coding Approach. New York: Oxford Uni-versity Press.

Rumelhart, David E., and McClelland, James L.1986. Parallel Distributed Processing: Explora-tions in the Microstructure of Cognition, Vol. 1:Foundations. Cambridge, MA: MIT Press.

Slamecka, Norman J., and Graf, Peter. 1978.‘‘The Generation Effect: Delineation of a Phe-nomenon.’’ Journal of Experimental Psychology:Human Learning and Memory 4:592–604.

Tulving, Endel, and Thompson, Donald M.1973. ‘‘Encoding Specificity and Retrieval Pro-cesses in Episodic Memory.’’ Psychological Re-view 80:352–373.

Yates, Francis A. 1966. The Art of Memory. Chica-go: University of Chicago Press.

Danielle S. McNamaraTenaha O’Reilly

LEARNING: KNOWLEDGE ACQUISITION, REPRESENTATION, AND ORGANIZATION 1435

NEUROLOGICAL FOUNDATION

Learning is mediated by multiple memory systemsin the brain, each of which involves a distinct ana-tomical pathway and supports a particular form ofmemory representation. The major aim of researchon memory systems is to identify and distinguish thedifferent contributions of specific brain structuresand pathways, usually by contrasting the effects ofselective damage to specific brain areas. Anothermajor strategy focuses on localizing brain areas thatare activated, that is, whose neurons are activatedduring particular aspects of memory processing.Some of these studies use newly developed function-al imaging techniques to view activation of brainareas in humans performing memory tests. Anotherapproach seeks to characterize the cellular code formemory within the activity patterns of single nervecells in animals, by asking how information is repre-sented by the activity patterns within the circuits ofdifferent structures in the relevant brain systems.

Each of the brain’s memory systems begins inthe vast expanse of the cerebral cortex, specifically inthe so-called cortical association areas (see Figure 1).These parts of the cerebral cortex provide major in-puts to each of three main pathways of processingin subcortical areas related to distinct memory func-tions. One system mediates declarative memory, thememory for facts and events that can be brought toconscious recollection and can be expressed in a va-riety of ways outside the context of learning. Thissystem involves connections from the cortical asso-ciation areas to the hippocampus via the parahippo-campal region. The main output of hippocampaland parahippocampal processing is back to the samecortical areas that provided inputs to the hippocam-pus, and are viewed as the long-term repository ofdeclarative memories.

The other two main pathways involve corticalinputs to specific subcortical targets that send directoutputs that control behavior. One of these systemsmediates emotional memory, the attachment of affili-ations and aversions towards otherwise arbitrarystimuli and modulation of the strength of memoriesthat involve emotional arousal. This system involvescortical (as well as subcortical) inputs to the amyg-dala as the nodal stage in the association of sensoryinputs to emotional outputs effected via the hypo-thalamic-pituitary axis and autonomic nervous sys-tem, as well as emotional influences over widespreadbrain areas. The second of these systems mediates

procedural memory, the capacity to acquire habitualbehavioral routines that can be performed withoutconscious control. This system involves cortical in-puts to the striatum as a nodal stage in the associa-tion of sensory and motor cortical information withvoluntary responses via the brainstem motor system.An additional, parallel pathway that mediates differ-ent aspects of sensori-motor adaptations involvessensory and motor systems pathways through thecerebellum.

The Declarative Memory System

Declarative memory is the ‘‘everyday’’ form of mem-ory that most consider when they think of memory.Therefore, the remainder of this discussion willfocus on the declarative memory system. Declarativememory is defined as a composite of episodic mem-ory, the ability to recollect personal experiences, andsemantic memory, the synthesis of the many episod-ic memories into the knowledge about the world. Inaddition, declarative memory supports the capacityfor conscious recall and the flexible expression ofmemories, one’s ability to search networks of epi-sodic and semantic memories and to use this capaci-ty to solve many problems.

Each of the major components of the declarativememory system contributes differently to declarativememory, although interactions between these areasare also essential. Initially, perceptual information aswell as information about one’s behavior is pro-cessed in many dedicated neocortical areas. Whilethe entire cerebral cortex is involved in memory pro-cessing, the chief brain area that controls this pro-cessing is the prefrontal cortex. The processingaccomplished by the prefrontal cortex includes theacquisition of complex cognitive rules and conceptsand working memory, the capacity to store informa-tion briefly while manipulating or rehearsing the in-formation under conscious control. In addition, theareas of the cortex also contribute critically to mem-ory processing. Association areas in the prefrontal,temporal, and parietal cortex play a central role incognition and in both the perception of sensory in-formation and in maintenance of short-term tracesof recently perceived stimuli. Furthermore, the orga-nization of perceptual representations in cerebralcortical areas, and connections among these areas,are permanently modified by learning experiences,constituting the long term repository of memories.

The parahippocampal region, which receivesconvergent inputs from the neocortical association

1436 LEARNING: NEUROLOGICAL FOUNDATION

areas and sends return projections to all of theseareas, appears to mediate the extended persistence ofthese cortical representations. Through interactionsbetween these areas, processing within the cortexcan take advantage of lasting parahippocampal rep-resentations, and so come to reflect complex associa-tions between events that are processed separately indifferent cortical regions or occur sequentially in thesame or different areas.

These individual contributions and their inter-actions are not conceived as sufficient to link repre-sentations of events to form episodic memories orto form generalizations across memories to create asemantic memory network. Such an organization re-quires the capacity to rapidly encode a sequence ofevents that make up an episodic memory, to retrievethat memory by re-experiencing one facet of theevent, and to link the ongoing experience to storedepisodic representations, forming the semantic net-work. The neuronal elements of the hippocampuscontain the fundamental coding properties that cansupport this kind of organization.

However, interactions among the componentsof the system are undoubtedly critical. It is unlikelythat the hippocampus has the storage capacity tocontain all of one’s episodic memories and the hip-pocampus is not the final storage site. Therefore, itseems likely that the hippocampal neurons are in-volved in mediating the reestablishment of detailedcortical representations, rather than storing the de-tails themselves. Repetitive interactions between thecortex and hippocampus, with the parahippocampalregion as intermediary, serve to sufficiently coacti-vate widespread cortical areas so that they eventuallydevelop linkages between detailed memories withouthippocampal mediation. In this way, the networkingprovided by the hippocampus underlies its role inthe organization of the permanent memory net-works in the cerebral cortex.

See also: Brain-Based Education.

BIBLIOGRAPHY

Eichenbaum, Howard. 2000. ‘‘A Cortical-Hippocampal System for Declarative Memory.’’Nature Reviews Neuroscience 1:41–50.

Eichenbaum, Howard, and Cohen, Neal J. 2001.From Conditioning to Conscious Recollection:Memory Systems of the Brain. Upper SaddleRiver, NJ: Oxford University Press.

FIGURE 1

Schacter, Daniel L., and Tulving, Endel, eds.1994. Memory Systems 1994. Cambridge, MA:MIT Press.

Squire, Larry R., and Kandel, Eric R. 1999. Mem-ory: From Mind to Molecules. New York: Scien-tific American Library.

Squire, Larry R.; Knowlton, Barbara; andMusen, Gail. 1993. ‘‘The Structure and Orga-nization of Memory.’’ Annual Review of Psychol-ogy 44:453–495.

Howard Eichenbaum

PERCEPTUAL PROCESSES

As Eleanor Gibson wrote in her classic text Principlesof Perceptual Learning and Development, perceptuallearning results in changes in the pickup of informa-tion as a result of practice or experience. Perceptionand action are a cycle: People act in order to learnabout their surroundings, and they use what theylearn to guide their actions. From this perspective,the critical defining features of perception includethe exploratory actions of the perceiver and theknowledge of the events, animate and inanimate ob-jects, and surrounding environment gained while

LEARNING: PERCEPTUAL PROCESSES 1437

engaged in looking, listening, touching, walking, andother forms of direct observation. Perception oftenresults in learning information that is directly rele-vant to the goals at hand, but sometimes it resultsin learning that is incidental to one’s immediategoals.

Perception becomes more skillful with practiceand experience, and perceptual learning can bethought of as the education of attention. Perceiverscome to notice the features of situations that are rel-evant to their goals and not to notice the irrelevantfeatures. Three general principles of perceptuallearning seem particularly relevant. First, unskillfulperceiving requires much concentrated attention,whereas skillful perceiving requires less attentionand is more easily combined with other tasks. Sec-ond, unskillful perceiving involves noticing both therelevant and irrelevant features of sensory stimula-tion without understanding their meaning or rele-vance to one’s goals, whereas skillful perceivinginvolves narrowing one’s focus to relevant featuresand understanding the situations they specify. Andthird, unskillful perceiving often involves attentionto the proximal stimulus (that is, the patterns oflight or acoustic or pressure information on the reti-nas, cochleae, and skin, respectively), whereas skill-ful perceiving involves attention to the distal eventthat is specified by the proximal stimulus.

Different Domains

Perceptual learning refers to relatively durable gainsin perception that occur across widely different do-mains. For example, at one extreme are studies dem-onstrating that with practice adults can gainexquisite sensitivity to vernier discriminations, thatis, the ability to resolve gaps in lines that approachthe size of a single retinal receptor. At the oppositeextreme, perceptual learning plays a central role ingaining expertise in the many different content areasof work, everyday life, and academic pursuits.

In the realm of work, classic examples includefarmers learning to differentiate the sex of chickens,restaurateurs learning to differentiate different di-mensions of fine wine, airplane pilots misperceivingtheir position relative to the ground, and machinistsand architects learning to ‘‘see’’ the three-dimensional shape of a solid object or house fromthe top, side, and front views.

In the realm of everyday life, important exam-ples include learning to perceive emotional expres-

sions, learning to identify different people andunderstand their facial expressions, learning to dif-ferentiate the different elements of speech whenlearning a second language, and learning to differen-tiate efficient routes to important destinations whenfaced with new surroundings.

In ‘‘nonacademic’’ subjects within the realm ofacademic pursuits, important examples involvemusic, art, and sports. For example, music studentslearn to differentiate the notes, chords, and instru-mental voices in a piece, and they learn to identifypieces by period and composer. Art students learnto differentiate different strokes, textures, and styles,and they learn to classify paintings by period andartist. Athletes learn to differentiate the different de-grees of freedom that need to be controlled to pro-duce a winning ‘‘play’’ and to anticipate what actionsneed to be taken when on a playing field.

Finally, perceptual learning plays an equallybroad role in classically academic subjects. For ex-ample, mathematics students gain expertise at per-ceiving graphs, classifying the shapes of curves, andknowing what equations might fit a given curve. Sci-ence students gain expertise at perceiving laboratorysetups. These range widely across grade levels anddomains, including the critical features of hydrolyz-ing water in a primary school general science setting,molecular structures in organic chemistry and ge-netics, frog dissections in biology, the functional re-lation of the frequency of waves and diffraction indifferent media in physics, and the critical featuresof maps in geology.

The borders separating perceptual learningfrom conceiving and reasoning often becomeblurred. And indeed, people perceive in order to un-derstand, and their understanding leads to more andmore efficient perception. For example, Herbert A.Simon elaborated on this in 2001 in his discussionof the visual thinking involved in having an expertunderstanding of the dynamics of a piston in an in-ternal combustion engine. When experts look at apiston or a diagram of a piston or a graph represent-ing the dynamics of a piston, they ‘‘see’’ the higherorder, relevant variables, for example, that morework is performed when the combustion explosionmoves the piston away from the cylinder’s base thanwhen the piston returns toward the base. The abilityto ‘‘see’’ such higher-order relations is not just aquestion of good visual acuity, but it instead de-pends on content knowledge (about energy, pres-sure, and work) and on an understanding of how

1438 LEARNING: PERCEPTUAL PROCESSES

energy acts in the context of an internal combustionengine. In a 2001 article, Daniel Schwartz and JohnBransford emphasized that experience with con-trasting cases helps students differentiate the criticalfeatures when they are working to understand statis-tics and other academic domains. In a 1993 article,J. Littlefield and John Rieser demonstrated the skillof middle school students at differentiating relevantfrom irrelevant information when attempting tosolve story problems in mathematics.

Classical Issues in Perceptual Learning andPerceptual Development

Perceptual development involves normative age-related changes in basic sensory sensitivities and inperceptual learning. Some of these changes are con-strained by the biology of development in well-defined ways. For example, the growth in auditoryfrequency during the first year of life is mediated inpart by changes in the middle ear and inner ear.Growth in visual acuity during the first two years ismediated in several ways: by changes in the migra-tion of retinal cells into a fovea, through increasingcontrol of convergence eye movements so that thetwo eyes fixate the same object, and through increas-ing control of the accommodate state of the lens sothat fixated objects are in focus. The role of physicalchanges in the development of other perceptualskills, for example, perceiving different cues fordepth, is less clear.

Nativism and empiricism are central to thestudy of perception and perceptual development.Stemming from philosophy’s interest in epistemolo-gy, early nativists (such as seventeenth-centuryFrench mathematician and philosopher René Des-cartes and eighteenth-century German philosopherImmanuel Kant) argued that the basic capacities ofthe human mind were innate, whereas empiricistsargued that they were learned, primarily through as-sociations. This issue has long been hotly debated inthe field of perceptual learning and development.How is it that the mind and brain come to perceivethree-dimensional shapes from two-dimensionalretinal projections; perceive distance; segment thespeech stream; represent objects that become cov-ered from view? The debate is very lively in the earlytwenty-first century, with some arguing that percep-tion of some basic properties of the world is innate,and others arguing that it is learned, reflecting thestatistical regularities in experience. Given that expe-rience plays a role in some forms of perceptual learn-

ing, there is evidence that the timing of theexperience can be critical to whether, and to whatdegree, it is learned effectively.

The ‘‘constancy’’ of perception is a remarkablefeat of perceptual development. The issue is that theenergy that gives rise to the perception of a particu-lar object or situation varies widely when the per-ceiver or object moves, the lighting changes, and soforth. Given the flux in the sensory input, how is itthat people manage to perceive that the objects andsituations remain (more or less) the same? Researchabout perceptual constancies has reemerged as animportant topic as computer scientists work to de-sign artificial systems that can ‘‘learn to see.’’

Intersensory coordination is a major feature ofperception and perceptual development. How is it,for example, that infants can imitate adult modelswho open their mouths wide or stick out theirtongues? How is it that infants can identify objectsby looking at them or by touching them and can rec-ognize people by seeing them or listening to them?

The increasing control of actions with age is amajor result of perceptual learning, as infants be-come more skillful at perceiving steps and other fea-tures of the ground and learn to control theirbalance when walking up and down slopes.

In 1955 James Gibson and Eleanor Gibsonwrote an important paper titled ‘‘Perceptual Learn-ing: Differentiation or Enrichment?’’ By differentia-tion they meant skill at distinguishing smaller andsmaller differences among objects of a given kind. Byenrichment they meant knowledge of the ways thatobjects and events tend to be associated with otherobjects and events. Their paper was in part a reactionto the predominant view of learning at the time: thatlearning was the ‘‘enrichment’’ of responses throughtheir association with largely arbitrary stimulus con-ditions. The authors provided a sharp counterpointto this view. Instead of conceiving of the world asconstructed by add-on processes of association, theyviewed perceivers as actively searching for the stimu-li they needed to guide their actions and decisions,and in this way coming to differentiate the relevantfeatures situated in a given set of circumstances fromthe irrelevant ones.

See also: Attention; Learning Theory, subentryon Historical Overview.

LEARNING: PERCEPTUAL PROCESSES 1439

BIBLIOGRAPHY

Acredolo, Linda P.; Pick, Herb L.; and Olsen, M.1975. ‘‘Environmental Differentiation and Fa-miliarity as Determinants of Children’s Memoryfor Spatial Location.’’ Developmental Psychology11:495–501.

Adolph, Karen E. 1997. Learning in the Develop-ment of Infant Locomotion. Chicago: Universityof Chicago Press.

Arnheim, Rudolph. 1974. Art and Visual Percep-tion: A Psychology of the Creative Eye. Berkeley:University of California Press.

Aslin, Richard N. 1998. ‘‘Speech and AuditoryProcessing during Infancy: Constraints on andPrecursors to Language.’’ In Handbook of ChildPsychology, 5th edition, ed. William Damon,Vol. 2: Cognition, Perception, and Language, ed.Deanna Kuhn and Robert S. Siegler. New York:Wiley.

Bahrick, Lorraine E., and Lickliter, Robert.2000. ‘‘Intersensory Redundancy Guides Atten-tional Selectivity and Perceptual Learning in In-fancy.’’ Developmental Psychology 36:190–201.

Baillargeon, Renee. 1994. ‘‘How Do Infants Learnabout the Physical World?’’ Current Directionsin Psychological Science 3:133–140.

Barsalou, Lawrence W. 1999. ‘‘Perceptual SymbolSystems.’’ Behavior and Brain Sciences 22:577–660.

Bransford, John D., and Schwartz, Daniel L.2000. ‘‘Rethinking Transfer: A Simple Proposalwith Multiple Implications.’’ Review of Researchin Education 24:61–100.

Bryant, Peter, and Somerville, S. 1986. ‘‘TheSpatial Demands of Graphs.’’ British Journal ofPsychology 77:187–197.

Dodwell, Peter C., ed. 1970. Perceptual Learningand Adaptation: Selected Readings. Harmonds-worth, Eng.: Penguin.

Dowling, W. Jay., and Harwood, Dane L. 1986.Music Cognition. New York: Academic Press.

Epstein, William. 1967. Varieties of PerceptualLearning. New York: McGraw-Hill.

Fahle, Manfred, and Poggio, Tomaso, eds. 2000.Perceptual Learning. Cambridge, MA: MITPress.

Garling, Tommy, and Evans, Gary W. 1991. Envi-ronment, Cognition, and Action: An IntegratedApproach. New York: Oxford University Press.

Gibson, Eleanor J. 1969. Principles of PerceptualLearning and Development. Englewood Cliffs,NJ: Prentice-Hall.

Gibson, Eleanor J., and Pick, Anne D. 2000. AnEcological Approach to Perceptual Learning andDevelopment. New York: Oxford UniversityPress.

Gibson, Eleanor J., and Walk, Richard D. 1961.‘‘The ‘Visual Cliff.’’’ Scientific American 202:64–71.

Gibson, James J., and Gibson, Eleanor J. 1955.‘‘Perceptual Learning: Differentiation or En-richment?’’ Psychological Review 62:32–41.

Goldstone, Robert L. 1998. ‘‘Perceptual Learn-ing.’’ Annual Review of Psychology 49:585–612.

Goodnow, Jacqueline J. 1978. ‘‘Visible Thinking:Cognitive Aspects of Change in Drawings.’’Child Development 49:637–641.

Granrud, Carl E. 1993. Visual Perception and Cog-nition in Infancy. Hillsdale, NJ: Erlbaum.

Haber, Ralph N. 1987. ‘‘Why Low-Flying FighterPlanes Crash: Perceptual and Attentional Fac-tors in Collisions with the Ground.’’ HumanFactors 29:519–532.

Johnson, Jacqueline S., and Newport, Elissa L.1989. ‘‘Critical Period Effects in Second Lan-guage Learning: The Influence of MaturationalState on the Acquisition of English as a SecondLanguage.’’ Cognitive Psychology 21:60–99.

Johnson, Mark. 1998. ‘‘The Neural Basis of Cogni-tive Development.’’ In Handbook of Child Psy-chology, 5th edition, ed. William Damon, Vol. 2:Cognition, Perception, and Language, ed. DeannaKuhn and Robert S. Siegler. New York: Wiley.

Jusczyk, Peter W. 2002. ‘‘How Infants AdaptSpeech-Processing Capacities to Native Lan-guage Structure.’’ Current Directions in Psycho-logical Science 11:15–18.

Kellman, Philip, and Banks, Martin S. 1998. ‘‘In-fant Visual Perception.’’ In Handbook of ChildPsychology, 5th edition, ed. William Damon,Vol. 2: Cognition, Perception, and Language, ed.Deanna Kuhn and Robert S. Siegler. New York:Wiley.

Littlefield, J., and Rieser, John J. 1993. ‘‘Seman-tic Features of Similarity and Children’s Strate-gies for Identifying Relevant Information inMathematical Story Problems.’’ Cognition andInstruction 11:133–188.

1440 LEARNING: PERCEPTUAL PROCESSES

McLeod, Peter; Reed, Nick; and Diences, Zol-tan. 2001. ‘‘Toward a Unified Fielder Theory:What We Do Not Yet Know about How PeopleRun to Catch a Ball.’’ Journal of ExperimentalPsychology: Human Perception and Performance27:1347–1355.

Postman, Leo. 1955. ‘‘Association Theory and Per-ceptual Learning.’’ Psychological Review 62:438–446.

Quinn, Paul C.; Palmer, Vanessa; and Slater,Alan M. 1999. ‘‘Identification of Gender in Do-mestic Cat Faces with and without Training:Perceptual Learning of a Natural CategorizationTask.’’ Perception 28:749–763.

Rieser, John J.; Pick, Herb L.; Ashmead, DanielH.; and Garing, A. E. 1995. ‘‘Calibration ofHuman Locomotion and Models of Perceptual-Motor Organization.’’ Journal of ExperimentalPsychology: Human Perception and Performance21:480–497.

Saarni, Carolyn. 1998. ‘‘Emotional Development:Action, Communication, and Understanding.’’In Handbook of Child Psychology, 5th edition,ed. William Damon, Vol. 3: Social, Emotional,and Personality Development, ed. Nancy Eisen-berg. New York: Wiley.

Saffran, Jenny R.; Aslin, R. N.; and Newport, E.L. 1996. ‘‘Statistical Learning by Eight-Month-Old Infants.’’ Science 274:1926–1928.

Saffran, Jenny R., and Griepentrog, G. J. 2001.‘‘Absolute Pitch in Infant Auditory Learning:Evidence for Developmental Reorganization.’’Developmental Psychology 37:74–85.

Schwartz, Daniel L., and Bransford, John D.2001. ‘‘A Time for Telling.’’ Cognition and In-struction 16:475–522.

Simon, Herbert A. 2001. ‘‘Observations on the Sci-ences of Science Learning.’’ Journal of AppliedDevelopmental Psychology 21:115–121.

Tighe, L. S., and Tighe, T. J. 1966. ‘‘DiscriminationLearning: Two Views in Historical Perspective.’’Psychological Bulletin 66:353–370.

von Hofsten, Claes. 1994. ‘‘Planning and Perceiv-ing What Is Going to Happen Next.’’ In The De-velopment of Future-Oriented Processes, ed.Marshall M. Haith, Janette B. Benson, andRalph J. Roberts. Chicago: University of Chica-go Press.

Walk, Richard D. 1966. ‘‘Perceptual Learning andthe Discrimination of Wines.’’ Psychonomic Sci-ence 5:57–58.

Walker-Andrews, Arlene, and Bahrick, Lor-raine E. 2001. ‘‘Perceiving the Real World: In-fants’ Detection of and Memory for SocialInformation.’’ Infancy 2:469–481.

Welch, Robert B. 1978. Perceptual Modification:Adapting to Altered Sensory Environments. NewYork: Academic Press.

John J. Rieser

PROBLEM SOLVING

Cognitive processing aimed at figuring out how toachieve a goal is called problem solving. In problemsolving, the problem solver seeks to devise a methodfor transforming a problem from its current stateinto a desired state when a solution is not immedi-ately obvious to the problem solver. Thus, the hall-mark of problem solving is the invention of a newmethod for addressing a problem. This definitionhas three parts: (1) problem solving is cognitive—that is, it occurs internally in the mind (or cognitivesystem) and must be inferred indirectly from behav-ior; (2) problem solving is a process—it involves themanipulation of knowledge representations (or car-rying out mental computations); and (3) problemsolving is directed—it is guided by the goals of theproblem solver.

The definition of problem solving covers abroad range of human cognitive activities, includingeducationally relevant cognition—figuring out howto manage one’s time, writing an essay on a selectedtopic, summarizing the main point of a textbooksection, solving an arithmetic word problem, or de-termining whether a scientific theory is valid by con-ducting experiments.

A problem occurs when a problem solver has agoal but initially does not know how to achieve thegoal. This definition has three parts: (1) the currentstate—the problem begins in a given state; (2) thegoal state—the problem solver wants the problem tobe in a different state, and problem solving is re-quired to transform the problem from the current(or given) state into the goal state, and (3) obsta-cles—the problem solver does not know the correctsolution and an effective solution method is not ob-vious to the problem solver.

LEARNING: PROBLEM SOLVING 1441

According to this definition a problem is per-sonal, so that a situation that is a problem for oneperson might not be a problem for another person.For example, ‘‘3 + 5 = ___’’ might be a problem fora six-year-old child who reasons, ‘‘Let’s see. I cantake one from the 5 and give it to the 3. That makes4 plus 4, and I know that 4 plus 4 is 8.’’ However,this equation is not a problem for an adult whoknows the correct answer.

Types of Problems

Routine and nonroutine problems. It is customaryto distinguish between routine and nonroutineproblems. In a routine problem, the problem solverknows a solution method and only needs to carry itout. For example, for most adults the problem ‘‘589x 45 = ___’’ is a routine problem if they know theprocedure for multicolumn multiplication. Routineproblems are sometimes called exercises, and techni-cally do not fit the definition of problem statedabove. When the goal of an educational activity is topromote all the aspects of problem solving (includ-ing devising a solution plan), then nonroutine prob-lems (or exercises) are appropriate.

In a nonroutine problem, the problem solverdoes not initially know a method for solving theproblem. For example, the following problem (re-ported by Robert Sternberg and Janet Davidson) isnonroutine for most people: ‘‘Water lilies double inarea every twenty-four hours. At the beginning ofthe summer, there is one water lily on the lake. Ittakes sixty days for the lake to be completely coveredwith water lilies. On what day is the lake half cov-ered?’’ In this problem, the problem solver must in-vent a solution method based on working backwardsfrom the last day. Based on this method, the prob-lem solver can ask what the lake would look like onthe day before the last day, and conclude that thelake is half covered on the fifty-ninth day.

Well-defined and ill-defined problems. It is alsocustomary to distinguish between well-defined andill-defined problems. In a well-defined problem, thegiven state of the problem, the goal state of the prob-lem, and the allowable operators (or moves) are eachclearly specified. For example, the following water-jar problem (adapted from Abrahama Luchins) is anexample of a well defined problem: ‘‘I will give youthree empty water jars; you can fill any jar with waterand pour water from one jar into another (until thesecond jar is full or the first one is empty); you canfill and pour as many times as you like. Given water

jars of size 21, 127, and 3 units and an unlimitedsupply of water, how can you obtain exactly 100units of water?’’ This is a well-defined problem be-cause the given state is clearly specified (you haveempty jars of size 21, 127, and 3), the goal state isclearly specified (you want to get 100 units of waterin one of the jars), and the allowable operators areclearly specified (you can fill and pour according tospecific procedures). Well-defined problems may beeither routine or nonroutine; if you do not have pre-vious experience with water jar problems, then find-ing the solution (i.e., fill the 127, pour out 21 once,and pour out 3 twice) is a nonroutine problem.

In an ill-defined problem, the given state, goalstate, and/or operations are not clearly specified. Forexample, in the problem, ‘‘Write a persuasive essayin favor of year-round schools,’’ the goal state is notclear because the criteria for what constitutes a ‘‘per-suasive essay’’ are vague and the allowable operators,such as how to access sources of information, are notclear. Only the given state is clear—a blank piece ofpaper. Ill-defined problems can be routine or non-routine; if one has extensive experience in writingthen writing a short essay like this one is a routineproblem.

Processes in Problem Solving

The process of problem solving can be broken downinto two major phases: problem representation, inwhich the problem solver builds a coherent mentalrepresentation of the problem, and problem solution,in which the problem solver devises and carries outa solution plan. Problem representation can be bro-ken down further into problem translation, in whichthe problem solver translates each sentence (or pic-ture) into an internal mental representation, andproblem integration, in which the problem solver in-tegrates the information into a coherent mental rep-resentation of the problem (i.e., a mental model ofthe situation described in the problem). Problem so-lution can be broken down further into solutionplanning, in which the problem solver devises a planfor how to solve the problem, and solution execution,in which the problem solver carries out the plan byengaging in solution behaviors. Although the fourprocesses of problem solving are listed sequentially,they may occur in many different orderings and withmany iterations in the course of solving a problem.

For example, consider the butter problem de-scribed by Mary Hegarty, Richard Mayer, and Chris-topher Monk: ‘‘At Lucky, butter costs 65 cents per

1442 LEARNING: PROBLEM SOLVING

stick. This is two cents less per stick than butter atVons. If you need to buy 4 sticks of butter, howmuch will you pay at Vons?’’ In the problem transla-tion phase, the problem solver may mentally repre-sent the first sentence as ‘‘Lucky = 0.65,’’ the secondsentence as ‘‘Lucky = Vons − 0.02,’’ and the thirdsentence as ‘‘4 x Vons = ___.’’ In problem integra-tion, the problem solver may construct a mentalnumber line with Lucky at 0.65 and Vons to the rightof Lucky (at 0.67); or the problem solver may men-tally integrate the equations as ‘‘4 x (Lucky + 0.02)= ____.’’ A key insight in problem integration is torecognize the proper relation between the cost ofbutter at Lucky and the cost of butter at Vons, name-ly that butter costs more at Vons (even though thekeyword in the problem is ‘‘less’’). In solution plan-ning, the problem solver may break the probleminto parts, such as: ‘‘First add 0.02 to 0.65, then mul-tiply the result by 4.’’ In solution executing, theproblem solver carries out the plan: 0.02 + 0.65 =0.67, 0.67 x 4 = 2.68. In addition, the problem solvermust monitor the problem-solving process andmake adjustments as needed.

Teaching for Problem Solving

A challenge for educators is to teach in ways that fos-ter meaningful learning rather than rote learning.Rote instructional methods promote retention (theability to solve problems that are identical or highlysimilar to those presented in instruction), but notproblem solving transfer (the ability to apply whatwas learned to novel problems). For example, in1929, Alfred Whitehead used the term inert knowl-edge to refer to learning that cannot be used to solvenovel problems. In contrast, meaningful instructionalmethods promote both retention and transfer.

In a classic example of the distinction betweenrote and meaningful learning, the psychologist MaxWertheimer (1959) described two ways of teachingstudents to compute the area of a parallelogram. Inthe rote method, students learn to measure the base,measure the height, and then multiply base timesheight. Students taught by the A = b x h method areable to find the area of parallelograms shaped likethe ones given in instruction (a retention problem)but not unusual parallelograms or other shapes (atransfer problem). Wertheimer used the term repro-ductive thinking to refer to problem solving in whichone blindly carries out a previously learned proce-dure. In contrast, in the meaningful method, stu-dents learn by cutting the triangle from one end of

a cardboard parallelogram and attaching it to theother end to form a rectangle. Once students havethe insight that a parallelogram is just a rectangle indisguise, they can compute the area because they al-ready know the procedure for finding the area of arectangle. Students taught by the insight methodperform well on both retention and transfer prob-lems. Wertheimer used the term productive thinkingto refer to problem solving in which one invents anew approach to solving a novel problem.

Educationally Relevant Advances in ProblemSolving

Recent advances in educational psychology point tothe role of domain-specific knowledge in problemsolving—such as knowledge of specific strategies orproblem types that apply to a particular field. Threeimportant advances have been: (1) the teaching ofproblem-solving processes, (2) the nature of expertproblem solving, and (3) new conceptions of indi-vidual differences in problem-solving ability.

Teaching of problem-solving processes. An impor-tant advance in educational psychology is cognitivestrategy instruction, which includes the teaching ofproblem-solving processes. For example, in ProjectIntelligence, elementary school children successfullylearned the cognitive processes needed for solvingproblems similar to those found on intelligence tests.In Instrumental Enrichment, students who had beenclassified as mentally retarded learned cognitive pro-cesses that allowed them to show substantial im-provements on intelligence tests.

Expert problem solving. Another important ad-vance in educational psychology concerns differ-ences between what experts and novices know ingiven fields, such as medicine, physics, and comput-er programming. For example, expert physicists tendto store their knowledge in large integrated chunks,whereas novices tend to store their knowledge as iso-lated fragments; expert physicists tend to focus onthe underlying structural characteristics of physicsword problems, whereas novices focus on the sur-face features; and expert physicists tend to work for-ward from the givens to the goal, whereas noviceswork backwards from the goal to the givens. Re-search on expertise has implications for professionaleducation because it pinpoints the kinds of domain-specific knowledge that experts need to learn.

Individual differences in problem-solving ability.This third advance concerns new conceptions of in-

LEARNING: PROBLEM SOLVING 1443

tellectual ability based on differences in the way peo-ple process information. For example, people maydiffer in cognitive style—such as their preferencesfor visual versus verbal representations, or for im-pulsive versus reflective approaches to problem solv-ing. Alternatively, people may differ in the speed andefficiency with which they carry out specific cogni-tive processes, such as making a mental comparisonor retrieving a piece of information from memory.Instead of characterizing intellectual ability as a sin-gle, monolithic ability, recent conceptions of intel-lectual ability focus on the role of multipledifferences in information processing.

See also: Creativity; Learning, subentry on Ana-logical Reasoning; Mathematics Learning,subentry on Complex Problem Solving.

BIBLIOGRAPHY

Chi, Michelene T. H.; Glaser, Robert; and Farr,Marshall J., eds. 1988. The Nature of Expertise.Hillsdale, NJ: Erlbaum.

Dunker, Karl. 1945. On Problem Solving. Washing-ton, DC: American Psychological Association.

Feuerstein, Reuven. 1980. Instrumental Enrich-ment. Baltimore: University Park Press.

Hegarty, Mary; Mayer, Richard E.; and Monk,Christopher A. 1995. ‘‘Comprehension ofArithmetic Word Problems: Evidence from Stu-dents’ Eye Fixations.’’ Journal of EducationalPsychology 84:76–84.

Hunt, Earl; Lunneborg, Cliff; and Lewis, J.1975. ‘‘What Does It Mean to Be High Verbal?’’Cognitive Psychology 7:194–227.

Larkin, Jill H.; McDermott, John; Simon, Doro-thea P.; and Simon, Herbert A. 1980. ‘‘Expertand Novice Performance in Solving PhysicsProblems.’’ Science 208:1335–1342.

Luchins, Abrahama S. 1942. Mechanization inProblem Solving: The Effect of Einstellung. Evans-ton, IL: American Psychological Association.

Mayer, Richard E. 1992. Thinking, Problem Solv-ing, Cognition, 2nd edition. New York: Freeman.

Mayer, Richard E. 1999. The Promise of Education-al Psychology. Upper Saddle River, NJ: Prentice-Hall.

Nickerson, Raymond S. 1995. ‘‘Project Intelli-gence.’’ In Encyclopedia of Human Intelligence,ed. Robert J. Sternberg. New York: Macmillan.

Pressley, Michael J., and Woloshyn, Vera. 1995.Cognitive Strategy Instruction that Really Im-proves Children’s Academic Performance. Cam-bridge, MA: Brookline Books.

Sternberg, Robert J., and Davidson, Janet E.1982. ‘‘The Mind of the Puzzler.’’ PsychologyToday 16:37–44.

Sternberg, Robert J., and Zhang, Li-Fang, eds.2001. Perspectives on Thinking, Learning, andCognitive Styles. Mahwah, NJ: Erlbaum.

Wertheimer, Max. 1959. Productive Thinking. NewYork: Harper and Row.

Whitehead, Alfred North. 1929. The Aims of Ed-ucation. New York: Macmillan.

Richard E. Mayer

REASONING

Reasoning is the generation or evaluation of claimsin relation to their supporting arguments and evi-dence. The ability to reason has a fundamental im-pact on one’s ability to learn from new informationand experiences because reasoning skills determinehow people comprehend, evaluate, and acceptclaims and arguments. Reasoning skills are also cru-cial for being able to generate and maintain view-points or beliefs that are coherent with, and justifiedby, relevant knowledge. There are two general kindsof reasoning that involve claims and evidence: for-mal and informal.

Formal Reasoning

Formal reasoning is used to evaluate the form of anargument, and to examine the logical relationshipsbetween conclusions and their supporting asser-tions. Arguments are determined to be either validor invalid based solely on whether their conclusionsnecessarily follow from their explicitly stated prem-ises or assertions. That is, if the supporting assertionsare true, must the conclusion also be true? If so, thenthe argument is considered valid and the truth of theconclusion can be directly determined by establish-ing the truth of the supporting assertions. If not,then the argument is considered invalid, and thetruth of the assertions is insufficient (or even irrele-vant) for establishing the truth of the conclusion.Formal reasoning is often studied in the context ofcategorical syllogisms or ‘‘if-then’’ conditional proofs.Syllogisms contain two assertions and a conclusion.

1444 LEARNING: REASONING

An example of a logically valid syllogism is: All dogsare animals; all poodles are dogs; therefore poodles areanimals. A slight change to one of the premises willcreate the invalid syllogism: All dogs are animals;some dogs are poodles; therefore all poodles are ani-mals. This argument form is invalid because it can-not be determined with certainty that the conclusionis true, even if the premises are true. The secondpremise does not require that all poodles are dogs.Thus, there may be some poodles who are not dogsand, by extension, some poodles who are not ani-mals. This argument is invalid despite the fact thatan accurate knowledge of dogs, poodles, and animalsconfirms that both the premises and the conclusionare true statements. This validity-truth incongru-ence highlights the important point that the concep-tual content of an argument or the real-world truthof the premises and conclusion are irrelevant to thelogic of the argument form.

Discussions of formal reasoning may sometimesrefer to the rules of logic. It is common for formalreasoning to be described as a set of abstract and pre-scriptive rules that people must learn and apply inorder to determine the validity of an argument. Thisis the oldest perspective on formal reasoning. Someclaim that the term formal reasoning refers directlyto the application of these formal rules.

However, many theorists consider this perspec-tive misguided. Describing formal reasoning as theevaluation of argument forms conveys a more inclu-sive and accurate account of the various perspectivesin this field. There are at least four competing theo-ries about how people determine whether a conclu-sion necessarily follows from the premises. Thesetheories are commonly referred to as rule-based per-spectives, mental models, heuristics, and domain-sensitive theories. People outside the rule-based per-spective view the rules of logic as descriptive rulesthat simply give labels to common argument formsand to common errors or fallacies in logical reason-ing. These theories are too complex to be detailedhere, and there is currently no consensus as to whichtheory best accounts for how people actually reason.A number of books and review articles provide com-prehensive discussions of these theories and theirrelative merits; one example is Human Reasoning:The Psychology of Deduction by Jonathan Evans, Ste-phen Newstead, and Ruth Byrne.

There is a consensus that human reasoning per-formance is poor and prone to several systematic er-rors. Performance on formal reasoning tasks is

generally poor, but can be better or worse dependingupon the particular aspects of the task. People per-form worse on problems that require more cognitivework, due to excessive demands placed on their lim-ited processing capacity or working memory. The re-quired cognitive work can be increased simply byhaving more information, or by the linguistic formof the argument. Some linguistic forms can affectperformance because they violate conventional dis-course or must be mentally rephrased in order to beintegrated with other information.

In addition, people’s existing knowledge aboutthe concepts contained in the problem can affectperformance. People have great difficulty evaluatingthe logical validity of an argument independent oftheir real-world knowledge. They insert their knowl-edge as additional premises, which leads them tomake more inferences than is warranted. Priorknowledge can also lead people to misinterpret themeaning of premises. Another common source oferror is belief bias, where people judge an argument’svalidity based on whether the conclusion is consis-tent with their beliefs rather than its logical relation-ship to the given premises.

The systematic errors that have been observedprovide some insights about what skills a personmight develop to improve performance. Making stu-dents explicitly aware of the likely intrusion of theirprior knowledge could facilitate their ability to con-trol or correct such intrusions. Students may alsobenefit from a detailed and explicit discussion ofwhat logical validity refers to, how it differs fromreal-world truth or personal agreement, and howeasy it is to confuse the two. Regardless of whetheror not people commonly employ formal rules oflogic, an understanding and explicit knowledge ofthese rules should facilitate efforts to search for vio-lations of logical validity. Theorists of informal rea-soning such as James Voss and Mary Means havemade a similar argument for the importance of ex-plicit knowledge about the rules of good reasoning.Errors attributed to limited cognitive resources canbe addressed by increasing reasoning skill, and prac-tice on formal reasoning tasks should increase profi-ciency and reduce the amount of cognitive effortrequired. Also, working memory load should be re-duced by external representation techniques, such asVenn diagrams.

LEARNING: REASONING 1445

Informal Reasoning

Informal reasoning refers to attempts to determinewhat information is relevant to a question, whatconclusions are plausible, and what degree of sup-port the relevant information provides for these var-ious conclusions. In most circumstances, peoplemust evaluate the justification for a claim in a con-text where the information is ambiguous and in-complete and the criteria for evaluation are complexand poorly specified. Most of what is commonly re-ferred to as ‘‘thinking’’ involves informal reasoning,including making predictions of future events or try-ing to explain past events. These cognitive processesare involved in answering questions as mundane as‘‘How much food should I prepare for this party?’’and as profound as ‘‘Did human beings evolve fromsimple one-celled organisms?’’ Informal reasoninghas a pervasive influence on both the everyday andthe monumental decisions that people make, and onthe ideas that people come to accept or reject.

Informal and formal reasoning both involve at-tempts to determine whether a claim has been suffi-ciently justified by the supporting assertions, butthese types of reasoning differ in many respects. Thevast majority of arguments are invalid according toformal logic, but informal reasoning must be em-ployed to determine what degree of justification thesupporting assertions provide. Also, the supportingassertions themselves must be evaluated as to theirvalidity and accuracy. Formal reasoning involvesmaking a binary decision based only on the given in-formation. Informal reasoning involves making anuncertain judgment about the degree of justificationfor a claim relative to competing claims—and basingthis evaluation on an ill-defined set of assertionswhose truth values are uncertain.

Based on the above characterization of informalreasoning, a number of cognitive skills would be ex-pected to affect the quality of such reasoning. Thefirst is the ability to fully comprehend the meaningof the claim being made. Understanding the concep-tual content is crucial to being able to consider whatother information might bear on the truth or false-hood of a claim. Other cognitive processes involvedin reasoning include the retrieval of relevant knowl-edge from long-term memory, seeking out new rele-vant information, evaluating the validity and utilityof that information, generating alternatives to theclaim in question, and evaluating the competingclaims in light of the relevant information.

Successful reasoning requires the understandingthat evidence must provide information that is inde-pendent of the claim or theory, and that evidencemust do more than simply rephrase and highlightthe assumptions of the theory. For example, the as-sertion ‘‘Some people have extrasensory perception’’does not provide any evidence about the claim ‘‘ESPis real.’’ These are simply ways of restating the sameinformation. Evidence must be an assertion that isindependent of the claim, but that still provides in-formation about the probable truth of the claim. Anexample of potential evidence for the claim that‘‘ESP is real’’ would be ‘‘Some people know informa-tion that they could not have known through any ofthe normal senses.’’ In other words, evidence consti-tutes assertions whose truth has implications for, butis not synonymous with, the truth of the claim beingsupported.

Without an understanding of evidence andcounterevidence and how they relate to theories,people would be ineffective at identifying informa-tion that could be used to determine whether a claimis justified. Also, lack of a clear distinction betweenevidence and theory will lead to the assimilation ofevidence and the distortion of its meaning and logi-cal implications. This eliminates the potential toconsider alternative claims that could better accountfor the evidence. People will also fail to use coun-terevidence to make appropriate decreases in the de-gree of justification for a claim.

Discussions of informal reasoning, argumenta-tion, and critical thinking commonly acknowledgethat a prerequisite for effective reasoning is a beliefin the utility of reasoning. The cognitive skills de-scribed above are necessary, but not sufficient, toproduce quality reasoning. The use of these skills isclearly effortful; thus, people must believe in theimportance and utility of reasoning in order to con-sistently put forth the required effort. The episte-mology that promotes the use of reasoning skills isthe view that knowledge can never be absolutely cer-tain and that valid and useful claims are the productof contemplating possible alternative claims andweighing the evidence and counterevidence. Putsimply, people use their reasoning skills consistentlywhen they acknowledge the possibility that a claimmay be incorrect and also believe that standards ofgood reasoning produce more accurate ideas aboutthe world.

Inconsistent, selective, and biased application ofreasoning skills provides little or no benefits for

1446 LEARNING: REASONING

learning. Greater reasoning skills are assumed to aidin the ability to acquire new knowledge and reviseone’s existing ideas accordingly. However, if onecontemplates evidence and theory only when it canbe used to justify one’s prior commitments, thenonly supportive information will be learned and ex-isting ideas will remain entrenched and unaffected.The development of reasoning skills will confer verylittle intellectual benefit in the absence of an episte-mological commitment to employ those skills con-sistently.

General Reasoning Performance

Reports from the National Assessment of Education-al Progress and the National Academy of Sciencesconsistently show poor performance on a wide arrayof tasks that require informal reasoning. These tasksspan all of the core curriculum areas of reading,writing, mathematics, science, and history.

Some smaller-scale studies have attempted topaint a more detailed picture of what people aredoing, or failing to do, when asked to reason. Peopledemonstrate some use of informal reasoning skills,but these skills are underdeveloped and applied in-consistently. Children and adults have a poor under-standing of evidence and its relationship to theoriesor claims. Only a small minority of people attemptto justify their claims by providing supporting evi-dence. When explicitly asked for supporting evi-dence, most people simply restate the claim itself ordescribe in more detail what the claim means. It isespecially rare for people to generate possible count-er-evidence or to even consider possible alternativeclaims.

The inconsistent application of informal rea-soning skills could have multiple causes. Some theo-rists suggest that reasoning skills are domain specificand depend heavily on the amount of domainknowledge a person possesses. Alternatively, under-developed or unpracticed skills could lead to theirhaphazard use. A third possibility is that people’slack of explicit knowledge about what good reason-ing entails prevents them from exercising consciouscontrol over their implicit skills.

Inconsistent use of informal reasoning skillsmay also arise because people lack a principled beliefin the utility of reasoning that would foster a consis-tent application of sound reasoning. People have ex-treme levels of certainty in their ideas, and they takethis certainty for granted. In addition, the applica-

tion of reasoning skills is not random, but is selectiveand biased such that prior beliefs are protected fromscrutiny. This systematic inconsistency cannot be ac-counted for by underdeveloped skills, but can be ac-counted for by assuming a biased motivation to usethese skills selectively. Regardless of whether or notpeople have the capacity for sound reasoning, theyhave no philosophical basis that could provide themotivation to override the selective and biased useof these skills.

Development of Reasoning Skills

There is only preliminary data about how and wheninformal reasoning skills develop. There is prelimi-nary support that the development of reasoningtakes a leap forward during the preadolescent years.These findings are consistent with Piagetian assump-tions about the development of concrete operationalthinking, in other words, thinking that involves themental manipulation (e.g., combination, transfor-mation) of objects represented in memory. Howev-er, younger children are capable of some key aspectsof reasoning. Thus, the improvement during earlyadolescence could result from improvements inother subsidiary skills of information processing,from meta-cognitive awareness, or from an increasein relevant knowledge.

A somewhat striking finding is the lack of devel-opment in informal reasoning that occurs from earlyadolescence through adulthood. Some evidence sug-gests that college can improve reasoning, but theoverall relationship between the amount of postse-condary education and reasoning skill is weak atbest. The weak and inconsistent relationship thatdoes exist between level of education and reasoningis likely due to indirect effects. Students are rarely re-quired to engage in complex reasoning tasks. How-ever, the spontaneous disagreements that arise in theclassroom could expose them to the practice of justi-fying one’s claim. Also, engagement in inquiry activ-ities, such as classroom experiments, could provideimplicit exposure to the principles of scientific rea-soning.

There are relatively few programs aimed at de-veloping informal reasoning skills; hence, there islittle information about effective pedagogical strate-gies. Where they do exist, curricula are often aimedat developing general reasoning skills. Yet, many be-lieve that effective reasoning skills are domain- ordiscipline-specific. Nevertheless, given the pervasiveimpact of reasoning skills on learning in general, it

LEARNING: REASONING 1447

is clear that more systematic efforts are needed tofoster reasoning skills at even the earliest grade le-vels. Of the approaches that have been attempted,there is some evidence for the success of scaffolding,which involves a teacher interacting with a studentwho is attempting to reason, and prompting the stu-dent to develop more adequate arguments. Anotherapproach is to explicitly teach what good reasoningmeans, what evidence is, and how evidence relates totheories. This approach could be especially effectiveif classroom experiments are conducted within thecontext of explicit discussions about the principlesof scientific reasoning. Also, if reasoning skills arediscussed in conjunction with the content of the coresubject areas, then students may develop an appreci-ation for the pervasive utility and importance of rea-soning for the progress of ideas.

A number of theorists have suggested that de-bate between students with opposing views couldfoster the basic skills needed for informal reasoning.Debates could give students practice in having toconsider opposing viewpoints and having to coordi-nate evidence and counterevidence in support of aclaim. Also, providing justification for one’s posi-tions requires some cognitive effort, and the normsof social dialogue could provide the needed motiva-tion. However, interpersonal debates are most com-monly construed as situations in which individualsare committed to a position ahead of time, and inwhich their goal is to frame the issue and any evi-dence in a manner that will persuade their opponentor the audience that their own position is correct.Students’ reasoning is already greatly impaired bytheir tendency to adopt a biased, defensive, or non-contemplative stance. Debate activities that reinforcethis stance and blur the difference between defend-ing a claim and contemplating a claim’s justificationmay do more harm than good. To date, there is noempirical data that compare the relative costs andbenefits of using interpersonal debate exercises tofoster critical reasoning skills.

See also: Learning, subentry on Causal Reason-ing; Learning Theory, subentry on HistoricalOverview.

BIBLIOGRAPHY

Baron, Jonathan. 1985. Rationality and Intelli-gence. Cambridge, Eng.: Cambridge UniversityPress.

Baron, Jonathan. 1988. Thinking and Deciding.Cambridge, Eng.: Cambridge University Press.

Boyer, Ernest L. 1983. High School: A Report onSecondary Education in America. New York:Harper and Row.

Cary, Susan. 1985. ‘‘Are Children FundamentallyDifferent Thinkers and Learners Than Adults?’’In Thinking and Learning Skills: Current Re-search and Open Questions, Vol. 2, ed. SusanChipman, Judith Segal, and Robert Glaser. Hil-lsdale, NJ: Erlbaum.

Evans, Jonathan St. B. T.; Newstead, StephenE.; and Byrne, Ruth M. J. 1993. Human Rea-soning: The Psychology of Deduction. Hillsdale,NJ: Erlbaum.

Johnson-Laird, Philip N., and Byrne, Ruth M. J.1991. Deduction. Hillsdale, NJ: Erlbaum.

Kuhn, Deanna. 1991. The Skills of Argument. Cam-bridge, Eng.: Cambridge University Press.

Means, Mary L., and Voss, James F. 1996. ‘‘WhoReasons Well? Two Studies of Informal Reason-ing Among Children of Different Grade, Ability,and Knowledge Levels.’’ Cognition and Instruc-tion 14:139–178.

Nickerson, Raymond S. 1991. ‘‘Modes and Modelsof Informal Reasoning: A Commentary.’’ In In-formal Reasoning and Education, ed. James F.Voss, David N. Perkins, and Judith W. Segal.Hillsdale, NJ: Erlbaum.

Perloms, David N. 1985. ‘‘Postprimary EducationHas Little Impact on Informal Reasoning.’’Journal of Educational Psychology 77:562–571.

Stein, Nancy L., and Miller, Christopher A.1991. ‘‘I Win–You Lose: The Development ofArgumentative Thinking.’’ In Informal Reason-ing and Education, ed. James F. Voss, David N.Perkins, and Judith W. Segal. Hillsdale, NJ: Erl-baum.

Voss, James F., and Means, Mary L. 1991. ‘‘Learn-ing to Reason via Instruction and Argumenta-tion.’’ Learning and Instruction 1:337–350.

Vygotsky, Lev S. 1978. Mind in Society: The Devel-opment of Higher Psychological Processes, ed. Mi-chael Cole. Cambridge, MA: Harvard UniversityPress.

Thomas D. Griffin

1448 LEARNING: REASONING

TRANSFER OF LEARNING

Imagine that every time that people entered a newenvironment they had to learn how to behave with-out the guidance of prior experiences. Slightly noveltasks, like shopping online, would be disorientingand dependant on trial-and-error tactics. Fortunate-ly, people use aspects of their prior experiences, suchas the selection of goods and subsequent payment,to guide their behavior in new settings. The abilityto use learning gained in one situation to help withanother is called transfer.

Transfer has a direct bearing on education. Edu-cators hope that students transfer what they learnfrom one class to another—and to the outsideworld. Educators also hope students transfer experi-ences from home to help make sense of lessons atschool. There are two major approaches to the studyof transfer. One approach characterizes the knowl-edge and conditions of acquisition that optimize thechances of transfer. The other approach inquiresinto the nature of individuals and the cultural con-texts that transform them into more adaptive partic-ipants.

Knowledge-Based Approaches to Transfer

There are several knowledge-based approaches totransfer.

Transferring out from instruction. Ideally, theknowledge students learn in school will be appliedoutside of school. For some topics, it is possible totrain students for the specific situations they willsubsequently encounter, such as typing at a key-board. For other topics, educators cannot anticipateall the out-of-school applications. When school-based lessons do not have a direct mapping toout-of-school contexts, memorization without un-derstanding can lead to inert knowledge. Inertknowledge occurs when people acquire an idea with-out also learning the conditions of its subsequent ap-plication, and thus they fail to apply that ideaappropriately. Memorizing the Pythagorean formu-la, for example, does not guarantee students knowto use the formula to find the distance of a shortcut.

Knowing when to use an idea depends on know-ing the contexts in which the idea is useful. The ideasthat people learn are always parts of a larger context,and people must determine which aspects of thatcontext are relevant. Imagine, for example, a youngchild who is learning to use the hook of a candy caneto pull a toy closer. As the child learns the action,

there are a number of contextual features she mightalso learn. There are incidental features—it isChristmas; there are surface features—the candy issmall and striped; and there are deep features—thecandy cane is rigid and hooked. Instruction fortransfer must help the child discern the deep fea-tures. This way the child might subsequently use anumbrella handle to gather a stuffed animal insteadof trying a candy-striped rope.

When people learn, they not only encode thetarget idea, they also encode the context in which itoccurs, even if that context is incidental. For a studypublished in 1975, Gooden and Baddeley askedadults to learn a list of words on land or underwater(while scuba diving). Afterwards, the adults weresubdivided; half tried to remember the words under-water and half on land. Those people who learnedthe words underwater remembered them better un-derwater than on land, and those people wholearned the words on land remembered them betteron land than underwater. This result reveals the con-text dependency of memory. Context dependency isuseful because it constrains ideas to appear in appro-priate contexts, rather than cluttering people’sthoughts at odd times. But context dependency canbe a problem for transfer, because transfer, by defi-nition, has to occur when the original context oflearning is not reinstated—when one is no longer inschool, for example.

Surface features, which are readily apparent tothe learner, differ from incidental features, becausesurface features are attached to the idea rather thanthe context in which the idea occurs. Surface featurescan be useful. A child might learn that fish have finsand lay eggs. When he sees a new creature with fins,he may decide it is a fish and infer that it too layseggs. Surface features, however, can be imperfectcues. People may overgeneralize and exhibit negativetransfer. For example, the child may have seen a dol-phin instead of a fish. People may also undergeneral-ize and fail to transfer. A child might see an eel andassume it does not lay eggs. Good instruction helpsstudents see beneath the surface to find the deep fea-tures of an idea.

Deep features are based on structures integral toan idea, which may not be readily apparent. To aphysicist, an inclined plane and scissors share thesame deep structure of leverage, but novices cannotsee this similarity and they fail to use a formulalearned for inclined planes to reason about scissors.

LEARNING: TRANSFER OF LEARNING 1449

Analogies are built on deep features. For exam-ple, color is to picture as sound is to song. On thesurface, color and sound differ, as do pictures andsong. Nonetheless, the relation of used to createmakes it possible to compare the common structurebetween the two. Analogy is an important way peo-ple discover deep features. In the 1990s, Kevin Dun-bar studied the laboratory meetings of cell biologists.He found that the scientists often used analogies tounderstand a new discovery. They typically madetransfers of near analogies rather than far ones. A faranalogy transfers an idea from a remote body ofknowledge that shares few surface features, as mightbe the case when using the structure of the solar sys-tem to explain the structure of an atom. A near anal-ogy draws on a structure that comes from a similarbody of knowledge. The scientists in Dunbar’s studyused near analogies from biology because they hadprecise knowledge of biology, which made for amore productive transfer.

Instruction can help students determine deepfeatures by using analogous examples rather thansingle examples. In a 1983 study, Mary Gick andKeith Holyoak asked students how to kill a tumorwith a burst of radiation, given that a strong burstkills nearby tissue and a weak burst does not kill thetumor. Students learned that the solution uses mul-tiple weak radiation beams that converge on thetumor. Sometime later, the students tried to solvethe problem of how a general could attack a fortress:If the general brought enough troops to attack thefortress, they would collapse the main bridge. Stu-dents did not propose that the general could split hisforces over multiple bridges and then converge onthe fortress. The students’ knowledge of the conver-gence solution was inert, because it was only associ-ated with the radiation problem. Gick and Holyoakfound they could improve transfer by providing twoanalogous examples instead of one. For example,students worked with the radiation problem and ananalogous traffic congestion problem. This helpedstudents abstract the convergence schema from theradiation context, and they were able to transfertheir knowledge to the fortress problem.

Transferring in to instruction. In school, transfercan help students learn. If students can transfer inprior knowledge, it will help them understand thecontent of a new lesson. A lesson on the Pythagoreantheorem becomes more comprehensible if studentscan transfer in prior knowledge of right triangles.

Otherwise, the lesson simply involves pushing alge-braic symbols.

Unlike transfer to out-of-school settings, whichdepends on the spontaneous retrieval of relevantprior knowledge, transfer to in-school settings canbe directly supported by teachers. A common ap-proach to help students recruit prior knowledge usescover stories that help students see the relevance ofwhat they are about to learn. A teacher might discussthe challenge of finding the distance of the moonfrom the earth to motivate a lesson on trigonometry.This example includes two ways that transferring inprior knowledge can support learning. Prior knowl-edge helps students understand the problems that aparticular body of knowledge is intended to solve—in this case, problems about distance. Prior knowl-edge also enables learners to construct a mentalmodel of the situation that helps them understandwhat the components of the trigonometric formulasrefer to.

Sometimes students cannot transfer knowledgeto school settings because they do not have the rele-vant knowledge. One way to help overcome a lackof prior knowledge is to use contrasting cases.Whereas pairs of analogies help students abstractdeep features from surface features, pairs of con-trasting cases help students notice deep features inthe first place. Contrasting cases juxtapose examplesthat only differ by one or two features. For example,a teacher might ask students to compare examplesof acute, right, and obtuse triangles. Given the con-trasts, students can notice what makes a right trian-gle distinctive, which in turn, helps them constructprecise mental models to understand a lesson on thePythagorean theorem.

Person-Based Approaches to Transfer

The second approach to transfer asks whether per-son-level variables affect transfer. For example, doIQ tests or persistence predict the ability to transfer?Person-based research relevant to instruction askswhether some experiences can transform people ingeneral ways.

Transferring out from instruction. An enduringissue has been whether instruction can transformpeople into better thinkers. People often believe thatmastering a formal discipline, like Latin or program-ming, improves the rigor of thought. Research hasshown that it is very difficult to improve people’sreasoning, with instruction in logical reasoning

1450 LEARNING: TRANSFER OF LEARNING

being notoriously difficult. Although people maylearn to reason appropriately for one situation, theydo not necessarily apply that reasoning to novel situ-ations. More protracted experiences, however, maybroadly transform individuals to the extent that theyapply a certain method of reasoning in general, re-gardless of situational context. For example, the cul-tural experiences of American and Chinese adultslead them to approach contradictions differently.

There have also been attempts to improve learn-ing abilities by improving people’s ability to transfer.Ann Brown and Mary Jo Kane showed young chil-dren how to use a sample solution to help solve ananalogous problem. After several lessons on trans-ferring knowledge from samples to problems, thechildren spontaneously began to transfer knowledgefrom one example to another. Whether this type ofinstruction has broad effects—for example, whenthe child leaves the psychologist’s laboratory—remains an open question. Most likely, it is theaccumulation of many experiences, not isolated,short-term lessons, that has broad implications forpersonal development.

Transferring in to instruction. When children enterschool, they come with identities and dispositionsthat have been informed by the practices and rolesavailable in their homes and neighborhoods. Schoolsalso have practices and roles, but these can seem for-eign and inhospitable to out-of-school identities.Na’ilah Nasir, for example, found that students didnot transfer their basketball ‘‘street statistics’’ tomake sense of statistics lessons in their classrooms(nor did they use school-learned procedures to solvestatistics problems in basketball). From a knowledgeapproach to transfer, one might argue that theschool and basketball statistics were analogous, andthat the children failed to see the common deep fea-tures. From a person approach to transfer, the cul-tural contexts of the two settings were so differentthat they supported different identities, roles, andinterpretations of social demands. People can viewand express themselves quite differently in schooland nonschool contexts, and there will therefore belittle transfer.

One way to bridge home and school is to alterinstructional contexts so children can build identi-ties and practices that are consistent with their out-of-school personae. Educators, for example, canbring elements of surrounding cultures into theclassroom. In one intervention, African-Americanstudents learned literary analysis by building on

their linguistic practice of signifying. These childrenbrought their cultural heritage to bear on schoolsubjects, and this fostered a school-based identity inwhich students viewed themselves as competent andengaged in school.

Conclusion

The frequent disconnect between in-school and out-of-school contexts has led some researchers to arguethat transfer is unimportant. In 1988, Jean Lavecompared how people solved school math problemsand best-buy shopping problems. The adults rarelyused their school algorithms when shopping. Be-cause they were competent shoppers and viewedthemselves as such, one might conclude that school-based learning does not need to transfer. This con-clusion, however, is predicated on a narrow view oftransfer that is limited to identical uses of what onehas learned or to identical expressions of identity.

From an educational perspective, the primaryfunction of transfer should be to prepare people tolearn something new. So, even though shoppers didnot use the exact algorithms they had learned inschool, the school-based instruction prepared themto learn to solve best-buy problems when they didnot have paper and pencil at hand. This is the centralrelevance of transfer for education. Educators can-not create experts who spontaneously transfer theirknowledge or identities to handle every problem orcontext that might arise. Instead, educators can onlyput students on a trajectory to expertise by preparingthem to transfer for future learning.

See also: Learning, subentries on Analogical Rea-soning, Causal Reasoning, ConceptualChange.

BIBLIOGRAPHY

Boaler, Jo, and Greeno, James G. 2000. ‘‘Identity,Agency, and Knowing in MathematicalWorlds.’’ In Multiple Perspectives on Mathemat-ics Teaching and Learning, ed. Jo Boaler. West-port, CT: Ablex.

Bransford, John D.; Franks, Jeffrey J.; Vye,Nancy J.; and Sherwood, Robert D. 1989.‘‘New Approaches to Instruction: Because Wis-dom Can’t Be Told.’’ In Similarity and Analogi-cal Reasoning, ed. Stella Vosniadou and AndrewOrtony. Cambridge, Eng.: Cambridge Universi-ty Press.

LEARNING: TRANSFER OF LEARNING 1451

Bransford, John D., and Schwartz, Daniel L.1999. ‘‘Rethinking Transfer: A Simple Proposalwith Multiple Implications.’’ In Review of Re-search in Education, ed. Asghar Iran-Nejad andP. David Pearson. Washington, DC: AmericanEducational Research Association.

Brown, Ann L., and Kane, Mary Jo. 1988. ‘‘Pre-school Children Can Learn to Transfer: Learn-ing to Learn and Learning from Example.’’Cognitive Psychology 3(4):275–293.

Ceci, Stephen J., and Ruiz, Ana. 1993. ‘‘Transfer,Abstractness, and Intelligence.’’ In Transfer onTrial, ed. Douglas K. Detterman and Robert J.Sternberg. Stamford, CT: Ablex.

Chi, Michelene T.; Glaser, Robert; and Farr,Marshall J. 1988. The Nature of Expertise. Hil-lsdale, NJ: Erlbaum.

Dunbar, Kevin. 1997. ‘‘How Scientists Think: On-line Creativity and Conceptual Change in Sci-ence.’’ In Creative Thought, ed. Thomas B.Ward, Stephen M. Smith, and Jyotsna Vaid.Washington DC: APA.

Gentner, Dedre. 1989. ‘‘The Mechanisms of Ana-logical Reasoning.’’ In Similarity and AnalogicalReasoning, ed. Stella Vosniadou and AndrewOrtony. Cambridge, Eng.: Cambridge Universi-ty Press.

Gick, Mary L., and Holyoak, Keith J. 1983.‘‘Schema Induction and Analogical Transfer.’’Cognitive Psychology 15(1):1–38.

Godden, D. R., and Baddeley, A. D. 1975. ‘‘Con-text-Dependent Memory in Two Natural Envi-ronments: On Land and Under Water.’’ BritishJournal of Psychology 66(3):325–331.

Lave, Jean. 1988. Cognition in Practice. Cambridge,Eng.: Cambridge University Press.

Lee, Carol. 1995. ‘‘A Culturally Based CognitiveApprenticeship: Teaching African-AmericanHigh School Students Skills of Literary Interpre-tation.’’ Reading Research Quarterly 30(4):608–630.

Moll, Luis C., and Greenberg, James B. 1990.‘‘Creating Zones of Possibilities: Combining So-cial Contexts for Instructions.’’ In Vygotsky andEducation, ed. Luis C. Moll. Cambridge, Eng.:Cambridge University Press.

Nisbett, Richard E.; Fong, Geoffrey T.; Leh-man, Darrin R.; and Cheng, Patricia W.1987. ‘‘Teaching Reasoning.’’ Science238(4827):625–631.

Novick, Laura R. ‘‘Analogical Transfer, ProblemSimilarity, and Expertise.’’ Journal of Experi-mental Psychology: Learning, Memory, and Cog-nition 14(3):510–520.

Peng, Kaiping, and Nisbett, Richard E. 1999.‘‘Culture, Dialectics, and Reasoning about Con-tradiction.’’ American Psychologist 54(9):741–754.

Schwartz, Daniel L., and Bransford, John D.1998. ‘‘A Time for Telling.’’ Cognition and In-struction 16(4):475–522.

Daniel L. SchwartzNa’ilah Nasir

LEARNING COMMUNITIES ANDTHE UNDERGRADUATE

CURRICULUM

Educational observers have long argued that studentinvolvement is important to student education. In-deed a wide range of studies, in a variety of settingsand of a range of students, have confirmed that aca-demic and social involvement, sometimes referred toas academic and social integration, enhances studentdevelopment, improves student learning, and in-creases student persistence. Simply put, involvementmatters. But getting students involved can be diffi-cult. This is especially true for the majority of collegestudents who commute to college, who work whilein college, or have substantial family responsibilitiesbeyond college. Unlike students who reside on cam-pus, these students have few, if any, opportunities toengage others beyond the classroom.

For that reason an increasing number of univer-sities and colleges, both two- and four-year, haveturned their attention to the classroom—the oneplace, perhaps the only place, where students meeteach other and the faculty. Researchers have askedhow that setting can be altered to better promotestudent involvement and in turn improve studenteducation. In response, schools have begun to insti-tute a variety of curricular and pedagogical reformsranging from the use of cooperative and problem-based learning to the inclusion of service learning inthe college curriculum. One reform that is gainingattention, that addresses both the need for studentinvolvement and the demands for curricular coher-ence, is the use of learning communities.

1452 LEARNING COMMUNITIES AND THE UNDERGRADUATE CURRICULUM