Reconceptualizing Change in the Cognitive Construction of Knowledge

Janice A. Dole and Gale M. Sinatra Department of Educational Studies

University of Utah

How is knowledg6 q u k d and repmenfed in memory? By what pmws do individuals come to change their &a, anreption$. or knowledge? Although the fint qnedon has been central to cognitivepsychologiats' nssarch Wdaa formsny years, relatively leas is known about the ohangeprocess. Wewarnine3 n s s a r c h l i l a l U u M t o h s d e a o u r ~ d t h e p m c e a s of kwwls4pc change. In partleular. we draw on models of conwphd chunge Prom cognitive psychology, social psycholo$y, and science education. Each model ad& a IWW perspcctiveon the change proms. B& on the 1ltemturc Prom these models of change, we developed a new model that q m m t r our monceptualization of the change procesrr. We d m d i the model aod use it to point out new anas of rascareh to be admesJed

P~shology has, at its foundation, an abiding interest in t m b t m d h g the construction of knowledge. How do indi- viduals come to perceive, know, remrnber, and l m ? How do individuals acquire new knowIedge? How is knowledge qmmted and organized in memory? How do individuals come to change their knowledge? TBwe questions have saved as wncq)hral framewoh fm psychological, as well as philowphid, thought for centuria (Hunt, 1993). A major wnaibution of cognitive psychology has been the

conoqrtaalization of knowledge as memory npnsentations in the fOllll of d p t s , fmmes, or achemta (Anderson & Fearson, lW,RumeIhart& Ortony. 19n; Shank&AbeIson, 1977, Spiro, 1980). Schemata am defined as "packets of intagmted information on various topics" (Hunt, 1993, p. 530). Throughout the 1970s and 1980s. cognitive psycholo- gists wan intmstcd in describing the name of these packets of iaformation. Spiro (1980) demonstratsd the constructive agd comp1ex nahm of achemam and highlighted contextual fmkm-incl&gtasks, texts, and situational contexts-that influcnccd how knowledge is orgenid in memory.

Recmtly, qn i t ive meamharp have come to view knowl- edge and schemata as multidinsional (Jetton, Rupley. & Willson, 1995). For example, meamhem havedifferentiated novics and expsrrpxpsrrp knowledge strumum~ in subject-matter domains (Chase & Simon, 1973; Chi, Glaser, & Rees, 1982;

Larkin, McDe.rmot& Simon. & S i o n , 1981; Vass, Greane, Post. & Pemer, 1983). Rtsearchers have wramined discourse knowledgc-k.nowledgs about language and how it works (hlcCutchen, 1986). Another aspect of knowledge that has been extensively studied is saategic knowledgtihrowkdge about proceduresfor acoompliihing ago4 or task (Alex& & Judy, 1988; 1. R Andason. 1983a; hwat, 1989).

Until recently, questiohs about the nature of knowledge and its representation have received far mope thaorctical and reserach attention than have questions about how iadividuals acquire knowledge. F'iaget distingubflbd himBOlfas a notablc exception to his cmatempo&es in that he spcnf his life sntdying knowledge acquisition. He dcswibed two pmcuim involved in acquisition. He used the term &mihion for the addition of mfmation to existing knowledge structureg and accommodarion for the modification or change of existing knowledge smtuns (Piaget, 1985). Mope recently, theae views of knowledge scqnisition are well captured by the pcrspecrive of wgnirive ~onstncctivism (Cow, 1994). Con- structivism placGs prime imp~tance on the individual's ac- tive role in the knowledge acquisition process.

Researchers afte~ Piaget fine-tuned the o~nstructs of as- similation and ~ccommodation. Schema theorists used the term accrerion for the assimilation of new factual informarion that fits into existing knowledge smctuns (Rumclhaa & Norman, 1981). Cognitive psychologi&t3 dssuibed various mechanisms of knowledge acquisition, such as addition, de- letion, discximihation, and generalization (Chi, 1992). that I result ih assimilation or "weak raptruenning" (Vosniadou &

Brewer, 1987). For example, students who attend a biology lecture may readily assimdate new knowledge about plant reproduction into their existing knowledge base about plants.

By contrast, the change pmcess, or accommodation, has been a much more thorny issue for researchers interested in knowledge acquisition. Suppose those same students who attended the biology lecture on plant reproduction attended another session on human reproduction. This time, students may not easily assimilate new information about how human beings reproduce. Students may have their own ideas about human reproduction. lhcy may not be willing to hear the instructor's ideas, or they may listen to them but reject them (Chinn & Brewer, 1993). If students do make radical shifts in their thinking, then accommodation would be said to have taken place.

Vosniadou and Brewer (1987) described the accommoda- tion process as "radical restructuring" or 'khange in knowl- edge that invofves the creation of new s t r u c m [which] are constructed either to reinterpret old information or to account for new information" (p. 52). Chi (1992) refwed to such knowledge reorganization as "radcal conceptual change." and lhagard (1992) used the term "'conceptual revolution" to describe how one conceptual system can replace another. Regardless of the tern used, radical changes in one's think- ing about any topic are usually difficult to attain.

Lkspite differences among theorists in their description of the accommodation process, there is one issue on which change theorists appear to agm. That is, we still do not know enough about the mechanisms that produce knowledge change (Siegler, 1996). Smith, di Scssa, andRoschelle (1993) argued that an adequate theory of learning must not only describe knowledge, but also provide. richer descriptions of how knowledge is transformed. Chi (1992) argued that too little is known about the conditions of knowledge repre- sentations that may lead to change, such as how learners come to remgnize that new information does not fit with their existing conceptions.

Under what conditions is knowledge change likely or unlikely to occur? What are potential sources of influences in bringing about change? Is change likely to be temporary or relatively long lasting? Weak or strong? What can educators do to facilitate the change p m s ? These questions are of central interest to educational researchers.

To explore these issues, we turn to models in cognitive psychology, science education, and social psychology for new insights into thechange process. First, we examine recent work on conceptual change within a cognitive psychological framework. How concepts come to be represented in memory and change in children's conceptual knowledge have been central issues of concern in cognitive psychology since its inception (for examples of classic works in these areas, see Mar- 1979, and Rosch & Mavis. 1978). Current cogni- tive psychological mearchers investigating change in con- cept reprosentation have elaborated on early Piagetian and schema-theoretic notlons of accommodation. These re-

searchers have attempted to more fully descnbe the process whereby conceptual knowledge comes to be reorganized (see Carey, 1992: Chi, 1992: Thagard, 1992).

Second, we tum to work in science education. Science- education researchers have been more influenced by the his- tory and philosophy of science and have described how knowledge acquisition and restructuring in scientific commu- nities is similar to knowledge acquisition and restructuring in learners (Carey. 1985; Driver & Easley, 1978; Osbome & Wiarock, 1983; Posner. Strike, Hewson. & Gertzog, 1982). These mearchers likened the paradigm shifts made by the scientific uwmunity (Kuhn, 1970) to conceptual changes made by students as they developed new frameworks through conslrwting their understanding of various scientific phe- nomena. Despite critiques that the process of knowledge restruchuing in students differs fundamentally from that of thescientific community (Pintrich,Marx. & Boyle, 1993). the Conceptual Change Model (CCM) still provides many useful insights into the nature of the change process.

We also turn to the field of social psychology. Social psychologists have had a long-standing interest in change and have generated an extensive body of research on how beliefs and attitudes can be changed. Researchers have studied fao- tors related to learners, presenters, and specitic contexts that makeindividualsmore or less likely to consider andlearn new information. Petty and Cacioppo's (1986) Elaboration Like- lihood Model (ELM) described the conditions necessary for change. Together, these pcrspectives and the reaeerch base provide a framework from which to reconccptualizcour view of the change process.

We begin by defining key concepts in the cognitive and social psychological literature as a context for describing current models of change. Next, we review change models with particular attention to their similarities and diierences. Then we present our reconceptualization of the change proc- ess. Finally, we present a discussion of unaddnssed rescarcb issues related to change models within a cognitive construc- tivist framework

OUT purpose in generating a new model of change is to providemearchers interested in the change procegs with new insights and perspectives fhm three fields of research. Our model was not developedtoresolve a specificempirical issue, but rather to integrate current theory and mearch from dif- ferent disciplinary domains. We believe such an integration will point to possible new and fruitful irreas for research for those interested in learning as change.

COGNITIVE CONSTRUCTIVIST PERSPECTIVES ON CHANGE

Cognitive psychologists have traditionally focused more at- tention on descrihingthe structure and function of knowledge than on change and growth in knowledge (e.g., Anderson & Pearson, 19% Rumelhart, 1980; Spuo. 1980). However,

Ccgnsthre PByohological Models of Changer

-J 2hologi--. . . .. -

Learning The6ri~s'

Concept :nrmattnn

'Cognltjvp' Information

, .~ , . Pi@cessing .

GoncEsprrrai Change Model

Chi (1992). Qtky (1W. Cany & Spelke, 1994). aed Tha@ (1992) have posited concGptual shifts as rsdical restroctoring. Sueh changes m profound and fundamental in that near conccptionsreplace dd. hoeording to this parspe~- tive, the new concept is so distiact from the old Brat the original wncept playe no role in funha rhinking within the domein 'Ihus, the change is revofutionary and multsin "the cmer$glce of new pinc'kples. incommensnrable with the old, which carve h e world at diffennt joints'' (Carey & SpaIke, 1994, p. 179).

Such h d m e n t a l change in tbe mental mpmmtations of coaccpts cannot occur through ss&niMion proctssts of ad- dition, deletion, digaimination, and pegalizstion. Rather, ~~aceptualeategoriesmust kmnfigured througha process Chi (19!Q)deucribedas eatogorkal remf8nmnt. According to Chi radical wnwphlal change occurs when concepts are reassigned to a new category that is fundamentally dif&rent in kind. For eategolical~ignment to oocor, a learner mwt lean a t h e new category's properks and about concepts that exist within the new categq. This can oanr through baditional assimilation mechanisms. Than, the leama must nassign a w m q t to the new category. An example of such aconceptusl shift would be a physics student wbrecatcgop izea heat as an process rather than a property of a material subotancc.

Saagerd (1992) described radical conceptual change as brunch jumping a me switching. If one considers concepts to be organizad birrenhically, then a radical c~~cdptual change would invoke moving a concept from one !me to another. Such a change in the hierarchical organizstion of concepts in memory is analogous to Darwin's repositioning of humans under the category of animd when p n v i d y huamswerewasidendtobeadifkmtltindoforeaauQ Not Wllie science educators imeresfed in change, Chi (m23 and Thagard (1992) drew analogies between the sci- entific rev01utiions described by Kuhn (1970) and the bowl- edge ehanpprcro*win learners. That is, in science, informa- tion aceumulates until auomalies am d i s c o v d that cannot beacco~~~tsdfOrwithe~gthdw.AnewtheotyrcCeives accaptance by sientists if it pv ides a more coherent *la- nation of phenomera So m, individuals emunter new information, notiw discrepdm bGtween their existing knowledge and new id-, and re~~ganizc their ktiow1edge struchm.8 to more coherently ~ccount f a the new informa- tion. 'Ibis wnceptuslization of the change process from cog- nitivc psychology is similar to concaptions of knowledge change in the %Id of science education.

In contrast to the view of knowledge change as revolution- ary, Smith et al. (1993) and Sieglcr (1996) described the change proms as evolutionary. 'IhePt theorists described how laowledge is gradually msformed from one state to another. AEconl'ito this view, concephutl change occurs by hBasforming and refiniig prior ktiowledge into more and morS sophisticated forms. Substantial change does not occur quickly, and relatively stable intermediate states of under-

standing exist f a some time hefore new concepts am fully formed.

Smith etal. (1993) a p e d that their position is b a d on a fundamental tmant of wnstn~tivism: that is, pioa howl- edge plays a significant roll in learning. According to this dew, mcepmalrep1"ementis "neitherplausiblcmalways des'uabIen @. 153). Ratha than wholesalcreplaccmsntofone concept with anotba, arsd iu l~&nof lmowledge , existing k n o w l e d & e f ~ the basis for new cmceptioas and rcwalnsimportant in the new undcrstaading.

Siegler (1996) borrowed a metaphor from m1utionary biology to d d b e incremental change in the coaceptual developme~~t ofchi1dren. Rs argued that children think about a phenomena in a variety of ways, rather thaa having a unitary understanding. In his dew, change results aschildren choose adaptively among these various ideas. In other words, a particular conception becomes the more prevalent w e when it proves to he more successful over time. Aeeordia$ to Siegler, t W n g changes continnously as it becomes more adaptive.

A View of Conceptual Change Flom Science Education

During the 1980s, science educators adopted a cognitive constructivist view of knowIedge consistent with cognitive psychology and Piagexian &my n, explain how children acquire knowledge about scientific phenomesa (White & Tishw, 1986). According to this consmtivist persptive, knowledge is not transmitted from teacher I0 laamex but constructed by individuals bough the processes of assimila- tion and acwmmadation. Bemuse the aecomt110dbti0n pioc- ess described change in existing knowI&ge structrrres, sc- wmmodation was viewed by science edocatots as the centrd mechanism for conceptual change.

Pmneret al. (1982)developadtheCCMtoexplain (a) why so many students held on to their existing conceptioris in spite of inshuctioa, and (b) what conditions were newssay for change to take place. As bs deve1oped the CCM, they drew fmm the litaatrne on the history and philosophy of science as well as cognitive theory and research. SEiencc educators' focus has beea on what they &an bodies of organ- iznd knowledge-such as Newtonian physics, the csrth aad gravity, and scientific explanations of photosynthapis, light, and heat mest & P., 1985). Tlrcse whereat bodies of knowledge consist of a set of concepts and interrelafad ideas that take time and effm to I m and are often inconsistant with intuitive experiences and evayday beliefs about the world. Basearch on conc8ptw.l change leaving ha$ b6a conductad on students of many d i i e n t *from young childm to adutts (fora review, see West &%, 1985).

To undersand mechanisms for change, dence educators, like cognitive psycho1o~ts intensted in change, drew paral- lels between the accommodation process that individuals

expQienc~ and the p m w scientists undergo as they do nomral science (Anderson & Smith, 1987). Philosophas and historians of science, such as T o w n (1972) and Kuhn (1970), argued that"scientificrevolutions," "paradiishifts" (Kuhn, 19701, or 'khauges in m m h pro$rams" (Lakatos, 1970) result from the accommodation proceap made by the scicntificwnuaunity. Toulmin (19721 refcnadto this m m - modation process, whether occurring in individuals or in scieacs, as "cooceptual change."

Posm et al. (1982) desaibed the similarity between the conceptual cbange preotss in scientific communities and the mmmodationpmocss in 1carner.s. For example, most young childnu, like many early Eumpm scientists, believe that the ceab is flat. Children's ideas come from nomea~ls 8e1180ry expsienecs with a flat horizon. Like the early European scientists, childrtn eventually enwunter new information suggesting that the earth is a sphere. At some point, children rejecttheii beliefthatthe eanb is flat and replace it with a new owctptualizsrion ofthe earth as a sphere (Nusshaum, 1985). Thr; proocss whereby the spherical notion of the eanh comes to replace children*~ flat earth views is called conceptual change. Posaa et al. (1982) ident8ed four critical variables in the

conoephlpl change pmcess (see F i 2). First, they argued that when scientists and studeno' knowledge s b u c t u ~ ~ are crystallid, coherent, and firmly enmched, they arc highly resistant to radical change. Of coma, not all conffip&ions are so resolute. Knowledge of scientific phenomena is o&n c8enrorcai$tically enmnched in the scientist because it is based on a coherent bady of knowledge and in the individual bccause of pasonal experience with scientific phenomena (the sua appears bo travel across the sb). Radical change of such entrenched idem nquireg cansidetable cognitive efforf and individuals are unlikely to engaae in such effort unless they believe that a less radical change will not work. Accord- ing to the concephlal change model, individuals must become dissatisfied with their existing wnctptions as explanations for data. Gecond, individuals must find that the new c~cy, t ions are intelligibleand make sense. Thatis, they must undmtand the new CMlCeptions if they are to adopt them. Third, scientists and studcnta must perceive the new conceptions to be plaosi- ble oasg Thus, the new conceptions must not only make sense and be comprehensible, but they must be believable. To be believahk, the new conceptions must fit into existing and related ideas. Last, scientists and students must find that the newwnccpti ons are ftuithd. or open to'hsw arcas of inquiry'' (p. 214). Thwthe new wnffiptionnmustlead to new insights and further hypotheses.

Science educators have used conqmal change as the thwretical framework for explaining a multitude of studies showing how individuals fail to develop conceptual under- standings about numemus scientific phenomena (e.g., Carey, 1985; Chinn &Brewer, 1993; West & Pin@, 1985). Ewther, themodelharibeenused t o p t o p i n s l ~ ~ t i o n a l interventions that have pmven promising in helping students change their

paconceptions or naive thoodes about scientiftcphenomena. We will not review these studies, as they have been reviewed extensively elsewhere (e.g., Guzze.t& Snyder, Glass, & Ga- mas, 1993; Nickerson, 1985, Nussbaum, 1985; West& Pines, 1985).

SOCIAL PSYCHOLOQICAL PERSPECTIVES ON CHANGE

Although cognitive psychologists have only m n t l y begun to examine the change ptgeesa, pocial psychologists have had a long-standing intenst in chan&.. Two construots of tradi- tional inmust to social psychologists arebeliefs andattitudes. Beliefs are chmterized as the thoughts that people have about attitude objaog. B e l i are the building blocks of attitudes (Eagly Bi Cbaiken. 1993) in fhat an d h d e consists of an interrelated and interwrmected set of beliefs that have a positive or negative valence for an individual. Thus, a person's attitude toward gun wnml exists along a positive to-negative continuum. It consist6 of a related set of beliefs about guns-who uscs them and for whet purposa, when and where they can and should be used- well as beliefs about gun owners, the National Rifle Association, and a host of gun-related issuur. It is iatapsting to note that even if indi- viduals have a neutral position about gun control, it would still be consided an attitude by social psychalogists m l y & Chaiken, 1993).

&lief8 and attitudss can be propositional in natum, or they can be npresaatcd by mental images or a set of experience6 and behaviors. Eor example, when asked about one's beliefs about guns, an individual may think about an image of a m i l i t i a - d d man shoothg at a target, a gang member in a drive-by shooting, or a hunting hip in Montnna. Such thoughts may be ssmiated with negative or positive evalu- ations about guns. Eagly and Chaiken (1993) argued that "these images no doubt carry evaluative meaning, just as pmpositional beliefs don @. 1W).

To provide a t h d c a l framework to undcraFtand beliefs better, social psychologists adopted some of the theory and methodology of cognitive psychology F ~ s k e & Taylor, 1991). In particular, the construct of schwa (J. Andexson, 1983% 1983b; R. C. AnderaDn & Pearson, 1984) was bor- rowed from cogpitive psychology to explain how beliefs are stored and organkd in memory. Social psycholagists con- cepfualiud beliefs as memory repmentations consisting of networks of associations or schemata (TssSar & Shaffer, 1990). just as cognitive psychologists concephlalizcd knowl- edge. L i e k n o w l ~ schemata, beliefs can be wnccptual- i d as simple or complex. Eor examplc, a less eomplex belief schema about a presidential candidate may contain beliefs about the candidate's position on 8 pin%& issue, such as the role of the federal government A more complex belief schema about that candidate, however, may entail aricher and more detailed understanding about who thecand i i i s , whar

Intelligible?

the candidate's personal and professional background is, how the candidate voted on important state and federal issues in the past, and knowledge abwt who is supporting and oppos- ing the &date and why.

This characterization of beliefs roncspnds with a eogni- tive view of knowledge representation. According to a schema-thsoaetic pcrspative on knowledge, an activist for a particular presidential candidate would hnve a rich schema with many knowledge nodes related to the candidate and many connections to other related knowledge nodes. A high school student studying civics, on the other hand, may have fewer knowledge nodes about the candidate snd fewer con- &om to other related concepts.

As far as we can tell, there are few differences in the way social pycblogists and cognitive psychologists use and apply the term schcmata It does appear, however, that social psychologisD include a full range of affective f w t m in their discussions about schemata. Although Bartlett (1932) and some LatathsMists may haveoriginally conceptuaIized sche- mata in the same way f.e.g., Spuo, 1980). the focus of cogni- tive pychologist~ on the information-pmcessling metaphor likely nsultedintheeliminationof guchaffectivecomponenta in tater discussions about schemata (Ftske & Taytor. 1991).

In sum, social psychologists have adopted some of the theory and methodology of cognitive psychology to explain how attitudes and beliefs are npresanted in memory. Social psychelogigtp view belieb similar to the way cognitive psy- cholagists view knowledge-as schemata with related, asso- ciative networks 'hm srmilarities between mcial and cog- nitive psychology lead us to contend that social psychology

research may well prove useful to educational reseamhem haditionally operating within a cognitive framework.

As part of tbcu interest in basic cognitive precessas, one branch of social psychoEagy4tude formation and change-has primarily focused on how people acquire their views, how likely they are to chat15 them, and under what conditions they do so. Thc social psychology firarmwork for undenftanding the change process has bacn concepmalizcd tbrough many different models and hundnds of stdies. Eor our purposes, we have chosen to look at information-pmc*rs- ing models of attitude and bskf change, because these mod- els were formed using theory and methods similar to those in cognitive psychology. The similarity of these models to the cognitive constructivist orientation make them particularly useful for our view of knowledge restructuring.

Models of attitude and belief change in social psychology come out of research on persuapion. 'hm models can be categorized into two types: combinatorialand pocess models (see Figure 3). Cambinatorid models consist of highly spe- cifrc quantitative desuiptions of how individuals combine penicular pieces of idormation to form new beliefs and change old ones (e.g.. McOuire 1960, Wyer. 1970). These models place numuioal weights on the components of change that reflect their relative contribution to the change prooess. Mathematical computations are then calculated using equa- tions that contain the weighted components. Eagb and Chaiken (1993) argued that cambiatorial models are useful for desctibiig attitude struchue. Further, they adequately explain how individuals integrate new information with ex- isting information once they have aocspted it. However,

~ t t i t d e and Belief 1

Change

Models Process Models

FIGURE 3 Models of altiNdc and belief dsaogo.

comhinatorial modele do not adequately explain how indi- viduals corn to accept new information in the first place, and therefore, they do not elaboraa fhe cognitive processeg un- derlying attitude change.

Pmcsas models, on the other hand, have a decidedly cog- nitive orientation in that they attempt to describe and expIain cognitive mechanisms for attitude and belief change (Eagly & Chrilmn, 1993). In contrast to the quantitative orientation of combinatorid modcls, pmces models ine qualitative de- sniptiom of ponruasion that show how individuals can come to aeccpt new information. ?hem rim saved type8 of prooess models. Dual ploccss models (e.g., Chailrea, 1980; Petty 8 Cacioppo, 1986) represent a synthesis of the findings from various procurs models. Dual proms models provide a cog- nitive pmcesing account of how individuals who hold a particular set of beliefs can change those beliefs based on exposme to a complex message.

Social psychology researchers have examined the m y factors involved in how individuals can be persuaded to chan~c thclr beliefs. A* a flurry of highly controlled l a b ratory studies from the 19% *ugh the 1960s. intenst in artitude and belief change diminished becausa '%here war so little agreement con-g if, when, and how [ d i h t variables] affected attitudechange" petty & Cacioppo. 1986,

p. 125). The lackof sgrcenaGnt stemmed from the diversity of findings and conflicting accounts of belief change in the litenhue.

Duting the 1970s and 1980s. social psychologists began to examine these conhndictory empirical Wigs. Several models of attitude change were considered, Some omdele assumed that thoughtful consideration of mcssagc content was necessary for attitude change (piske & 'layor, 1991). Other models assumed that individuals did not always need to engage in thoughtful evaluation for attitude change to occur. A new m of modole, called dual pmcass modeb, wete developed to account for both of thew passibilltia. Dual process models postulated two mutes to attitude change. One route involved an individual's thoughtful consideration of issue-mlevant infarmation, and the the route involved an individual's quick and cursory judgment based on variables peripheral to the message content. Peny and Cacioppo's (1986) ELM is one such dual process model (see Figure 4).

Elaboration Likelihood Model

The ELM specifies the processes underlying belief change, known in the social psychology literature as pcasuasion (see

1 Persuasive Communication

dotivated to Process

NaturwA Gognitiv-e Processing; --

,,.. cosnifm.' ~ ~ r u & r ~ ' Change?:, +..-+, ., . ~~ , ..

2entral Attitude Change

Peripheral Attitude Shifi

'eripheral CUE Present?

Initial Attitud

also the bclnistic-systematic model, Chiken, 1987; Chaiken, Liberman. 8i Eagly, 1989). The key construct in the model is that of elaboration. Elaboration is the extent to which indi- viduals think about "issue-relevant arguments contained in a message" (Petty & Cacioppo, 1986, p. 128). Elaboration exists on a continuum from high to low. High elaboration oecurs when individuals engage in issue-relevant thinldng; low elaboration oecurs when individuals do not.

The two key factors contributing to high elaboration are motivation and ability. Petty and Cacioppo (1986) use moti- vation as an overarching eategory for several constructs. For am, personally =levant topics, or topics in which individuals have a stake in the outcome, will promote motivation to process a message (for a review, see Chaiken & Stangor, 1987). Petsonally relevant topics am in t r ins ' iy important because they have signif~cance for individuals' personal lives (Sherif & Hovland. 1961). For example, Petty, Ostrom, and Brock (1981) manipulated the personal relevance of a policy change that would d i i y affffit university students. They told undergraduafes that the proposed change would take place in 1 year or in 10 y m . Students who were told that the

change would take place in 1 year were more likely to be motivated to pmoess infomation about the policy change. In a meta-amlysis of "invalvement effects'' ssuch as personal rele vance, Johnson and Fhgly (1989) showed thai individuaLP who wen more involved because they c a d about the outcnns of a debatedissueweremon:pcrslLsdedbystrongargnments.In genaslsocialpsychologistspositthatpmswally relevanttopica stimulate thought, which, in turn, can lead to high dabomtion. Petty and Cacioppo (1986) noted two caveats, however,

regardii the influence of personal relevance. First, a topic may be so intensely tied into individuals' values that they are not willing to process the message at all (Oreenwald, 1980; 1982). For example, individuals who espouae a creationist view of the origins of humans may be unwilling to process a message related to evolutionary theory. The second caveat involves message repetition. Sometimes individuals feel that they have heard a message and its arguments and counterar- guments so many time8 that they are unwilling to listcnto and process the new ideas in the message. For example, smokers may be unwilling to p m e s new information about added nicotine in cigarettes because they have heard so many argu-

meats against smoking in the past (for the effect of message repetition, see Cacioppo & Petty, 1979,1985).

Anothe~ oonmct related to individuals? motivation to pmcsss a mesage is a personality trait that social psycholo- gists dcsaibe as a "need for cognitionw (Cacioppo & Petty, 1982). Social pychologists d&be this individual differ- ence as to struchlrc relevant situations in meaningful, integrated ways" or a need to "understand and make reason- ablotheeqdential world" (Cohen, Stotland, & Wolfe., 1955, p. 291). Some indiiiduals have a high need for cognition. whae they enjoy the effortPul engagement of arguments, the. evaluation of ideas, and the analysis of problems and their solutiws. These individuals by their very nature are more likely to engage in high elaboration. 0 t h individuals will. by their very nature, not be motivated to eagagc in effomul, thoughtful evaluation and analysis of ideas. Theindividuals will be mom likely to prwcess the information heuristically, that is, with low elaboration (Chaiken, 1987).

Finally, high interest in a topic can promote motivation to process a message. This is not unlike the finding in educational research that situational intereat (interest stem- ming from environmental conditions) can promote text ~ ~ m p r e h e r i ~ i ~ n , learning, and achievement (Alexander, Kulikowich, & Schulze, 1994; Shraw, Bmning, & Svoboda. 1995).

The other key factor contributing to high elaboration is ability. For a message to be processed. it must be compr* hensible and not too difficult conceptually. Further, indi- viduals most have sufficient background knowledge to make sense of the message. Research in this area has shown that the more comprehensible the message, the more the recipients of the message can think about what was said. Then, if the arguments are good, recipients can agree with the presentex (Ratneshwar & Chaiken, 1991: Wood & Eagly, 1981). In contrast, an incomprehensible argument can elicit fmsuation and negative feelings. These negative feelings can become associated with the message and the presenter-in effect discouraging high elaboration (Fiske &Taylor, 1991).

Research has also shown that message effects interact with the degree to which the listener has a stake in the outcome to produce different levels of elaboration. For examplc, if the message is difficult, the extent to which someone will try to pmcess the message will depend on their involvement. Fiske and Tayor (1991) explained how involvement affects the likelihood that students will prao ess a difficult statistics lecture with high elaboration. A student who has alnady given up on the course will likely daydream and not process the message. A student who will be a teaching assistant for the course next semester will make the effort to understand the lecture, despite its diffi- culty. Students moderately involved are the ones most likely to be affected by the difficulty of the lecture. The added diiculty may induce greater processing of the in- formation and thus high elaboration.

Two mutes to persutwion. Tbrougb positing two routes to psuasio-tral and peripheral---~~~ial psy- chologists rcsolved the apparently conUadictmy Endings in the literature on attitude change. Research participants in social psychological studies changed their beliefs either though the thoughtful weighing of idess or through attention to a peripkal cuc. Undez conditions of high elaboration, individuals engage in thoughtful and effonful prooessing, which may lead to relatively strong and long-lasting change in beliefs. Petty and Cacioppo (1980) called this the central route topersuasion. Howevsr, the model also predicts that a weakchange may ocau. Under conditions of low elaboration, individuals engage in quick and superficial evaluations that may lead to achangein beliefsthat is w&aodcan,over time, dissipate. Petty and Cacioppo called this weak change the peripheral r o ~ ~ to persuoswn

The central route to persuasion describes a process lesding to relatively strong and long-lasting change in attitudes or beliefs, not unlike the process of conceptual change. This route begins with a message that is intended to persuade individuals to adopt a parrieular attitude or belief. For exam- ple, a health teacher presents information ngsrding the risk of AIDS to high school seniors. If students am motivated and able to process the information. then the ELM predicts they can consider the issues. ?his involves effottfd, thoughtful prowssingof theinformation studentsfinddevant Thistype of evaluation and analysis of arguments is anaIogous to "dcsp processing of informati6n"nsulting in the formation of mem- ory repmentations or lemming (Craik & Lockhart, 1972).

At this point, the quality of the argument is a key factor in whether or not individuals will be pernuadd to change mGir minds. If students engage in deep processing of a message on AIDS, a strong argument will be n d to promote change. The ELM, like the CCM, suggests that the process leadiing

to relatively strong and pgmanent belidchange is a difficult one. Individuals may not be motivated toprooess the message because they may we be intwe6ted in the topic. Individuals may not see the paecnal relevance of the issue, or they may not have a stake in the outcome of the issue.

Even if they am motivated. individuals may not have the ability to p m c ~ s the information. Individuals may be dis- tracted; they may not have the backg~~und knowledge; or the message may be incomprehensible. Faaher, becauee process- ing the message takes time, cognitive effort, and energy, individuals may not be willing to process the information at all.

One of the unique features of the ELM is that it p i t s another set of facrors that can lead to the peripheral mute to persuasion. F a examplo, students listening to a lecture on AIDS may not be motivated or committed to engage in high elabaation ofthe issuesand arguments. Portheseindividuals, thequality or strength oftheargumentsdoes notmatter@gly & Chaiken, 1593). However. these same students may be persuaded by an attmctive.credible, or trustworthy source; by a pleasant context (Petty & Cacioppo, 1980); or by a simple

messagethatthey understandeasily (Eagly & Chaiken, 1993). Time students are not persuaded by the strength of the arguments, butrather by the peripheral cue iwlf. Under these circumstancas, the resulting beliefs are less strong than cen- trally pmwsed ones. In time, individuals are l i l y to retm to their previously held beliefs (Petty & Cacioppo. 1986).

Although the attention to psripheral cues with low elabo- ration can lead to weak belief change, peripheral cues may still be useful in promoting change. The ELM predicts that pexipheral cuts can serve w an impctus toward h i e1abw8- tion, which may lead to central change and strong beliefs (Petty & Cacioppo, 1986). For example, consider unmoti- vated shldsnts in a health class who heur a lecture from someone they admire who is HIV positive. The speaker may act as a pwiphcral cue that then leads to d a p and thoughtful promsing of information about HIV safety precautions. The thoughtfulproeessing of informationcould then 1 4 students toward change in their thinking about taking precautions against infection.

In sum, the U M predicts change in one of two ways-through high or low elaboration. A quick. hewistie evaluation of a message that occurs because of aperipheral cuc-an attractive, credible, or trustworthy source-can lead to change. But the change is weak and may not be long-lasting. The only way skong and relatively long-last- ing change can occur, and the hallmark of the model, is through high elaboration of the issues and arguments in- volved in a message.

A RECONCEPTUALIZATION OF CHANGE: THE COGNITIVE

RECONSTRUCTION OF KNOWLEDGE MODEL

Because of the similarities among tbe theoretical assumptions ofthe various change models, cognitive pychdogy, science education and soeial psycholopical nsesrch can be used to reconccptualii change within a cognitive cwskuctivist framework. We present hen our reconceptualization of change. We call our reconceptualization thecognitin Recon- skuction of Knowledge Model (CRKhQ. Our model takes into account nsearch within a cognitive psychological per- spective. science education re.se~n:h, and social psychology.

Ollr reconceptualization of the change process begins by examining the interdon of lcamcr and m c 8 ~ ~ g e charso- t e r i b (see Figure 5). The first laarna characteristic of importance in the change prooess is an individual's exisling conception m g d g an idea, topic, event, or phenomenon. Social psychol&ts have often ignored existing concaptions or pior knowlsdge in their nseatch on persuasion (P. White. personalcemmunication. 1996). Sometimesthese~e~eurchms s e l d subjects who self-reported that they held particular stances on issuas, and in M doing, provided some control of prio~ knowledge in their T C S ~ ~ T C ~ . However, for the most part,

individuals' existingconceptions werenot assesscdbeforeths persuasion intervention (Eagly & Chaiken, 1993).

Research in science education (West & Pines. 1985). as well as in reading research using ashma-dmxetic approach (Andem &Pearson, 1984). and mon recently, in mnsmc- tiviat theories Of knowledge (cg., Steffe & Gale, 1995) has shown that understanding students' existing conceptions in critical to u-ding the knowledge recollgtnrction proc- ess. This research indicates that one's prior knowledge re- gard'mg an issw is not an ali-or-none wrmuct. b a d , individuals diier on how much prior knowledge they have and also on the quality of that knowledge. Sometimes prior knowladge can get in the way of processing new information (Guzzetti et al., 1993; Lipson, 1982). Prior knowledge also influences indtviduals' intapretation of the same events (Gaskins. 1996).

Therefore, we note three relevant qualities of a learner's existing conception that influence the Iikelihood of change: strength, coherence, and commitment. Strength refers to the richness of a learner's existing idea, that is, is it well f o d and detailed or spame and fragmented? Research suggests that the strength of an existing conception is a factor inflwncing the likelihood of change (Eagly & Chaiksn. 1993). The stronger the idea. the less likely it is that change will occur.

Coherencenftrs to the conceptual coherenae of the indi- vidual's existing knowledge. That is, does the existing con- ception provide a explanation of the phenomenon, idea, or event that fits together all the evidence, leaving no loose ends? Thagard (1992) argued that the likelihwd of change is af- fected by the conceptual coherenccof theexistingconccption. When existing ideas lackconceptual coherence, they are mom susceptible to change.

Finally, individuals can be mom or less committed to their existing conception, regatxiless of the idea's skength, coher- ence, or both. Individuals' commitment to their ideas can come from a variety of sources, such as sensory experience, social group membuship, or cultural background. Sinakaand Dole (1998). in their work examining creationists' views of evolutionary biology, identified what they called a ''need to believe." Individuals in thew study reported fhat they could not accept the idea that humans wae animals because they were committed to the conflicting notion that humans beings are a special creation. This nscd to believe cuts a c m id i - viduals with high and low background knowledge in both creationism and evolution. Individuals who are very commit- ted to theii existing ideas ate less likely tochange thoseideas. Thus, it is imperative that a model of knowledge change examine the skength, coherence, and commitment of an individual's existing conception to assess the likelihood of conceptual change.

Another critical. albeit interrelated, characteristic of leam- em is whether they are motivated to process the new informa- tion. Motivation has been identified indii t ly in the CCM and mon d i i y in Strike and Posner's reconceptualization of theCCM (Skike & Posner, 1992). Posmret al. (1982) used

Learner r L A I J L I ~ ~ Conceptiu~~

trength Coherence Commitment 1 -.,.Motivation !.. - -.

Relevance?

Social Need for '

Context? Cognitio~

~~~~~ ~

Strong Conceptu Change

F

MNo - If Yes

Plausible? 1

I Engagement Continuum

If Hiah If LOW

lo Conceptual Change

Periphera Cue

Present?

FIGURE 5 cognidversonsrmctiw of knowledge.

the term dimtisfacriDn for describing what motivates indi- viduals to change their ideas. lhey argued that individuals, wbdhar scientists or students, must experience something like cognitive conflict ordisonance (West & Pines, 1985) to consider change. The conflict or dissatisfaction comes from seeing anomalies b e t w n their existing conceptions and new data (Chinn & Brewer, 1993). Cognitive conflict can also arise when individuals hold two views simultanwusly that conlradict one snother (Heed & S u m , 1985).

In contrast to the CCM, our model posits dissatLt//aftion asjust one ofmany reasons why individualsmay bemotivated to prous8 new information. In fact, the CRKM includes four facets of motivation. of which dissatisfaction is one. Dissat- isfaction may be the critical motivating element when scien- tific ideas conflict with intuitive ideas about how the world works. Inmitive ideas me formeddirectly from senmy expe risnccs--like the sun @ng on the horizon-and are thcnfon particularly resistant to change.

However, it seems clear from the research in both cognitive and social psychology that at kast in domains 0 t h than science, individuals can be motivated to change their minds without being dissatisfied with the'iexisting id-. For exam- ple, they may haveastakein the outcome(Chaiken& Stangor,

1987) or an interest in the topic (Alexander. KuUkowich. & Jetton. 1994, Hidi, 1990: Schiefele, 1991). They m y be emotionally involved (Gaskins, 1996) or have high self-effi- cacy related to the topic or task at hand (Dole, Brown. & Trathen, 1996: Parajeut 1997). Thus. we consider these el* ments as possible alternative sourns of motivation to process new and conflicting information. We use the twn personal relevance from thc ELM to cap- the motivation factor stemming fnnn these personal factms.

The third fact^ we subeuine under motivation is a host of social contexts that may persuade an individual to process information with high engagement. Students may be moti- vated to process a message because their peers show interest in the topic. Often, school administrators or teachera invite guest speakem to give talks to their students to convince them not to d d c and drive, not to smoke, or to pmtice safe sex. Depending on the credibility of this individual, the students' respect for this individual, or their peers' nactims to the speaker, the guest could provide a motivating context to p-s a message that otherwise studen0 would not want to hear. Alternatively, students in a group discussion may be motivated to consider new or conflicting information they have disregarded in the past because they value theii peers'

viewpoints. In sum, the context in which the message is considered eonfains a variety of social omtext variables. Interactions with members of q. community, school, or pem group may motivate individuals to pmcess information they would not othwwise consider.

Finally, some individuals, by their very nature, are inher- ently motivated to process information. These individuals have a need for cognition (Petty & Cacioppo, 1986). lhGy are. challenged by copsidering many s i b of an issue and enjoy the cognitive procedng inherent in high elaboration. Moti- vation mfmchers might characterizt someone with a high need for cognition as "intrinsically motivated" to engage with ideas, infamation, or concepts. Pincrlch and Schunk (1996) dwcribed intrinsic motivation as willingmgs to "engage in an activity for its own sake" @. 257). Individuals with a high nted for cognition are also willing to persist in their consid- &on of ideas. This persistence is akin to the motivational construct of 'Colitim." Como (1993) described volition as "a dynamic system of psycholo@cal wntrol promass thnt pro- tect concentmion and directed sffort in the faca of personal and/or 6nvironmental distraction" @. 16).

Certainly, we do not consider these four elements to be the only motivators that may contribute to the change process. However, baaed on existing nscarch in cognitive and social psychology and science education, we think then is Bufficient evidence. to support them as elements that may lead toward motivation to proocss information.

I n m m t i ~ with the leamerCTs existing conception and his or her motivation to process a message an critical features of the message itself. The message canies with it a set of variables that are unique to that particular message, such as the format. organization, and the task implied by the mes- sage. 'Ihcse. variables intMact with l6arms' existingconcep- tions and motivation to make the mwage more or less comprehensible, plausible, coherent. and rhetorically com- pelling to each individual.

Following the CCM, we hypothesize that a messagemust be ~mtgrehibk and p h i b k to a Wcula r individual. For the message to be comprehensible, it must not be too conceptually difficult for a paiticular individual to grasp. Fmther, the individual must have sufficient bsck.$round knowledge to d a t e to the message.

For a message to be eonsidered plausible, an individual must decide that the message could reasonably be me. That is, the individual would have to weigh the quality of the evidence and decide on the probability of its truthful- ness. For example, recently in California, 14 members of the Heaven's Gate cult committed suicide. They believed that their souls would be saved by aliens in spaceships flying behind Halley's comet. Although it is not hard to comprehend the explanation offered by the cult members for their plan, mom of us would de&de that the explanation fails to meet the plausibility criterion. Specifically, the evidence supporting their views seems tenuous at best, and the probability that their explanation pomays a likely sce-

nario seems to be quite small. Unless information is both understandable (comprehensible) and credible (plausi- ble), an individual will not be motivated to process it.

Individuals must also find the massage to have a csrtain level ofexplanatory eohomnce('Riagard, 1992) in accounting for the phenomena. In other words, it must provide an expla- nation of the phenomenon that linlrs ideas into a wnoeptual whole. Last, amessage must be rhbtoricalIy compelling to an individual. That is, the language usage, tbe sources of infor- mation that form the arguman5 and the justi4ications psovided must be convincing and persuasive to the individual. Some individuals may be more compelled by an impassioned and emotional s p h given by a politician on the plight of the h o m e h , whueas anotbR individual may be more e m - pelled by a candidate who V n t s dam on growing housing costs and statistics on population growth to make his or her point. Conversely, a message cannot be Morioally compeI- ling if it is ambiguous, confusing, or disjointed.

We c o n c e ~ existing wl~~eptioas, motivation, and a given mosgage to form an inwraeting, dynamic system. A comprehensible message may be personally meaningful to one individual and yet meaningless to another. Some indi- viduals who are initially dintac?ated in a topic may become motivated to attend to a message if thsy find the musage rhetorically compelling, but athas may not find the nrgument persuasive. Students may be more willing to work at compre- hending a confusing message about a univusity's conversion from qu- to semesters if Ihe information is personally relevant-that is, if the conversion will take place before they graduate. Thewfom the qualities of a message can only be considmad as they intexaet with a given individual's existing conceptions and motivation.

It is interesting to note that the ELM @ins with mes- sage effects that in time, may lead to motivation, whereas the CCM begins with dissatisfaction-a kind of motivation to process--and then message effects (comprehensibility and plausibility) come into play. The reverse order of these components in the ELM and the CCM may bathe result of differences in the bodies of research the two models try to explain. The ELM synthesizes experimental research, which typically begins by introdwing paaicipants to a message in a laboratory setting. The paaicipaots are there because they have generally received some sort of compen- sation for their rime. Thus. they may be motivated by the experimental conditions. If the message is comprehensible to the pacticipant, the experimenter can then ass= whether the individual is motivated to process the information further. But this order of events may be an artifact of the experimental situation.

On the other hand, the CCM tries to explain conceptual change within a classroom context. h a classroom, stu- dents are not always rnetivated to process every bit of new information they hear. The CCM 5uggest9 that if students are not motivated to process new and conflicting informa- tion, they are udiiely to attend to the mmsage. Thus, the

process bsgins with motivation for the CCM but with the message in the ELM.

In our model, we hypothesize that learner characteristics and mwsage effects are not linear at all. Raehor than suggest- ing that the procass starts at one particularpoint and proceuls in a step-by-step fashion, we wish to capture $he iterative ~atursofthechangcpncpur.Weima@nethatthenare chmmmm under which the change proccss can be initi- ated at the 1- or at the message level.

So, @?he intendon betwecn lcamsr and message charac- terislice occur in a W t i v e manner, then the individual will lilccly process the information. However, this processing, like pria knowledge, is not an alla-none phenomenon. We suggest that the processing ofthe infomation itself lies on a continuum from knv wgni'tivc engagement to high nrstacog- niriyb engagmm. ' he notion of acontinuumof en@ganent relates to what persuasion nscmhess call thb continuum of elaboration. High elaboration aQcu18 when individuals en- gage in e W , analytical proceasing of information. At the other end of tlie continuum individuals may engage in more supsrficial or heuristic pmco6sing of information or low elaboration.

We use the term engagsmetat to nflffit a cognitive-con- sttuctivist view of informstion processing, strategy use, and

(for Mi notions of nadzng engagement, m Guthrie, 1996; Outhrie & Wigfield, 1997). We bypothep'i thar engagement ean exM on a conflnuum from low to high, Wvidualsmg@inalow~m~~~tofpmces~may attend to new information and actively process that information through datively simple strategies, such as maintenance reheanat or mnemonics, p m m m requiting little reflective or mcfacopitiva thought. This may lead to the assimilation of ths new infnmaton into existing conceptions without sisnificant change to those conwptions, 'Zhese new idea could also be membered but not integated into existing coaecptio~, leadiig individuals to wrapsnmentalixc the in- formation (Chinn & Erewer, 1993). Compartraental'ization would result in no change to fhe existing conception. Tbup,

as depicted in Figure 5, low engagement will often result in no ooRecptual change.

Individuals also may engage in a more moderate level of engagement. They may mekc meaningful connections with their existing conceptions through such activities as answer- ing l ehn t ia l questions during reading (Benton, Glwa, & &'wing, 1983) or drawing pgsoael umnffitions to the con- tent (Alvemmn& Hayes, 198% M m w , Rcssley, Smith, & Smib, 1997). lhis moderate amount ofengagement involves gnatsr ckpth of pmcassing, more elaborate suategy use, and somcrcflectivethought ormetac~&nitive ragulation. Herr. we tslathcELMonestepfuaha, largely basadon thesubstsntial bndy of nscarch on metaoognitien ( B a h & Brown, 1984, Oarnet, 1987).

We posit that engagement at the hillpast end of the contin- uum would involve deep procassing, elaborative shatc&y use, and significant mefawgnltive ntlwtion. We propose that

individuals invalved in the highest form of elaboration would not only engage in thc cognitive srrategles of connening and comparing existing conceptions with new informiaton, but they would be reflective about what they were thinkkg and why. Thcy would think deeply about rhe arguments and c o w ~ t s related rt, the message. This is the higtrost form of mctacbgnitive en$agement. It is only through Bigh metecognitive engegmant that smng conceptual chsngc is possible (mFigum 5).

In what context would such high engagemmt occur? Bere- i w (1990) described a leaning situation that captung what we mean by mefacognitive engagement He called rhis situ- ation intcntwnal iearncm in inquieg ciasmwoms. In these clasaooms, students are awareef their ideas and have control ovar their own h i n g . Such classro~ns become swings where stu&nts knlr: Haw do I I-? How do I solve prob- 1-4 How do I chinlt things t6rough critically 7 Am I aware of whaf I believe? What would it fake to convince me other- wise? Studants in such classrooms would I% taught to think critically, possibly tlmugh instructid mchniqucs such as disoussion (Alvcmmn & Ham, 1989). critical inquiry, or questioning the intent of the author in communiosting a message (Bffik, McKmwn, Hamilton, & Kucan, 1997).

It seems tous tharthecontinunm ofengagement is the most impntant element of the change prows& In4ividuels' proe- essing of a message is not an all-ofinone phenomeaon, even if they are m o t i v d to pocess the massagp. Instcad, in& viduals' procasshg lies on a continuum from low cognitive angagement-active wndderation of i&w-to high msta- cognitive engag~mcr ,~p ,ana ly t i e , criticul dection and awareness and regulation of thought proeessas.

Note that even if individuals process a message at a high matacopitive l e d ofengagement, concepml changeis Dtin not guaranteed, For example. after rcpats of cod fwsion had been publicizrd, many scienthts engaged in high daeoration by weighing the issues and arguments caretully and rhor- oughly. In fact, many conduotcd experiments in anattempt to replicate the qmted results. But sciLntist8 finally rejedcd the new infomation about cold fusion, even though they engaged in high elaboration. Whether high ebboration Isads to conceptual change depends on s complex intcrsction among existing wnctpi.ions, motivation, and the incoming message. (e.g., m u h from the new experiments). W~entists may have been vexy motivated to considex rhe information, butthey may havefouadthenewmcssageaboutcddfusjon to be less cdmmt or less dmtatioally compelling than their own existing concepdon about fusion. In fact, their existing conceptions were likely btrenptbned by the'u failures to replicate the cold fusion cffcct.

Depend'ig on the outcome of the engagement with the issuts, an individual may w p n d in one of several different Ways (Chinn & Brewer, 1993). In the cold fusion example,

scientis$ rejected the new infomation and held onto their existingconceptim. But they wuMhavehad otherreactions. Por example, if fheir own W i g s pmved inconclusive, they could have suspended their judgmant until additional infor- mation was available. Thus, wncepEual change depends on the outcome of the engapnent process as it ~ a c l s with the lcama and massage characteristics.

Our t e c m z a t i o d of the d t s thw men if Mi- - . viduals an not motivated to change and do not fmd the new information comprehensible or compelling, change is still p i b l e . As depicted in Figure 5, a peripheral cue can induee low c w t i v e engagement, which can, in turn, c a w B supsr- ftcial change in conceptions. This is anaIog6w to the @ph- eral m t e to attitude change identified in tha ELM U is impwtant to keep in mind that engagement that occurs through pnipheral cues is often low, and thuscbange is l i l y we& temporary, and smngly susceptible to funher change. Science education research is replete with studies showing smdents who pafono well on measures d knowledge & insbvctien but re* back to tbe'u old emoqtions ova time (Uuueai et at., 19PR West Bc Pines. 1985). Tlms, paiphnal cues that i n d a law engag,ementmay producetheappearanee of change or weakconceptuul change (mPigureS), but such change is likely superficial and trmporary.

However, as rasea~ch on the ELM suggs8ts, a peripheral cue has the potentid to draw an individual into high engage- ment with the isaua and arguments. For example, if a group of ranchus who are opposed ro the ninwduction of wolves into Yellowstone National Park hear new information about &m to owaol the wolves' predation on livestock from a fellow rancher, tlrey might be willmg to take a c k look et the issue. They might be iaters8ted in gaining fruthw infor- mation about tfmecontrol efforts and the sucamratcs of such measu~cs. In ether words, the fellow rancher m l d sefve as a periphmal cue, drawing the ranchers into the type of high metaeugnitive mgaItcment that could laad to conceptual c*

The CRKM deserih a complex interaction of existing lmowledge, motivation, message effects, and engagemenk lbs, it illpotrates why radical conoqmaldutngeasdasrribed by Chi (1992) is so wrz Chahghg naive or intuitive existing conosptions o h quires individuals m reassign wnoepts acms what Chi cancd ontological mtepries. This type of change is nmksssry in the study of physics, for example, if studem lrpe ta wmpt w W am often c c w ~ ~ h z i t i v e scie~t tific explenasions for phenomena One mason such pdound changee are exfmndy difficult to achieve, according to the CRIEM, is that naive or intuitive existing c o ~ ~ ~ p t i o n s are l i l y strong and coherent Pnrther, individuals may be very committed to thcse conceptions beeause they &re based on their everyday expaienm.

What would it fnke to prinIuce radical conceptual change sccording to the CRKM? It would take an exceptionally well-formed massage that an individual foand to be very cohepeaf comprehena'ble. plausible. and rk.torically em-

pelling. It would also require the individual to b motivated to interact with the message at the highest end of the engage ment continuum. It seem clear why radical amptual change in so diffcult for t e d ~ ~ to pnunote: The coaditiw necessary for radid change are challenging to pmduce in a classroom setting.

In sum, our l t ~ ~ ~ l c c p h E a l ' i o n of the change pmes

RESEARCH IMPLICATIONS OF THE CRKM

We have m n t e d a model of the conditims that nead to mew for h e cogniti* ncanstnrction of knowladgc. Itl db vebping the model. wednw from theory andmearch in three areas. In this way, we hope to wmbinc the best feahues of these perspectives to develop a more general model of con- ceptual change. Next, we examine each c o m p m t of the CRlCM as it relam to the &sting r-h and suggest arras for future remarch that need to be itddresised to more. M y undcrstandthec~eproa%s.

Existing Conceptions

fdentifying existing conceptions about a topic remains a basic probhnfsr~csearch~r~. One issue that has madethispmMem particulady complex is thst existing conceptions cons& of d i f f m t types of Bnowitdge, e.g, topic knowledge. do& knowledge, and metawgnitive and strategic knowledge (Por arcview of knowtedge tmninology, ses Alexanda, Shallat, 8t Hare, 1991). The implications of this issue for thosc of us intmsted in conceptual chgertseatr;h is whether we o d d identify d i f ferem id thc conccpual change prowsdepand- em on the type of knowledge.

Given that them are these d i f f m t types of knowledge, another issue. comes to the fore: Can thesu different types of howledge he assessed? Alexander, Jeiton, and Kulikowich (1993) developed twe types of q W o n s to separately assess domain and topic knowledge. Questions used to sasess dc- main knowledge consisted of items measuring key concepts and principles in the domain. Questions used to assess topic knowledge consistad of detailed knowledge tied specifically to a given twrt on a topic.

Thsoridcal issue for the CRKl.4, however, is not only haw to aeaqs diffemat types of knowledge, but how to asses the s t l~g th , coherence, and commitment of sn individual's ex- isting conceptions. Clearly, empirieal wearch is needed to establish whether theseconstructs can he measurrd.However, much wisting work in knowledge representation and assess- ment suggests it may be p s i b l c (see Sonassen. Beissuer, 8t Yacci, 1993). Recall that strength refen to the richness of an individual's wnceptualkmwledge. whetherit is well formed and i n ~ n a c c t . d or sparse and ffagmentsd. Shangth could bc d d n e d by asking individuals to C O I L S ~ ~ I C ~ network raprcsantatim, such as webs or semantic maps (F'iher, 1990). It may be possible to assess the relative shangth of individuals' c o n a p ~ a l knowledge by dctcrmining the number of Laanchas and instawes they produce in generating thentsps.

SiIarly,conceptual coherence could bc assessed by ask- ing individual# to produce concept maps or causal maps (Jonaracn a al, 19!B). In rhew network replwsntations, individualsmost~otonly generarenodesandconnections, but they mwt label the links in terms of therelarionthey repment (using W l s such as caused by, fjpa ox characterisric c$ neceJsttatcb, c@crs, etc.) Thtnodes and labelimg of the links could bcjudged asdepicting more or l m explanatory coher- ence of a phenomenon, idea, or event. Another method that may p e useiul for assessing cohgGnce is the use of simi- larity retine (Diekhoff & Dickhoff, 1982). This method involves asking Wviduals to genuatc a list of o f s nlatsd to a cancegtand judge the degree of simiiaity or dissimilarity between ~ a o h pair of listad tcmls. ' lbss ratings can be as- ssssed using staristical beehniqucs, such as clweranaiysis or multidimoosiomal sealing, and individuals' ratings can be compared to each other's a to an

W i n g individuals' commihuent to their ideas may pow to be the most challenging. One technique we are upIoring is asking questions such as T h a t could someone tell you thst would change your vkwl" and then scoring rsspons~s bnaed on a N ~ S ~ C ranging &om highly committed (reapembits indicate that them is nothing anyone wuld say that would change their thinking) ro weakly committed (re spoadents iadicate an openness to bearing evidence in o w sition to their view).

Clearly, diffarent t y p of measurea may be needed to identifyttlGsbngth, wherence,and wmmibcmtofindividu- als' existing knowledge. A program of rss~arch examining possible measures is l i l y to produce new and helpful infor-

mation regarding thecharacterization of existing cowqtlons. As well, it may bc that different measures may work be* in some domains than o w . Identifying the interactions h e tween measures and domaim may be an important conhibu- tion toward understanding how to tap the nuances of students' existing conceptions as they affect the likelihood of concep tual change.

A critically impormt construct dad to learner charac- teristics in the CRKM is motivation. Thc conarnct of moti- vation has a long-standing 'tradition of research in the field of psychology, imcludingsocialpaychology. Within educational psychology as well, rasca~ch has been extpnsive, particularly as motivation relates to academic achievement and schooi learning (s.g., ( k n h a ~ ~ & Weiner, 1996, Stipek, 19%).

What does the body of nscaroh on motivation within fhe8e traditions tell us abwt what might motivate imdividuals to consider new information about a topic about which they ~yhavcbcliefs7Toc~idnchangingons's mindmeans to be wilding to engage in the deep cognitive and metawDpi- tive processing - s a q to reexamine issues andmgummts. For example, one of the authors is cumntly working with teachers and students in the Southwest. Some students come to school with negative stereotypes abnd Native Americans (or Anglos). In faet, some teachers come in& classroom with the same stereatypes. The stereotypes influence the way teachers view students and their work. 'Ih sterrotgpes also influence the way students view each other and the many topics that come up in social studies, science, and hismy classes.

What happens when Anglo students read a text (or beat a lecture or audiorape) describing the negativeeffects of settlers of the American West on the Nativo American culture7 Or. what happen8 when Anglo teachers rend about (or ate told) thenegetim impact of Native American sbcnotJlpesmNative American childnn's learning? Will Anglo satdcntsand teach- ers considcr tbe measage w k n it contradicts existing bftliefs and wncsptim?

Existing social psychology theory and-ch has shown that individuals can be motivated by the personal relevance of an issue or bscause they have a stake in the outcome, Thus, students and teachers in the Southwest may be m m willing to consider a messsgc about Native American (or Anglo) cultlne than about other cultures. In this w e , Native Ameri- cancul'hlremay have more personalrclevanue to some South- we% students and teachers than, for example, other cultures may have. As weU, the ELM pdic t s that certain peripheral cw-aclascfriend o r a m t w o h y source, perhaps+&@ serve to facilitate later d q processing and rethinking of issues related to thew topics.

What additional insights w motivation to change do cm- rent cognitive perspectives give us? Motivation wearchers

have developed congtructs mlated to aehiovement and lean- ing in @n&ral (Dcmbo & Eaton, 1997). We sa the value of this nstrnch ss helping us envision a general di isi t ion toward &-regulated h i n g and deep mQacOBnitive proc- essing, Studan& who have self-confidence in their abilities to laem, who have the ability to self-monitor and self-regulsre Wi learning, and who are intrinsically motivated and have a mastery gaal orientation would bc in a better position ta reexamine theiu e & h g beliefs (pmhirh et el., 1993). Such behaviors would be consistent with Beniter's (1990) inquir- ing, self-regulating lmers (see also Paris, Wasik, & Turner, 1991; Zimmannan & Risemberg, 1997) and with social psy- chologists' msm ofa "need for cognition." Oa rhe other hand, 8tudef1ts who are motivated by the need to get a gaod gradh who want to l a ~ k good to their pcsrs, or who view the classroam ss a oompetitiVe envircmmant are lass likely to =gage in the deep processing neccssray to change their wnoopions ar beli~fs (Guzzeni & Hynd, 1998).

Thus, the cognitive rascarch on motivation identifies char- gctaistics zhet pmmte in student% s gcntral disposition to I-. However, W g e n d dtspitlon may or may not be the same as a disposition to re%wmina issues critically aad to engage in deep metacognitive p~ocessiqg. Thw, the tesearch has not focused on motivation to roleam or recomider wmc- thing that is inconsistent with existing beliefs. It is one thing togain knowledge and beliefs about some&ing. But itis very diffaontto~idermdmdi~koritically new and conflict- ing ideas for wbieh you slnady haw knowledge and belicfs. Bacs it mke a d i h t level of motiwatien to be motivated to msxamina existing ideas than it does to examine something with which one is unfamiliar? Does this type of motivatioQ differ only in d m or in kind?

A sccMd foam of cognitive rsscapch on motivation is the identaicstian of contextud variables that might promote mo- tivation b lerun something that conflicts with existing beliefs. Motivation ressarch has @en edwatm important ideas a W contextual variabIes that we more or lsss l i l y to promote motivation to kern. These can be used to u n M hetIerthecmtextthatcanbecratedtomaximizGthe~i- bility that sNrlsnts can reconsider theiu &sting Wtefs. For exampla, classroom cuntkxts thatpromow u)operarion rather then eompttition, that use intrinsic rather than extrinsic incen- tives, that promote mastery rather thaa puformanw goals, and that provkdtsmdents withmoderately d i c u l t tasks tend to mate Itamem who are more likely to edgage in strategies that foster self-regulated leasing @embo & Eaton, 1997; StipeL 1996).

As we examined the cognitive and social psychology rcsearoh on motivation as it relates tochange, it isdear that we are looking at dierent types of motivation We nted knowledge about general motivation to I-, motivation to consider a plntioular message on a particular topic, motivation to change, and motivation to engage in high. met%eognitiv~~~eotion. Although we have more research in some. enss than in others (e.g., motivation to learn vs.

motivatioatoconsidetaparticularme~sage).~eg~netal~ of motivation for reconsidering one's ideas in light of new and conflicting inf~mation remains a topic for further research.

Comprehensibility and Plausibility

Comprohensibiiity has been studied extensively by cognitive psychologists and social psychologists. It is mtaesting that both areas have examined written texts (messages) form- prchcnsibility. Meyet and Rice (1984) rcvlcwed a body of early wgnitiw work on text structures and how b y &t recaU and l m i n g . Mditional mwwh witbin e cognitive perspu,tive identified numerous ways in which text can be "inconsidmate" (Atmb~stcr, IS&+).

However, more meamh is needad on the inteaaction of wmpreheusibility and text difficulty. Tho regearch has fb cused on comprehensibility as easing the amount ofeognitive effort needed to understwd. The pemiontaseerchsu$gests that sometima easier may not be better. This wMlld be. consistent with nsaarch by Chall, Conard, a n d l k i s (1991), who found that high schod students considered some text- books "too easy!' These students reposed los* interant in boaks that they found to be too simplistic. Further, Remhger (1992) found that students who were intuaMsd in a topic comprehended text bettcr even when the material w s more difficult for them to read. Thus, motiv8tion may intemct nith comprehensibility in intemting ways. It would be ltsaful to know how mis finding relates to conceptual change.

It is inmesting that plausibility has not bsea studied in detail. But Poslntr and his coUeaguw (Posr~er at 81.. 1982) recognized that not only must a claim be undgstandable, it mustmake sen= a d belogical. Thiisconaistentwith astudy by W e , S i t r a , and Reynolds (1991). who found that an important determiner of studants' accoptancc of creationism or evolution is whether the argument8 "made sense." We believe this holds tme for domains of knowledge otbg than science. But how do individuals know if something 'bakes sense"? We think this is an interesbg inea to explorein teamg of conoepfual change. U n h an argument "makes malrc9'' it is unlikely to be coneidwed by individuab-no matter how motivated they may b.

High Metacognitive Engagement

F l y , we believe the CRKM builds on the strengths of the pe~sugsion and cognitive research to identify an m g w t continuum that has not been discussed in previous nsaarch. Ihc CRKM predicts that students can engage in thinldns about an issue at many different IevcIs, from very low levels of engagement. to medium, to high levels of engapwrit Consistent with the hc on mefmopition, we propose that the highest level of engagement involves deep thinldng

and p c d n g of the information and reflwtion on one's progmsion throagh the pmess.

tiowevex, ne&a~ the wgnitive nor the social psychology ha M b c d the tngpgmmt plowis very well.

Csso Pbrdiw of students who en vadously engagad in a topic m y gcaarstG SOBM idcrrs about what v t looks like in a cbuunxrm. GuaGni and Hynd (1998) pregented claw room dgneE(as in which skudents varied in thek levels of enmgsent in science acFidw. Thcn, students' levels of aapaOemeat and nd thcns b e h ' i t h e ievtls wart pDMu- l a t d b y l w w ~ m h n a ~ d i ~ ~ o a l ~ t i v e s . Studias .such as these should add SignidoanUy to om under- rtondidg of engagement DevcIopmentsl data may be appro- ~bgO,~wd.Itislikalythathighengag~tatthc first-grde level will look very different from high engage- mestatthssscoadarylovel.

Ttreiamrdetion&weenh&h tneascopitive engagement abut lwrning and high met~oogni t i~ engagement abwl a topic needs to be examined more dasely. Webetieve them is rnac to 1- rbout motivation to 1sam psrticule~ topics. This is eln~ently bcing done a$ the Nationel I\oading Remmh Csntor (e.g.. Chihie & W i d , 1997), where m h e m am sxanafniag atudGntat8 motivation m read. But we need to c o n d u c t ~ m ~ ~ ~ ~ ~ h on how motivation to learn inicracts ~ e ~ w i t h s p e c ~ c ~ .

A m t h fraitfu1 a m for rcaaarch on wnocpeutl change is to exphe et111p@tompnt for leaning about a topic in one Wvidual bva time. 'IB'i would be productive for examia- iag @saxpieition afbbtiofs a8 WOll as oonqtual change. Two isMlCr are hpmteat hnr. First, we aeed m Learn how klicfs am dewloped initially. What level8 of engagement are nod& to form a bslief? R d in neither d a l nor cognitive psychoIogy has ddRsgcd this important issue. Sewnd, once a belief is formed, what particular a r p c t i e m ~ u t e to kUef change? Row are i w w and arguments weighed, mjsctsd, cr cDunteB evcr time9 9 would be inter- esting to follow the tlrinliing of individuals who are in the p.oaess ofcWging their id-. Studies can be done with edults, butillsy alsOmu~t bedone with tniddkaad s c ~ ~ ~ d a r ) r

atadcne. &y y e e ~ s a g o , ~ ~ ~ o n ( ~ ~ ~ ~ n o t e d 'hadoubt 11~hcrnsM chaqp Ly gradual extension, articulation and re- finement" @. 424). It would be interesting and informative to & a finegained analysis of extension, articulation, and FeRnement of the chanp proms.

A f@ contributioa of the CRKM, and one that we think is d d y imporrant, is the idsntiftFetion of eltwnstive vad- eMaa rdrtad to change. Tho CRKM wts that change can oomsabotit duou@hthepsriphsrsl mate. When this hsppsns, sardeatll am not involxed in high engagement and eIaboratim but in quick hatdstic judgmentsthat do not lead to smng and l e n s . m o b a a g c .

B u t t h e m o d u l ~ ~ e $ l t b a t a ~ ~ c u e c a a l s l d to high engagement. that will evmtudly W to oonc-cphtal change. Socia2 paycho1ogy research has v ~ d o s b m ; e a y educafors know intuitively. Outside s e 6JRWiPlly those to whom sndants can relate, can oftea have-dm effeet on stodsras than OGachers caa S i g n i f i c a n t o u c ~ ~ a a aa €riggem to iaducc or promote high copitive which eventually can lead to change. Ths CIUCM offem an explanafion for why .sMm@

to thajr pvioasly held w e 0

Niederhawer, 19W). It is likely they en rial at a fairly low cagpitive level, rathsFthan at a@& laucil. l'l~erefore, when memhem evaluate studca$' Wn&g b u p h oral or Bvritfeu assessments, they nwd to be atwax that change may appear just aftcs instnwion but overtime. W e t h i n k a n i m p m t ~ t ~ ~ E a i i s a u D i s t o

We do no! want to lenve the impredm that we believe schools are all ebcut comcptual change. llh eortainly is not rhccapc. Most knowledge is assimilated by students mugh extending expan&, and mfiniig existin$ coaseptions. Uoreeva, when *ts are wnfmntad with a n d data, wedonotlnsanto~lythatthe0atymsponscshould beoonocphlalchange. Quitethc w ~ , e d u c a t o r J h o ~ ~ students have inquiring minds and that they cballengc, ni- tiwe, apal~na~ &criticaUy evaluate whattliey cxpekenceas conflicmingdata

So hew impo~tant is conceptual change for sduostors? Afieccdl,howoftonanstlldcnrPconfronti?dwithtcdlnformation that wnfli* with lhsir prior knowI.dee? fn fsch hofh adults and children plr wII&Mtcd with anomalous data aa a part of tBeir everyday snpgienccu. We believe trend will in- onasc as technology, computers, and the tbaarrrmtion Age bwomemoreea~inoursodetg.Tbsmopewecorh- municats with olbcbs mund UIC wid, the mms l W y we will be amfronted with DIIOBMIOUS data.

Re~mofstud~ts'axistintingview, aducataaohopstha students will gain laon bigh t into critical issuea king society and bc able to view them from multiple panspactivw. lhig is p i b l e , howeva, only to the extent that studem bsoome highly engaged with the issnrs awl arguments. A ~ u o d s r s t a n d i n g o f t k e c h a n g e p r ~ w i l l ~ l p e d ~ create an enuironment in which sNdents can engage with multiple pqwcdveb.

ACKNOWLEDGMENTS

We thank Patricia Akader and our wlleaguss in the De- partment of MEIcationai SWea for hdpM feedback on

~ ~ . r ; l ~ ~ . & d $ s . ~ . i I e ~ ~ ~ . ~ ~ ~ ~ k (W. 1Mcm. New Yarlr: CnmWBC tJahmby FRa.

. 32.54%.

Nickcma R. S. 11985). Un-g AnurlclaZavrrJ $ Education. 93.201-239.

Nussbarrm, I. (1985). Tho senh 06 conraic body. h R Drlw, E. OPtslg &

Oabom* R. I.. & w ~ - M . C. (1983). l.emhg ~samiapscicm: A @n pnaespl. Science Eduuaicqh W, &B-J(W.

Pmju. P. (199T). Sclf&apl bellda in acsdsmic acvings. RNLnv of - ~ u s c v c h a ~3-na.

Tbce~stofirsmodcntiqgofkawtbs~ivcimpactofsamxeucs. J w d a f CunncnrrRap~(pch 18,5262.

R d n g r . R. A (1992). ladlvidud ia~uesi and dmlopmcDE lmpliePima failag,~dPIIICtLS.InI.A.Rs&ga.S.Hldi&ADnpp(Bdr). Thrmkoflnlenn~kMu'ngamidmlop~(pp.MI-3%). lMI&k NJ: L D w r c o c c B r l b a m ~ . kc.

Rmm. E.. & Morvis. C. (1978). ChUdrm'a 8odng A ffiin-011 bssrd ontbshcnatunofnbstrsdimin~~~~~~.laM.S.Sman&RC.

Schicfek, U. (1991). lnknn, lenming, and mMivation. Educational Psy- chologirt, 24 299-323.

S W R. C.. &Aklsoo. R. (197%. Plons, scriprs, gi~&andundcrsuvldIng. Hilladale, NI: Lawnmc PrIbaum krraeiates, ine.

Shaif. M.. &Hovlanb C. 1. (1961). SocinIJudgmnr. New Haven, C T Yale Udvedty M.

Slnnw.G.,BNning,R..&Svtrboda C.(1995). Sovncsofsauntionalinknst. J w d of Reading Behavior. 27.1-17.

Sicglsr. R. S. (1996). Ennrging minds: Thcpmcus ofchange in children's fhinLIng. New Yorlr: Oxford University Ress.

Sinaim, G. M., & Dok. 1. A. (1998). Knmrkdgc versus bclicfs: An nnipcinl in rhc srvdy ofevolution Uanwripf in pnparation.

Smi~h, I. P.. di Sessq A. A,, & R 4 e . 1. (1993). Misconceptions nrrmaivcd: A wosauctivia analynis of knowledge in W t i o n . Thc

InuUigem, and educafion @p. 245-2781. Hiisdlle, N1: La- Prlbaum Assaciatcs, Inc.

Steffe, L. P.. & Gale, I. (1995). Const~ucmilsm in edueaion. Hilladale. N1: LaWDaCC ~1~ Ascciatg, l a .

Stipk. D. 1. (19%). Motivation and instruction. In D. C. Bcrlimr & R. C. Cslfcc(EdP.), Hrmdbwkof ed~~mwnalpsychology (pp. 85-1 13). New YOIIC sirnotla Sehusm.

Strike, K. A. & P m w , 0. 1. (1992). A revisionist tbcay of co~cepusl shsogc. lo R. A. DvsN and R. 1. HMiltcd~ @UP..). Philosophy of sciancc, c o g n i r i w ~ h o l o g y , andcdvcofiond rhcory~dprncnce (pp. 147-176). New Yo* SIaIc Univsrsily of New York M.

TeBTr, A., & Shaffcr, D. R. (1990). Altitudes snd .flindc change. Ann& RevinvqfP~hology. 41.479-523.

Tbagprd, P. (1992). C o n c e p ~ / revolutions. Rlaocton, NI: Rinmm Ud- vemity Rcss.

Toulmis S. (1972). Hwnmrodcr&ng. PI~IIcM~,NI: Pthmon Uoiva- sity mss.

Voaniadou, S.. & B m W. P. (1987) .~of laowladgercwuauring in development. Review of E d v c n f i o ~ R c s e m h 57.5147.

Voss. I. F., Gpknc, T. R.. Post, T. A,, & Ponner, B. C. (1983). Problem-soh- ing &ill in ibc said scimocs. In G. H. Bower W), %-psychology of knmlng andnwrivol io~ A d v m in msemh andthcory (VoI. 17, pp. 16S213). New YON: Acsdcmic.

Wesf L. H. 7.. & Pines, A. L. (1985). Cognirivl s m b n and c o ~ a p N n l chmgc. Oltsndo. PL: Academic.

While. R. T.. & Tisher. R. P. (1986). R u r M h in science education (Vd. 16). Pspcrs from the annual d ~ r e ~ ~ of ibc Avsbsliaa Sdcne. PdwtionRuiealch Associaioa. *dslpid+ Sauh Aosaslia: Ausaaliae Sciwa EducPtion Rcaearch AmccWm.

Wood, W., BEagly, A. H. (1981). Stages in the analysisof persuasive messages: The mls of causal attributions and message compre- hension. Journnl of Personality nnd Social Psychology. 40. 246-259.

Wyer. R. S.. Ir. (1970). Quantitative prrdidiooofbeUefandopinionchangc: A further test of a subjective probability md4. J w d ofPersonalify andSocinl Psycholoff. 16.559470.

Zimmrman. B. I.. & Risetuberg. R. (1997). Self-regulatory dimen- sions of a&mic learning and motivation. la G. D. Wys W.). Handbook of ucndcmic lenrning (pp. 106-125). San Diego. CA: Academic.