~l Routledge
EDUCATIONAL PSYCHOLOGIST, 44(3),193-197,2009 Copyright © Taylor & Francis Group, LLC ISSN: 0046-1520 print 11532-6985 online DOl: 10.1080100461520903029022
~~
Taylor & Francis Group
Icep_ Pin , NJ:
Cognitive Scientists Prefer Theories and Testable Principles With Teeth
iver_ :am
s for logy
Arthur C. Graesser Department ofPsychology & Institute for InteUigent Systems
University ofMemphis
eraU and
ome ds.), . NJ:
Alexander, Schallert, and Reynolds (2009lthis issue) proposed a definition and landscape of learning that included 9 principles and 4 dimensions (what, who, where, when). This commen tary reflects on the utility of this definition and 4-dimensionallandscape from the standpoint of educational psychologists who have a cognitive science perspective. Their analysis has practical value in positioning different research programs in the landscape, planning research road maps, and identifying the scope of research efforts. However, it is argued that the learning definition is underspecified and that the learning landscape is both cumbersome and insuffi ciently constrained. Cognitive scientists are more likely to be inspired by theories and testable principles that have more teeth.
I re rnal
ltel tua! .-sity
-ker. eve
68 ndi The article :du ork:
by Alexander, Schallert, and Reynolds C2009/this issue) has the ambition of offering a useful definition oflearn ing. They begin by identifying nine principles of learning (such as learning is change, learning can be resisted, learning is interactional) that are shared by different research commu nities. They also specify what learning is not. Learning is not merely an innate capacity, a maturation of a biological/ neurological mechanism, or the recall of prior material. Alexander et al. subsequently characterize human learning "in a topographical framework, a quadrangulation based on the convergence of the what, where, who, and when dimen sions of learning" (p. 189). I refer to this topological quad rangulation as a learning landscape in this article. Alexander et a!. apply this learning landscape to three cases that include a child biting into a cherry with a pit, an adult crossing a busy street in Italy, and a student writing a paper in a new discipline. There are many motives for a group of researchers to define learning. It promotes coherence in the discipline to the extent that enough researchers are on board and some semblance of convergence is achieved. The learning land scape provides a perspective to position different theories . Researchers who are engaged in theoretical disputes often are merely positioned at different regions of the landscape, so
Correspondence should be addressed to Arthur C. Graesser, Depart ment of Psychology & Institute for Intelligent Systems, University of Memphis, 202 Psychology Building, Memphis, TN 38152-3230. E-mail:
[email protected]
the debates degenerate into pointless turf battles (in this case, "My turf is more important than your turf"). When projects are positioned in the broad learning landscape, empty regions become apparent and help guide the development 0 [- research road maps. The learning landscape undoubtedly broadens everyone's perspective, emphasizes interactions b(tween di mensions, and strengthens common ground among different research communities. The goal of this commentary is to analyze the Alexander et a!. C2009/this issue) definition of learning and the learning landscape from the standpoint of educational psychologists who are influenced by contemporary cognitive science. Cog nition in the cognitive science arena is not limited to "cold" cognition but strongly embraces motivation, em:)tion, and social interaction. It is important to point out thal cognitive scientists are interdisciplinary in the sense that they adopt the ories and methodologies from multiple discipline ~i : psychol ogy, computer science, linguistics, anthropology discourse processing, education, neuroscience, the list goe ~ ; on. From the standpoint of education, cognitive science has a strong presence in professional groups that are explicit'y interdis ciplinary, with labels such as Learning Science, Cognition and Instruction, Artificial Intelligence in Education, Intelli gent Tutoring Systems, and Computer-Supported Collabo rative Learning. Thus, the research community of relevance here is not limited to the prototypical cognitive psycholo gist who works in a narrow laboratory paradigm on cold cognition. With this context in mind, what would an educa tional psychologist with cognitive science leanings have to
".
oJ
l
194
ORAESSER
say about the Alexander et al. definition and landscape of
HOW USEFUL IS THE ALEXANDER ET AL.
DEFINITION OF LEARNING?
A first
is to whether the nine principles are techadequate for a definition The definition of propose has that Alexander et aL
1. is
2, is inevitable, and Ubiquitous
3, Learning can be resisted
4. may be disadvantageous 5, can be tacit and incidental as well as conscious and intentional is framed our humanness 6. refers to both a process and a product is different at different in time 8. 9, is interactional The purported of this definition is that diverse groups ofeducational researchers allegedly as true, One concern about this definition is that it does not have teeth and may not go the distance in what is versus what is not A classical definition of any C has a set of features that are necessary andjointly sufficient to discriminate C from non-C entities, A definition of a concept has a set of features that are more associated with C than non-C entities. There are also definitions, prototypical definitions, and definitions of concepts (Smith & Medin, 1981), all of which attempt to discriminate C from non-C entities, The problem with the Alexander et aL issue) definition is that is pitched at such an abstract level that it could be applied to many sister constructs in the area, All of these would be true if the word was substituted with any of the and other components of cognition, In this sense, the Alexander et aL definition is abstract and issue) take a giant step in ofAlexander et aL a more discriminating definition oflearning when they tell us what is not. That is, an innate a maturation of a l)lV'lVI';lC,Q,,' or the recall of prior materiaL Here we have some claims with teeth. And as with all interesting there will be members of the community who will the dismissal of the innate capacities, biological/neurological maturation, and mere recall of prior material. For there are models
of neuroplasticity and maturation are bound to (Damasio, 2003; Elman et is essential for V'V'lV''''-''W That uration, and vice versa. However, it is beyond the of this commentary to count the ways that biological/neurological mechanisms are some theories rely mechanisms in their explanations VIO,lliil". & l'UJlW,I."U,
and
ing claims with teeth in the Alexander et aL UvJlUU,LlUll probably debatable in the science community, Aside from the technical adequacy of a we can inquire whether a definition is usefuL The \\,arry is that the Alexander et aL (2009/this issue) definition is an abstract level that it will not invite concrete programs and -"'1'" ."... to compare their nine with the seven
of that were identified in
Instruction and to Improve Student Learning (Pashler et an initiative of the Institute of Education Sciences (IES) of the 1 of Education, A consensus for the community, 'H""\."'~" among researchers in the I, over time. 2. Interleave worked exercises, 3. Combine ntpar~tp abstract and 4.
problem·
con,~rete
rer)re:,enxa·
5, Use
6. 7. motivated and empirical evidence that varied from low to There were also concrete in the how teachers would apply these Evidence-based
a community of researchers affiliated with
at Work and at Home an initiative launched by the Association of Psychological joint initiative between Association Sciences and an of Sciences and the Ameri::an Psychological Association. The 25 included the IES plus an additional 18, such as number Ul"nT",,.,
(3) Dual Code and Multimedia and multimedia form
LANDSCAPES AND THEORlE:l
to
is enbanced when learners them an
irrele
hard or too easy, but at the student's level of skill or
,Iem
ent!!
These nitive foundations of
as Dunlosky, & would be needed to cover other of such as motivation, emotion, dis course, social interaction, personality, development, and neu· roscience. All of the \H.a."~~vl,
and eon
,om. )sed . ex·
are more seductive in inThese are claims with teeth that prompt curious souls to conduct research that have testable to conditions where break down, to debate those who and to the wisdom to educational In contrast, the definitional are truisms that are but hardly stimulus for inquiry.
tion ieal
ned the
HOW USEFUL THE LEARNING
LANDSCAPE?
landscape is a with the dimensions of what, and when. Building a landscape with
195
dimensions is an 'ITln",-t","t on any phenomenon, so it is wise to do so for out, the various Alexander et a1. aptly and educational programs can be of the landscape. The notion of a multidimensional landscape to "WJlVie,'",''' research and practice has a very than 3 decades ago, Jenkins (I model of and memory which four inter acting factors that determined learning outcome~;: cri.terial characteris,ics of the of nature of
the four dimensions were the ques tion the type of knowledge, and the tyVe of cog nitive process. Sometimes there are three dmensions in these as in the case of models that range from the triarchic theory of to the of love 1986). I am unaware of any five-dimensionallCulu~"''''fJ'v0 haps because it is hard to get humans to dimensions. Researchers will on sions to include in any multidimensional issue) a Alexander et al. developmental dimension whereas that absent in Jenkins' tetrahedral modeL The landscape load" a lot of information into the what dimension, whereas the four olf ~_,._,,_._ dimensional processes, and/or criterion measures. The relevant dimensions differ among research communities. The Alexand:r et al. ar to the views of the different ticle has educational communities and ended up on the what-who-where-when solutIOn. Time will tell whether this four-dimensional space ends up being adopted. It is important to the an atheoretical space, not a One ~'C.~"'~" this commentary is to ther. The educational needs to move frJm merely the landscape to what their theo ries predict within the landscape. The theories to offer claims with testable
desert. As an I am somewhat dis vvc,<""",,u that the Jenkins tetrahedral model end:ld up hav such a modest on cognitive research in learning and memory. The tetrahedral model has had an
L
196
GRAESSER
on some researchers from the standpoint of widenthe landscape beyond narrow laboratory and educational However, it is not that model that drives research is driven by more focused theories So this raises a Wby hasn't the landscape driven our research over the last 30 years? There are at least four reasons, as elaborated next.
the particular researcher can hand I,;; and that to the under Some ries are narrow and others very broad. The scope of a terrain in the four-din:ensional trunp breadth and scape. complexity.
a four-dimensional landscape is difficult to ine. Researchers are not prone to follow a metaphor that is
THE VALUE OF THE ALEXANDER AL.
DEFINITION OF LEARNING AND THE
LEARNING LANDSCAPE
difficult to visualize and that strains A dimension is easy to handle and explains why main effects and constructs like g are so in the field. Two dimensions are also as in the case oftwo-way interactions and tradeoffs between variables. Three dimensions push the boundary for most but a interaction can be achieved with thought, time, and effort. four dimensions for everyone and four dimensions is to in both text and researchers tum to a different metaphor when a model is com plex, as in the case of production connectionist dynamical systems, and other models that require simulations.
empirical tests create a combinatorial
Consider the sim case where there is only one variable per dimension and two values per variable. The number of cells in the factorial arrangement would be 24 ] 6. So if there were 20 observa the minimum total size would be 320. How tions per ever, quite clearly, a researcher would want to consider three values per variable, as in tests of curvilinear and contrasts between a treatment and mUltiple ('(ymr,,,rl conditions. In this case, the number ofcells in the factorial ar would be 34 = 81 and the number of observations would be 1 A researcher consider two variables per dimension which would 6561 cells 38 and 131 observations. There are not enough on the planet to run a between-subjects design with 4 variables on 4 dimensions. Tests of a four-dimensional are ifnot
interactions are
'U>,'~~VUf'V
than
Researchers most excited when they test their pet theories that generate testable prin with predictions. The theories can vary in whatever
is
the cats" under some of a comIt is needed to communicate what the field is faculty, politicians, and educaIt is needed when a research I02lorrlap for that want to a research for the next 5 or 10 years. The uncovers salient gaps in the terrain that stimul2te new research initiatives. We can identify when there is a between two research two theories the same turf, but with opposing predictions) versus the more where two research --0-----0 different of the TI,e '''''''"'-''0 assessmg the scope and of a theory or testable These are the that Alexander et contribution has tremendous value. "",\.U,HIS
REFERENCES
to
and Researchers who have tried to pin down these complex interactions are faced with the difficult task of communicating them to fellow researchers. The inter actions do not unless the values of the variable are precisely tuned. At that point the defensible conclusion is that there are sometimes interactions among the four dimensions. This is not new because it has been known for centuries that there are interactions among levels in any system. theories with constraints are more
The central argument in this commentary is that ---~-'~'V''''H researchers are most by theories and testable and pies. Moreover, the most seducti',re principles are those that have teeth. That the predictions are decisive, readily sometimes c ;mnterintuitive, and/or incompatible with the prevailing wisdom of communities. Theories and testable research
Alexander, P. A, Schallert, D, L., & Reynolds, R. E. learning anyway? A topological perspective con;;idere
Damasio, A (2003). Looking/or Spinoza: Joy, sorrow, New York: Harcourt.
de Vega, M" Glenberg, AM" & Graesser, A, C. and embodiment,· Debates on meaning and cognition Oxford, University Press, Elman, J. L., Bates, E. A, Johnson. M. H., Plunkett, K, (J 996). Rethinking innateness A conr on development. Cambridge, MA: MIT Press. Gienberg, A M. (1997). What memory is for. "o,.n"I{,7'" ences, 20, 1-19. Graesser, A c., Halpern, D. E, & Hakel, M. learning. Washington, DC: Taskforce on Lifelong
LANDSCAPES AND THEORlES
1
y
! d
at Home, Retrieved from httpllwwwpsyc,memphis,edu/Jearning/ whatweknow/index,shtml Graesser, A Ozuru, Y, & Sullins, 1. (2009). What is a good question? M McKeown (Ed.), Festscrifi jor Isabel Beck Cpp. 361-382). Mahwah, NJ: Erlbaum. Hacker, D 1, Dunlosky, I, & Graesser, A. C (Eds} (2009), Handbook oj melacognilion in education, New York: Taylor & Francis. Jenkins, 1. 1. (1979). Four points to remembec A tetrahedral model of
memory experiments, In L S. Cermak & F L M. Craik (Eds.), Levels
memory (pp, 429-446). Hillsdale, NJ Erlbaum.
Kintsch, W (1998), Comprehension,' A paradigmJor cognition, Cambridge,
UK: Cambridge University Press, Landauer, McNamara, D. S., Dennis, S., & Kintsch, W (Eds.). (2007), Handbook ojlatent seman lie analysis. Mahwah, NJ: Erlbaum.
197
Pashler, H., Bain, P, Bonge, 8., Graesser, A., Koedinger, K, McDaniel, M" et al. (2007), Organizing instruction and study to improve student learning (NCER 2007-2004). Washington, DC: National Center EducatlOn Research, Institute of Education SCIences, U,S, Department of Education. Retrieved from http://ncer.ed,gov Smith, E. E., & Medin, D, L. (1981). Categories and concepts. MA: Harvard University Press, Snow, C. (2002). Reading Jor understanding' Toward R&D program in reading comprehension. Santa Monica, C.'\: RAND Corporation, Sternberg, R.l (1985), Beyond IQ.· A triarchie theory ojinleilig,mee, Cam bridge, UK: Cambridge University Press, Sternberg, R (1986). A triangular theory of love. Psychological Revie'fl, 93, 119-135,
i
~,
".'