This article was downloaded by: [University of Otago]

On: 25 December 2014, At: 11:27

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Educational Psychologist
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/hedp20

Aptitude Theory: Yesterday, Today, and Tomorrow

Richard E. Snow
Published online: 08 Jun 2010.

To cite this article: Richard E. Snow (1992) Aptitude Theory: Yesterday, Today, and Tomorrow, Educational Psychologist, 27:1,
5-32, DOI: 10.1207/s15326985ep2701_3

To link to this article: http://dx.doi.org/10.1207/s15326985ep2701_3

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the
publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations
or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any
opinions and views expressed in this publication are the opinions and views of the authors, and are not the
views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be
independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses,
actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever
caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 EDUCATIONAL PSYCHOLOGIST, 27(1), 5-32
Copyright o 1992, Lawrence Erlbaum Associates, Inc.

Aptitude Theory:
Yesterday, Today, and Tomorrow

Richard E . Snow
Stanford University

The concept of aptitude is reviewed and reconstructed. Its original sense of

reciprocity between person and situation and appropriateness of person-
Downloaded by [University of Otago] at 11:27 25 December 2014

situation fit is restored. Modern interpretation thus emphasizes readiness to

learn in particular instructional situations and recognizes conative and
affective as well as cognitive sources of aptitude. Limitations of old aptitude
theories are noted. Requirements for new aptitude theories are listed. A new
conceptual language for aptitude theory is suggested by the Thorndike-
Thomson sampling theory, Gibson's (1966, 1979) affordance theory, and
Simon's (1969) artifact theory, in combination with implications from current
research.

Aptitude is an old term for an old concept still widely used, but also widely
misused and misunderstood in much scientific, professional, and public
parlance today. I believe it to be a central concept for educational
psychology (s~odid B. L. Thorndike, although he seems not to have used the
term), and for many other psychologies as well. In educational, industrial,
engineering, clinical, counseling, and school psychology, researclhers and
practitioners alike face problems of aptitude every day, whether or not these
are recognized or labeled as such. Although I concentrate here om educa-
tional issues, it is helpful to keep the larger sphere in mind, because thinking
of aptitude only in education can lead us into local traps.
This article addresses three interrelated questions: What is aptitude?
What form might theories of aptitude take? How can we describe and
understand aptitude in research on learning, teaching, and instruction?
These broad questions go back to the ancient Chinese and the Israelites and
can also be traced through the ages in many European philosophies (Snow,
---
Requests for reprints should be sent to Richard E. Snow, School of Education, Stanford
University, Stanford, CA 94305.
 1982). However, they have been badly distorted, in my view, by English-
language psychology in this century (Snow, 1991b). Thus I first review and
contrast old and new thinking on these questions. I then propose a common
form for conceptualizations of aptitude across this broad front, and define
a domain called aptitude theory within the framework of a person-situation
interactional paradigm. The result is to suggest that a radical departure
from yesterday's styles of theory and research has long been needed and has
been taking shape in recent years; this leads to some projections about the
nature of aptitude theories for tomorrow.

RECONSTRUCTING THE CONCEPT OF APTITUDE

Definitions New and Old

Downloaded by [University of Otago] at 11:27 25 December 2014

Aptitude is a present concern in virtually all goal-directed human activities;

whenever one thinks about the antecedents of observed individual differ-
ences in some valued goal attainment, some concept of aptitude is needed.
The concept is especially close to readiness (as in reading readiness), but
also to suitability (for a purpose or position), susceptibility (to treatment or
to persuasion), and proneness (as in accident proneness). All these concepts
carry the implication of predisposition for differential response by persons
to some situation or class of situations. The common thread through these
and other related terms is potential-a latent, present, inferred quality or
power that makes possible the development, given specified conditions, of
some further quality or power, positive or negative.' The person-situation
interaction implied by this concept is especially to be noted. Aptitudes are
initial states of persons that influence later developments, given specified
conditions. Furthermore, they are initial states that are not merely corre-
lates of learning, but rather are propaedeutic to (i.e., needed as preparation
for) learning in the particular situation at hand. The combination of
aptitude and learning in this situation is then propaedeutic to later learning
in related situations; thus, learning and aptitude development are cumula-
tive.
The term aptitude came to us from Latin via French, and in various early
writings from the Roman Quintilian to the Frenchman Binet the notion of
person-situation reciprocity and adaptation in the development of aptitude
for later accomplishment is clear (Snow, 1982). In origin, aptitude means
"apt, appropriate, suitable," so the definition of some particular aptitude

'Potentials can be negative; that is, there are aptitudes for doing evil (see Scheffler, 1985).
These must be ignored in the present article, although they are also an important topic for
educational psychology.
 APTITUDE THEORY 7

has to be situation dependent. Thus, in French usage it is associated with

apropo. The reciprocity can be between person and person, as well as
person and situation, and even in early English poetic writing this usage is
apparent: consider Milton's (1643) reference to "that sociable and helpful
aptitude . . . between man and woman". Note also that in much early
writing the person characteristics that are apt or apropo are not limited to
the cognitive, but include conative and affective characteristics as well.
Unfortunately, however, in English the concept was gradually equated with
intelligence and capacity through the 17th and 18th centuries. Then it was
generalized in the 19th century through a misinterpretation of Darwinian
theory to mean a biologically fixed, single rank order of general intellectual
fitness for any situation. It was captured in this distorted condition by the
mental testing movement in the 20th century (see Cronbach & Snow, 1977;
Snow, 1991b).
With the growth of mental testing technology in this century, applied
Downloaded by [University of Otago] at 11:27 25 December 2014

psychologists increasingly ignored the substantive roots of the concept.

Aptitude became nothing more than the predictions made from conven-
tional ability tests. General aptitude became synonymous with intelligence.
Scholastic aptitude became synonymous with verbal and quantitative
ability. Beyond this there were special aptitudes defined according to the
conventional distinctions among either school subject matter or vocational
domains. There might be isolated research on aptitude for mechanics, or
art, or computer programming, for example. But these were regarded as
atheoretical and relatively unimportant applied measurement issues. In
yesterday's theoretical writing, the interpretation of aptitude differences
typically relied on one or another kind of entity theory. Aptitudes were
reified as things in the head of the person. They were not things actually -
the old phrenology and faculty psychology had been soundly rejected (by
Thorndike, among others)-but they were the products of things genetic
and physiological, and they were described metaphorically as things in the
head (e.g., mental energy, mental engines, functional unities, instinctive
responses, and stimulus-response bonds) that the person possesses. With
only a few exceptions, questions about aptitude development; the mixture
of cognitive, conative, and affective differences in response to instruction;
and the reciprocal, adaptive relationship between persons and situations
were left largely unexplored.
Thus, through the first half of this century, and still today in some
quarters, British and American psychology has limped along ,with this
narrow and stultifying misconstrual of aptitude. In my view, this enterprise
ignored Quintilian altogether, hung Darwin upside down in the closet, and
then closed most of Binet in there with him.
To be fair, many of yesterday's theorists tried to write carefullly about
their interpretations. Spearman (1927) reviewed the pros and cows of his
 mental energy hypothesis and included conation in his otherwise cognitive
view. Thurstone (1924, 1947) also included conative and affective states in
his thinking, and saw that his functional unities might arise from many
possible sources. Thorndike (1921) thought of intelligence as behavior
appropriate to situations -the collective result of many learned connections
developed in previous situations. He also studied relations among intelli-
gence, achievements, attitudes, and interests. Thorndike's view was least
like an entity theory, and I shall return later to a version of his
connectionism when considering theories for today and tomorrow. Some of
yesterday's applied psychologists, notably Bingham (1937), even held to the
original de,inition of aptitudes as aspects of personality that signify
readiness to profit from particular situations. But the picture of aptitude
most psychologists and educators carried around with them was an entity
theory of a fixed, single rank order, general-purpose cognitive trait called
intelligence.
Downloaded by [University of Otago] at 11:27 25 December 2014

It is strange to note, furthermore, that as the original concept of aptitude

lost its meaning in English, the concept of readiness gained this meaning
(which it did not have in English to begin with). By the 1950s, readiness was
being defined as

preparedness to respond . . . a state or condition of the person that makes it

possible . . . to engage profitably in a given learning activity. . . . it is a
composite of many personal qualities and conditions and differs from one
learning task to another. (English & English, 1958, p. 441)

In particular, readiness for reading was

the totality of personal factors conducive to satisfactory progress in learning

to read under given conditions of instruction. . . . the relevant factors may be
intellectual, emotional and motivational, or physiological. Both general
maturation and effective specific previous experiences play a part. A child
may be ready for one kind of reading method and not for another. (English
& English, 1958, p. 441)

These definitions of readiness hark back to Binet. They clearly identify a

complex of properties, qualities, states, or conditions of persons (not just a
traitlike cognitive unity) that enable profitable learning or development
under specified situational conditions. Different composites are relevant to
different learnihg or developmental situations, and these likely include
conative and affective as well as cognitive properties. A complex reciprocity
between person and situation is expected wherein a person may be ready to
profit from one kind of treatment and not from another aimed at the same
achievement goal.
 APTITUDE THEORY 9

Apparently then, as aptitude became intelligence, readiness became

aptitude. But why should aptitude be defined more narrowly than reading
readiness? If reading readiness is an inferred mixture of cognitive, emo-
tional, motivational, physiological, and experiential properties expected to
differ for different instructional methods, should aptitude for reading mean
something less or something else? Would tests of readiness to learn from
phonics reading instruction not be aptitude tests? Apparently, English-
language psychology at midcentury believed that an aptitude test is merely
an ability test, that one kind of test can be called an aptitude test whereas
another cannot, and that the situation for which a test is supposed to
indicate aptitude need not be specified beyond, for example, a special
aptitude for mechanics.
In short, the term aptitude has fallen into such frequent misuse that
Anastasi (1980) would have it abolished in favor of the term akveloped
ability. In a subset of applications, her proposed substitution might serve
Downloaded by [University of Otago] at 11:27 25 December 2014

well enough. As argued here, however, the broad concept of aptitude

includes conative and affective characteristics of persons, not just cognitive
abilities. Also, the concept is useful in fields of human science where
aptitude or some synonym such as readiness is a concern, but developed
abilities are clearly not. Think of aptitude for a particular kind of
psychotherapy or medical treatment, for example. Even in education, where
aptitude, ability, and achievement are most easily confused (see Snow,
1980), the broader definition is to be preferred; it allows new conceptions of
aptitude for school learning to draw research attention to other personal
properties such as achievement motivations, interests, and attitudes about
self and school and to the mixtures of these properties that connect
individual learners to particular learning tasks in different ways, profitably
or unprofitabtly (Snow, 1986).

Some Boundary Conditions

One might object that this broad, reconstructed concept of aptitude
includes too much -all possible individual differences even remotely related
to learning would need to be considered. The counterargument is threefold.
First, as a public stance, we ought to be persuading the world that it is
wrong to think of human potential in yesterday's narrow way. Any aspect
of a person that can predict his or her response to instruction ought to be
examined as relevant to important personal and instructional goals. Second,
as a research strategy, the concept ought to be broad enough, at least
initially, so as not to miss aspects of the person that may be propaedeutic to
learning in some important situation. There may even need to be as many
kinds of aptitude as there are important situations. Finally, for theoretical
purposes, it may be combinations of person characteristics that most need
 to be understood in relation to these important situations. Human beings
are not lists of independent variables; they are coordinated wholes. Limiting
the initial list arbitrarily risks missing important combinations.
It is thus especially important to represent in the concept of aptitude the
full spectrum of individual differences related to learning. Boundary
conditions are needed in some other directions, however. Two such limits
are noted here. These concern the stability, malleability, or ephemerality of
person and situation characteristics.
Characteristics of persons that count as aptitude ought to be those that
are of lasting concern. Person characteristics that are easily changed ought
not to count as aptitudes. This much is consistent with yesterday's aptitude
theory. But stability is a matter of degree and thus a question for research;
aptitudes ought not to be limited to those assumed to be stable. If some
persons know a particular fact and others do not, the difference may predict
learning. If this prediction is easily removed by teaching the needed fact to
Downloaded by [University of Otago] at 11:27 25 December 2014

all, then initial possession of the fact is ephemeral and thus not an indicant
of aptitude. On the other hand, not knowing or understanding a particular
fact may be an indicant of deep-seated, ingrained beliefs or naive personal
theories that may inhibit new learning over a long period. Such beliefs
ought to be counted as aptitude (or in this case inaptitude) even if with
substantial instructional effort the beliefs can be changed. There is thus a
continuum from stable to malleable to ephemeral, and the line that bounds
the domain of aptitude is necessarily fuzzy. Even when potential inaptitudes
are easily remediated, aptitude research may still be needed to pinpoint
what needs remediation and how this may best be done for each person in
need.
Similarly, characteristics of situations that are ephemeral or superficial
may produce differential learning in the short term. But if the effect is easily
erased by minor changes in the situation, it too is of little lasting concern.
As in the case of person characteristics, aptitude research ought to focus on
the major, relatively stable dimensions along which situations vary or can
be made to vary. Yesterday's aptitude theory concerned only the situational
variations that defined school subject matter or vocational domains.
Today's focus has added many other dimensions of instructional treatment
design.

Research on Aptitude-Treatment Interaction

With this reconstructed concept of aptitude, Cronbach and I set out in the
late 1960s to investigate aptitude-treatment interactions (ATI) in instruc-
tional research. Cronbach (1957) had recognized that yesterday's aptitude
research was limited to demonstrations of test-test and test-criterion
correlations that ignored situational variations, just as experimental re-
 APTITUDE THEORY '1

search on instructional situations had ignored aptitude variations; the two

approaches had to be united in the study of AT1 if a new aptitude theory
was to be reached. I had done some AT1 research of my own (in a cloctoral
dissertation; see Snow, Tiffin, & Seibert, 1965). We thus joined forces to
seek new theories of aptitude and new schemes for adapting instruction to
aptitude differences.
The history and details of AT1 research since then have been reviewed
elsewhere (Cronbach, 1975; Cronbach & Snow, 1977; Snow, 1977b, 1989a),
but the major conclusions we published in 1977 still hold: They can be
summarized briefly in five points: (a) AT1 are ubiquitous in education. (b)
Measures of general ability do indeed reflect an important aspect of
aptitude and show many AT1 but interact especially when one treatment can
be characterized as highly structured, complete, and direct and another can
be characterized as relatively unstructured, incomplete, and indirect. (c)
Measures of specialized abilities show relatively few ATI, but there are
Downloaded by [University of Otago] at 11:27 25 December 2014

notable exceptions. (d) Measures of conative and affective aptitudes enter

an enormous variety of AT1 patterns, including some that identify the same
treatment dimension of structure and completeness that enters cognitive
ATI. (e) The ubiquitous complexity of AT1 makes conventional hypothesis-
testing methodologies inadequate, not only for AT1 research, but for
educational psychology in general.
We had also hoped to demonstrate a form of macroadaptation of
instruction wherein students with differing aptitude patterns (could be
assigned on a weekly or monthly basis to alternative treatments matched to
their prevailing strengths and weaknesses. This would complement the kind
of microadaptation then growing from work on computerized instruction
(e.g., Atkinson, 1972) and now blossoming in research on int~elligent
tutoring. This is still a long-range goal, and there are new demonstrations of
the continuing need for it. Shute (1990a, 1990b) recently reported strong
AT1 results using two versions of a computer-tutoring system in electricity,
each of which is also microadaptive to the minute-by-minute details of
learner performance. Swanson's (1990) study of human tutoring in optics
also showed AT1 between alternative overall tutoring styles, despite tutor
microadaptation in each; her results further suggest that tutors differ in
their ability to use different overall styles. In short, then, both macro- and
microadaptations seem to be needed to reach optimal instruction. And AT1
concepts and procedures seem to be needed in research toward this end.
Unfortunately, through the 1980s, the complexity of the probllem and
failure to heed our methodological conclusions led (I would say tnisled)
many educational psychologists to ignore or reject the AT1 phenomenon.
Critics said there were no ATI, or they were not replicable, or if tl~eywere
it only happened at Stanford University, or they were not practically useful
anyway. Most recently, in a speech heard at the 1990 American Educational
 12 SNOW

Research Association meetings, AT1 was listed as just one of many fads that
educational researchers had lived through and discarded in recent years.
Space does not permit me to address these criticisms here. It should be
clear to anyone who studies the literature, however, that the AT1 phenom-
enon is no fad, even if the theory and methodology used to study it in some
particular era or domain proved inadequate. The just-noted Shute (1990a,
1990b) and Swanson (1990) results taken alone are enough to show that AT1
are easily found and can help in understanding instructional effects more
deeply, when investigators bother to include provision for this sort of
analysis in their work. Furthermore, educational researchers continue to
make good use of ATI-style theory and research in evaluations of large-
scale educational programs (e.g., Corno, 1988; Peterson, 1988). Note also
that in research on alternative psychotherapies, AT1 is now a booming
frontier (Dance & Neufeld, 1988; Shoham-Salomon & Hannah, 1991;
Snow, 1991a). In short, aptitude research via the AT1 paradigm is hardly
Downloaded by [University of Otago] at 11:27 25 December 2014

dead.

Aptitude Constructs New and Old

Moreover, if my reconstructed definition of aptitude is accepted, then it can

be said that many new kinds of aptitude constructs have been developed in
the past decade, although they have not been labeled or evaluated as such.
Table 1 gives a rough sketch of a network of aptitude constructs that might
be candidates for research on, say, high school science instruction, partic-
ularly where treatments are expected to vary in degree of completeness,
directness, and structure. Table 1 is an expansion of one part of a layout of
aptitude, learning, and achievement constructs relevant to Glaser's (1976)
view of instructional theory (see Snow, 1989c, 1990). It is not to be
interpreted as a full classification of aptitudes, but rather as an example of
the sort of provisional list one might sketch out in planning a research
program. At least, it identifies most of the important kinds of student
differences in prior learning and development that may need to be assessed
and examined in relation to new attempts at instructional improvement in
this domain.
Table 1 is a matrix of five rows (for different categories of aptitude
constructs) and three columns (for different degrees of apparent stability).
The aptitude constructs in the left column of Table 1 reflect relatively stable
characteristics of persons that may be difficult (although not impossible) to
alter in the short term. Most are well-known, molar constructs that have
shown AT1 with instructional treatments varying in structure and complete-
ness. Other potential aptitude differences, in the middle and right columns,
may be malleable if they are detected and addressed by instruction. All are
initial states of preparation for learning, but each may carry different
Downloaded by [University of Otago] at 11:27 25 December 2014

TABLE 1
Aptitude Constrllcts That Might Be Studied in Relation to High School Science Learning
Aptitude Constructs
Aptitude
Category Sfable? Malleable? Ephemeral?
Conceptual structures Crystallized intelligence Prior knowledge of physics Believes concept X
Explains concept X
Quantitative ability Computes automatically
Procedural skills Fluid intelligence Selective attention Infers relations
Detects cues
Visualization ability Imagery Graphs relations
Representation of goals Tests hypotheses
Learning strategies Deep versus surface Heuristics, and mnemonics Checks progress
processing
Adaptation of strategy Works alone
Self-regulatory functions Independence-flexibility Action orientation Protects intention
Conformity-responsibility Content motivation Works extra
Motivational orientations Extraversion-ascendancy Investigative interest Explores resources
Need for achievement Mastery orientation Seeks explanation
Test anxiety Self-orientation Worries about failure
 implications for assessment and research. Some "new" aptitude constructs
in the middle column may also be interpreted as constituents of the more
molar constructs. The right column lists even more molecular aspects of
aptitude differences; some of these may identify learning skills or strategies
that can easily be changed, but some may also be seen as constituents or
behavioral manifestations of constructs listed to their left.
In the rows of Table 1, five categories of cognitive and conative
constructs are identified: conceptual structures; procedural skills involved
in learning, thinking, and reasoning; learning styles, strategies, and tactics;
self-regulatory and action control functions; and motivational orientations.
I consider the first two categories cognitive, the last two conative, and the
middle one a mixture. But distinctions within the resulting matrix are rather
arbitrary. Knowledge, skill, strategy, control, and motivation intermingle in
learning, and all have both cognitive and conative aspects. Thus, some
constructs straddle categories, and some could be shifted easily to other
positions. Continuing research would be expected to revise this matrix
Downloaded by [University of Otago] at 11:27 25 December 2014

substantially, collapsing distinctions between some constructs, changing the

positions of some, and adding others.

Aptitude Complexes

Part of today's problem is that we need theories of aptitude complexes, not

just of single constructs (Snow, 1987). There are many hypothetical
connections one might draw between the rows and columns of Table 1. But
today's research tends to work on single constructs in isolation. This is
sometimes a practical necessity, but it is always a severe limit for theoretical
purposes. If we want to understand aptitude for discovery-oriented versus
direct instruction in physics, say, or if we want to improve either kind of
instruction and use aptitude information as part of the evaluation, we have
to see aptitude as multivariate. At least, we have to study the most likely
interrelations in Table 1 and imagine the most likely aptitude complexes.
Again, human learners are not just lists of variables.
There is ample evidence that some ability and personality aptitudes enter
higher order ATI. Ability x Anxiety x Treatment structure is one oft-used
example (see Snow, 1977b, 1987, 198%). But there are many other cognitive
and conative interactions in need of joint study. To illustrate what is
needed, let me criticize my own first AT1 study as an example (Snow et al.,
1965; see also Snow, 1989b). It involved some of the constructs listed in
Table 1, notably prior physics knowledge (PK) and quantitative ability
(QK), but also conformity-responsibility (CR) and extraversion-
ascendancy (EA) as conative aptitudes.
The experiment contrasted live performance versus filmed versions of a
series of physics demonstrations across a semester-long introductory college
 APTITUDE THEORY 15

course. Achievement measures were both immediate and delayed problern-

solving exercises. Performance immediately after demonstrations was su-
perior under live conditions for students high in QK and low in PK, but
superior under film conditions for students low in QK and high in I'K. The
effect persisted on delayed recall measures. Apparently, the live condition
permitted and required the exercise of mathematical ability in underr-
standing the demonstrations, particularly their mathematical formulations,
which were often done incompletely or hurriedly by the live instructor on an
overhead projector. That is, the relatively incomplete and less structured
live demonst~rationsdemanded more complex and inferential mathematical
processing to comprehend, especially if PK was low. In the films, the
mathematics was presented in a complete and simpIer, structured fashion,
well integrated with the physics of the demonstration. This was a particular
benefit for the mathematically weak student with more background in
science, but no benefit for the more mathematically able student who had
Downloaded by [University of Otago] at 11:27 25 December 2014

less background in science.

However, the film-live comparison also interacted with CR and EA
scales from a personality inventory (as well as with measures ireflecting
experience with film learning). Live instruction was best for students high in
EA or low in CR; film instruction was best for students with the opposite
patterns. The results were clear for immediate but not delayed measures.
The interpretation was that the film condition had an adverse effect on
learning in some individuals because it removed the active "interpersonal"
character of instructor demonstrations and fostered a passive "s]pectator"
state of mind. Ascendant, self-assertive, more socially dominant lpersonal-
ities were frustrated by the private nature of the film learning tasks vvhereas
those described as relatively irresponsible were allowed to escape mentally
from learning in them. The effect did not show on delayed ]measures
because the intervening instruction allowed those who did not get the film
messages to recoup their temporary losses.
In sum, live learning tasks were best for students described as high in
mathematical ability but low in prior physics knowledge; or as ascendant,
assertive, self-assured, active, and independent; or as flighty, irresponsible,
and unable to stick to tasks that do not interest them; or as unlikely to seek
film learning experience on their own. Film learning tasks were best fctr
students described as low in mathematical ability but high in prior physics
knowledge; or as passive observers lacking in self confidence; or as
responsible and conscientious; or as likely to seek film learning experience
on their own.
Now, this study was strong in important respects: It used a substantial1
amount of real instruction, treatments were clearly defined, measures were
of good quality, and the student sample was large. But the methodological
criticism is obvious (at least it is today)-aptitude complexes shodd have
 been studied by considering all the individual variables jointly. There might
well be moderating relations among PK, QK, CR, and EA aptitudes. There
should also be patterns of relations among these variables that define
profiles or categories of persons well served by one or another treatment.
These are defined in a multivariate aptitude space, but they are categories of
whole persons, not lists of interacting variables. Furthermore, in hindsight,
a much deeper interpretation might have been gained. If I were doing this
research today, I would next interview successful and unsuccessful students
from these groups about their learning activities and collect think-aloud
protocols of their work on the physics problems to reach a richer descrip-
tion of psychological processes in each kind of person-situation
c~mbination.~
Thus, AT1 experiments can be used to dissect functionally different
aspects of aptitude operating jointly in learning from instruction as well as
functionally different aspects of the instructional situation. But they need to
be aimed at understanding the kinds of persons (as aptitude complexes or
Downloaded by [University of Otago] at 11:27 25 December 2014

profiles) that are served well or poorly by an instructional treatment, not

merely at discovering the list of individual variables that interact. Also, one
need not think of using this result only as a decision rule for assigning
students to different treatments. One could also use it to redesign either
treatment or to make a combined treatment with the aim of eliminating
ATI, as Bunderson (1969) argued long ago. Either way, the result is both
practically and theoretically useful.

BUILDING A FRAMEWORK FOR APTITUDE

THEORIES

Let us move back now to consider in the abstract what aptitude theories are
today and should be tomorrow. The question is, What kind of framework
for aptitude theories should we construct and pursue?

The Place of Aptitude Theory in Psychology

Aptitude theory should be thought of as a linking science, aimed at

descriptive and explanatory concepts that connect the characteristics and
capabilities of persons to features of treatment environments, real or

'There are important connections to be built between AT1 research and the methods used in
cognitive analyses of expert and novice performance. Experiments with task conditions chosen
to characterize expert-novice performance differences are formally equivalent to within-
person AT1 designs; the expert-novice contrast is the aptitude variable. Unfortunately, little
work that integrates the two approaches has yet been done.
 APTITUDE THEORY 17

desired, so as to reach goals of field achievement. Any purposefully

designed instructional program, therapy, or person-machine system is
implicitly a theory of how a goal can be reached given existing or expected
conditions, both personal and environmental. The usage here is akin to
Simon's (1969) vision of design science as a link between abstract general
principles and the achievement of real-world goals. However, in psychology
to date and particularly in instructional psychology, most attempts to forge
such links have been built on the strong simplifying assumption that
individual differences could be ignored. Research on aptitude can thus be
fin>-1bw3-sketched into Smon,S picture o f an instructibniil design
science. The connection of aptitude constructs to constructs of field
achievements has been largely empirical and atheoretical. The connection
with treatment design has been virtually nonexistent. Still today, most
instructional designs, including intelligent tutors, routinely assume that
learners come in as blank slates or that they are relatively uniform in abilif,y
Downloaded by [University of Otago] at 11:27 25 December 2014

and prior knowledge. The connection of aptitude constructs to generd

psychological! theory has also remained vague and unanalyzed, for the most
part.
But aptitude theories could be aimed at linking differential constructs
into learning, achievement, and instructional design theories (Snow,
1991b). As already argued, they could also be particularized to the
important field achievements and treatment designs of concern in a specific
situation.

Constituents of an Aptitude Theory

What are the constituents of an aptitude theory in such a situation? Some

general features of such a theory are listed in Table 2, even though we do

TABLE 2
Constituents of an Aptitude Theory
1. Convergent and discriminant validity
2. Predictive validity
3. Differential validity
4. Process explanation
5. Other boundary conditions
6. Short-term malleability
7. Long-term development
8. Measurement model
9. Methodological model
10. Selection decision rules
11. Classification decision rules
12. Education decision rules
 not yet have any theories in this full form today. Here I identify these points
briefly in order to devote the remainder of this article to Point 4 in Table 2,
which I think is the most important for theory-oriented aptitude research
today and tomorrow.
Points 1 and 2 are obvious. Any aptitude theory must include convergent
and discriminant validation of its constructs and measures. To be consid-
ered aptitude, there must also be evidence of prediction to some valued
criterion. Yesterday's aptitude research pretty much stopped with these first
points.
Point 3 adds differential validation with respect to varying situations; this
is the basic AT1 view. Because the concept of aptitude always implies
prediction, it always implies some particular educational (or job or thera-
peutic) treatment situation wherein persons will function. Describing the
situation is thus part of defining the aptitude. Unless the aptitude in
question is truly general, there must be situations in.which that aptitude is
less propaedeutic to success or not propaedeutic at all. Thorough differen-
Downloaded by [University of Otago] at 11:27 25 December 2014

tial validation research is required to set substantive boundary conditions

on the aptitude construct.
Point 4 is process description of the person-situation interface. De-
scribing an AT1 pattern as an aptitude complex for learning from discovery-
oriented or direct instruction in introductory physics is a step in the right
direction. But person and situation description has to be carried to much
further detail to reach a process explanation of the propaedeutic relations
involved.
Point 5 concerns the need to identify other boundary conditions beyond
those explicit in the primary AT1 demonstration of Point 3. There will be
boundaries on the ranges of aptitude variables and other person character-
istics in the populations of interest. Also to be considered are boundaries
around the school and community contexts within which the research is
situated. All of the limits that make a particular aptitude theory a local
theory (Snow, 1977a) need to be spelled out eventually, although many will
not be investigated empirically.
Points 6 and 7 address the malleability and development issues. Deter-
mining the extent to which the aptitude differences in question can be
changed by direct intervention, and the nature of that intervention, adds to
the interpretation of the aptitude construct. Tracking the natural courses of
aptitude development and differentiation across a relevant span of years
similarly contributes meaning.
Points 8 and 9 justify the choices of measurement and methodological
models that have gone into the development of an aptitude theory. There
are alternative choices to be entertained, which might appreciably change
the form of the theory or continuing research on it.
Beyond these descriptive constituents of an aptitude theory, one would
 expect one or more prescriptive constituents eventually. Points 10-12 thus
suggest the development of decision rules for using the aptitude th~eoryin
practical selection, classification, or education systems.

REINTERPRETING APTITUDES AS INTERACTIONAL.

PROCESSES

TIk for New Language

L~etus return now to Point 4-the process description of the person-

situation interface. This is the step on which today's aptitude research has
been stumbling. Although there has been progress in several directions
(Snow, 1989a; Snow & Lohman, 1984; Tobias, 1989), there also has
emerged a conceptual problem that I believe limits theoretical adlvance in
Downloaded by [University of Otago] at 11:27 25 December 2014

psychology in general, not just in aptitude research. Simply put, the

conceptual limitation derives from our tendency to think of persons and
situations as independent variables, rather than as persons-in-situations as
integrated systems. To build the aptitude theories of tomorrow, we need a
language for describing the interactional processes that connect persons and
situations-the processes that operate in their interface. Put another way,
we need a language for aptitude constructs that is common to concepts of
individual differences in learning, achievements from learning, and imstruc-
tional treatment designs, locating aptitude in the nexus between them.
It is noteworthy that general cognitive psychology has come to recognize
a similar problem. The implication from an increasing number of cognitive,
developmental, cross-cultural, and anthropological studies is that situa-
tional specificity is an important feature of cognitive skill. To cite just one
early collection of these views, Rogoff (1984) noted that

Cognitive skills seem to fluctuate as a function of the situation, which suggests

that skills are limited in their generality. . . .
Thinking is intricately interwoven with the context of the problem to be
solved. The context includes the problem's physical and conceptual structure
as well as the purpose of the activity and the social milieu in which it is
embedded. . . .
One must attend to the content and the context of intellectual activity in order
to understand thought processes. This is the case for any situation in which
thinking is studied, including the laboratory context, which is not context-free
as researchers frequently assume. (pp. 1-3)

Thus, in the line of research that has developed from this view, thinking
skills reside in the person-situation interaction, not solely in the mind of the
 20 SNOW

person. This is a radical departure for general cognitive theory; it seems to

be pushed by these context- and culture-sensitive studies toward the same
conception of cognitive skill that comes from the push of AT1 research. The
aim appears to be the same: to create a new conceptual language for
theories of person-situation interactions. However, the problem for apti-
tude theory is even more difficult because the language we build must cover
conative and affective as well as cognitive interactional processes, and
individual differences in them besides.
There is a candidate theoretical language to propose for tomorrow's
aptitude research, although today it may appear to be a patchwork of
several different languages. That is reasonable enough in the interim, if
different languages have particular strengths for different parts of the
problem. The goal for tomorrow's theory, then, will be to replace today's
quilt with seamless fabric.
Specifically, the proposed language uses some of Gibson's (1966, 1979)
Downloaded by [University of Otago] at 11:27 25 December 2014

theory of affordances in perception and some of Simon's (1969) science of

the artificial in treatment design, as well as language from cognitive
information-processing research on comprehension (Kintsch, 1988) and
reasoning (Sternberg, 1977, 1985). But I also reach back to Thorndike's
(1921) connectionism and its use by Thomson (1919, 1939) to formulate a
response sampling account of cognitive task intercorrelations; the modern
version of this comes through Humphreys (1979, 1981). Perhaps it even
reaches ahead to the new connectionism of Rumelhart, McClelland, and the
PDP Research Group (1986). There are connections as well to current
thinking about situated learning and distributed intelligence (Greeno, 1988;
Pea, 1990; Perkins, 1990; Salomon, 1990) and also to certain aspects of
modern work on phenomenological perspectives (Marton, 1983), person-
ality theory (Hettema, 1979, 1989; Mischel, 1984), and ability factor theory
(Carroll, 1989; Gustafsson, 1984, 1989). There are also substantial meth-
odological implications (see Snow, 1991b).
There is no space here to explicate all of these views separately or to detail
the interconnections that led me to cite these sources in particular.3 Rather,
to reach a concise summary, it is easiest to trace the development of this
view in our Stanford Aptitude Research Project, starting in 1975.

Strategic Assembly and Control Processes

My colleagues and I (for a list of project participants, see Snow, 1989a) had
accumulated by the mid-1970s many AT1 results that cried out for detailed
analysis. We thus began information-processing studies of key aptitude
differences, hoping to reach deeper interpretations. We concentrated on

3This is one subject of a book in preparation entitled Aptitude Theory.

 cognitive differences despite important AT1 results on personality because
it seemed the simplest way to start. However, we emphasized somie
hypotheses and goals that differed from those pursued by other cognitive
investigators, in part to leave room for these other processes, and we have
tried to bring conation and affect explicitly into the picture in the last 5
years.
First, we expected that the most important individual differences in
cognitive task performance in relation to learning would show up in
within-person as well as between-person strategic differences in information
processing. In other words, there might be individual differences in
elementary processes, as Hunt (1980) suggested, or in the performance
components identified by Sternberg (1985). Persons might also differ in the
sequence in which they executed component steps or in the kinds of
component steps they included- we labeled these sequence and route
differences. But most important, we thought, would be a person's facility in
Downloaded by [University of Otago] at 11:27 25 December 2014

assembling component sequences and routes into a strategy for task

performance and in reassembling or adapting this assembly as task perfor-
mance proceeded. Intraindividual differences in these functions would
show adaptation (i.e., learning) within cognitive performance tasks (e.g.,
ability tests) that should be the key to understanding similar differences in
learning tasks. We labeled these adaptive functions assembly and' control
processes (for details, see Snow, 1978, 1981; Snow & Lohrnan, 19134). Our
results from eye movement analyses as well as Sternberg-style componential
analyses seemed to bear out the promise of this view. In short, our
continuing hypothesis is that aptitude differences in learning appear in the
person-task interface as differences in within-person adaptation to the
stream of continuing changes in within-task demands and opportunities.
Learners construct their performances in instructional situations by
drawing on their resources and assembling, reassembling, and controlling
them to adapt to perceived needs and opportunities in the situation. We alsa
think there are thresholds of task novelty and complexity for each
person-task interface near which learning is optimal and flexible assembly
and control functions are most needed. No single information-processing
model can account for performance near this threshold, even if such models
are adequate to characterize the automatic performance observed when
persons are working far below their thresholds.
Second, our goal was to integrate an information-processing account of
individual differences in aptitude for learning with the correlationall evi-
dence amassed over this century in factor analytic study of cognitive ability
organization. Most of the other investigators were studying one ability at a
time. We produced multidimensional scalings of old ability matrices as well
as our own new ones because we wanted to represent the multivariate
interrelations among ability tests in a way that provided a more direct. map
 22 SNOW

of cognitive interrelations and yet was also consistent with the old factor
analytic results (see Snow, Kyllonen, & Marshalek, 1984). With this sort of
map in hand we could study the degree to which information-processing
models of single tasks could be stitched into continua of tasks in the
multivariate space. After all, an information-processing theory of partic-
ular abilities has to depict the similarities and differences between tasks, just
as the correlations between tests do. An information-processing theory of
cognitive abilities has to be consistent with the correlational structure of
ability tests in general. This would also be true for a theory of personality
differences in learning, and we are currently at work on similar multivariate
maps in this domain.
At least for cognitive abilities, we believe we succeeded in demonstrating
that a Radex structure provides a fitting and useful map (see Snow et al.,
1984). It distinguishes the more complex and general ability constructs in
the center of the Radex from the simpler, more specialized and domain-
specific abilities distributed around the periphery and suggests that the
Downloaded by [University of Otago] at 11:27 25 December 2014

arrays of the Radex represent continua of increasing information-

processing complexity as one moves from periphery to center. The central
ability constructs show much stronger correlations with learning from
instruction, and more potent and reliable AT1 results, than do the periph-
eral abilities. Information-processing analyses also suggest that they involve
more variance due to adaptive, strategic assembly, and control processes. In
other words, cognitive aptitude differences in the center of the Radex are
differences in within-task flexible adaptation of processing, at least in
significant part.

A Sampling-Assembly-Control Model of
Person-situation Interaction

Putting these views and results together with evidence from many other
investigators as it accumulated, we reached a way of describing both
intraindividual adaptation and interindividual differences, both within and
across situations, that seemed flexible enough to capture the dynamics of
processing in person-situation interactions. An early version was called
simply a response sampling model to honor the Thorndike-Thomson
origin, although the whole is better described as a sampIing-assembly-
control model, and the sampling is assumed to be bidirectional (see Snow &
Lohman, 1984).
The basic event at the interface of person and situation is a sampling, of
person by situation and situation by person, governed by associative
networks of stimulus and response components residing in the inner
environment of the person and the outer environment of the situation.
 Although based on associative networks of components, the model is by no
means limited to a stimulus-response association theory. Rather, the
associative network provides the base from which a variety of structures
and representations can be assembled as needed in a particular
person-situation match. The neutral term component is used to cover
stimulus-response bonds but also other kinds of hypothesized mental units
or connections, including information-processing components, plans, im-
ages, learning sets, schemata, nodes in semantic networks, productions in
production systems, and the like. The model is thus not restricted to any one
cognitive representational construct and accommodates nonrepresenta-
tional constructs as well. Furthermore, although components are conve-
niently described in the model as bits and pieces of knowledge (and skill),
the term component can apply as well to key aspects of conative and
affective aptitudes.
First consider the inner environment of the person. Each individual3
Downloaded by [University of Otago] at 11:27 25 December 2014

inner environment contains a vast assortment of response components.

They are probabilistically interconnected in multiple associative n~etworks;
that is, the connections vary in strength to reflect the person's learning
history. Many sorts of assemblies of these components can be constructed
in different ways for different situations. These assemblies are also clecom-
posable, so pitrts can be used in other assemblies as needed. The products
of past learning are components already assembled into units to be triggered
anew by situations similar to those previously faced. The products of
continuing learning are additional components, new assemblies of both new
and old components, and strengthened connections between them. But
learning also exercises and thus strengthens the assembly and control
functions themselves. In short, the human mental system is designed to be
loosely coupled and flexible in assembling and reassembling components
into performance programs to meet varying situational needs. Because ilt
reflects personal learning history in this regard, it is also highly idiosyn-
cratic.
Now consider the person-situation interface. Each performance situatio~~l
samples from each person, in the sense that the demands and opportunities
it presents draw forth whatever relevant response components and assem-.
blies of components each person can muster. But the person also samples
the situation, in the sense that stimulus components are perceived and
selected. Stimulus components may suggest a demand for particular
response components or assemblies. They may also provide an opportunity
to use particular response components or assemblies. And the situation may
contribute components to the performance that the person then need not
provide. Of course, each person's learning history will influence this
perception-selection process to some degree. Thus, the sampling is designed
 in part by the demands and opportunities afforded by the performance
situation presented and in part by the possibilities and constraints afforded
by the assembly and control history of the performing person.
Considering the person's learning history suggests some further distinc-
tions between several kinds of situational demands and opportunities.
There are those involving the retrieval and application of old, familiar
component assemblies versus those involving the construction and applica-
tion of novel component assemblies. Also, there are those that call forth
familiar or novel performance assemblies from the learner versus those that
supplant the need for such performance assemblies, by providing stimulus
components as prosthetics that can substitute for response components.
Complex tasks will typically involve some of each kind of component and
will further require the flexible reassembly of interconnections within and
between them as various parts of the learning task proceed. But all these
situational components are there to be perceived and used as such, at least
Downloaded by [University of Otago] at 11:27 25 December 2014

for persons who are tuned to do so.

Thus, situations can be described as consisting of networks of stimulus
components that (a) represent either demands for or opportunities to use
particular response components or assemblies, but (b) may also supplant
the need for particular response components or assemblies, and (c) may be
either familiar or novel with respect to the person's learning history. Note
that these essential aspects of situations are defined by their connections
with aspects of person performance, just as the essential features of person
performance were defined earlier by their connections to situations. We can
now simplify the language, and bring out some other implications, by
recognizing that all these person-situation connections are affordances in
Gibson's (1966, 1979) sense of that term and also artifactsin Simon's (1969)
sense of that term.

Affordances, Artifacts, and Aptitude Differences

Gibson's concept of affordances addresses the mutuality of person and

situation in the control of perception-action sequences. To paraphrase
Gibson (1979, pp. 127-129, 138-139), the affordances of a situation are
what it offers the person, what it provides or furnishes, for good or ill. The
term implies a complementarity of person and situation, as in an ecological
niche. A niche is a place or setting that is appropriate for a person-a
combination of situational components into which the person "fits."
Likewise, a situation is an assembly of affordances with respect to some
particular person or kind of person. Affordances reflect the invitation,
demand, or opportunity structure of a situation for those persons who are
tuned or prepared to perceive it. Particular affordances invite particular
actions. Gibson's concept of affordances is thus at many points close to the
 old meanings and roots of aptitude previously noted: Milton's (1643)
"sociable and helpful aptitude between man and woman" becomes Gibson's
(1979) "what the male affords the female is reciprocal to what the female
affords the male" (p. 135).
This implication of aptitude in Gibson's theory has also been elaborated
by others in the movement now called ecological realism.

To say that affordances are perceived means that information specifying the
affordances is available in the stimulation and can be detected by a properly
attuned perceptual system. To detect affordance is, quite simply, to detect
meaning.
Different animals engage in very different behaviors. The potential purposive
behaviors are called its effectivities. . . . Whether an animal flies, swims,
walks, or slithers; whether it pecks, nibbles, sucks, or licks; whether it
smokes, watches television, or mugs old people will "determine" the
Downloaded by [University of Otago] at 11:27 25 December 2014

affordances it can detect. Because information specifies behaviors that are

afforded and because different animals have different sets of effectivities,
affordances belong to animal-environment systems and nothing less. [Tlhus]
information about affordances is "personal"; it is unique to particular
animal-environment units. (Michaels & Carello, 1981, pp. 42-43)

Translating this view of ecological realism into AT1 terns, research on

aptitude requires a detailed analysis of the affordance-effectivity matches
of different learners and different instructional treatments. This analysis
would emphasize the opportunities offered by a particular treatment to be
detected and capitalized on by a particular person to achieve a goal. The
analysis also would remain at a level that identifies the unique
person-situation synergy in local ecological terms, rather than reducing to1
physical or biological description or abstracting to generalized principles.
Because ecological information is personal, it is unique to particular
person-situation units. There is therefore no detached or abstracted List of
qualities of instructional treatments that will be equally important for all
persons or similar list of qualities of persons that will be equally important
for all treatments. Aptitude is the unique coalition of affordances and
effectivities in particular person-treatment systems.
The analysis of aptitudes as affordances emphasizes the important ways
in which person and situation are tuned to one another to be in harmony for
successful performance. An equally important question for aptitude tlheory
is the analysis of inaptitudes, that is, the disharmonies in the
person-situation interface that result in failure. Some aspects of these
disharmonies can be described as failures of tuning to perceive affordances.
Other aspects seem better described in Simon's language of artifacts and
interface redesign.
 To paraphrase Simon (1969, pp. 7-13), artifacts are interfaces between
inner and outer environments. If these inner and outer environments are
appropriate to one another (i.e., they are adapted or designed to fit one
another optimally), then the artifact serves its purpose essentially unno-
ticed. Often, however, interface design can be only approximate. Then the
limiting properties of the inner system will appear in the failure to match the
demands of the taxing outer environment.
For a particular person in a particular environment, the empirical
evidence of aptitude arises from the inabilities of the behavioral system to
adapt perfectly to its environment. Aptitude differences between persons in
particular environments show through at the interface as a presence-
absence (or degree) of these inabilities. For a person who is perfectly suited
to a treatment or a treatment that is perfectly suited to a person, the goal is
reached successfully. The presence of aptitude is inferred from this success,
but it is attributable to both person and environment, that is, to their benign
Downloaded by [University of Otago] at 11:27 25 December 2014

interface. For a person who is not perfectly adapted to a treatment or a

treatment that is not perfectly adapted to a person, the goal is not
successfully reached. This failure shows that inaptitude of some kind is
present. But again, inaptitude is attributabIe to the interface: Either the
inner system or the outer system, or both, need to be redesigned to bring
them into adaptive harmony. Research aimed at system redesign needs to
find the key inabilities in the interface that constitute the mismatch and
correct them. System redesign proceeds by reshaping the treatment to
eliminate demands, thereby circumventing limitations, or by removing
limitations directly by retraining the person.

SOME SUMMARY IMPLICATIONS FOR THE

APTITUDE THEORIES OF TOMORROW

In conclusion, at least the main implications for the aptitude theories of

tomorrow can be summed up as follows:

1. Aptitudes are affordances (Gibson, 1966, 1979). They are properties

of the union of person and environment that exhibit the opportunity
structure of a situation and the effectivity structure of the person in fitting
that situation, that is, in taking advantage of the opportunities afforded for
learning. Particular persons are tuned or prepared to perceive particular
affordances in a situation that invite the particular actions they are able to
assemble.
2. Inaptitudes are also artifacts (Simon, 1969). An aptitude is an
interface between an inner environment (the person) and an outer environ-
ment (the instructional treatment situation). Aptitude differences are
 invisible when inner and outer environments are perfectly adapted to one
another. When the outer environment is demanding, however, limiting
properties of the inner environment show through at the interface as
aptitude differences. Instructional treatment redesign seeks to circumvent
these inner limiting properties (inaptitudes) by adapting the outer environ-
ment or by changing the inner environment (removing the inaptiitudes by
direct training).
3. From the view of Simon's (1969) artifact design, future research on
aptitude requires a detailed analysis of the treatment design features that
seem mismatched to the person when limiting properties of the person show
through in the performance interface. The analysis is geared to detect
inaptitudes (weaknesses) so as to remove or circumvent them in treatment
redesign. From the view of Gibson's (1966, 1979) affordance theory, future
research on aptitude requires a detailed analysis of the affordance-
effectivity matches in different person-treatment unions. The analysis is
Downloaded by [University of Otago] at 11:27 25 December 2014

geared to detect aptitudes (strengths) so as to capitalize on them in

treatment redesign. The two views are complementary because the most
successful instructional treatments will be those that both capitalize on
strengths and compensate for weaknesses, for each individual to be treated.
4. Aptitude is situated. It is reflected in the tuning of particular persons
to the particular demands and opportunities of a situation. It therefore
resides in the union of person in situation, not in the mind alone. It is a
two-way sampling of performance components and their assembly between
person and situation and is thus also distributed between person and
situation; the situation contains some pieces of what the person needs or can
use to accomplish a given task. But individuals need to be tuned to perceive
and use those pieces and need also to supply some pieces from their own
learning history. Some individuals are prepared to perceive these
affordances-to use the pieces provided by the situation-but others are
not. Among those who are so tuned, each may supply slightly different
pieces, although each piece thus supplied may be equally effective. The
result is that some persons succeed in learning in a given situation; they are
in harmony with it. Others do not, because they are not tuned to use what
the situation affords or to produce what it demands. The tuning is at once
cognitive, conative, and affective; learners who are not tuned in vvill tune
out, and ultimately turn off.
5. Persons assemble their performances in response to these perceived
affordances from vast banks of potential response components org'anizedl
into associative networks. As a function of learning history, parts of' these:
networks may be coupled tightly and triggered as units; others may be
coupled loosely and disconnected easily when parts are triggered. Actually,
each component might itself be thought of as a network, also coupled either
tightly or loosely. Continuing this reduction might eventually reach the
 neural networks of new connectionism. But that would go beyond the ken
here.
6. Persons also control or adapt these component assemblies as
affordances change in a dynamic situation. In a heterogeneous group of
persons, some components and assemblies will be held in common and
some will not. The connections among components will differ in strength.
The assembly and control history of these components and assemblies will
also differ from person to person and therefore so will their facility for
adaptive assembly and control during performance in a present situation.
7. In this view, valid aptitude measures (whether maximum performance
tasks or typical response tasks) are situations that evoke some semblance of
the sample of components and assemblies and their adaptations that are
also evoked by some learning or achievement task to which the aptitude
measures are therefore correlated. The test situation and the criterion
situation involve affordances that are invariant across these situations; that
Downloaded by [University of Otago] at 11:27 25 December 2014

is, the samples drawn by the two situations overlap substantially. The size
of the correlation suggests how large is the overlap.
8. Aptitude-treatment interaction occurs because the affordance profiles
of person-situation interfaces differ. To continue the earlier example, a
relatively incomplete and unstructured situation samples just the kinds of
assemblies that persons described as able, independent, mastery oriented,
flexible, and the like are tuned to produce. Highly structured and complete
instruction does not provide that opportunity for them. It does, however,
provide some of the assemblies that less able, less independent, less
mastery-oriented learners cannot provide for themselves, and it does not
sample what such learners cannot produce.
9. The three major cognitive aptitude constructs can be distinguished in
this assembly and control process. Fluid intelligence reflects the more
flexible assembly and adaptation of strategies for performance in novel
unfamiliar tasks. Crystallized intelligence reflects more the retrieval and
adaptation of old assemblies for familiar tasks. Visualization ability reflects
a collection of specialized skills that pop in and out of relevance in a variety
of tasks that afford their use; other special abilities can be similarly
described.
10. Finally, one can think of a performance assembly pathway, from
activation in and retrieval from the person's bank of experience, to
adaptation in the person-situation interface, to action in the task or
instructional situation. Performance is assembled and reassembled along
this path to meet the characteristic affordance profile of this situation. An
analysis of this profile with respect to familiarity-novelty, structure com-
pleteness, and the use of special knowledge and skills will provide a picture
of its cognitive aptitude requirements and opportunities for each person.
However, each person's mental bank contains not only bits and pieces of
 knowledge and skill, but also wishes, wants, needs, intentions, interests,
attitudes, etc. These are also component networks that may differ in
strength and that can be triggered in whole or in part by situational
affordances. In parallel with the performance assembly pathway, one can
think of a performance commitment pathway, from activation to action,
that accounts for the appearance of conative and affective aptitude
differences. The operation of such aptitude constructs as achievement
motivation, anxiety, mastery versus performance orientation, conformity-
responsibility, independence-flexibility, and extraversion-ascendancy may
also be describable in terms of the assembly and control of sampling
processes and the tuning of these processes to affordances in the
person-situation interface (see Snow, 1989b), The job for tonncrrow's
aptitude theorists will be to convert each of these constructs, along with the
cognitive constructs, into more detailed affordance sampling descrilptions in
the treatment situations with which they have been associated m i AT1
Downloaded by [University of Otago] at 11:27 25 December 2014

research.

ACKNOWLEDGMENTS

This article represents my address in receipt of the E. L. Thorndike Award

for Distinguished Psychological Contributions to Education, 1990. It was
delivered at the annual meeting of the American Psychological Association,
Boston, August 12, 1990. It is also an update of my APA Division 15
Presidential Address for 1982, which was never published. For these
honors, I hereby thank APA Division 15 and also the many students and
colleagues who have contributed to the Aptitude Research Project over the
years.

REFERENCES

Anastasi, A. (1980). Abilities and the measurement of achievement. In W. B. Schrader (Ed.),

Measuring achievement: Progress over a decade (pv. 1-10). San Francisco: Jossev-Bass.
Atkinson, R. C. (1972). Ingredients for a theory ofihstruction. American ~sychol&ist,27,
921-93 1.
Bingham, W. V. (1937). Aptitudes and aptitude testing. New York: Harper & Brothers.
Bunderson, C. V. (1969, February). Ability by treatment interactions in designing inssbruction
for a hierarchical task. Paper presented at the meeting of the American Educational
Research Association, Los Angeles.
Carroll, J. B. (1989). Factor analysis since Spearman: Where do we stand? What do we know?
In R. Kanfer, P. L. Ackerman, & R. Cudek (Eds.), Abilities, motivation, and methodology
(pp. 43-67). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Corno, L. (1988). More lessons from aptitude-treatment interaction theory. Educational
Psychologist, 23, 353-356.
 Cronbach, L. J. (1957). The two disciplines of scientific psychology. American Psychologist,
12, 671-684.
Cronbach, L. J. (1975). Beyond the two disciplines of scientific psychology. American
Psychologist, 30, 116-127.
Cronbach, L. J., &Snow, R. E. (1977). Aptitudes and instructional methods: A handbookfor
research on interactions. New York: Irvington.
Dance, K. A., & Neufeld, R. W. J. (1988). Aptitude-treatment interaction research in the
clinical setting: A review of attempts to dispel the "patient uniformity" myth. Psychological
Bulletin, 104, 192-21 3.
English, H. B., & English, A. C. (1958). A comprehensive dictionary of psychological and
psychoanalytical terms. New York: Longmans Green.
Gibson, J. J. (1966). Thesenses considered asperceptualsystems. Boston: Houghton Mifflin.
Gibson, J. J. (1979). The ecological approach to visualperception. Boston: Houghton Mifflin.
Glaser, R. (1976). Components of a psychology of instruction: Toward a science of design.
Review of Educational Research, 46, 1-24.
Greeno, J. G. (1988). Situations, mental models, and generative knowledge. In D. Klahr & K.
Kotovsky (Eds.), Complex information processing: The impact of Herbert A. Simon (pp.
285-318). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Downloaded by [University of Otago] at 11:27 25 December 2014

Gustafsson, J. E. (1984). A unifying model for the structure of intellectual abilities.

Intelligence, 8, 179-203.
Gustafsson, J. E. (1989). Broad and narrow abilities in research on learning and instruction.
In R. Kanfer, P. L. Ackerman, & R. Cudeck (Eds.), Abilities, motivation, and method-
ology (pp. 203-207). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Hettema, P. J. (1979). Personality and adaptation. Amsterdam, The Netherlands: North-
Holland.
Hettema, P. J. (Ed.). (1989). Personality and environment. Chichester, England: Wiley.
Humphreys, L. G. (1979). The construct of general intelligence. Intelligence, 3, 105-120.
Humphreys, L. G. (1981). The primary mental ability. In M. P. Friedman, J. P. Das, & N.
O'Connor (Eds.), Intelligence and learning (pp. 87-102). New York: Plenum.
Hunt, E. (1980). The foundations of verbal comprehension. In R. E. Snow, P. A. Federico,
& W. E. Montague (Eds.), Aptitude, learning, and instruction: Vol. I . Cognitiveprocess
analyses of aptitude @p. 87-104). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Kintsch, W. (1988). The role of knowledge in discourse comprehension: A construction-
integration model. Psychological Review, 95, 163-182.
Marton, F . (1983). Beyond individual differences. Educational Psychology, 3, 291-305.
Michaels, C. F., & Carello, C. (1981). Direct perception. New York: Appleton-Century-
Crofts.
Milton, J. (1643). The doctrine and discipline of divorce (Book I, Chapter IV). London:
Printed by T.P. & M.S.
Mischel, W. (1984). Convergences and challenges in the search for consistency. American
Psychologist, 39, 35 1-364.
Pea, R. D. (1990, April). Distributed intelligence and education. Paper presented at the
meeting of the American Educational Research Association, Boston.
Perkins, D. N. (1990, April). Person plus: A distributed view of thinking and learning. Paper
presented at the meeting of the American Educational Research Association, Boston.
Peterson, P. L. (1988). Selecting students and services for compensatory education: Lessons
from aptitude-treatment interaction research. Educational Psychologist, 23, 313-352.
Rogoff, B. (1984). Introduction: Thinking and learning in social context. In B. Rogoff & J.
Lave (Eds.), Everyday cognition (pp. 1-8). Cambridge, MA: Harvard University Press.
Rumelhart, D. G., McClelland, J. L., &the PDP Research Group. (1986). Parallel distributed
processing: Vol. I . Foundation. Cambridge, MA: MIT Press.
Salomon, G. (1990, April). If intelligence is distributed, what about the cultivation of
 individuals' abilities? Paper presented at the meeting of the American Educational Resear~ch
Association, Boston.
Scheffler, I. (1985). Of human potential. Boston: Routledge & Kegan Paul.
Shoham-Salomon, V., & Hannah, M. T. (1991). Client-treatment interactions in the study of
differential change process. Journal of Consulting and Clinical Psychology, 59, '217-225.
Shute, V. J. (1990a, April). A comparison of rule-induction and rule-application learning
environments: Which is better for whom and why? Paper presented at the meeting of the
American Educational Research Association, Boston.
Shute, V. J. (1990b). Aptitude-treatment interactions and cognitive skill diagno~iis.Unpub-
lished report, Air Force Human Resources Laboratory, Brooks Air Force Base, San
Antonio, TX.
Simon, H. A. (1969). The sciences of the artificial. Cambridge, MA: MIT Press.
Snow, R. E. (1977a). Individual differences and instructional theory. Educational Researcher,
6(10), 11-15.
Snow, R. E. (1977b). Research on aptitudes: A progress report. In L. S. Shulman (Ed.),
Review of research in education (Vol. 4, pp. 50-105). Itasca, IL: Peacock.
Snow, R. E. (1978). Theory and method for research on aptitude processes. Intelligence, 2,
225-278.
Downloaded by [University of Otago] at 11:27 25 December 2014

Snow, R. E. (1980). Aptitude and achievement. In W. B. Schrader (Ed.), Measuring

achievement: Progress over a decade. Proceedings of the 1979 ETS Invitational Ccmference
(pp. 39-59). San Francisco: Jossey-Bass.
Snow, R. E. (1981). Toward a theory of aptitude for learning: Fluid and crystallizedl abilities
and their correlates. In M. P. Friedman, J. P. Das, & N. O'Connar (Eds.), Intelligence and
learning (pp. 345-362). New York: Plenum.
Snow, R. E. (1982). Intelligence and education. In R. J. Sternberg (Ed.), Handbook of human
intelligence (pp. 493-585). Cambridge, England: Cambridge University Press.
Snow, R. E. (1986). Individual differences and the design of educational programs. American
Psychologist, 41, 1029-1039.
Snow, R. E. (1987). Aptitude complexes. In R. E. Snow & M. J. Farr (Eds.), Aptitude,
learning, and instruction: Vol. 3. Conative and affective process analyses (pp. 13-59).
Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Snow, R. E. (1989a). Aptitude-treatment interaction as a framework of research in individual
differences in learning. In P. L. Ackerman, R. J. Sternberg, & R. Glaser (Eds.)~,Learning
and individual differences: Advances in theory and research (pp. 11-34). New York:
Freeman.
Snow, R. E. (1989b). Cognitive-conative aptitude interactions in learning. In R. Kanfer, P. L.
Ackerman, & R. Cudeck (Eds.), Abilities, motivation, and methodology (pp. 435-474).
Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Snow, R. E. (1989~).Toward assessment of cognitive and conative structures in learning.
Educational Researcher, 18(9), 8-14.
Snow, R. E. (1990). New approaches to cognitive and conative assessment in education.
International Journal of Educational Research, 14, 455-473.
Snow, R. E. (1991a). Aptitude-treatment interaction as a framework for research on
individual differences in psychotherapy. Journal of Consulting and Clinical Psychology,
59, 205-216.
Snow, R. E. (1991b). The concept of aptitude. In R. E. Snow & D. F. Wiley (Eds.), hproving
inquiry in social science (pp. 249-284). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Snow, R. E., Kyllonen, P. C., & Marshalek, B. (1984). The topography of ability and learning
correlations. In R. J. Sternberg (Ed.), Advances in the psychology of human intelligence
(Vol. 2, pp. 47-104). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Snow, R. E., & Lohman, D. F. (1984). Toward a theory of cognitive aptitude for learning from
instruction. Journal of Educational Psychology, 76, 347-376.
 Snow, R. E., Tiffin, J., & Seibert, W. F. (1965). Individual differences and instructional film
effects. Journal of Educational Psychology, 56, 315-326.
Spearman, C. E. (1927). The abilities of man. New York: Macrnillan.
Sternberg, R. J. (1977). Intelligence, information processing, and analogical reasoning: The
componential analyst of human abilities. Hillsdale, NJ: Lawrence Erlbaum Associates,
Inc.
Sternberg, R. J. (1985). Beyond IQ: A triarchic theory of human intelligence. Cambridge,
England: Cambridge University Press.
Swanson, J. (1990, April). The effectiveness of tutorialstrategies:An experimentalevaluation.
Paper presented at the meeting of the American Educational Research Association, Boston.
Thomson, G. H. (1919). On the cause of hierarchical order among correlation coefficients.
Proceedings of the Royal Society. A, 95, 400-408.
Thomson, G. H. (1939). The factorial analysis of human ability. London: University of
London Press.
Thorndike, E. L. (1921). Educational psychology (3 vols.). New York; Teachers College,
Columbia University.
Thurstone, L. L. (1924). The nature of intelligence. Westport, CT: Greenwood Press.
Thurstone, L. L. (1947). Multiple factor analysis. Chicago: University of Chicago Press.
Downloaded by [University of Otago] at 11:27 25 December 2014

Tobias, S. (1989). Another look at research on the adaptation of instruction to student

characteristics. Educational Psychologist, 24, 213-227.