Content uploaded by Kathleen Slaney
Author content
All content in this area was uploaded by Kathleen Slaney on Sep 11, 2019
Content may be subject to copyright.
1
Running head: HYPOTHETICAL CONSTRUCT
What makes a hypothetical construct “hypothetical”? Tracing the origins and uses of the
‘hypothetical construct’ concept in psychological science
Nathalie Lovasz & Kathleen L. Slaney
Department of Psychology, Simon Fraser University, Burnaby, BC, Canada
Copyright Elsevier
https://doi.org/10.1016/j.newideapsych.2011.02.005
This article may not exactly replicate the authoritative document published in the
Elsevier journal. It is not the copy of record
Correspondence concerning this article should be addressed to Kathleen Slaney,
Department of Psychology, Simon Fraser University, 8888 University Drive, Burnaby, B.C.,
V5A 1S6. Electronic mail may be sent to klslaney@sfu.ca.
This study was supported in part by a Social Sciences and Humanities Research Council
of Canada (SSHRC) Standard Research Grant 410-2010-0194 awarded to the first author.
2
Abstract
The “hypothetical construct” has been an important concept in philosophy of science for the past
half century. References to the concept date as far back as 1943 (Loomba, 1943). Inconsistencies
in the use of the term and the related ‘intervening variable’ concept prompted MacCorquodale
and Meehl (1948) to distinguish the two concepts and propose conventions for their employment
in psychological discourse. They recommended that ‘hypothetical construct’ designates
theoretical concepts that “refer to processes or entities that are not directly observed” and, thus,
fail to meet the requirements of intervening variables (p. 104). It is interesting to speculate what
makes a hypothetical construct “hypothetical.” The motivation for attaching “hypothetical” to
constructs is not always immediately apparent. The aim of this paper is to trace the origins of the
expression ‘hypothetical construct’, delineate its employment in psychology, and explore the
ontological and epistemological presuppositions that underlie conceptions of hypothetical
constructs.
Keywords: Hypothetical construct, intervening variable, construct, epistemology, ontology,
concepts.
3
What Makes a Hypothetical Construct “Hypothetical”? Tracing the Origins and Uses Of
the ‘Hypothetical Construct’ Concept in Psychological Science
The “hypothetical construct” is and has been an important concept in the philosophy of
psychological science for the past half century. References to the concept date at least as far back
as 1943,when Loomba appealed to the fact that sensations are hypothetical constructs and thus
could not cause perceptions, as had been a premise of the Woodwardian position on the
relationship between perceptions and sensations (Loomba, 1943). MacCorquodale and Meehl
(1948), based on what they took to be a commonly occurring confusion between “abstractive”
and “hypothetical” concepts, suggested that psychological theorists adopt the linguistic
convention that ‘intervening variable’ (after Tolman, 1938) be used in reference to quantities
obtained from the manipulation of empirical (i.e., observed or potentially observable) variables;
they claimed ‘hypothetical construct’, conversely, should be used to designate theoretical
concepts that “refer to processes or entities that are not directly observed” and, thus, fail to meet
the requirements of intervening variables (p. 104). More recently, Colman (2006) describes a
‘hypothetical construct’ as “a conjectured entity, process, or event that is not observed directly
but is assumed to explain an observable phenomenon”, with the further caveat that “It is not
merely a summary of the relationships between observable variables but contains surplus
meaning over and above such relationships” (p. 359; emphasis added). In comparison, he defines
‘construct’ as “a model based on observation guided by a theoretical network”, but also as a
“psychological attribute, such as intelligence of extroversion, on which people differ from one
another”, or “more generally any complex concept synthesized from simpler concepts” (p. 166).
In current psychological research it is rare to see references to intervening variables but the term
‘hypothetical construct’, and especially the diminutive ‘construct’, is referenced frequently.
4
In light of this, it is interesting to speculate as to what makes a ‘hypothetical construct’
“hypothetical.” The hypothetical construct concept is clearly presumed to be related to the
notions ‘theoretical concept’, ‘theoretical term’, ‘hypothetical entity’, and the like in important
ways. However, the motivation for attaching “hypothetical” to constructs is not always
immediately apparent. It appears that sometimes what makes a construct hypothetical is simply
that it appears in a scientific hypothesis. In other contexts, the role of the hypothetical rider is to
convey the property of unobservability with respect to an entity which is theorized to exist and,
typically, to have generative powers in reference to a particular set of observable phenomena. In
such a case, the hypotheticality presumably is a consequence of the fact that its referent is
unobservable and, so, its existence may not be directly and conclusively demonstrated. The
“hypothetical” descriptor at other times seems intended to convey a merely abstracting or
heuristic role, wherein constructs are explicitly created by the scientists in order to make theories
more efficient and facilitate ease of communication about a given class of phenomena. The
possibility of there being multiple senses of hypotheticality where constructs are concerned
raises a further question about the extent of overlap between these various senses. In this present
paper we will trace the origin of the expression ‘hypothetical construct’ in the philosophy of
science, delineate the various ways in which it has been and is currently employed in
psychological science, and explore the ontological and epistemological presuppositions that
might underlie different conceptions which are held regarding the nature of hypothetical
constructs.
However, before attempting a description of the various ways in which the ‘hypothetical’
descriptor appears to operate, it is of course first prudent to remind the reader that the more
general concept ‘construct’ has been employed in various ways in scientific discourse and has
5
both a longer history and broader scope of use than does ‘hypothetical construct.’
1
. In the most
general sense, constructs constitute a general class of scientific concepts, including but not
limited to hypothetical constructs, intervening variables, operational definitions, and other
classes of theoretical concepts (e.g., technical concepts, names of apparatus, etc.). Although we
recognize that the notion of ‘hypothetical construct’ is related in a very clear way to current
conceptualizations of psychological “constructs”, here we will not explicitly examine the
relationship between the two. Rather, the major aim of the current work is to explore the history
of the hypothetical construct concept and determine what advantage, if any, the ‘hypothetical’
rider provides in specifying the role this particular type of scientific concept is meant to play.
The Historical Context: Intervening Variables and Hypothetical Constructs
Although we note a single reference to the term ‘hypothetical construct’ prior to 1948,
this particular employment of the term appears to accord more with what others referred to as an
intervening variable (cf. Loomba, 1943). To our knowledge, MacCorquodale and Meehl are the
formal originators of the term that has been in use within psychology (and in a limited way also
within the philosophy of science) since they introduced it in their well-known 1948 paper.
The Philosophical and Disciplinary Context
The earliest papers on hypothetical constructs contain implicit and explicit references to
the disciplinary and philosophical context that forms the backdrop for the development of this
concept in psychological discourse, including the growing influence of empirical (scientific)
realism as a general approach to science and within psychological science the emergence of
neobehaviorism and neuropsychology. At the time, the influence of logical positivism and strict
operationism, which had been the predominant philosophy of science underlying most
1
Orton (1987) speculates that the origins of the former may lie in Bertrand Russell’s (1929) plea to philosophers to
attempt whenever possible to “construct” individual terms by logical operations from concrete experiences.
6
psychological theory for several decades, was slowly waning. A growing recognition of the
limitations of operationalist reductionism precipitated a move towards empirical realist
conceptions of science and an attendant emphasis on and appreciation for the importance of
theoretical concepts. The latter, it was now thought, could not be strictly reduced to
observational terms and played an essential role in denoting those entities (or properties thereof)
that featured in theoretical explanations for the existence of particular phenomena.
At the same time, psychological scientists were losing faith in the purely behaviorist
approach that had dominated psychological practice in North America for several decades and
that restricted explanations of behavior to series of direct stimulus-response relationships. The
failure of traditional behaviorist paradigms to provide adequate explanatory frameworks
contributed to a general shift towards neobehaviorism, which allowed for investigations of the
roles motivational or cognitive variables had to play in stimulus-response relationships.
Psychology thus required a language to describe variables and their role in previously established
stimulus-response relationships.
Finally, the commonly held belief that observable behavior constitutes the only legitimate
unit of analysis was challenged through the development of new technologies for making
“observations” (e.g., neurosurgical or brain imaging techniques) and thus recognition of the role
of neurological determinants of behavior. Attention to neurological processes further
necessitated consideration of the legitimacy of real but unobservable psychologically relevant
entities and a way to talk about them and distinguish them from operationalized variables. We
contend that the promotion and development of the hypothetical construct concept provided a
means of denoting real but potentially unobservable structures, mechanisms, and processes that
might play a mediating role in stimulus-response relationships. Furthermore, in contrasting
7
hypothetical constructs from intervening variable constructs, a clear distinction could be made
between theoretical concepts that are reducible to observables and those that are not.
The Early Theories
Within this historical context, theorists such as Tolman (1938) and Hull (1943) began to
champion the use of “logical or theoretical constructs” and “intervening variables” to liberate
psychological science from the strict confines of behaviorism, which reduced all interesting
questions to explicitly defined stimulus-response relationships. In his 1938 paper, “The
Determiners of Behavior at a Choice Point,” Tolman characterized intervening variables as
intermediaries in the relationship between independent and dependent variables; he argued that a
“theory is a set of intervening variables” and “intervening variables are ‘constructs’ which we,
the theorists, evolve as a useful way of breaking down into more manageable form the original
complete…function.” (p. 9). Hull (1943) in his Principles of Behavior argues that logical
constructs or intervening variables facilitate scientific thinking and that an “intervening variable
is never directly observed but is an inference based of the observation of something else and the
inference is critically dependent upon the experimental manipulations (operations) which lead to
the observations.” (p. 30). Although both theorists assigned intervening variables a central role in
psychological theory, Tolman defined intervening variables by their functional role in the
relationship between independent and dependent variables whereas Hull characterized
unobservability as their hallmark feature.
The concept of the ‘hypothetical construct’ itself arguably entered theoretical
psychological discourse most formally with the publication of MacCorquodale and Meehl’s
(1948) paper, “On a Distinction Between Hypothetical Constructs and Intervening Variables”,
which was expressly aimed at clearing up conceptual confusions plaguing the use of the term
8
‘intervening variable.’ MacCorquodale and Meehl suggested three formal ways to state the
distinction between hypothetical constructs and abstractive concepts such as intervening
variables: a) Whereas abstractive concepts contain only words which can be explicitly defined in
terms of empirical relations, hypothetical constructs contain words that cannot be defined strictly
in terms of observables in this way; b) the truth of the empirical laws that justify sentences
containing hypothetical constructs is necessary but not sufficient to establish the truth of the
sentence; however, in the case of sentences containing abstractive constructs, the truth of the
empirical laws is both necessary and sufficient for the truth of the sentence; c) for abstractive
concepts, the construct itself is reducible to the quantitative representations of the construct but
the hypothetical construct is not reducible to its quantitative representations (i.e., laws of
relations between empirical properties can be deduced from a hypothetical construct but the
construct cannot be deduced from the laws); in other words, explicit definitions of a hypothetical
construct convey at best only part of the concept’s meaning. MacCorquodale and Meehl also
argued that the use of intervening variables is justified by their utilitarian value while the use of a
hypothetical construct is justified by its truth (i.e., the actual existence of that which it denotes).
Thus, MacCorquodale and Meehl recommended the above mentioned convention of using the
expression ‘hypothetical construct’ to denote in principle observable but currently unobservable
theoretical entities and ‘intervening variable’ for concepts which fulfill a strictly heuristic or
summarizing function.
Far from providing but a simple conceptual clarification, MacCorquodale and Meehl’s
account spurned a lively debate that quickly illuminated further dimensions of the problem that
would require explication and clarification. Some theorists (e.g., Krech, 1949; 1950; Tolman,
1949) thought MacCorquodale and Meehl’s account implied that hypothetical constructs
9
designate neurological or physiological entities,
2
and promoted the use of such constructs in
psychological discourse. However, others voiced concerns either about the legitimacy of
admitting into science terms that contain “hypothetical content”, i.e., theoretical terms which
cannot be sufficiently operationalized (Marx, 1951), or of the constrained conception of
theoretical concepts as falling into only the two classes identified by MacCorquodale and Meehl
(Lindzey, 1953). Still others (Adcock, 1954; Plutchik, 1955) defended the value of both
hypothetical constructs and intervening variables for scientific practice on grounds of parsimony
and simplicity of theoretical explanations. Finally, attributing much of the intervening variable-
hypothetical construct debate to conceptual confusion and logical inconsistency, theorists such as
Ginsberg (1954) and Hilgard (1958) suggested alternate ways of conceptualizing the difference
between hypothetical constructs and intervening variables, e.g., emphasizing their role in relation
to theories and laws as opposed to predominantly with regard to interpreting empirical
observations. From these early debates, however, it is clear that in the decade following the
publication of MacCorquodale and Meehl’s paper, the presumed nature of hypothetical
constructs and their role in psychological science was still unclear and therefore subject to
debate.
Interestingly, by 1955 when Cronbach and Meehl published what would become a much
more influential work, “Construct Validity and Psychological Tests”, outside of theoretical
discourse the ‘hypothetical’ descriptor had been all but dropped in favor of adopting the
2
It should be noted that in much of the literature pertaining to hypothetical constructs, both historically and currently,
the hypothetical construct concept is conflated with the class of theoretical entities for which such constructs are reserved as
designators. That is, many will speak of a hypothetical construct as being an “entity”, “mechanism”, or “process”, etc., which
exists but cannot be observed. We believe that this common conflation betrays certain confusions regarding the proper roles of
definitional, ontological, and epistemological issues in psychological science, a point to which we return later in the paper.
10
diminutive ‘construct.’ Whether this was the result merely of a relatively natural “pruning” that
tends to occur with expressions over time, or an attempt to define a broader class of concepts of
which hypothetical constructs are to be considered merely a subclass, is difficult to extrapolate
from the vague characterization of hypothetical constructs provided by Cronbach and Meehl in
their 1955 paper.
What Makes a Hypothetical Construct “Hypothetical”?
We suspect that part of the equivocation regarding the appropriate role of hypothetical
constructs within psychological science (and by pain of contrast that also of intervening variable
constructs) is that what is to be taken as the “hypothetical” part of a hypothetical construct is not
abundantly clear. A variety of characterizations have been given of hypothetical constructs, and
these are not always consistent with respect to what feature or component of a hypothetical
construct is hypothetical—some aspect of the construct (i.e., as a concept) itself, or of the entities
(processes, states, mechanisms, etc.) that are in the construct’s extension, or some combination
of both. Such inconsistencies imply multiple and possibly distinct senses of ‘hypotheticality’
when it comes to hypothetical constructs.
3
The task here is to attempt to shed light on these
different senses and to reveal certain philosophical presuppositions on which they are based.
A perhaps sensible starting point to an explication of the meaning of ‘hypothetical
construct’ is to elaborate the distinction that has been made within the philosophy of science
between so-called “empirical” and “theoretical” constructs. Although there exists some
equivocation over what constitutes an empirical construct, in general terms one could take
empirical constructs to be concepts which are defined in terms of observation predicates;
theoretical concepts, conversely, are not defined explicitly, i.e., strictly in terms of observables,
3
For expository purposes, in the sequel we shall employ “entity” to represent generally both spatio-temporal entities
and states, processes, mechanisms, etc.
11
and thus may be characterized as involving some degree of abstraction. The relationship that is
presumed to exist between these two classes of concepts varies, with some (e.g., Bergmann &
Spence, 1941) taking empirical constructs to be strictly reducible to the observation language
with no further implicit reference to unobservable entities and theoretical constructs as a class of
concepts reserved for referring to such unobservables. Others (e.g., Chaplin, 1968; English &
English, 1958; Hempel, 1950) assume a greater degree of overlap between empirical and
theoretical constructs, with the former essentially providing a mechanism for defining theoretical
concepts in terms of observation predicates, but whose reference is not restricted to the realm of
observable entities. The relevant question for the current work is where hypothetical constructs
are presumed to reside with respect to this distinction. We believe the answer hinges in important
ways on what is meant by “hypothetical” in this context and aim to explore this question.
In service of this aim, we examined definitions and uses of the ‘hypothetical construct’ concept
in dictionaries and encyclopedias of psychology, psychological measurement texts, and various
published works within the philosophy of psychological science. We found that, indeed,
hypothetical constructs might be considered “hypothetical” in a number of different senses,
which are briefly summarized below:
Hypothetical = Theoretical, Constructed, Heuristic
In the most general sense, hypothetical constructs may be characterized as a class of
theoretical concepts that are used by the scientists of a given discipline or research domain to
communicate about some phenomenon or set of phenomena of interest. In this case, what makes
them “hypothetical” is that they occur in scientific theories (and in the propositions deduced
from them), and scientific theories themselves are tentative descriptions of some possible state of
affairs. If true, the latter are correct descriptions, but if false they are not. From this perspective,
12
hypothetical constructs are the core concepts that appear in scientific propositions which once
subjected to empirical test may change from hypothetical descriptions to factual (or, more
realistically, empirically supported) ones. Definitions of ‘hypothetical’ appearing in dictionaries
of psychology—e.g., “still unproved” (Drever, 1952, p. 126), “characteristic of ideas or
explanations which are tentative or which take the form of ‘as-if’ propositions” (Chaplin, 1968,
p. 244)—appear to be consistent with the notion that one aspect of the hypotheticality of
hypothetical constructs is that they appear in theoretical propositions and thus are a class of
theoretical concepts.
In a related way, the hypotheticality of hypothetical constructs could refer to the
constructed nature of these concepts, i.e., that the terms in which they are expressed are to a large
extent chosen by a community of scientists for the explicit purpose of designating some
constituent of the domain under study. Similarly, hypothetical constructs are sometimes credited
with having a heuristic value, in that they enable the scientist to designate in an efficient manner
a particular class of observations whose underlying origins are as yet not well-specified but
which are presumed to have a common causal source. Hence, hypothetical constructs are
“hypothetical” in the sense that they are created by scientists to function either as linguistic
summaries of classes of observables or as linguistic place-holders of sorts for yet to be
discovered entities. Along these lines, hypothetical constructs have been said to be “literally
constructed” (Thorndike and Hagen, 1969, p. 174), or the “products of…informed scientific
imagination” (Crocker and Algina, 1986, p. 4).
Importantly, hypothetical constructs are characterized as being “heuristic” in two
different senses: First, as indicated above, they are characterized as being “summary concepts”,
“fictions”, or “stories” which help to put a label on an otherwise large and difficult to interpret
13
set of observations. Statt (1998) for example, defines the hypothetical construct as “a fiction or
story put forward by a theorist to make sense of a phenomenon.” (p. 67) In this case, no or little
existential importance is placed on the referent. In the second case, hypothetical constructs are
heuristic in the sense of being useful, or even essential, to science because they “allow
investigations to be made which would otherwise be virtually impossible” (Eysenck, Wurzburg,
& Berne, 1972, p. 97) due to the largely unknown and thus tentative nature of the phenomena
under study.
Hypothetical = Unobservable, Existential, Inferred
The senses of hypotheticality that have been thus far described imply that hypothetical
constructs are, first and foremost, of a linguistic nature in that they are theoretical concepts
which are created and used by scientists to communicate about particular objects and/or areas of
study. Up to this point, it may be surmised that their hypotheticality comes primarily from the
fact that they feature in scientific theories from which descriptions of proposed but perhaps not
actual states of affairs or predictions are deduced, i.e., that they appear in statements whose
content is of a hypothetical rather than established nature. However, more commonly the
hypotheticality of hypothetical constructs is implied to be a feature not of a particular class of
scientific concepts, but of those entities to which hypothetical construct concepts are thought to
extend. Specifically, hypothetical constructs are thought to constitute a special class of
theoretical concepts by virtue of the fact that the entities (or properties thereof) to which they
refer are unobservable. In some cases, these entities are characterized as being real, but in
principle unobservable, as in the following definition of a hypothetical construct as: “any concept
referring to a process or phenomenon, the existence of which cannot be empirically demonstrated
but which nevertheless seems to be required on theoretical grounds or for pragmatic descriptive
14
purposes” (Richards, 2009, p. 99). In other cases, hypothetical constructs are taken to denote
entities that have spatio-temporal existence, are in principle observable, but have yet to be
observed directly (e.g. English & English, 1958; Reber, 1985). English and English for example,
define the hypothetical construct as “a construct referring to an entity or process that is inferred
as actually existing (though not at present fully unobservable) and as giving rise to measurable
phenomena, including phenomena other than the observables that led to hypothesizing the
construct” (p.116).
Importantly, what makes hypothetical constructs a special class of theoretical concepts is
not simply that they do not have observable referents, but rather that they refer to real but
unobservable entities or processes. In other words, they are not merely theoretical abstractions—
which either do not refer at all, or refer only to circumscribed sets of observables—but are
instead “abstract” only in the sense that the real entities and processes to which they refer are, for
whatever reason, perceptually remote and our knowledge of them thus necessarily partial.
Furthermore, it is these very entities and processes whose existence is hypothesized to explain
the observable behaviors that gave rise to the initial articulations of the hypothetical construct,
i.e., those phenomena which the scientists seeks to explain in his or her theories. Thus, whereas
as truly abstractive concepts (such as intervening variables) merely provide the theorist with
efficient descriptive aids, hypothetical constructs enable him or her to test existential hypotheses
about the entities and processes which are theorized to underlie the relevant observables. It is for
this reason that hypothetical constructs are described as having “factual reference” and “surplus
meaning” (Cronbach & Meehl, 1955; Feigl, 1950; Reichenbach, 1938).
It is clear from this perspective that hypothetical constructs are presumed to fulfill some
sort of inferential role in scientific theorizing: If the entities in their extensions truly exist but are
15
unobservable, then their existence must be inferred by the observed phenomena to which they
are hypothesized to give rise. In contrast to intervening variables, which may be explicitly
defined strictly in terms of empirical relations, hypothetical constructs are defined in terms of
entities that can only be indirectly measured, viz., in terms of the empirical variables to which
they are theoretically connected, either causally or concomitantly. Thus for example,
MacCorquodale and Meehl (1948) argue that the resistance of a wire, an intervening variable,
can be strictly empirically defined by the flow of an electric current through it, which can be
measured empirically, while a hypothetical construct such as an electron can only be defined by
its hypothesized causal relations with various empirically observable variables but cannot itself
be directly observed. MacCorquodale and Meehl themselves explicitly tie hypothetical
constructs to Reichenbach’s “illata”, or “inferred things” (1948, p. 96). The fact that the
existence and properties of hypothetical entities must be inferred as opposed to directly accessed
attaches to them a certain degree of uncertainty and the perception that their designation be
restricted to a special class of theoretical terms. Thus, it would seem that here hypotheticality
pertains not to the class of constructs per se, but to the class of entities these constructs are meant
to signify and is an ontological hypotheticality in the sense that claims about the existence of
those (presently or in principle) unobservable entities remain at least to some extent provisional.
Hypothetical = Conjectural, Provisional, “Open”
Hypotheses are by nature conjectural in the sense that they are propositions whose truth
or falsity has up to this point been neither confirmed nor falsified. Thus, it is not surprising that
hypothetical constructs are frequently touted for the indispensable role they play in enabling the
scientist to posit the existence of entities whose presence would explain the occurrence of
particular observable phenomena. However, because these entities are (as yet) unobservable,
16
their existence remains as conjectural as the propositions which contain the terms and
expressions (i.e., the hypothetical constructs) that denote them. It is for this reason that
hypothetical constructs are often described as having “explanatory” value or power—if a given
hypothetical construct in fact denotes the actual entity that is hypothesized to underlie the
relevant set of observable phenomena, then this entity is in reality as it has been hypothesized to
be. Such conceptions are apparent in some definitions given of hypothetical constructs. For
example, Coleman (2006) defines a hypothetical construct as “A conjectured entity, process, or
event that is not observed directly but is assumed to explain an observable phenomenon.” (p.
359). Richards (2009) argues that “Many scientific concepts begin as hypothetical constructs, in
the physical sciences however these typically end by being empirically confirmed or validated,
or, alternatively, abandoned”(p.99). Both explicitly and implicitly, the hypotheticality of
constructs is thus often taken to be related to their provisional and conjectural status. Unlike
intervening variables, which denote mathematically defined functions among observables,
hypothetical constructs are reserved for representing entities that are not strictly reducible to such
formalized empirical relations, but have existence independent of them. Rather, their relations
are ontological in nature in that the hypothetical entities denoted by hypothetical constructs are
those that, if they do, in fact, exist (and they may not), would provide an explanation for the
particular observations which have been made in the specific scientific context at hand.
However, because the existence of the theorized entities designated by hypothetical concepts
cannot be settled merely by appeal to observable phenomena, these concepts are thus said to be
“open.” In other words, because these concepts designate unobservable (and thus potentially
non-existent) entities, their meanings are at least to some extent ineffable—we cannot define
them in explicit terms and thus their meanings are “fuzzy” and difficult to articulate in exact
17
terms. Furthermore, as a science progresses and more is learnt about the hypothesized entities
themselves, the meanings of the corresponding hypothetical construct(s) becomes progressively
more complete. The hypotheticality of these constructs captures their preliminary and tentative,
but revisable, nature.
It would seem from our survey that the above-described senses of hypotheticality do not
stand alone, but rather overlap substantially. Taken together, hypothetical constructs are
theoretical concepts that have real, but unobservable referents whose existence is posited to
explain the relations among a specified class of observable variables. What makes them
“hypothetical” and thus distinguishable from other theoretical concepts is that they are neither
pure abstractions from nor reducible to a specified set of observed variables. Rather, knowledge
of the existence and properties of their unobservable but real referents is still at the earliest
stages of accumulation and thus their full meanings cannot be completely articulated at this time.
As long as the existential status of the denoted theoretical entity in question remains
hypothetical, the extent to which the complete meaning of the associated construct may be
articulated will be constrained. Both the existence of the theoretical entity and the meaning of the
construct denoting it must be inferred and thus remain “hypothetical.”
Taking Root: The Philosophy of Hypothetical Constructs
Despite the debates surrounding the particular role played by and importance placed on
hypothetical constructs, MacCorquodale and Meehl’s original characterization appears to have
held up over time. Hypothetical constructs are thought to fulfill an essential role by virtue of
denoting those entities which, if they truly exist, would explain the occurrence of particular
relations among observed variables. However, given the unobservability of their referents, and
the fact that they cannot be defined simply in terms of empirical relations, their meanings are to
18
some extent open to revision as science progresses and more is discovered about the entities they
denote. This characterization of the special nature of hypothetical constructs carries with it the
following implications: First, the meaning and definability of a given theoretical concept is
contingent on both the ontology and observability of the theoretical entity (process, mechanism,
relation, etc.) it denotes; that is, the extent to which it can be defined will depend on whether it
refers to some real entity—in the sense of existing autonomously of the scientist’s imagination—
and also whether that entity is observable or is merely inferable from other observables. Second,
the meaning and definability of a theoretical concept admits of degree—the greater the
knowledge of the theoretical entity which it designates, the more completely (i.e., explicitly) can
the concept be defined. And, third, the full meaning of a theoretical concept is given by complete
knowledge of the theoretical entity it denotes with respect to its existence, properties, and
relations to other entities.
We contend that this conceptualization is tied not only to a realist ontology, but also to a
particular conception of language and meaning and of the nature of concepts generally and,
furthermore, fuels certain common misconceptions about the proper roles of definitional (i.e.,
conceptual or grammatical), ontological, and epistemological dimensions of science. We further
contend that this conception underlies the common practice of conflating theoretical concepts
and their referents. In the final section of the paper, we briefly describe a particular picture of
language in which the received view of hypothetical constructs is firmly embedded and then
attempt to disambiguate the definitional, ontological, and epistemological dimensions of science
and the role each plays with respect to treating hypothetical constructs as a special class of
theoretical concepts.
The Meaning and Definability of Hypothetical Construct Concepts
19
It is argued here that the view that hypothetical constructs constitute a special class of
theoretical concepts, whose semantics cannot be fully explicated, is reflective of particular
commitments regarding language and the nature of concepts. Specifically, it relies on a general
picture of language known as the Augustinian conception of language (ACL; cf. Baker &
Hacker, 2005; Maraun, Slaney, & Gabriel, 2009), the contours of which were first sketched by
Wittgenstein in his later work (e.g., Wittgenstein, 1953). The ACL is based on three fundamental
beliefs: a) all words are names, b) word meaning is assigned by correlating a word with the
thing(s) it names and, therefore, c) ostensive definition is the fundamental form of explanation of
words. A corollary of the ACL is that there are two types of words: so-called “indefinables”
(e.g., ‘red’), which denote immediately and directly observable phenomena and “definable”
words, whose meanings may only be explained via an analysis of other words and expressions
(e.g., ‘colour preference’). Whereas the meanings of indefinables are under ACL deemed
unproblematic, the extent to which definables have meaning is contingent on whether they can
be defined strictly in terms of indefinables. In other words, if a definable term is reducible to a
set of necessary and sufficient indefinable terms, its meaning is clear; otherwise, its meaning is at
least to some extent “open.” According to this view, the full and complete meaning of a word is
contained in the entity to which it refers—i.e., the meaning of a word is the thing to which it
refers.
The ACL has the following implications for how the hypothetical construct concept is
conceptualized: First, since hypothetical constructs are reserved for denoting real (but
unobservable) theoretical entities, all such concepts are names, and an explicit definition of the
concept is given by the theoretical entity it denotes. From this standpoint, there would be little
need to distinguish between a hypothetical construct and its referent. Thus, common references
20
to hypothetical constructs themselves being “actually existing” or “real”, but “unobservable” or
“inferred” entities would not be cause for alarm.
A second implication of the ACL is that, whereas the semantics of intervening variable
concepts are clear, hypothetical constructs are semantically underprivileged because they refer to
unobservables and cannot therefore be defined strictly in terms of indefinables (as can
intervening variables). This points to a third implication: As a science advances and more and
more discoveries are made about the theoretical entity denoted by a hypothetical construct, the
full meaning of the construct will be progressively more completely defined. Thus discovery of
the meaning of a given hypothetical construct is part and parcel with discoveries made about the
theoretical entity denoted by the concept. The ultimate aim would be to give a complete
definition of the hypothetical construct by ostension, i.e., by direct reference to the previously
unobservable but now observed theoretical entity in question. However, in cases in which
hypothetical entities are in principle unobservable, the aim would be to provide an explicit
definition of the hypothetical construct in terms of an articulation of the network of empirical
relations into which the relevant theoretical entity enters (i.e., strictly in terms of the full set of
indefinables to which the hypothetical construct is theoretically connected). The latter aim, in
fact, provided the motivation for Cronbach and Meehl’s well-celebrated endorsement of the
nomological network view of theories.
Finally, these three implications together imply a fourth implication of the ACL, viz., the
common practice in psychological scientific discourse of using ‘hypothetical construct’ (and
more recently ‘construct’) to refer simultaneously to a class of theoretical concepts and the class
of theoretical entities to which such concepts refer. On the one hand, hypothetical constructs are
characterized (properly) as the linguistic theoretical concepts that are used to refer to the entities,
21
or abstract classes thereof, which are focal to a given hypothesis or theory more generally. On
the other hand, hypothetical constructs are frequently explicitly defined or otherwise
characterized as “unobservable” or “inferred” entities that give rise to the observable phenomena
and thus have existential status over and above those observable effects. Examples of the latter
abound in both the theoretical and applied literature (e.g., Colman, 2006; Helmstadter, 1964;
Moore, 1998).
Here, we contend that the intervening variable-hypothetical construct distinction, and so
too the ACL which informs it, betrays a confused conception of the proper roles of definitional,
ontological, epistemological dimensions of scientific theorizing and of how to appropriately
construe the relations among them. We now briefly describe each in turn and the implications of
their relations—properly construed—for the legitimacy of distinguished hypothetical constructs
as a special class of theoretical concept.
The definitional dimension of science.
Definitional issues are conceptual in nature, as definitions are explanations of meanings
of concepts and their modes of expression. They need not—and, in fact, seldom are—expressible
in terms of necessary and sufficient criteria and, as a result, the grammars of some concepts can
indeed be very difficult to lay down in explicit terms. Rather, the meanings of concepts and the
words and phrases in which they are expressed are manifest in their use within a given language.
Concepts can, and in many cases do, denote objective entities, some of which cannot be observed
for one reason or another. However, denotational relations are grammatical and, thus, must be
settled by conceptual rather than empirical investigations. Likewise, the specification of
denotational relations between a concept ‘X’ and an entity X is logically prior to empirical
claims about X and thus one cannot appeal to the empirical in order to settle definitional matters
22
(i.e., “We will know what ‘X’ means once we’ve clearly identified X”). For these reasons, it
makes little sense to say (or imply) that the meaning of a concept, and so too the word that
expresses it, is contained in the referent. This is to confuse the matter of clarifying the meaning
of a concept (a definitional and thus conceptual matter) and the task of making discoveries about
the existence and properties of an entity (a bona fide empirical matter). And, although one is free
to cast one’s empirical net far and wide in service of the latter, one cannot do so fruitfully until
one has clarified the meanings of the relevant concepts—i.e., no amount of empirical observation
or experimental test will clarify what ‘X’ means. This does not, however, preclude the possibility
that empirical discoveries about X will lead to changes or modifications in the concept ‘X’
(Baker & Hacker, 1982), but this is to say only that the impetus for conceptual change was some
empirical discovery (or set of discoveries) and not that the meanings of the concepts are
themselves those empirical facts.
The ontological dimension of science.
With respect to the ontological dimension, one must distinguish between the reality of
theoretical concepts and that of the theoretical entities they denote. The former is trivial—as long
as there are rules for the use of a given concept (i.e., as long it has a grammar in one or more
languages) then the concept “exists.” Here ontological claims about concepts are really
definitional and are thus conceptual in nature. From the current authors’ perspective, the only
meaningful sense in which one could speak of an “ontology” of concepts (theoretical or
otherwise) is to note that there are different types of concepts and that these different types may
play distinctly different roles within a language (e.g., subject versus predicate nouns, etc.).
Conversely, the ontology of theoretical entities is a different matter. Whereas one cannot
dispute the existence of a concept (if one is using it), the existence of a theoretical entity or of
23
some subset of properties thereof is a disputable matter, and the existence of a theoretical
concept is of course insufficient for making existential claims about entities that are designated
by them. Such ontological questions are properly construed as being empirical in nature and thus
are part and parcel with the aims of science. However, one cannot embark on such ontological
queries in the absence of concepts which denote the particular entities under investigation,
otherwise how would one identify or recognize the entity when one came across it? This does not
mean that all the concepts that will ultimately become relevant to the entity (e.g., those
signifying particular properties, etc.) need to exist prior to such investigations, or that new
entities will not be discovered and subsequently named. Science is in the business of discovery
and novel concepts are frequently created in order to communicate scientific findings and name
discovered entities. Here, we argue only that the reality of theoretical concepts be kept distinct
from the reality of theoretical entities and that the existence of the latter is not sufficient for the
“truth” (i.e., meaningfulness) of the former.
In the same vein, concepts can refer to entities that have objective existence, and may be
observable or unobservable, or may merely be abstractions over classes of observations or serve
some other summary function. The meanings of concepts, however, are not compromised by the
observability or nonobservability of their referents. That is, their meanings—which are manifest
in their use—are no less “complete” than the meanings of concepts which refer to observable
entities.
4
To believe otherwise is to confuse definition with existence, grammatical criteria with
properties of objects, and the conceptual issue of clarifying the meanings of theoretical concepts
and terms with the epistemological problems that may be encountered when attempting to
describe and explain the properties of theoretical entities which may be denoted by such terms.
4
Although, they may certainly be impossible to lay down in terms of necessary and sufficient criteria. This
fundamental point was elaborated by Wittgenstein (1953) in his well-known work on the family resemblance nature
of most concepts.
24
The epistemological dimension of science.
In the most general sense, science is in the epistemology business, i.e., it is driven by the
fundamental aim of generating and accumulating knowledge. In service of this aim, theoretical
concepts are articulated and developed such that propositions can be put to the empirical test.
However, as with the ontological dimension, the extent to which any importance may be attached
to epistemological claims is contingent on the conceptual clarity of the relevant propositions and
their constituents. Furthermore, although it is true that epistemological problems can and do arise
from observability problems, this pertains to the entities under study and not to the particular
class of concepts that may be used to denote them. The history of science is replete with
examples of difficulties encountered by scientists interested in studying entities, properties,
mechanisms, etc., which are unobservable and indeed many clever apparatuses have been
developed in order to circumvent such problems (e.g., alpha chambers, electron microscope).
Another such strategy, one that has been wholly embraced within psychological science,
has been to operationalize the theoretical concepts of interest, by laying down stipulative
definitions so as to render the relevant theoretical propositions empirically testable. However, for
some—viz., positivist thinkers—operational definitions and other “empirical constructs” are
sufficient for defining the phenomena of interest. In this case, operational definitions are viewed
as a means of giving explicit definitions to an otherwise abstract theoretical concept which does
not itself refer to an unobservable theoretical entity but, rather, to a class of related observations.
Here, operational definitions may themselves be considered theoretical concepts in that they
provide a heuristic or summarizing function.
Realist thinkers, on the other hand, view operationalization of theoretical concepts as an
epistemological tool that provides a means of studying those unobservable or otherwise
25
inaccessible theoretical entities which “underlie” the observations resulting from the
measurement operation. Here, operational definitions satisfy merely an epistemological aim and
are not meant to replace the relevant theoretical concepts and are certainly not meant to be taken
as a sufficient definition of the phenomena under study. In such cases it is thought that, although
concretely defined, the operational definition gives only partial definition to the theoretical
concept at hand and, as such, it is insufficient for specifying its (complete) meaning, the latter of
which may only be specified by discovering the theoretical entity to which the concept refers
(Hempel, 1965).
Thus, it would seem that operational definitions may serve more than one role in
scientific theorizing—sometimes they are used to denote a (hypothesized) theoretical entity and
at other times they merely designate a relatively abstract class of observations or relations among
them, the former motivated by the desire to circumnavigate epistemological barriers, the latter by
the need to communicate scientific findings in an efficient and consistent manner. It would seem
that the impetus behind MacCorquodale and Meehl’s paper hung on this distinction.
What Really Makes a Hypothetical Construct “Hypothetical”:
Meaning, Existence, and Observability.
A number of important implications for commonly held notions regarding hypothetical
constructs follow from the characterization of the definitional, ontological, and epistemological
components of science that have been provided above: First, irrespective of the entities that
hypothetical constructs may denote, definitions are embedded in linguistic practices, or
conventions, and thus are not contained in the referent of a concept. Even ostensive definitions,
i.e., those given by pointing, are themselves rules and as such are autonomous of the objects
defined. Although concepts (and their particular modes of expression) are, to be sure, internally
26
related (i.e., conceptually) to their referents, they are not the same thing as their referents and, so,
their meanings are not given completely by their referents. Furthermore, unless one is speaking
about the different roles that different types of concepts fulfill in a language, there is no ontology
of concepts, per se, outside of the trivial sense in which a concept either has a sense (i.e., a use in
a language) or it does not. There may be concepts for which no object referent exists; there may
also be an objectively existing object for which there is no concept. But, in the latter case, we can
say nothing about the object.
A second implication is that meaning and ontology are separate and need not inform one
another at all. A theoretical construct may well refer to something real but unobservable, or may
refer to nothing at all. But, its meaning and any definitions that are given of it are quite
autonomous of the existence, or observability for that matter, of its referent. On a related note,
the meaning of a concept is not contingent on the status of knowledge about its referent, the
former being a conceptual issue, the latter epistemological. But, this is not to say that linguistic
conventions do not change or that empirical discoveries cannot motivate conceptual change
(Baker & Hacker, 1982): particular modes of expressions which were once in common use may
cease to be used and other, perhaps new, modes of expressions may be adopted; the empirical
grounds typically appealed to for ascribing a given concept at one point in time may be
abandoned in favor of other grounds due to some relevant empirical discovery. Empirical
discovery may alter or influence the meaning of a construct, but such discoveries themselves
presuppose that the concept has a meaning.
Third, although operational definition may serve multiple functions in science, as with
any concept, the specific role a given operational definition plays in a given context has no
bearing on its meaning. Regardless of whether it denotes, or merely designates, specifying and
27
clarifying its meaning is a definitional (conceptual) issue which is separate from the ontology of
its referent and from the epistemological problems (or ease) the scientist encounters when
engaged in scientific inquiries into the nature of its referent. Once again, epistemological issues
concerning the empirical accessibility of a theoretical entity, however it may be defined, do not
bear on the definitional dimension. Said another way, stipulating what a hypothetical construct
concept means in a given research context is distinct from specifying the properties of the entity
to which it is thought to refer. Thus, although operationalization may render an unobservable
theoretical entity epistemologically accessible, it need not fulfill only this role—i.e., it can serve
a merely heuristic or summarizing function—and specification of its meaning is not constrained
by the particular role it serves. In the case of hypothetical constructs, we believe that it is
important to remember that if epistemological properties such as unobservability are posited,
those are relevant only to the entity denoted by a given hypothetical construct and not the
construct itself.
Finally, clearly one cannot sort out the epistemological issues relevant to a given
hypothetical entity until one clarifies the meanings of the terms and expressions employed in the
theoretical propositions involved. In other words, in order even to recognize that one may
encounter difficulties in confirming the existence of some entity or in describing its properties
one must already be in possession of concepts that denote the entity and its various properties.
Moreover, the epistemological issues that bear on investigations of a theoretical entity do not
impinge on conceptual matters regarding the clarification of the meaning of a concept. Laying
down or stipulating definitions, clarifying the meaning of a given scientific term or expression,
and explicating the relations among different kinds of scientific concepts is a conceptual matter.
28
Discovering the features or properties of a given scientifically relevant entity is an empirical
matter. However, empirical investigations presuppose conceptual clarity.
From this vantage point, there is nothing hypothetical about hypothetical constructs;
rather, what is hypothetical is the existence of the entity or class of entities designated by such
constructs. The meaning of a hypothetical construct is no more semantically “open” than the
meaning of any other concept and although the unobservability of the referent of a hypothetical
construct (an epistemological issue) may limit the kinds of definitions that can usefully be
provided for such a construct (e.g. it cannot be given strictly in terms of indefinables), as long as
there are conventions regarding the correct employment of a concept, those conventions
constitute the meaning of the concept. Scientific discoveries about the referent of a hypothetical
construct do not bring to light the full meaning of the concept, since the former is an
empirical/ontological issue and the latter a conceptual one. In other words, although one can
make discoveries about hypothetical entities, one cannot discover their meanings. Rather, the
meanings of hypothetical concepts are given in the grammatical criteria which stipulate their
appropriate uses (sometimes but not always in terms of necessary and sufficient criteria). The
“surplus meaning” that is often attributed to hypothetical constructs is really a reference to the
fact that the existence and description of the full set of properties and relations of the theoretical
entities they are believed to denote have yet to be empirically verified.
Clearly, then, the hypothetical constructs cannot simultaneously be a class of theoretical
concepts and the class of theoretical entities to which such concepts refer. The properties of a
hypothetical construct are separate from the properties of its referent and although it may well
make sense to classify hypothetical constructs as a class of concepts that refer to entities which
are thought to be real but unobservable, it makes no sense to describe the hypothetical constructs
29
themselves as either real or unobservable. Hence, while a certain amount of conflation between
concepts and entities may be trivial in common parlance, the distinction between concepts and
entities is paramount in the conceptualization of the definitional, ontological and epistemological
issues that pertain to hypothetical constructs.
In sum, it makes little sense to distinguish hypothetical constructs and intervening
variables on ontological grounds, as proposed by MacCorquodale and Meehl. A given
hypothetical construct has meaning quite independently of whether or not is does in fact denote
something real, and specifying its meaning is, as with all other concepts, a conceptual matter.
Nothing about the class of entities, real or not, thought to be denoted by hypothetical constructs
makes them semantically special. We contend that meaning problems do not stem from the
ontology or observability of referents and that the intervening variable-hypothetical construct
distinction is based on epistemological constraints as opposed to differences in our ability to
specify the meanings of the two different classes of concept.
Conclusion
The ‘hypothetical construct’ concept came to prominence in psychology during a time
when philosophical shifts away from logical positivism and strict behaviorism towards empirical
realism, neobehaviorism, and cognitive and neuropsychological investigations necessitated a
vocabulary to distinguish various concepts based on their role in theorized relationships. Early
debates surrounding the ‘hypothetical construct’ concept and its distinction from intervening
variable constructs raised a number of issues that remained unresolved and are reflected in the
multitude of senses of ‘hypotheticality’ that are evident in psychological reference works and
textbooks. We attribute this disparity at least in part to some of the definitional, ontological, and
epistemological philosophical issues that underlie the various conceptualizations of this concept.
30
In terms of definitional matters, entities and the concepts that denote them are often conflated
due in part to the widespread embracing of the ACL as well as confusions concerning the
“ontology” of concepts more generally. As regards ontology, it is important to distinguish
between the hypotheticality of concepts, which is trivial, and hypotheticality of theoretical
entities, which is not. With respect to epistemology, the different roles of operationalization may
further contribute to a failure to distinguish epistemological issues related to the entities denoted
by hypothetical constructs from meaning issues as they bear on the concept itself. The
‘hypothetical construct’, which has provided a foundation for the currently employed ‘construct,
is a psychological concept with a rich historical, conceptual, and philosophical foundation that,
once elaborated, can aid in clarifying the implications of its employment, and that of its
successor, in psychological discourse.
31
References
Adcock, C. (1955). Hypothetical constructs. Australian Journal of Psychology, 7, 77-81.
Baker, G.P., & Hacker, P.M.S. (1982). The grammar of psychology: Wittgenstein’s Bemerkugen
über die Philosophie der Psychologie. Language and Communication, 2(3), 227–244.
Baker, G.P., & Hacker, P.M.S. (2005). Wittgenstein: Understanding and meaning. Oxford:
Blackwell.
Bergmann, G., & Spence, K.W. (1941). Operationism and theory in psychology. Psychological
Review, 48(1), 1-14.
Chaplin, J.P. (1968). Dictionary of Psychology. New York: Dell Publishing Co. Inc.
Colman, A.M. (2006). A dictionary of psychology (2nd Edition). Oxford: Oxford University
Press.
Crocker, L., & Algina, J. (1986). Introduction to classical and modern test theory. Fort Worth:
Harcourt Brace Javanovich.
Cronbach, L., & Meehl, P. (1955). Construct validity in psychological tests. Psychological
Bulletin, 52(4), 281-302.
Drever, J. (1952). A Dictionary of Psychology. Middlesex, England: Penguin.
English, H.B., & English, A.C. (1958). A comprehensive dictionary of psychological and
psychoanalytic terms: A guide to usage. New York: David McKay Company.
Eysenck, H.J., Wurzburg, W.A., & Berne, R.M. (Eds.) (1972). Encyclopedia of Psychology. Vol.
2. New York: Herder and Herder.
Feigl, H. (1950). Existential hypotheses: realistic versus phenomenalistic interpretations.
Philosophy of Science, 17, 36-62.
32
Ginsberg, A. (1954). Hypothetical constructs and intervening variables. Psychological Review,
61(2), 119-131.
Helmstadter, G.C. (1964). Principles of Psychological Measurement. New York: Appleton-
Century-Crofts.
Hempel, C. G. (1950). Problems and changes in the empiricist criterion of meaning. Revue
Internationale de Philosophie, 41, 41-63.
Hempel, C. G. (1965). Aspects of Scientific Explanation and other Essays in the Philosophy of
Science. New York: Free Press.
Hilgard, E. (1958). Intervening variables, hypothetical constructs, parameters, and constants.
American Journal of Psychology, 71, 238-246.
Hull, C. (1943). Principles of behavior: an introduction to behavior theory. Oxford England:
Appleton-Century.
Krech, D. (1949). Notes toward a psychological theory. Journal of Personality, 18, 66-87.
Krech, D. (1950). Dynamic systems, psychological fields, and hypothetical constructs.
Psychological Review, 57(5), 283-290.
Lindzey, G. (1953). Hypothetical constructs, conventional constructs, and the use of
physiological data in psychological theory. Psychiatry: Journal for the Study of
Interpersonal Processes, 16, 27-33.
Loomba, R.M. (1943). Sensation, perception and immediate experience. Indian Journal of
Psychology, 18, 125-128.
MacCorquodale, K., & Meehl, P.E. (1948). On a distinction between hypothetical constructs and
intervening variables. Psychological Review, 55(2), 95-107.
33
Maraun, M.D., Slaney, K.L., & Gabriel, S.M. (2009). The Augustinian methodological family of
psychology. New Ideas in Psychology, 27(2), 148-162.
Marx, M. (1951). Intervening variable or hypothetical construct?. Psychological Review, 58(4),
235-247.
Moore, J.M. (1998). On behaviorism, theories, and hypothetical constructs. The Journal of Mind
and Behavior, 19(2), 215-242.
Orton, R. (1987). The foundations of construct validity: Towards an update. Journal of Research
and Development in Education, 21(1), 23-35.
Plutchik, R. (1954). Further remarks on the hypothetical construct. Journal of Psychology:
Interdisciplinary and Applied, 37, 59-64.
Reber, A.S. (1985). The Penguin dictionary of psychology. Middlesex, England: Viking.
Reichenbach, H. (1938). Experience and prediction. Chicago: University of Chicago Press
Richards, G. (2009). Psychology: The key concepts. London: Routledge.
Russell, B. (1929). Sense-data and physics. In B. Russell (Ed.), Mysticism and logic. New York:
W.W. Norton & Co.
Statt, D.A. (1998). The concise dictionary of psychology (3rd. edition). London: Routledge.
Thorndike, R.L, & Hagen, E. (1969). Measurement and evaluation in education and psychology
(3rd edition). New York: John Wiley and Sons.
Tolman, E.C. (1938). The determiners of behavior at a choice point. Psychological Review, 45,
1-41.
Wittgenstein, L. 1953. Philosophical investigations. (Anscombe, G.E.M., trans.). Oxford: Basil
Blackwell.
34
Author Note
This study was supported in part by a Simon Fraser University, Vice President Research -
Social Sciences and Humanities Research Council of Canada 4a Grant awarded to the second
author.
A version of this paper was presented at the 2nd Biennial Midwinter Meeting of the
Society for Theoretical and Philosophical Psychology in Miami, Florida, February 2010.