Content uploaded by Eiko Fried
All content in this area was uploaded by Eiko Fried on Sep 27, 2017
Content may be subject to copyright.
What are psychological constructs? On the nature and statistical
modelling of emotions, intelligence, personality traits and mental
Eiko I. Fried
Department of Psychology, University of Amsterdam, Amsterdam, Netherlands
ARTICLE HISTORY Received 25 February 2017; Accepted 10 March 2017
Many scholars have raised two related questions: what are psychological constructs (PCs) such as
cognitions, emotions, attitudes, personality characteristics and intelligence? And how are they best
modelled statistically? This commentary provides (1) an overview of common theories and statistical
models, (2) connects these two domains and (3) discusses how the recently proposed framework
pragmatic nihilism (Peters & Crutzen, 2017) fits in.
For this overview, I use an inclusive definition of the term ‘psychological construct’that also
encompasses mental disorders, similar to Cronbach and Meehl (1955). This is consistent with
recent efforts such as the research domain criteria (RDoC) that aim to refine such constructs (Cuthbert
& Kozak, 2013), and is relevant given many recent discussions on the nature of psychopathology.
Four common accounts have been put forward: PCs are natural, social, practical or complex kinds.
Natural kinds are unchanging and ahistorical entities that exist whether or not they are recognised as
such. They have intrinsic properties that establish a natural set of kind members, making them the
thing they are. The element gold, for instance, has 79 protons, and everything with 79 protons is
gold; this internal feature is necessary and sufficient to define kind membership. In psychology,
basic or primary emotions such as fear, anger or disgust are often seen as natural kinds (Barrett,
2006). Among others, Ekman and Cordaro (2011) suggested that ‘emotions are discrete [and] can
be distinguished fundamentally from one another’(p. 364). Mental disorders are a second
example of PCs. In recent years, they have been described increasingly as brain disorders (Insel
et al., 2010), and the search for biological markers presupposes that these disorders exist as
natural kinds that can in principle be discovered; that is, the notion of brain disorders assumes a
realist ontology about mental disorders. Personality characteristics have also been conceptualised
as natural kinds by, among others, McCrae and Costa (e.g., McCrae et al., 2004; cf., Mõttus & Allerhand,
2017, for a detailed discussion).
The idea that psychological kinds are socially constructed –socially agreed upon definitions –is more
common in the social sciences. Emotions, personality domains or mental disorders do not carve
nature at its joints: they are produced, not discovered. Berger and Luckmann’s seminal book on the
© 2017 Informa UK Limited, trading as Taylor & Francis Group
CONTACT Eiko I. Fried firstname.lastname@example.org
HEALTH PSYCHOLOGY REVIEW, 2017
VOL. 11, NO. 2, 130–134
topic states that all knowledge, even the most basic insight, is derived and maintained by social inter-
actions (1966). As an example for PCs, Szasz posited that all mental disorders are socially constructed
(e.g., Szasz, 1961; cf., Hacking, 1999, for a more nuanced perspective).
Instrumentalists such as James, Dewey and Zachar understand PCs as practical kinds, a position
referred to as pragmatic nominalism. From this perspective, emotions (Zachar & Bartlett, 2001), per-
sonality characteristics (Zachar, 2002) or mental disorders (Kendler, Zachar, & Craver, 2011) should be
judged in terms of practical scientific success, not whether they correspond to an independent
reality. Above anything, constructs should be useful. Socioeconomic status (SES) hardly represents
a‘real’entity in a metaphysical sense, but provides important insights because it predicts adverse
social and health outcomes such as injury, poverty, morbidity and mortality. Instrumentalists argue
that unlike social kinds, practical kinds are not just ‘made up’: the goal is to identify scientifically
useful categories. Many PCs such as emotions (Zachar & Bartlett, 2001) or mental disorders (Zachar
& Kendler, 2007) have been discussed as practical kinds.
Boyd (1991,1999) understood biological species as homeostatic property clusters (HPC). Biological fea-
tures often occur together in nature because the presence of one property tends to favour the pres-
ence of another, forming clusters. This means that beavers are aggregations of features that reliably
occur together because of underlying causal processes. The same can be argued for depression:
symptoms co-occur because the presence of one likely leads to others. Relationships between prop-
erties are probabilistic and not deterministic, and imperfect aggregations of properties are common:
not all beavers have identical biological features, and not all depressed patients have the same symp-
toms. In this aspect, HPCs are different from natural kinds (which have necessary and sufficient fea-
tures), but like natural kinds, they are non-arbitrary. Recent work suggests that a wide range of PCs
may best be understood as HPCs: mental disorders (Borsboom, 2017; Kendler et al., 2011), personality
characteristics (Mõttus & Allerhand, 2017), attitudes (Dalege et al., 2016) and intelligence (van der
Maas et al., 2006; van der Maas, Kan, Marsman, & Stevenson, 2017).
The question what a PC is guides the type of statistical model that is appropriate. Three common
models are described, along with their relationships to psychological kinds. Since constructs in
Figure 1. Schematic visualization of three types of statistical models for psychological constructs (PC). Left: reflective model where
the latent variable (thick border) ‘PC’is the common cause for the 10 observed indicators 1–10 (thin borders). Centre: formative
model where the latent variable is constructed from the indicators. Right: network model where the co-occurrence of all observed
items is due to causal processes; there is no latent variable, and self-loops indicate that items cause each other over time.
HEALTH PSYCHOLOGY REVIEW 131
psychology are usually unobserved, they are referred to as latent variables in the statistical literature.
The three models are summarised in Figure 1.
Conceptualising a PC as a natural kind implies a clear causal model: the latent variable causes a set of
observable indicators, and a person’s position on the latent variable can be inferred by measuring
these indicators (Figure 1, left). To assess mathematical intelligence, for instance, a battery of math
items (e.g., ‘what is 4 + 4’) can be used. For extraversion and schizophrenia, researchers can
measure extraversion items (e.g., ‘I like going to parties’) and schizophrenia symptoms (e.g., ‘I hear
voices’), respectively. Statistically, such models are called reflective latent variable models, because
the items reflect the manifestation of an underlying PC (Edwards & Bagozzi, 2000). While the use
of reflective models follows from a natural kind perspective, not all reflective models represent
natural kinds. Being rich can be understood to be socially constructed: some societies attribute mon-
etary value to pieces of paper or digits on bank accounts, while others attribute value to oddly shaped
wooden sticks. It is possible to infer if someone is affluent (the latent variable) via observable indi-
cators such as the quality of clothing, the frequency of vacations, or by counting how many oddly
shaped sticks a person carries.
The formative model is the causal opposite (Figure 1, centre): indicators determine the latent variable
(Edwards & Bagozzi, 2000). I used SES as an example for a practical kind, and it fits the bill of a com-
posite variable. SES is constructed via indicators such as occupation, education and income: if a
person’s income increases, so does the SES, but it is not possible to increase a person’s income by
changing the latent variable. The opposite holds for reflective models: changing mathematical intel-
ligence in the brain (if possible) would change a person’s performance on math questions, but inter-
fering with a person’s ability to do well on a test (e.g., lack of sleep) does not diminish the person’s
actual mathematical ability. Note that such interventions have been proposed as a way to determine
whether a PC is ‘real’(Hacking, 1983): if researchers feel comfortable manipulating it in an experiment
to investigate something less known, it can be considered real.
From the perspective of HPC, extraversion items do not co-occur because of an underlying PC –they
are related because they influence each other (Mõttus & Allerhand, 2017)(Figure 1, right). Likewise,
mental disorders such as schizophrenia or psychosis can be conceptualised as clusters of symptoms
with causal interactions and vicious circles that can form stable systems (Borsboom, 2017; Cramer,
Waldorp, van der Maas, & Borsboom, 2010; Fried et al., 2016), rather than as passive indicators of a
brain disorder. Instead of modelling PCs as a reflective or formative latent variable, network
models for cross-sectional (Epskamp, Borsboom, & Fried, 2017) and longitudinal (Bringmann et al.,
2013) data have been developed recently that allow for modelling complex systems of observable
items. It has also been suggested that the environment plays an important role in such networks,
and that it may lead to stable networks (Mõttus & Allerhand, 2017; van der Maas et al., 2017).
A major shortcoming of the empirical literature on PCs is the lack of clarity regarding how researchers
understand what they study. For personality traits and mental disorders, for instance, researchers pre-
dominantly use reflective latent variable models, but they remain largely silent about what ‘extraver-
sion’or ‘neuroticism’factors are, or what a ‘cognitive depression’factor is with symptoms like
132 E. I. FRIED
worthlessness, hopelessness, guilt and pessimism. Do these latent variables cause their indicators,
and what is their ontology?
The question what psychological kinds are is also at the core of Peters and Crutzen’s(2017) idea of
pragmatic nihilism (from here on P&C). The authors do not refer to the large corpus of literature on
the topic, and I hope the brief overview above is helpful to embed their ideas into prior work.
The first tenet of pragmatic nihilism is that PCs are ‘useful metaphors in understanding and mod-
elling the world’, but need not actually exist. This resembles a practical kinds view: pragmatic nomin-
alists have long argued that ‘realness’does not add anything relevant to a construct (Fine, 1984), and
that constructs should be, above all, useful. Second, pragmatic nihilism pertains to definition,
measurement and operationalisation of constructs: how researchers define them determines how
they should be measured. This calls for attention to operationalisation: if PCs are not necessarily
real, and the goal is to create useful metaphors rather than to discover truths, the operationalisation
of a construct defines it. Depression rating scales, for instance, differ considerably in symptom content
(Fried, 2017), and P&C would likely argue that these different scales measure somewhat different
‘depressions’. This focus seems to call for formative instead of reflective latent variable models,
which contrasts with the widespread use of exploratory and confirmatory (i.e., reflective) factor
models in the psychological literature. Third, P&C argue that ‘cognition and emotion […] exist as
emergent properties in a distributed network of neurons’, but they ‘cannot be pinpointed physically
in people’s brains’. This raises the question whether the authors understand PCs as identical to their
neurological realisations, or whether they supervene on neurological processes (cf., Kievit et al., 2011).
My main concern is this: there is nothing better than a true theory. Discovering truths about the
universe, such as the table of elements, has facilitated –and in many cases enabled –scientific pro-
gress. If there is nothing real about psychological processes, psychology can only ever describe, but
never understand, which may greatly limit prediction and intervention. I’m not quite ready to give up
on the notion that psychology can discover truths, and find the idea that all PCs are made-up dis-
heartening …but that’s what you get messing about with nihilism, I suppose.
I would like to extend my sincerest thanks to Denny Borsboom for comments on a prior version of this paper, and to the
editor for soliciting this commentary.
No potential conflict of interest was reported by the author(s).
This work was supported by European Research Council [ERC Consolidator Grant no. 647209].
Barrett, L. F. (2006). Are emotions natural kinds? Perspectives on Psychological Science,1(1), 28–58.
Berger, P., & Luckmann, T. (1966). The social construction of reality: A treatise in the sociology of knowledge. New York City:
Borsboom, D. (2017). A network theory of mental disorders. World Psychiatry,16,5–13.
Boyd, R. (1991). Realism, anti-foundationalism and the enthusiasm for natural kinds. Philosophical Studies,61, 127–148.
Boyd, R. (1999). Homeostasis, species, and higher taxa. In R. A. Wilson (Ed.), Species: New interdisciplinary essays (pp. 141–
185). Cambridge, MA: A Bradford Book/The MIT Press.
Bringmann, L. F., Vissers, N., Wichers, M., Geschwind, N., Kuppens, P., Peeters, F., …Tuerlinckx, F. (2013). A network
approach to psychopathology: New insights into clinical longitudinal data. PloS One,8(4), e60188. doi:10.1371/
HEALTH PSYCHOLOGY REVIEW 133
Cramer, A. O. J., Waldorp, L. J., van der Maas, H. L. J., & Borsboom, D. (2010). Comorbidity: A network perspective. The
Behavioral and Brain Sciences,33(2–3), 137–50. doi:10.1017/S0140525X09991567
Cronbach, L. J., & Meehl, P. E. (1955). Construct validity in psychological tests. Psychological Bulletin,52(4), 281–302.
Cuthbert, B. N., & Kozak, M. J. (2013). Constructing constructs for psychopathology: The NIMH research domain criteria.
Journal of Abnormal Psychology,122(3), 928–937. doi:10.1037/a0034028
Dalege, J., Borsboom, D., van Harreveld, F., van den Berg, H., Conner, M., & van der Maas, H. L. J. (2016). Toward a formal-
ized account of attitudes: The causal attitude network (CAN) model. Psychological Review,123(1), 2–22. doi:10.1037/
Edwards, J. R., & Bagozzi, R. P. (2000). On the nature and direction of relationships between constructs and measures.
Psychological Methods,5(2), 155–174.
Ekman, P., & Cordaro, D. (2011). What is meant by calling emotions basic. Emotion Review,3(4), 364–370. doi:10.1177/
Epskamp, S., Borsboom, D., & Fried, E. I. (2017). Estimating psychological networks and their stability: A tutorial paper.
Behavior Research Methods,1–34. doi:10.3758/s13428-017-0862-1
Fine, A. (1984). The natural ontological attitude. In J. Leplin (Ed.), Scientific realism (pp. 83–107). Berkeley, CA: University of
Fried, E. I. (2017). The 52 symptoms of major depression: Lack of content overlap among seven common depression
scales. Journal of Affective Disorders,208, 191–197. doi:10.1016/j.jad.2016.10.019
Fried, E. I., van Borkulo, C. D., Cramer, A. O. J., Boschloo, L., Schoevers, R. A., & Borsboom, D. (2016). Mental disorders as
networks of problems: A review of recent insights. Social Psychiatry and Psychiatric Epidemiology,1,1–32. doi:10.1007/
Hacking, I. (1983). Representing and intervening: Introductory topics in the philosophy of natural science. Cambridge:
Cambridge University Press.
Hacking, I. (1999). The social construction of what? Cambridge: Harvard University Press.
Insel, T. R., Cuthbert, B., Garvey, M., Heinssen, R., Pine, D. S., Quinn, K., …Wang, P. (2010). Research domain criteria (RDoC):
Toward a new classification framework for research on mental disorders. The American Journal of Psychiatry,167(July),
Kendler, K. S., Zachar, P., & Craver, C. (2011). What kinds of things are psychiatric disorders? Psychological Medicine,41(6),
Kievit, R., Romeijn, J.-W., Waldorp, L. J., Wicherts, J. M., Scholte, H. S., & Borsboom, D. (2011). Mind the gap: A psychometric
approach to the reduction problem. Psychological Inquiry,22(2), 67–87. doi:10.1080/1047840X.2011.550181
van der Maas, H. L. J., Dolan, C. V., Grasman, R. P. P. P., Wicherts, J. M., Huizenga, H. M., & Raijmakers, M. E. J. (2006). A
dynamical model of general intelligence: The positive manifold of intelligence by mutualism. Psychological Review,
113(4), 842–61. doi:10.1037/0033-295X.113.4.842
van der Maas, H. L. J., Kan, K., Marsman, M., & Stevenson, C. E. (2017). Network models for cognitive development and
intelligence (January), 1–14. doi:10.20944/preprints201701.0107.v1
McCrae, R. R., Costa, P. T., Martin, T. A., Oryol, V. E., Rukavishnikov, A. A., Senin, I. G., …Urb??nek, T. (2004). Consensual
validation of personality traits across cultures. Journal of Research in Personality,38(2), 179–201. doi:10.1016/S0092-
Mõttus, R., & Allerhand, M. (2017). Why do traits come together? The underlying trait and network approaches. In V.
Zeigler-Hill & T. Shackelford (Eds.), SAGE handbook of personality and individual differences: Volume 1. The science of
personality and individual differences (pp. 1–22). London: SAGE.
Peters, G. Y., & Crutzen, R. (2017). Pragmatic nihilism: How a theory of nothing can help health psychology progress.
Health Psychology Review,16,1–19. doi:10.1080/17437199.2017.1284015
Szasz, T. (1961). The myth of mental illness: Foundations of a theory of personal conduct. New York, NY: Harper & Row.
Zachar, P. (2002). The practical kinds model as a pragmatist theory of classification. Philosophy, Psychiatry, & Psychology,9
(3), 219–227. doi:10.1353/ppp.2003.0051
Zachar, P., & Bartlett, S. (2001). Basic emotions and their biological substrates: A nominalistic interpretation. Consciousness
& Emotion,2, 189–221. Retrieved from http://www.ingentaconnect.com/content/jbp/ce/2001/00000002/00000002/
Zachar, P., & Kendler, K. S. (2007). Psychiatric disorders: A conceptual taxonomy. American Journal of Psychiatry,164(4),
557–565. Retrieved from http://ajp.psychiatryonline.org/article.aspx?articleid=98178&RelatedWidgetArticles=true
134 E. I. FRIED