Full Terms & Conditions of access and use can be found at
Health Psychology Review
ISSN: 1743-7199 (Print) 1743-7202 (Online) Journal homepage: https://www.tandfonline.com/loi/rhpr20
The reasoned actions of an espresso machine: a
comment on Peters and Crutzen (2017)
Stefan L. K. Gruijters
To cite this article: Stefan L. K. Gruijters (2017) The reasoned actions of an espresso machine:
a comment on Peters and Crutzen (2017), Health Psychology Review, 11:2, 125-129, DOI:
To link to this article: https://doi.org/10.1080/17437199.2017.1306716
Accepted author version posted online: 13
Published online: 01 May 2017.
Submit your article to this journal
Article views: 117
View Crossmark data
Citing articles: 3 View citing articles
The reasoned actions of an espresso machine: a comment on
Peters and Crutzen (2017)
Stefan L. K. Gruijters
Department of Work and Social Psychology, Maastricht University, Maastricht, Netherlands
ARTICLE HISTORY Received 6 February 2017; Accepted 19 February 2017
In ‘Pragmatic nihilism: How a theory of nothing can help health psychology progress’, Peters and
Crutzen (2017) discuss a question relevant to many in psychology: What are we actually measuring?
According to Peters and Crutzen (hereafter, P&C), determinants are merely metaphors that allow pre-
diction of behaviour, but ‘cannot be pinpointed physically in people’sbrain’(p. 3), i.e., have physical
form, substance, or a location. Pragmatic nihilism arguably promotes a variant of behaviourism, one
allowing ‘black-box’metaphors but not requiring measurement of its content. P&C argue that ‘the
assumption that the variables hypothesised by theories in fact exist as more than flexibly employable
metaphors is not necessary for theories to be useful’(p. 3). For all practical intents and purposes, it then
seems, a theory of everything for health psychology is unnecessary, unpractical, and perhaps utopic.
Pragmatic nihilism raises several issues that carry undiscussed consequences. Briefly, pragmatic
nihilism faces a problem similar to what philosophers of mind have dubbed a problem of mental cau-
sation (e.g., Baker, 1993; Bennett, 2007; Kim, 2007): How can a mental state cause behaviour, if this
mental state is itself not physical (e.g., my desire for coffee causes me to walk over to the espresso
machine)? Accepting that determinants are metaphors describing non-existing states in the ‘black-
box’implies that these variables are incapable of causally explaining behaviour. Theories in the
business of predicting behaviour by describing variables that cannot be reduced to physical mech-
anisms do not explain anything. It could even be argued that this disqualifies them as scientific the-
ories, and rather puts them closer to magic.
If generating predictions were sufficient to qualify as a
scientific theory, then on a good month, astrology also fits the bill (but see Thagard, 1978). I suggest
modifications to pragmatic nihilism that might solve this problem of mental causation; granting
determinants more theoretical status than metaphors.
Losing causation inside the espresso machine
When requested, an espresso machine will usually make a steaming cup of coffee. Sometimes,
however, it does not. Using a Reasoned Action Approach, I could predict its behaviour by attributing
normative, control, and behavioural beliefs to the machine (Fishbein & Ajzen, 2010). When ‘start’is
pressed, the machine perceives a desire for coffee in its environment. It can be assumed that the
machine has a favourable attitude towards coffee making, since most of the time this perception
seems a sufficient reason for it to do so. This single determinant, however, does not perfectly
predict the machine’s behaviour; its coffee-making behaviour hinges on more. First, the machine’s
behaviour correlates with my refrigerator’s behaviour: it will only make coffee when the refrigerator
is also keeping the milk at proper temperature. Second, it will not make coffee when it has the belief
that it is physically unable to do so (e.g., lack of water in the reservoir); it will even communicate this
lack of behavioural control by flashing some lights.
© 2017 Informa UK Limited, trading as Taylor & Francis Group
CONTACT Stefan Gruijters firstname.lastname@example.org
HEALTH PSYCHOLOGY REVIEW, 2017
VOL. 11, NO. 2, 125–129
Since it helps me to predict its coffee-making behaviour on a daily basis, I consider the Reasoned
Action Approach to be highly beneficial in the morning. The merits of this ‘intentional strategy’,
is described in the works of the philosopher Daniel Dennett (e.g., 1971,1988,1995,1997), applies not
only to coffee machines, but to animals and people too. This strategy of behaviour prediction, in its
most basic mode, could make use of a Desire–Belief–Action principle (Kim, 1989,2007). According to
this principle, a desire and a corresponding belief about how to secure that desire are sufficient to
explain behaviour. It has the logical form: If person X desires Y (and believes that doing A will
secure Y) then person X will do A.
Does it matter whether something real corresponds to the mental states ascribed to an object? In
line with Dennett, P&C are pragmatic: if the intentional strategy generates accurate predictions, then
it is appropriate to use the strategy. As Dennett (1997) argued, whether we are talking about the
behaviour of trees, animals, machines, or humans, the realism of determinants is not assumed and
is not relevant for the success of the strategy. This strategy is often pragmatic, because one alterna-
tive strategy would require us to unravel all physical causal processes involved in generating the
behaviour. One would need to predict the espresso machine’s behaviour by opening it up and
fully examining its wiring; this would make for a long wait (cf. neuroscience).
Health behaviour theories (HBT) embrace this intentional strategy by formalising and expanding it
into models that effectively predict behaviour (e.g., Ajzen, 1991; Fishbein & Ajzen, 2010; Ryan & Deci,
2000; Stretcher & Rosenstock, 1997). There can be little doubt that infection screening and other
health behaviours have been far better predicted by the intentional strategy than its physical
counterpart (e.g., Gruijters, Tybur, Ruiter, & Massar, 2016), despite the metaphors described by
HBT. However, there is a difference between behavioural prediction and explanation, one that is
made clear by the espresso machine example. The causal explanation why the espresso machine
is normatively influenced by the refrigerator is that they draw from the same power source. When
my refrigerator does not function, neither will the espresso machine. For predictive purposes, it suf-
fices to attribute the espresso machine normative beliefs about conduct, but this does not get us any-
where near the causal explanation of this relationship. Metaphors cannot serve causal roles in models,
since causation boils down to the interplay of two ‘things’that need to have a size, location, matter,
and other such physical properties. Determinants, according to pragmatic nihilism, do not possess
these qualities because they do not refer to discrete psychological states. Yet HBT explicitly (and
often implicitly), rely on the notion that they do.
One example of researchers’consistent reliance on the causal assumption of such ‘metaphors’is in
the types of measurement models commonly used. HBT considers determinants to be so-called
latent variables (see Bollen, 2002; Borsboom, 2005,2008). Because we cannot directly tap into
such variables and assess their states (say, an ‘attitudemeter’), measurement models rely on the
notion that questionnaires instead measure reflections of latent variables. In reflective factor
models, answers to particular items (e.g., ‘How pleasant is behaviour or object X?’)causally reflect
an individual’s position on the latent attitude variable (e.g., Bollen & Bauldry, 2011; Edwards &
Bagozzi, 2000). Following P&C’sclaim that determinants need not exist beyond their operationalisa-
tion, how can self-reported scores causally reflect an individual’s position on a latent determinant?
One conclusion follows, scale unidimensionality cannot exist in a world of pragmatic nihilism.
mow’s(2017) commentary (this issue) converges on a conclusion that follows by extension: opera-
tionalism makes it impossible to differentiate measures in terms of validity, since any measure by
definition perfectly describes a construct that it defines (see also Borsboom, Mellenbergh, & van
Heerden, 2004). This would yield an unwieldy number of incomparable attitude measurement instru-
ments. It could however be the case that one operationalisation (defining construct i) turns out to be
more predictive of certain outcomes than others, but then we end up losing any grasp as to why this
is the case.
Borsboom, Mellenbergh, and van Heerden (2003) conclude on these considerations that an oper-
ationalist view is incompatible with latent variable theory. It also invalidates common methods of
factor analysis (e.g., confirmatory factor analysis) and most indices of reliability (e.g., Peters, 2014)
126 S.L.K. GRUIJTERS
which assume reflective measurement (Bollen & Bauldry, 2011; Edwards & Bagozzi, 2000). Lastly,
researchers also require determinants to play causal roles in experiments. If effect Y does not exist,
it cannot be an effect, and X did not cause anything. What use then are randomised controlled
trials, those using determinants as outcomes, to pragmatic nihilists?
The discussion thus far need not worry P&C, because these consequences do not, in principle,
object to pragmatic nihilism. Perhaps researchers in health psychology simply need to drop causal
assumptions, and forget about reflective models and experimenting on determinants. Alternatively,
aformative measurement model may be suited to P&C’s operationalism because it relies on different
assumptions about construct-indicator relationships
(e.g., Bollen & Bauldry, 2011; Borsboom et al.,
2003; Diamantopoulos, Riefler, & Roth, 2008; Edwards & Bagozzi, 2000; Jarvis, MacKenzie, & Podsakoff,
Putting causation back inside the espresso machine
However, the conundrum imposed by pragmatic nihilism can be avoided with few additions to the
model. While beliefs, attitudes, and perceived norms do not need to physically exist anywhere
(except in the form of utterances), we do need some correspondence between measured constructs
and discrete, in principle discernable, psychological states. For example, when a person says ‘I think
that unsafe sex increases the risk of STI-contraction’, and re-expresses the same belief the next day,
we observe temporal consistency, maybe not of the belief, but at least of a system that is causally
involved in people’s claims to possess belief X. Such a real system can at least in principle be causally
held responsible for our behaviour. In this view, both belief statements and behaviour share a
common cause responsible for determinant-behaviour covariance (see Figure 1). This assumes that
people say ‘A’because they think ‘C’, and that determinant-behaviour correlations arise because
people also tend to do ‘B’because they think ‘C’. The success of the determinant strategy (A predicts
B) may rest on a spurious causal model. Nonetheless, this model gives a causal grounding to HBT’s
Figure 1. A comparison of pragmatic nihilism (Panel A) and the proposed extended model (Panel B). The direction of relationships
is indicated by arrow-heads; non-causal associations are indicated with the dashed lines. Within pragmatic nihilism (Panel A), the
metaphorical nature of attitude and norms implies that items A1–A3 and N1–N3 (e.g., semantic differentials) cannot be causally
related to individuals’position on latent variables; but appear to create the determinant construct. Panel B also assumes a formative
model of determinants, but with some additions. The system that generates belief output (B1–B5), e.g., ‘I think that unsafe sex
increases the risk at contracting STI’, then, forms the basis of the answers to direct measures of attitude (A1–A3) and norms
(N1–N3), e.g., ‘I think it is [important-unimportant] to have safe sex’. The attitude construct then, in line with P&C, is merely a
helpful summary of the underlying belief output but does tell us something about real cognitive processes. Additionally, by
adding a common cause for individuals’belief expressions and behaviour, Panel B explains why these constructs predict health
HEALTH PSYCHOLOGY REVIEW 127
predictions. By contrast, pragmatic nihilism’s commitment to leaving the black-box dark, sheds no
light on why the determinant strategy so effectively predicts behaviour.
The modified model gives determinants more theoretical status than metaphors, since now these
variables can be considered reflections of an underlying belief-generating system. In this model, a
theory of everything for health psychology is certainly possible, but it would need to describe the
functioning of a belief-generating system, instead of the system’s output (beliefs) that constitutes
our measurement of determinants. In line with pragmatic nihilism, then, beliefs and determinants
do not exist in the form of ‘units’or representations, but they can certainly be seen as the behavioural
output of representations and algorithmic processes operating on these representations, the same
ones that are linked to actions. Figuring out the nature of these representations and computations
(and its physical basis) are central tasks of cognitive science and neuroscience, which eventually
need to supply health psychology with a theory on how people come to ‘have’certain beliefs.
While these fields are studying the mechanisms, HBT can keep progressing with a pragmatic nihilistic
approach in the meantime –getting to the coffee a whole lot quicker.
1. Philosophers, at this point, may start to feel a ‘shiver running down their spine’. Agreement with this statement
depends on various philosophical tastes. Thagard (2009), for example, discusses (in an accessibly written article)
some of these issues.
2. The term intentional is not used by Dennett (and other philosophers) in its ordinary language form, referring to
intentions as in ‘planned’. Instead, the term intentional has to do with the notion that mental states such as
‘beliefs’,‘desires’,‘hopes’,or‘attitudes’have content, they are about something.
3. Unidimensionality of measurement implies that multiple indicators are replications of a single latent variable
score. Since for pragmatic nihilists the latent variable equals its operationalisation, it is impossible to formulate
a unidimensional measurement model. Put differently, remove a single item from a measurement model contain-
ing n= 10 items of attitude, and the scale measures a different latent variable because its operationalisation is
different (Borsboom et al., 2003).
4. Formative models do not assume that indicators are ‘effects’of a latent variable, but conversely assume that indi-
cators create the construct. An example of such a formative model applies to socioeconomic status (SES): vari-
ables that measure this construct (such as occupation, income, or education background) are not reflections of
an individual’s position on a latent SES variable (no such thing exists), but rather form the construct (e.g.,
Edwards & Bagozzi, 2000). As Borsboom et al. (2003) explain, this model does not assume that SES exists indepen-
dently of its measurement. The theoretical value of such formative, purely descriptive constructs for theory in
health psychology is, however, far from clear.
I am grateful to Bram Fleuren and Rachel Brown for helpful comments on a draft of this paper.
No potential conflict of interest was reported by the author.
Ajzen, I. (1991). The theory of planned behavior. Organisation Behavior and Human Decision Process,50, 179–211. doi:10.
Baker, L. R. (1993). Metaphysics and mental causation. In J. Heil & A. R. Mele (Eds.), Mental causation (pp. 75–96). New York:
Oxford University Press.
Bennett, K. (2007). Mental causation. Philosophy Compass,2(2), 316–337. doi:10.1111/j.1747-9991.2007.00063.x
Bollen, K. (2002). Latent variables in psychology and the social sciences. Annual Review of Psychology,53, 605–634. doi:10.
Bollen, K., & Bauldry, S. (2011). Three Cs in measurement models: Causal indicators, composite indicators, and covariates.
Psychological Methods,16(3), 265–284. doi:10.1037/a0024448
128 S.L.K. GRUIJTERS
Borsboom, D. (2005). Conceptual issues in contemporary psychometrics. Measuring the mind: Conceptual issues in contem-
porary psychometrics. Cambridge: Cambridge University Press.
Borsboom, D. (2008). Latent variable theory. Measurement: Interdisciplinary Research & Perspective,6(1–2), 25–53. doi:10.
Borsboom, D., Mellenbergh, G. J., & van Heerden, J. (2003). The theoretical status of latent variables. Psychological Review,
110(2), 203–219. doi:10.1037/0033-295X.110.2.203
Borsboom, D., Mellenbergh, G. J., & van Heerden, J. (2004). The concept of validity. Psychological Review,111(4), 1061–
Dennett, D. C. (1971). Intentional systems. The Journal of Philosophy,68(4), 87–106. doi:10.2307/2025382
Dennett, D. C. (1988). Précis of the intentional stance. Behavioral and Brain Sciences,11(3), 495–505. doi:10.1017/
Dennett, D. C. (1995). Do animals have beliefs? In H. Roitblat & J.-A. Meyer (Eds.), Comparative approaches to cognitive
sciences (pp. 111–118). Cambridge: MIT Press.
Dennett, D. C. (1997). True believers: The intentional strategy and why it works. Mind Design,57–79. doi:10.1007/s13398-
Diamantopoulos, A., Riefler, P., & Roth, K. P. (2008). Advancing formative measurement models. Journal of Business
Research,61(12), 1203–1218. doi:10.1016/j.jbusres.2008.01.009
Edwards, J., & Bagozzi, R. (2000). On the nature and direction of relationships between constructs and measures.
Psychological Methods,5(2), 155–174. doi:10.1037/1082-989X.5.2.155
Fishbein, M., & Ajzen, I. (2010). Predicting and changing behaviour: The reasoned action approach.New York, NY: Taylor &
Gruijters, S. L. K., Tybur, J. M., Ruiter, R. A. C., & Massar, K. (2016). Sex, germs, and health: Pathogen-avoidance motives and
health-protective behaviour. Psychology & Health,31(8), 959–975. doi:10.1080/08870446.2016.1161194
Jarvis, C. B., MacKenzie, S. B., & Podsakoff, P. M. (2003). A critical review of construct indicators and measurement model
misspecification in marketing and consumer research. Journal of Consumer Research,30(2), 199–218. doi:10.1086/
Kim, J. (1989). Mechanism, purpose, and explanatory exclusion. Philosophical Perspectives,3,77–108. doi:10.2307/2214264
Kim, J. (2007). Causation and mental causation. In B. P. McLaughlin & J. D. Cohen (Eds.), Contemporary debates in philos-
ophy of mind (pp. 227–242). Oxford: Blackwell.
Peters, G.-J. Y. (2014). The alpha and the omega of scale reliability and validity. The European Health Psychologist,16(2),
Peters, G.-J. 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
Ryan, R. M., & Deci, E. (2000). Self-determination theory and the facilitation of intrinsic motivation, social development,
and well-being. American Psychologist,55,68–78. doi:10.1037/0003-066X.55.1.68
Stretcher, V., & Rosenstock, I. M. (1997). The health belief model. In K. Glanz, F. M. Lewis, & B. K. Rimer (Eds.), Health behav-
iour and health education: Theory, research and practice. San Francisco, CA: Jossey-Bass.
Thagard, P. (1978). Why astrology is a pseudoscience. PSA: Proceedings of the Biennial Meeting of the Philosophy of Science
Thagard, P. (2009). Why cognitive science needs philosophy and vice versa. Topics in Cognitive Science,1(2), 237–254.
Trafimow, D. (2017). Why I am not a fan of pragmatic nihilism. Health Psychology Review,1–5. doi:10.1080/17437199.2017.
HEALTH PSYCHOLOGY REVIEW 129