Should psychiatry abandon “natural
science” based methods?
Introduction
Critical psychiatrists claim that natural science methods
are of little use in studying humans and their problems that form the subject
matter of mental health due to their
ignoring important factors such as intentionality, meaning and being embodied
in social contexts and that social science methods such as hermeneutics are preferred (Middleton & Moncrieff,
2019).
I will argue that natural science methods
include descriptions such as observation of outcomes of interest to provide the
information on causes, complications and treatment responses needed for medical
practice. Natural science techniques even in unquestioned natural sciences such
as physics may also struggle to discover universal laws because of the effect
of multiple interactions, yet we still use them to explain relationships
between observable entities hence this is no reason to disbar them in mental health.
They can be used to establish causations of organic mental disorders. Even for functional mental health problems
where natural science methods are not used to explain causation these problems
are negatively defined using natural science methods. Finally, it may be
possible to use natural science methods to study relationships between social
contexts and mental health problems. Psychiatry should still use natural
science methods.
What is the attitude of critical psychiatry to the use of natural science
techniques?
An article describing critical psychiatry (Middleton
& Moncrieff, 2019) stated one key attitude was the applicability of
scientific methodology in psychiatry. This is quoted in full to avoid any
summarising on my part being misleading about their position.
“Medical knowledge is identified
with the scientific approach that was developed to study the natural world;
systematically investigating assumed-to-be immutable truths by measurement and
manipulation of particular elements in a controlled environment. The
application of this form of knowledge-seeking to the world of human affairs is
referred to as ‘positivism’, an approach that has been criticised for
oversimplifying human affairs. Earlier scholars distinguished between
Geisteswissenschaften, human or moral sciences, and Naturwissenschaften,
natural sciences. The distinction remains a core feature of social science,
where it is argued that the study of human beings is irreducibly different from
the study of the natural world. Human behaviour is intentional, interactive and
inextricable from its social setting. It has meaning, rather than causes, that
can only be discerned by reference to its context.
Therefore, if mental disorders
are to be understood as human reactions rather than as physical diseases, a
positivist perspective cannot provide adequate or comprehensive knowledge.
Different approaches are needed that can study social phenomena appropriately.”
(Middleton & Moncrieff, 2019: 49)
Alternative social science approaches such as
hermeneutics and social constructivism are suggested as alternatives (Middleton & Moncrieff, 2019: 49)
The article summary stated:
“Critical psychiatry does not reject science, but
questions the applicability of positivist research paradigms to the study of
complex human responses, suggesting that other epistemologies may often be more
enlightening.” (p. 53)
An article on the ‘Critical Psychiatry Network’ also
described the views of critical psychiatry (Double, 2019) including their view
on the role of “natural science” methods.
“However, critical psychiatry’s
challenge to reductionism and positivism, including mechanistic psychological
approaches, does create a framework which focuses on the person and has
ethical, therapeutic and political implications for clinical practice. It also
has consequences for psychiatric research, which has become too focused on
speculative neurobiological notions” (p.62)
These articles particularly
the former promulgate the view that “natural science” methods
“systematically
investigating assumed-to-be immutable truths by measurement and manipulation of
particular elements in a controlled environment” (Middleton &
Moncrieff, 2019 p.49).
are not suitable for “the study of human beings” (Middleton
& Moncrieff, 2019 p.49).
Critical psychiatry regards the subject of psychiatric
research and clinical practice – what problems people present with and are seen
by psychiatrists– as “human reactions” or “meaningful responses” (Middleton
& Moncrieff, 2019 p.49). The article by Double places a limit on what is
referred to is mental health problems not clearly linked to somatic disease
such as hypothyroid induced depression or Alzheimer’s disease
“the essential position of
critical psychiatry is that functional mental illness should not be reduced to
brain disease...Functional mental illness is a personal experience that does
not have an underlying brain pathology. No definite biomarkers have been linked
to functional mental illness” (Double, 2019 p. 62).
“Natural science” techniques are described
here as controlling for potentially confounding variables whilst altering the
experimental variable with results being used to produce “immutable laws”. For
the problems that come to psychiatrists this method is regarded as not being
able to provide “adequate or comprehensive knowledge” because they are “human
reactions” and as such are driven by intentions and are meaningful in response
to multiple contexts. Thus, controlling for these contextual factors – regarded
as confounding by the “natural sciences” – in order to examine the effects of a
single experimental factor (even if it is a social factor such as class) cannot
explain why the behaviour happened. There is also a view that meaningful
intentional behaviour belongs to the “space of reasons” and may not conform to
the “realm of laws” (McDowell, 1994) and hence natural science methods cannot
thus provide explanations.
What kinds of information do psychiatrists need for clinical practice?
To address the question of which type of scientific
methodology is best for psychiatry we need to know why they need science.
“The link between psychiatry and
medicine confers legitimacy on psychiatry as a professional enterprise because
its practitioners are seen to hold and exploit expert medical knowledge”
(Middleton & Moncrieff, 2019. P. 48-49).
Science is thus used to gain
information – for critical psychiatry this is to cement psychiatry’s
professional status. Another reason, which may still apply with the previous
reason, is that doctor’s professional role to advise on, coordinate, or deliver
interventions for health improvement should be based on the best possible
evidence gained from science (Shah & Mountain, 2007).
The type of information doctors need for clinical
practice can be ascertained by descriptions of ‘clinical utility’ that is
stated to come from diagnostic constructs. Diagnostic constructs are used as
containers of information – the question of their suitability as carriers of
information in psychiatry is not relevant here but given the centrality of
making a diagnosis is to medical practice, the purported information attached
to a diagnosis is clearly thought to be essential.
Examples of the type of information include
“nontrivial information about prognosis
and likely treatment outcomes, and testable propositions about biological and
social correlates” (Kendell & Jablensky,2003. P.9)
A more exhaustive list of useful information is listed
below
“1. Conceptualizing diagnostic entities
2. Communicating clinical
information to practitioners, patients and their families,
and health care systems administrators
3. Using diagnostic categories and criteria sets
in clinical practice (including for diagnostic interviewing and differential
diagnosis)
4. Choosing effective interventions to improve
clinical outcomes
5. Predicting future clinical management needs”
(First et al, 2004. P.947)
In this list points 2 and 3
relate more to the use of diagnosis in other functions so will be ignored.
Therefore, the types of
information that doctors including psychiatrists require include information on
likely outcomes (prognosis), the effectiveness of various interventions for
this type of problem, what problems may occur in future (complications) and
what possible causative mechanisms may be involved (aetiology). The other
important thing to note is this type of information is in the form of a range
of probabilities not binary information. Even with conditions with almost
certain mortality, it is important to know the likely range of possible
survival time.
What is the best way to acquire this information?
Doctors can use the patient’s history to provide some of
this information in an idiographic fashion. If they have had episodes of low
mood in the past in response to interpersonal stressors then you can predict
that an upcoming family wedding mixing with hated in-laws may trigger low mood;
if they found that a certain type of medication helpful then it may be helpful
in the future. If patient had an abscess and they were allergic to penicillin
in the past then even If the microscopy culture and sensitivity showed
penicillin was the most effective antibiotic you would not prescribe
penicillin. This relies on the patient’s past predicting the future – so long
as the patient has all the relevant experience.
Often though this information is not available – the patient
may not have experienced this type of health problem before to provide guidance
on treatments or prognosis, the patient may not be able to recall accurately
the information, the patient may be at risk of a complication or problem they
have never experienced so cannot be guide in themselves as to the risk of this
problem (they may have never had a heart attack but the doctor may be
interested in their risk of a heart attack or whether vaccination to prevent a
disease is advisable) or they may not have tried a potential treatment so
cannot tell you whether it will be effective or if they experienced any side
effects.
The commonest way in current medical practice to get this
useful information is research on people whose problems resemble the problems
of the patient in the clinic in front of them in some way (Kendell in Shepherd
& Zangwill (1983): 191-198). This type of research may be simple
observation by a clinician that is recorded or remembered in some fashion all
the way up to multi-million pound projects involving many researchers. The
information gained from this research can then be used to give some idea of
probabilistic information of the types described in the previous section.
The natural scientific method is described as a method of
investigation to ascertain laws (Middleton & Moncrieff, 2019) but
scientific methods are used to describe phenomena as well as explain what is
described using laws (Hempel in Sadler et al, 1994: 317-18). Descriptive goals
can be achieved by observation and use of standardised terminology to
facilitate communication, replication and application by others such as doctors
in their clinical practice. For medical practice, important information can be gained
by observation of participants with some commonality of their problems - at
repeated intervals if necessary – of classified events to gain knowledge on prognosis/
complications of mental health problems, effectiveness and side-effects from
treatment.
For this nomothetic approach from research to be useful it
relies on several possible mechanisms. One is that the similarities between
research participants and the patient in the clinic increases the chances of
making successful inferences - the probabilities of recovery or rates of
response to treatment -as they have an increased chance of sharing some
fundamental quality even if it is unknown. Practical kinds that perhaps share
external features such as similarities in symptoms rather than common mechanisms
or causes have a greater degree of uncertainty making inferences from research
to the clinic. Another possibility is that even if the mechanisms are unknown
and the participants are heterogenous there maybe some common elements of
mechanisms (mechanistic property clusters) allowing greater degrees of
certainty. For natural kinds with common mechanisms and/or causes the certainty
is greater still (Kendler et al, 2011). The knowledge gained from research has
a degree of uncertainty, greatest when relying on “practical kinds” but also
influenced by the effect of multiple contextual factors (such as participants
in research differing markedly from patients seen in clinical practise) but
information even with great degrees of uncertainty is still better than no
information at all so long as this degree of uncertainty is taken into account
in decision-making.
If general laws are demonstrated that can offer explanation
of observed descriptive data – such as increase in neurochemical x leads to
increase in symptom y then this can allow a greater degree of explanation and
extrapolation from research participants to the patient in the clinic. With
this higher degree of explanation comes increased predictive ability – it can
lead to increased understanding and explanation of the patient’s situation; an
increased knowledge of the mechanisms can allow increased predictive ability
for complications and prognosis and more rational choices in treatment and development
of more effective treatments in future.
It is important to recognise that there are certain
limitations to using natural science methods in mental health. The information
gained even if only descriptive uses classifications varying from a “black box”
type where we have little explanation to types with a great degree of
explanation. There is always a degree of uncertainty but even in the most
uncertain cases, information with high degrees of uncertainty is better than no
information at all so long as one is aware of the degrees of uncertainty (see
evidence pyramid and subsequent strengths of recommendation).
Do natural science techniques have limits for explaining and establishing
causation in mental health problems?
Middleton and Moncrieff state that human beings cannot be
studied by natural science but later specifies “behaviour” and “human reactions”.
It makes the case that behaviour, thoughts, emotions and so forth cannot be
studied in a natural science way because this ignores context, personal
history, meaningfulness that do not apply to the subjects of natural sciences
such as subatomic particles and animals.
Middleton and Moncrieff describe natural science’s
“immutable truths” derived from “controlled” “manipulation” of “particular
elements”. It is true that scientific practice often tries to generate
observations from experiments aiming to minimise the effects of confounding
variables to clarify the role of the experimental variable. Are these
experiments in natural sciences such as physics or chemistry producing
“immutable truths”? No, they provide results that operate under certain
conditions. For example, determining the state of water as a liquid depends on
certain conditions of temperature and external pressure. Even what we may
regard as relatively stable phenomenon in physics such as the passage of time
are changed by near-relativistic speeds and the laws of physics were different
at the time of the big bang and continued to evolve even during tiny fractions
of a second afterwards.
We cannot expect “immutable truths” very often, if at all,
in the hard sciences. What we get is contextual-influenced observations.
Hacking suggests that whilst proving that theories (these explanatory
underpinning “laws”) are real may be an ultimate goal of scientists but in
practice most scientific work is concerned with manipulations of observable
entities and the relationships between them (Hacking in Boyd et al, 1999:
247-260). So in psychiatry we natural scientific methods may be used to observe
interactions between say giving an antidepressant and effects on depression
without necessarily proving an underpinning law predicting the entire effects
of the antidepressant on speech and behaviour (it might demonstrate a
relationship between a reduced tendency to have less negative thoughts but not
the exact content of the negative thought).
In terms of explanations of human behaviour and speech there
is always going to an effect of culture, history and environment. Delusions in
the 19th Century were often religious due to the importance of religion in
culture; nowadays people often have delusions around technology such as
television or the internet which were not around in the 19th
Century. Now if we posit a biological causation for delusions then a purely
biological model cannot explain this variation – it may say delusions around
available technology have a biological basis but not why the patient talks
about being controlled by wifi now and emissions from natural gas used in
lighting in the 19th Century - that is due to historical contingent
factors as to what technology is available.
Psychiatric symptoms and signs have been described as
‘hybrid objects’ (Markova and Berrios in Zachar et al, 2015) where this model
leaves aside the question of causation. The term ‘objects’ describes
constructs that are used to depict or explain aspects of the world. The
generation of psychiatric symptoms as signs are said to be ‘hybrid’ because
they involve combinations of biological and semantic elements—a biological
element/kernel is contained within two layers of configurating envelopes.
The cause of the biological element is not depicted in this
model. The biological element refers to the brain activity associated with the
symptom or sign. This might be, for example, nervous
activity that is a corollary of the patient’s thoughts.
The two configurating envelopes
describe the semantic and contextual elements that act to configure and modify
the interpretation and expression of the biological element successively. The
first envelope involves individual and sociocultural forces (such as personality traits and
culture) whilst the second envelope consists of interactional forces (e.g.
between doctor and patient or within a broader social context) (Markova and Berrios in Zachar et al, 2015)).The second type of configurating envelope describes interactions
including that between the patient discussing their experiences with somebody
else.
|
C4.P4
|
Thus, any explanation of a patient’s behaviour or speech
will necessarily include contextual social and interpersonal factors. This
means any explanation from laws derived natural scientific methods will also
need to explain these contextual and interpersonal factors in order to claim a
full explanation of the mental health problems expressed in behaviour and
speech. For laws derived from natural
science methods to be able to explain the entire reasons for behaviour they
have to be able to explain these contextual social and interpersonal/
interactional factors.
Even in what seem paradigmatic natural sciences such as
physics it has been argued that laws derived from the uncontested use of
natural science methods fail to explain as much as people think. They often
have certain assumptions such as “all things being equal” which in practice is
hard to achieve to control for all contextual factors absolutely even in
experiments let alone in naturally occurring situations (Cartwright, 1983). Cartwright
puts forward the argument that natural science methods even in physics only
allow us to make predictions about relationships between observable events
rather than hidden laws. Interestingly, Cartwright makes a direct comparison
between the behaviour of particles and people.
“I imagine that natural
objects are much like people in societies. Their behaviour is constrained by
some specific laws and by a handful of general principles but is not determined
in detail, not even statistically. What happens on most occasions is dictated
by no law at all.” (Cartwright, 1983: 49).
Even in physics it is
also hard to establish causality.
“All the counter examples I
know to the claim that causes increase the probability of their effects work in
this same way. In all cases the cause fails to increase the probability of its
effects for the same reason: in the situation described the cause is correlated
with some other causal factor which dominates in its effects. This suggests
that the condition as stated is too simple. A cause must increase the
probability of its effects; but only in situations where such correlations are
absent” (Cartwright 1983: 25).
Given these problems with uncontested use of natural science
methods in paradigmatic natural sciences such as physics in being able to
establish laws that can explain everything and to establish causation, it would
be reasonable to infer that for the experiences/behaviours patients have that
are disvalued that result in them receiving psychiatric care may not be able to
have laws discovered that predict and explain them totally or establish
causation using natural science methods. There may be at least a gap where
different types of methods may be useful, a “space of reasons” perhaps where
the social science methods advocated by Moncrieff and Middleton may be
advantageous.
Can we really do without natural science methods at all in psychiatry?
There may be certain situations or certain areas when
natural science methods may fail to explain fully the speech, thoughts or
behaviour of people with mental health problems. It is possible that in all
patients the contextual broader sociocultural factors and interpersonal factors
that affect speech/ thought/ behaviour (as described by Markova and Berrios in
Zachar et al, 2015) are the “space of reasons” requiring different methods than
the natural science methods. On the other hand, it is well recognised there are
cases when mental health problems are at least in part “the psychological
consequences of cerebral disorder” such as general paralysis of the insane,
hypothyroid induced depression or the dementias (Lishman, 1997). It is untenable
that in cases of what would be regarded as ‘organic mental disorders’ that
natural science techniques would be of no use. As Berrios & Markova
suggests yes the exact content of speech and behaviour is affected by broader
contextual factors but surely the presence of neurosyphilis is germane to the
patient talking of their fantastical schemes or an underactive thyroid in a
patient with depressed mood and paranoid psychosis? Natural science surely has
a role to play when clear-cut biological factors seem to be strongly related to
the mental health problems – not only for judging aetiology but also prognosis
and treatment even if they cannot fully explain all the details of someone’s
speech or behaviour.
Double suggests that ‘critical psychiatry’ is only referring
to ‘functional’ mental health problems. But what is a ‘functional’ mental
health problem? It is a negatively defined state – one where there is an
absence of a biological cause that has been demonstrated in the research to
cause the defined problem. How do know what biological causes cause these
problems? By using natural scientific methods to demonstrate an absence of a
demonstrated biological cause. So even if we adopt the position of only using
the social science methods suggested for functional mental health problems
these need to be defined by using natural science methods to confirm they are
functional mental health problems. Furthermore, one cannot state that with
further advances in scientific techniques we will not identify in future biological
causes for some cases of mental health problems that are not apparent now. We
already have an example in antiNMDA encephalitis that in the time of Szasz he
would describe as a ‘problem in living’ as the ability to detect the relevant
antibodies was unavailable.
It may be possible to use natural science methods to study
the effects of social contextual and interpersonal factors on a variety of
outcomes such as the relationship of employment to ethnic minority status. In
line with Hacking’s view, Brown and Harris used a standardised interview with
reasonable reliability – the Life Events and Difficulties Schedule – to measure
the relationships between described observable entities: social contexts, life events and interpersonal
difficulties and the risk of developing depression in women (Brown &
Harris, 1978). This demonstrated that social contexts and interpersonal factors
(three or more life events, lack of social confidants) and their effects on
depression (an increased risk in women) could be measured and a relationship
demonstrated even though the full meanings and a satisfactorily predictive law
of all resultant phenomena were not demonstrated.
Conclusion
Critical psychiatry states that natural science methods
should be replaced by social science methods such as hermeneutics because such
techniques cannot explain the speech and behaviour of people with mental health
problems as this requires understanding of social contexts and interpersonal
interactions. The critical psychiatry view ignores that natural science has
descriptive and explanatory aims. Descriptive methods can be used to gain
information on information that is important to medical practice including
psychiatry such as probabilistic information on prognosis, complications, and
treatment responses.
Natural science methods can struggle to establish laws that
fully explain behaviours of bodies and particles in physics due to the
difficulty of fully accounting for contexts and this may apply to establishing
laws that can fully explain speech and behaviour due to the effect of social
contexts. Natural science methods can
have some explanatory value in ‘organic’ mental disorders. Even if we accept
that natural science methods cannot be used to explain speech and behaviour at
all in ‘functional’ mental health
problems then we still need natural science methods to negatively define these
‘functional’ mental health problems and in the future scientific advances may
allow us to identify biological cases of what seemed to ‘functional’ mental
health problems for which natural
scientific methods could be applied to. Natural science methods can still be
used to establish relationships between reliably described and observed
contextual and interpersonal factors even if the full meanings cannot be
described or explanatory laws established.
In conclusion, natural science methods should not be
abandoned by psychiatry.
Bibliography
Boyd, R., Gaspar, P. and Trout, J.D., 1999. The
Philosophy of Science. Cambridge, MA: MIT Press
Brown
G.W. and Harris T. (1978). Social Origins
of Depression: A Study of Psychiatric Disorder in Women. London; Tavistock
Cartwright, N., 1983. How the laws of physics lie.
Clarendon.
Double, D.B., 2019. Twenty
years of the Critical Psychiatry Network. The British Journal of
Psychiatry, 214(2), pp.61-62.
Fine NOA
First, M.B., Pincus, H.A.,
Levine, J.B., Williams, J.B., Ustun, B. and Peele, R., 2004. Clinical utility
as a criterion for revising psychiatric diagnoses. American Journal of
Psychiatry, 161(6), pp.946-954.
Hacking
Sadler, J.Z., Wiggins,
O.P. and Schwartz, M.A., 1994. Philosophical perspectives on
psychiatric diagnostic classification. Johns Hopkins University Press.
Kendell, R. and Jablensky, A., 2003. Distinguishing
between the validity and utility of psychiatric diagnoses. American journal
of psychiatry, 160(1), pp.4-12.
Kendler, K.S., Zachar, P. and Craver, C., 2011. What
kinds of things are psychiatric disorders? Psychological medicine, 41(6),
pp.1143-1150.
Lishman
W.A. (1997). Organic Psychiatry. The
Psychological Consequences of Cerebral Disorder, 3rd edn. Oxford: Blackwell
Science.
McDowell, J., 1994. Mind
and world. Harvard University Press.
Middleton, H. and
Moncrieff, J., 2019. Critical psychiatry: a brief overview. BJPsych
Advances, 25(1), pp.47-54.
Shah, P. and Mountain, D., 2007. The medical model is
dead–long live the medical model. The British Journal of Psychiatry, 191(5),
pp.375-377.
Shepherd,
M. & Zangwill. O.L. (eds) (1983). Handbook
of Psychiatry 1: General Psychopathology. Cambridge: Cambridge University
Press.
