A quantum-like model of homeopathy clinical
trials: importance of in situ randomization
91, Grande Rue, 92310 S? evres, France
clinical pharmacology. However, for many practitioners of homeopathy, blind RCTs
are an inadequate research tool for testing complex therapies such as homeopathy.
Methods:Classical probabilities used in biological sciences and in medicine are only a
special case of the generalized theory of probability used in quantum physics. I describe
homeopathy trials using a quantum-like statistical model, a model inspired by quantum
physics and taking into consideration superposition of states, non-commuting observ-
ables, probability interferences, contextuality, etc.
Results:The negative effect of blinding on success of homeopathy trials and the
‘smearing effect’ (‘specific’ effects of homeopathy medicine occurring in the placebo
group) are described by quantum-like probabilities without supplementary ad hoc hy-
potheses. The difference of positive outcome rates between placebo and homeopathy
groups frequently vanish in centralized blind trials. The model proposed here suggests
a way to circumvent such problems in masked homeopathy trials by incorporating
in situ randomization/unblinding.
Conclusion:In this quantum-like model of homeopathy clinical trials, success in open-
This model suggests that significant differences between placebo and homeopathy in
blind RCTs would be found more frequently if in situ randomization/unblinding was
Homeopathy (2013) 102, 106e113.
The randomized controlled trial (RCT) is the ‘gold standard’ of modern
Keywords: Quantum probabilities; Entanglement; Contextuality; Non-local
“Inexplicable observations are not always signs of the
Homeopathic remedies are considered by many scien-
tists and physicians as implausible and ineffective. At best
they consider that homeopathy works, but only because of
the consultation.2,3For many detractors of homeopathy,
the final word has been spoken with the study of Shang
et al.4e6The authors of this study reported a comparison
of randomized placebo-controlled trials of homeopathy
and allopathy; they concluded that e despite comparable
cebo effects. In contrast with allopathy, blinding of trials of
homeopathic drugs strongly decreased the probability of
success compared to open-label setting.
Although this study was heavily criticized,7e9its
conclusions were not completely unexpected since the
main reason for rejection of homeopathy is the difficulty
for homeopathic remedies to pass successfully the test of
blind randomized controlled trials (RCTs), which is the
‘gold standard’ of modern clinical pharmacology. Some
supporters of homeopathy argue that blind RCTs are an
inadequate research tool for testing complex therapies
*Correspondence: Francis Beauvais, 91, Grande Rue, 92310
S? evres, France.
Received 16 November 2012; revised 28 January 2013; accepted
20 February 2013
Homeopathy (2013) 102, 106e113
? 2013 The Faculty of Homeopathy
http://dx.doi.org/10.1016/j.homp.2013.02.006, available online at http://www.sciencedirect.com
such as homeopathy.10,11In the present article, I propose a
possible way to increase the chance of demonstrating a
but there is a debate among them about the way in which it
works.12For many homeopaths, there is ‘something’ in the
homeopathic medicine, which explains the patient’s
response. In other words, there is a specific factor or cause
located in the water or in the granules that acts locally as
does a pharmacological compound. The ‘local’ explana-
veniste’s experiments e is supported by some laboratory
investigations.13e17In these experiments, the states of
biological systems were significantly different in the
presence of highly diluted pharmacological compounds
or corresponding controls. Shaking between each dilution
appeared to be necessary for ‘memory’ while some
physico-chemical treatmentssuch as heating were reported
information storage in high dilutions remains to be
convincingly demonstrated by physical methods.17
Besides the ‘classical’ hypothesis of local causality,
other authors have more recently proposed that the cause
of homeopathy effectiveness is not specifically located in
water samples or remedy. Instead they used concepts
derived from quantum physics, such as non-locality and
entanglement.18e21Entanglement is a central concept of
quantum theory: two quantum systems isolated from
environment become entangled after interaction and as a
consequence they share a single quantum state. This
means that when a measurement occurs, the respective
outcomes of the two quantum systems are correlated.
Entanglement is also expected in systems that need both
local and global descriptions; if these descriptions are
complementary, then theoretical models predict non-local
correlations between the elements of the system.22,23
The authors who apply quantum concepts to homeopa-
thy differ in what is entangled among practitioner, patient
and medicine.12Moreover, quantum physics describes
particles and atoms and quantum phenomena are supposed
to vanish in macroscopic world due to the decoherence
mechanism, which is related to the numerous interactions
of any macroscopic object with its environment. To over-
come this obstacle, Walach proposed applying a ‘general-
ized’ version of quantum theory to homeopathy, which
makes the theory applicable in more general contexts
than the original quantum physics.22,24
In the present article, I present a simple model for global
description of homeopathy trials. This model describes the
cognitive states of practitioner and patient using notions
from quantum physics such as superposition and non-
commuting observables. Operations are non-commutative
if changing the order of operations does not change the
result. For instance washing and drying clothes are not
commutative, the order in which the operations are carried
is commutative, the order in which they are put on makes
no difference to the outcome.
This modeling is in the spirit of an emerging discipline
named ‘quantum cognition’ at the frontiers of artificial in-
research areas, which have in common the description of
cognition mechanisms and information processing in the
problems that were unresolved in a classical frame.25A
quantum-like formalism has thus been applied to human
memory, information retrieval, decision making, opinion
forming, personality psychology, etc.26e30This new
approach does not rest on the hypothesis that there is
something quantum mechanical about the physical brain.
The quantum formalism is simply used as a source of
entanglement. In these studies, the mental states of
agents were characterized by state vectors in Hilbert
space and, in several experimental models, quantum
probabilities had better predictive power than classical
particularly in psychology and cognitive sciences, can be
modeled by this method.26,27,31e33
In order to make the notions of quantum physics more
easily understandable, I will draw a formal comparison be-
tween homeopathy trials and single-particle interference in
quantum physics. In both cases, contextuality has been
shown to play a chief role.
Single-particle quantum interference illustrates the su-
perposition principle and some characteristics of quantum
probabilities. The classical two-slit interferometer of
Young is usually used for such a description, but the
MacheZehnder device has the advantage of including
only two detectors (D1 and D2) as depicted in
Figure 1.34Light is emitted from a monochromatic light
source: 50% of light is transmitted by beam splitter
(BS1) in path T and 50% is reflected in path R. In BS2,
the two beams are combined and 50% of light is trans-
mitted by beam splitter in detector D1 and 50% in detector
waves from the two paths are constructivewhen theyarrive
in D1 and destructive in D2. Therefore, clicks after light
detection are heard only in D1. This is indeed what exper-
iment shows and it is an argument for the wave-like nature
On the contrary, if we consider light as a collection of
small balls (photons), they should randomly go into path
T or R (with a probability of 0.5 for each path) and then
in BS2 they go into D1 or D2 randomly (again with a prob-
ability of 0.5 for D1 or D2). As a consequence D1 should
click in 50% of cases and D2 in 50% of cases. However,
even if photons are emitted one by one (by decreasing light
intensity), the interference pattern persists (100% of clicks
in D1). This is a quite counterintuitive result. Even more
Quantum probabilities and homeopathy
appears if theinitial path(Tor R) is detectedbyanymeans:
then either D1 or D2 clicks, each in 50% of cases (classical
probabilities apply) (Figure 1; lower drawing).
I draw an analogy between the one-particle interference
experiment and homeopathic trials, which appear to have
comparable mathematical structures. Indeed, according
and pair concordance e measured by patient/practitioner
vs. ‘external’ measure of labels), either only concordant
pairs (CP) (equivalent to detection in D1) or both CP/
discordant pairs (DP) (i.e., equivalent to random detection
by D1 and D2) are obtained (Figure 1 and Table 1). This
suggested that a quantum (or more precisely quantum-
like) model could be built to describe homeopathy trials,
including the ‘paradoxical’ failure of homeopathy blind
In classical probability theory, probabilities add; if P1
and P2 are the probabilities for two events E1 and E2
(for example head or tail in a coin toss), the probability
for either event to occur is Prob(E1 or E2) = P1 + P2.
With quantum probabilities, there is a key difference. First,
we must write P1 = a2and P2 = b2; the letters a and b are
their squares allow calculation of the corresponding prob-
abilities. The second key point is that for quantum proba-
bilities, probability amplitudes, not probabilities add:
Prob?E1 or E2?¼ ða þ bÞ2
¼ P1 þ P2 þ interference term:
The interference term is thus added to or subtracted to the
classical probabilities to give quantum probabilities.
The objective of my study was to describe the possible
outcomes of the cognitive states of patient/practitioner in
different contexts. Mathematically, a state is represented
by a vector in a Hilbert space. Using the quantum
formalism, the cognitive state A of an agent (observer,
experimenter, practitioner or patient) is represented by a
state vector jjAi, which summarizes all the information
on the quantum system. The linear combination of any
states is itself a possible state (superposition principle).
Thus, if jA1i and jA2i are two possible states of the system,
then jjAi ¼ l1jA1i þ l2jA1i is also a possible state of A
(with l1and l2real or complex numbers). Therefore, a
quantum system exists in all its particular and theoretically
possible states. When the system is ‘measured’, only one
state among the possible states is observed. In the quantum
formalism, the probability to observejA1i is the square of
the probability amplitude l1associated with this state.
An example of superposition that is directly observable
is the interference pattern observed in the two-slit experi-
ment. Interferences are the hallmark of superposed states
Figure 1 Interpretation of the outcomes of homeopathy trials as a
consequence of quantum-like interferences (for an experiment
with optimal interference term). A. In an open-label trial or with
in situ randomization and unblinding, the cognitive state of pa-
tient/practitioner (described by the state vector jPPi) is able to
interfere with itself (as a single-particle interferes with itself) and
the rate of correlated pairs is high. B. In a centralized blind RCT,
the cognitive state of PP cannot interfere with itself (there is no su-
perposition) and the rate of correlated pairs is not better than
random; ‘specific’ effects from homeopathy group are described
in placebo group (‘smearing effect’).
experiment with interferometer and randomized placebo-controlled
Parallelism between single-photon interference
Path T and path RPPPLAand PPHOM
100% Detector D1
0% Detector D2
Path T or path RPPPLAor PPHOM
50% Detector D1
50% Detector D2
PP, cognitive state of the couple practitionerepatient; PLA, placebo
label; HOM, homeopathy label; Y, negative outcome; [, positive
outcome; T, transmission; R, reflection; CP, concordant pairs; DP,
* For an experiment with optimal correlations between labels and
outcome (and with l1
yPPPLAwith PPYor PPHOMwith PP[.
zPPPLAwith PP[or PPHOMwith PPY.
Quantum probabilities and homeopathy
and are the heart of quantum physics. In a single-photon
interference experiment, if one can (even in principle)
distinguish the path each photon has taken, then interfer-
ences vanish andclassical
Figure 1, the initial path (R or T) cannot be distinguished
in upper drawing, and interferences occur; in lower draw-
ing, paths are distinguished by measurement and conse-
quently classical probabilities apply (without interference
The notion of ‘non-commuting observables’ is a key
concept of quantum probabilities. Technically speaking,
physical observables are mathematical ‘operators’ and
for each operator there is a spectrum of possible results,
which are named the ‘eigenvectors’ of the operator (they
constitute an orthogonal basis in the vector space). When
an operator is applied to a random state vector, the vector
is split into different components, which are the eigenvec-
tors of the operator. If the original state vector to be
observed is an eigenvector of the operator, then it is not
affected (this means that the value of the parameter to be
measured was already fixed before measurement). Two ob-
servables are said to commute with each other when they
by the measure of the other observable). As a consequence,
the outcomes will not differ according to the order of the
When two observables are non-commuting, the set of ei-
genvectors of one observable (orthogonal basis) can be ex-
the other observable; as a consequence, there are two
different bases for the same vector space and the outcomes
will be different according to the order of the measure-
Put more simply, different results according to experi-
mental devices (e.g., presence or absence of interference
pattern in Young’s experiment) are the consequence of
non-commuting observables. The different experimental
outcomes are said to be complementary, since both as-
pects (wave and photon) are necessary to describe the
system in terms of classical physics. Another conse-
quence is the establishment of non-local correlations be-
tween the different parts of the system, which are thus
probabilities apply. In
The purpose of a randomized controlled clinical trial is
to assess whether a therapy is associated more frequently
with a positive outcome than a placebo or a reference treat-
homeopathy trial, one calculates whether the homeopathy
therapy (labelled HOM) is more frequently associated
with positive outcome denoted ‘[’ than placebo (PLA).
A negative outcome is denoted ‘Y’.
The aim of this formal description is to calculate the
probabilities of the possible cognitive states of the pa-
tient/practitioner before and after the trial. We define the
cognitivestates of practitioner andpatientbyusing asingle
state vector: jjPrijjPai ¼ jjPrPai ¼ jPPi with eigenvectors
(i.e., state vectors of possible outcomes) written with in-
dexes corresponding to cognitive state. For example,
jPP[i summarizes the cognitive states of the couple pa-
tient/practitionerthat are associatedwithpositiveoutcome.
We define ‘CP’ as the association of the label PLAwith
as the association of PLA label with outcome ‘[’ or the as-
sociation of HOM label with outcome ‘Y’ (i.e., DP are
Using this notation, Probclass(PPCP) is defined as the
classical probability for the cognitive states of patient/
practitioner associated with CP; Probquant(PPCP) is the
quantum probability of the cognitive states of patient/
practitioner associated with CP. Probclass(PPDP) and
Probquant(PPDP) are the respective classical and quantum
probabilities for DP.
It is important to note that we adopt the point of view of
an observer who describes the system formed by the cogni-
tive states of patient/practitioner, using concepts of quan-
tum mechanics. The observer knows the initial state of
the system and does not perform any measurement on
this system during its evolution (from randomization to
assessment of the CP).
First case: patient/practitioner measure both
In this situation, we consider that patient/practitioner
‘measure’ both labels and pair concordance (Figure 1A).
The state vector of the cognitive state of patient/practi-
tioner is described in terms of the eigenvectors of the first
observable (PP indexed with labels PLA or HOM). From
the point of view of the observer defined above, for each
treatment chosen for a given patient by a given physician:
jjPrPai ¼ l1jPPPLAi þ l2jPPHOMi;
l12and l22are the probabilities associated with label PLA or
label HOM after randomization, respectively.
The eigenvectors of the first observable (labels) are
developed on the eigenvectors of the second observable
(concordance of pairs). It is postulated that PP indexed
with ‘labels’ and PP indexed with ‘concordance of pairs’
are non-commuting observables:
jPPPLAi ¼ m11jPPCPi þ m12jPPDPi;
jPPHOMi ¼ m21jPPCPi þ m22jPPDPi:
Therefore, jjPrPai can be expressed as a superposed state
of jPPCPi and jPPDPi:
jjPrPai ¼ ðl1m11þ l2m21ÞjPPCPi þ ðl1m12þ l2m22ÞjPPDPi:
The probability of PPCPis the square of the probability
amplitude associated with its state:
Quantum probabilities and homeopathy
ProbquantðPPCPÞ ¼ jl1m11þ l2m21j2:
Similarly, Probquant(PPDP) is calculated:
ProbquantðPPDPÞ ¼ jl1m12þ l2m22j2:
Probquant(PPCP) + Probquant(PPDP) = 1,we can easilycalculate
Therefore, we can replace the probability amplitudes in
the equations calculated above:
jPPPLAi ¼ cos qjPPCPi ? sin qjPPDPi;
jPPHOMi ¼ sin qjPPCPi þ cos qjPPDPi;
jjPrPai ¼ ðl1cos q þ l2sin qÞjPPCPi
þ ðl2cos q ? l1sin qÞjPPDPi;
ProbquantðPPCPÞ ¼ jl1cos q þ l2sin qj2;
ProbquantðPPDPÞ ¼ jl2cos q ? l1sin qj2:
Second case: one observable is measured by ‘outside’
supervisor (centralized blind RCT)
In a centralized blind RCT, the context changes; labels
are now measured/observed by a statistician or by the
main investigator who supervises the trial (Figure 1B).
This experimental situation is formally equivalent to a
which-path measurement in single-particle interference
experiment. Indeed, the information on the path must be
taken into account; therefore, conditional classical proba-
bilities, which include path data, must be used for calcula-
tion of the probability for the cognitive states of patient/
practitioner to be associated with CP:
ProbclassðPPCPÞ ¼ ProbðPPPLAÞ ? ProbðPPCPjPPPLAÞ
þ ProbðPPHOMÞ ? ProbðPPCPjPPHOMÞ:
ProbðPPCPjPPPLAÞ ¼ cos2q
PPHOMÞ ¼ sin2q (see Eqs. (4) and (5)), we calculate:
ProbclassðPPCPÞ ¼ l2
1cos2q þ l2
ProbclassðPPDPÞ ¼ l2
We conclude that ProbquantðPPCPÞsProbclassðPPCPÞ in
the general case. In the squaring of the sum, we have ob-
tained an additional term 2l1l2cos qsin q, which is typical
of quantum probability interferences:
2cos2q þ l2
ProbquantðPPCPÞ ? ProbclassðPPCPÞ ¼ 2l1l2sin qcos q:
Therefore, we dispose now of a simple description of
clinical trials for homeopathy (Table 2). Assuming non-
commuting observables and superposition of cognitive
states, this model describes 1) the establishment of corre-
lations when patient/practitioner measure both observ-
ables and 2) correlations not better than random in
trials with labels blinded by external supervisor. The
higher probability of CP in the first experimental situation
compared to the second one is due to the interference
This formal description supports Walach’s suggestion
that if homeopathy effects are based on some form of
entanglement, then these effects cannot be treated caus-
ally.35As a consequence, he predicted that placebo-
controlled trials and experiments that force the non-local
effect into a causal framework are not adequate.
Table 2Summary of the quantum-like model describing open-label vs. double-blind placebo-controlled randomized trials of homeopathy
Non-commuting observables (q s 0)Commuting observables (q = 0)
With interference term
term (no superposition)
Concordance of labels
Probability of CP: Prob(PPCP)
jl1cos q + l2sin qj2
jl2cos q ? l1sin qj2
Open-label trial or in
situ randomization and
l12cos2q + l22sin2q
l22cos2q + l12sin2q
Probability of DP: Prob(PPDP)
Centralized blind RCT Only negative outcomes
NA: not applicable; RCT: randomized clinical trial.
* Probquant(PP[) = l12? Probquant(PPDP) + l22? Probquant(PPCP).
yProbclass(PP[) = l12? Prob(PP[jPPPLA) + l22? Prob(PP[jPPHOM) = l12sin2q + l22sin2q = sin2q.
zObservables commute with cos q = 1 and sin q = 0; then Prob(PP[) = 0 and Prob(PPY) = 1 (only negative outcome is observed by practitioner
and patient; there is no positive outcome).
xCP: PPPLAassociated with PPYor PPHOMassociated with PP[.
jjFor sin q = l2(and consequently cos q = l1), the quantum interference term is maximal with Probquant(ACP) = 1 and Probquant(ADP) = 0.
Quantum probabilities and homeopathy
In the proposed model, q is the only parameter to be
adjusted. This parameter allows the passage from classical
to quantum probabilities. When q = 0, the observables
jPPPLAi ¼ 1 ? jPPCPi ? 0 ? jPPDPi ¼ jPPCPi;
jPPHOMi ¼ 0 ? jPPCPi þ 1 ? jPPDPi ¼ jPPDPi:
In this case, the two observables share their eigenvec-
tors: jPPPLAi ¼ jPPCPi and jPPHOMi ¼ jPPDPi. Thus, if
q = 0, PLA label is always associated with CP and HOM
label is always associated with DP; no positive outcome
is observed. This indicates that q s 0 (non-commuting ob-
servables) is necessary not only for high rate of CP,butalso
for high rate of positive outcome. Note also that positive
outcome should be present in the experimental back-
to be spontaneously observed is low but not equal to zero
and thanks to entanglement, the rate of positive outcomes
The only postulates of the model are 1) non-commuting
observables (q s 0) and 2) superposition of the states of
PP. We simply enlarge the description of clinical trials
from classical probabilities to quantum probabilities;
indeed, classical probabilities are only a special case of
quantum probabilities (with q = 0).
I did not hypothesise what q stands for; this parameter
connects together expected effects (symbolized by labels)
and observed success/failure (concordance of pairs)
without making hypotheses on the underlying mecha-
nisms. Thus, q could summarize cognitive and mental phe-
nomena as different as empathy, physician’s experience,
expectation of patient and/or practitioner related to beliefs
about treatment effectiveness and other cultural beliefs,
Pavlovian conditioning, implicit learning, unconscious
mechanisms and unknown mechanisms.
This model does not rule out an entanglement of homeo-
pathic medicine with patient and/or practitioner as pro-
posed by some authors, but it does not support either
assumption.18,19,21,24The homeopathic medicine appears
in the equations of the formalism through its label, i.e.,
its meaning or, in other words, the expected effect. The
only hypothesis that appears to be excluded is a strict
pharmacological effect. In this case, label blinding by
external supervisor would be without consequence on the
result of the RCT.
such asan ordinary
Non-local correlations cannot be used to transmit infor-
mation; it has been pointed out that one consequence of
non-local theories was that e if the effects are treated caus-
ally e “they go away, change channel or do something
crazy”.35Indeed, with entangled quantum objects, the
same randomness is observed at two distant locations,
but cannot be controlled and used to send useful informa-
Pilot studies of blind homeopathic pathogenetic trials
(HPTs, provings) suggest that such ‘channel change’
does indeed occur. In these HPTs, symptoms typical of
one remedy were found in another study group.36,37One
of the reasons for the difficulties in obtaining significant
differences between placebo and homeopathy remedy in
blind trials could be related to ‘smearing’ between
placebo and homeopathy groups since the difference
between groups is then erased.6
‘Channel change’ between treatment groups emerges
from the formalism without additional hypotheses. For
example, we can calculate the probability of positive
outcome. For simplicity, we suppose that the probabilities
for a given patient to be randomized in homeopathy group
or in placebo group are equal (l12= l22= 0.5) and that the
concordance of pairs is optimal (cos q = l1and sin q = l2).
In open-label setting, we calculate that as expected
Probquant(PPCP) = jl1cos q + l2sin qj2= 1: all patients
who receive homeopathy have positive outcome and all pa-
tients who receive placebo have a negative outcome. In
blind setting, we calculate:
ProbclassðPPCPÞ ¼ l2
1cos2q þ l2
2sin2q ¼ 0:5;
ProbclassðPPDPÞ ¼ l2
2cos2q þ l2
1sin2q ¼ 0:5:
In other words, the proportion of patients in homeopathy
group associated with positive outcome is decreased to 0.5
in blind setting and half of patients in placebo group are
associated with positive outcome. The ‘smearing’ of posi-
tive outcomes from ‘active’ group to placebo group is thus
a direct consequence of the proposed formalism.
veniste’s experiments. They were particularly obvious
when Benveniste’s team used the Langendorff system in
experimental protocols that were similar to blind RCTs;
these experiments and their ‘anomalies’ have been exten-
sively described elsewhere.38e42Because Benveniste
searched for a local explanation to explain his results on
high dilutionsand ‘digital
changes’ were considered as failures due to molecular
contamination, electromagnetic interferences, ‘jumps’ of
activity from sample to sample, errors of manipulation,
etc. Benveniste’s later work, based on ‘digital biology’
concepts and using an automated analyzer could not be
replicated by an independent team despite promising
initial results.43Therefore, although the successive experi-
mental systems of Benveniste’s team benefited from
important methodological improvements, the weirdness
persisted and has been an obstacle to convincing other sci-
entists of the validity of the results.
The experiments with high dilutions in the basophil sys-
tem also exhibited comparable spreading of activity be-
tween samples in large-scale blind experiments. Both
Benveniste’s team15,44and Sainte-Laudy and Belon (see
biology’, the ‘channel
Quantum probabilities and homeopathy
Figure 2 in Ref. 45) have reported numerous experiments
with regular ‘waves’ of basophil degranulation (or inhibi-
tion of degranulation). In large-scale blind experiments
from thesame authors, themean difference betweenhighly
diluted control and ‘active’ samples strongly decreased
and the usual regular patterns of degranulation according
to dilution titer vanished.13,15Despite the small mean
statistical significance was nevertheless achieved thanks
to the high statistical power of the experiments. The
detrimental consequences of blinding in these in vitro
experiments also suggest nonlocal correlations.40,46
to be dependent on experimental context: as for a quantum
object, we can decide to observeeither ‘waves’ (open-label
trial) or ‘particles’ (centralized blind trials). In the single-
particle interference experiment, there is no ‘failure’
when particles are observed and there is no ‘success’
when waves are observed. Waves and particles are two
complementary aspects of the same quantum object. Due
to these complementary aspects, if homeopathy trials
follow quantum logic, we are then faced with the impossi-
bility of getting significant correlations if we want to
‘prove’the efficacyof homeopathy byblind RCTs,the cur-
rent ‘gold standard’ for evaluating treatments.
A recent double-blind study in patients with rheumatoid
arthritis concluded that homeopathic consultations, but not
homeopathic remedies, were associated with clinical
improvement.2As pointed out by Milgrom, although these
results did not prove entanglement during homeopathy
therapy, theyneverthelessrevealtheefficacyof the homeo-
pathic consultation.3This comment is valuable because it
means that homeopathic consultation could be a starting
point to study which parameters influence clinical
Our quantum-like model suggests that the correlations
between positive outcomes and homeopathy treatment
observed in open-label conditions vanish in conditions
comparableto blind RCT. The crucial pointis the measure-
ment/observation of the two observables: labels of treat-
ments and clinical outcomes. If both observables are
assessed by patient/practitioner, then non-local correla-
tions of cognitive states are possible (with the hypothesis
of noncommuting observables).
This suggests a method to overcome the ‘glass ceiling’
of blind RCTs in homeopathy. In this method, each practi-
tioner would receive two sets of treatment units (homeop-
athy medicine and placebo) under code names. After a
random choice of treatment by the physician, the patient
would receive the chosen treatment (A or B). After assess-
ment of clinical outcome, the result of each individual trial
would be recorded in an electronic devicewith no possibil-
ity of erasing or modifying it. Then, the treatment name
(stored in the memory of the device) e placebo or verum
e administered to patient would be unblinded to the pa-
tient/practitioner. By this method, the two observables
(treatment label and concordance of pairs) would be
‘measured’ locally. Therefore, all operations from random-
office in a defined order without need of centralized super-
In an editorial in Homeopathy, Fisher noted that the hy-
potheses for nonlocal effects in homeopathy were not
proven and added: “A decisive ‘Aspect’ experiment has
not even been proposed, much less conducted”.12Indeed,
the experiment performed by Aspect et al. in 1982 was a
convincing proof that the microscopic world was nonlocal
as predicted by quantum physics.47Comparing in situ
randomization/unblinding vs. centralized supervision of
clinical trials could be theequivalent of theAspect’sexper-
iment for quantum theories of homeopathy. Thus, positive
results in homeopathy blind RCTusing the in situ method-
ology would be not only a means to circumvent the gold
standard, but also a very strong argument in favor of
nonlocal theories for homeopathy. Nor can we exclude
the possibility that non-local effects also occur in allopathy
trials and add to the causal/local pharmacological effect;
comparing in situ and centralized randomization/unblind-
ing in the same trial would give evidence of such effects.
blind RCTs emerge logically from the formalism in this
quantum-like model of homeopathy clinical trials. This
model suggests that theprobability of observing significant
differences between placebo and homeopathy therapy in
blindRCTswouldbe increasedbyusing insitu randomiza-
The author has no conflict of interest.
1 Maddox J. When to believe the unbelievable. Nature 1988; 333:
2 Brien S, Lachance L, Prescott P, McDermott C, Lewith G. Homeop-
athy has clinical benefits in rheumatoid arthritis patients that are
attributable to the consultation process but not the homeopathic
remedy: a randomized controlled clinical trial. Rheumatology (Ox-
ford) 2011; 50: 1070e1082.
3 Milgrom LR,ChatfieldK.“It’stheconsultation,stupid!” .Isn’tit?
J Altern Complement Med 2011; 17: 573e575.
4 The end of homoeopathy [Editorial]. Lancet 2005; 366: 690.
5 Vandenbroucke JP. Homoeopathy and “the growth of truth”. Lancet
2005; 366: 691e692.
6 Shang A, Huwiler-Muntener K, Nartey L, et al. Are the clinical ef-
fects of homoeopathy placebo effects? Comparative study of
placebo-controlled trials of homoeopathy and allopathy. Lancet
2005; 366: 726e732.
placebo effects? Lancet 2005; 366: 2081. author reply 2083e2086.
8 Fisher P, Berman B, Davidson J, Reilly D, Thompson T. Are the
clinical effects of homoeopathy placebo effects? Lancet 2005;
366: 2082e2083. author reply 2083e2086.
Quantum probabilities and homeopathy
9 Ludtke R, Rutten AL. The conclusions on the effectiveness of ho- Download full-text
meopathy highly depend on the set of analyzed trials. J Clin Epide-
miol 2008; 61: 1197e1204.
10 Milgrom LR. Gold standards, golden calves, and random reproduc-
ibility: why homeopaths at last have something to smile about. JAl-
tern Complement Med 2009; 15: 205e207.
11 Weatherley-Jones E, Thompson EA, Thomas KJ. The placebo-
controlled trialasatestofcomplementary andalternativemedicine:
observations from research experience of individualised homeo-
pathic treatment. Homeopathy 2004; 93: 186e189.
12 Fisher P. Entangled, or tied in knots? Homeopathy 2004; 93:
13 Belon P, Cumps J, Ennis M, et al. Inhibition of human basophil
degranulation by successive histamine dilutions: results of a Euro-
pean multi-centre trial. Inflamm Res 1999; 48(Suppl. 1): S17eS18.
14 Benveniste J, Davenas E, Ducot B, Cornillet B, Poitevin B, Spira A.
L’agitation de solutions hautement dilu? ees n’induit pas d’activit? e
biologique sp? ecifique. C R Acad Sci II 1991; 312: 461e466.
15 Davenas E, Beauvais F, Amara J, et al. Human basophil degranula-
tion triggered by very dilute antiserum against IgE. Nature 1988;
16 Davenas E, Poitevin B, Benveniste J. Effect of mouse peritoneal
macrophages of orally administered very high dilutions of silica.
Eur J Pharmacol 1987; 135: 313e319.
17 Walach H, Jonas WB, Ives J, van Wijk R, Weingartner O. Research
18 Hyland ME. Extended Network Generalized Entanglement Theory:
therapeutic mechanisms, empirical predictions, and investigations.
J Altern Complement Med 2003; 9: 919e936.
19 Milgrom LR. Patientepractitionereremedy (PPR) entanglement.
Part 1: a qualitative, non-local metaphor for homeopathy based on
quantum theory. Homeopathy 2002; 91: 239e248.
20 Walach H. Magic of signs: a non-local interpretation of homeopa-
thy. Br Hom J 2000; 89: 127e140.
21 Weingartner O. What is the therapeutically active ingredient of ho-
meopathic potencies? Homeopathy 2003; 92: 145e151.
22 Walach H, von Stillfried N. Generalised Quantum Theory e basic
idea and general intuition: a background story and overview. Axio-
mathes 2011; 21: 185e209.
23 Walach H. Generalized entanglement: a new theoretical model for
understanding the effects of complementary and alternative medi-
cine. J Altern Complement Med 2005; 11: 549e559.
24 Walach H. Entanglement model of homeopathy as an example of
generalized entanglement predicted by weak quantum theory.
Forsch Komplementarmed Klass Naturheilkd 2003; 10: 192e200.
25 Bruza P, Busemeyer JR, Gabora L. Introduction to the special issue
on quantum cognition. J Math Psychol 2009; 53: 303e305.
26 Mogiliansky AL, Zamir S, Zwirn H. Type indeterminacy: a model
of the KT(KahnemaneTversky)-man. J Math Psychol 2009; 53:
27 Pothos EM, Busemeyer JR. A quantum probability explanation for
violations of ‘rational’ decision theory. Proc Biol Sci 2009; 276:
28 Khrennikov A. Quantum-like model of cognitive decision making
and information processing. Biosystems 2009; 95: 179e187.
29 Khrennikov A. Quantum-like brain: “Interference of minds”. Bio-
systems 2006; 84: 225e241.
30 Busemeyer JR, Wang Z, Townsend JT. Quantum dynamics of hu-
man decision-making. J Math Psychol 2006; 50: 220e241.
sure-thing principle: the use of probability interference and other
concepts. J Math Psychol 2009; 53: 378e388.
32 Conte E, Todarello O, Federici A, Vitiello T, Lopane M,
Khrennikov A, et al. A preliminary evidence of quantum like
behavior in measurements of mental states. In: Khrennikov AYu
(ed). Quantum Theory: Reconsideration of Foundations. V€ axj€ o:
V€ axj€ o University Press, 2004, p. 679e702. Available from: http://
33 AtmanspacherH,Filk T,RomerH.QuantumZenofeatures ofbista-
ble perception. Biol Cybern 2004; 90: 33e40.
34 Scarani V, Suarez A. Introducing quantum mechanics: one-particle
interferences. Am J Phys 1998; 66: 718e721.
35 WalachH.Entangled eandtiedinknots!Practicalconsequences of
an entanglement model for homeopathic research and practice. Ho-
meopathy 2005; 94: 96e99.
36 Mollinger H, Schneider R, Loffel M, Walach H. A double-blind,
randomized, homeopathic pathogenetic trial with healthy persons:
comparing two high potencies. Forsch Komplementarmed Klass
Naturheilkd 2004; 11: 274e280.
37 Walach H, M€ ollinger H, Sherr J, Schneider R. Homeopathic patho-
genetic trials produce more specific than non-specific symptoms:
results from two double-blind placebo controlled trials. J Psycho-
pharmacol 2008; 22: 543e552.
38 Schiff M. The memory of water: homoeopathy and the battle of
ideas in the new science. London: Thorsons Publishers, 1998.
39 Benveniste J. Ma v? erit? e sur la m? emoire de l’eau. Paris: Albin Mi-
40 Beauvais F. Emergence of a signal from background noise in the
“memory of water” experiments: how to explain it? Explore (NY)
2012; 8: 185e196.
41 Beauvais F. Memory of water and blinding. Homeopathy 2008; 97:
42 BeauvaisF. L’^Ame des Mol? ecules e Une histoire de la “m? emoire de
l’eau”. Collection Mille Mondes; ISBN 978-1-4116-6875-1; 2007.
Available from: http://www.mille-mondes.fr.
43 Jonas WB, Ives JA, Rollwagen F, et al. Can specific biological sig-
nals be digitized? FASEB J 2006; 20: 23e28.
44 Poitevin B, Davenas E, Benveniste J. In vitro immunological
degranulation of human basophils is modulated by lung histamine
and Apis mellifica. Br J Clin Pharmacol 1988; 25: 439e444.
45 Sainte-Laudy J, Belon P. Inhibition of basophil activation by hista-
ical activity of high dilutions. Homeopathy 2009; 98: 186e197.
46 Beauvais F. Description of Benveniste’s experiments using
quantum-like probabilities. J Sci Explor 2013; 27: 43e71.
using time-varying analyzers. Phys Rev Lett 1982; 49: 1804e1807.
Quantum probabilities and homeopathy