Towards an Ontology of Pain

Barry Smith1,2, Werner Ceusters3, Louis J. Goldberg4, Richard Ohrbach4

1 Department of Neurology, University at Buffalo
School of Medicine and Biomedical Sciences
Buffalo General Hospital, 100 High Street, Buffalo, NY 14203, USA
phismith@buffalo.edu
http://ontology.buffalo.edu/smith
2 Department of Philosophy, University at Buffalo
130 Park Hall, Buffalo, NY 14260, USA
3 Department of Psychiatry, University at Buffalo
School of Medicine and Biomedical Sciences
ceusters@buffalo.edu
http://www.referent-tracking.com
Erie County Medical Center, 462 Grider Street, Buffalo, NY 14215, USA
4 Department of Oral Diagnostic Sciences, University at Buffalo
School of Dental Medicine, 355 Squire Hall, Buffalo, NY 14214, USA
{goldberg,ohrbach}@buffalo.edu
http://www.sdm.buffalo.edu/home.asp?id={524,579}

ABSTRACT. We present an ontology of pain and of other pain-related phenomena, building

on the definition of pain provided by the International Association for the Study of Pain
(IASP). Our strategy is to identify an evolutionarily basic canonical pain phenomenon,
involving unpleasant sensory and emotional experience based causally in localized tissue
damage that is concordant with that experience. We then show how different variant cases of
this canonical pain phenomenon can be distinguished, including pain that is elevated relative
to peripheral trauma, pain that is caused neuropathically (thus with no necessary peripheral
stimulus), and pain reports arising through deception either of self or of others. We describe
how our approach can answer some of the objections raised against the IASP definition, and
sketch how it can be used to support more sophisticated discrimination of different types of
pain resulting in improved data analysis that can help in advancing pain research.

1 Background: The Physical Basis of Disease

Increasingly, ontologies are being used to support the retrieval, integration and analysis
of a variety of different kinds of biomedical data. Ontology-based technology has been
successful especially in support of data-driven research in the basic biological sciences
and in model organism studies, and efforts are now being made to extend these suc-
cesses to the domain of human disease and diagnosis. The most successful ontologies,
above all the Gene Ontology [Bodenreider 2008], rest on objective classifications of
biological phenomena primarily at the molecular and cellular levels, and we face diffi-
culties in applying the same approach where we are dealing with clinical data pertaining
to pain and to other symptoms of human disease marked by the feature of subjectivity.
The goal of this communication is to provide the beginnings of an ontological account
of pain and of those phenomena closely related to pain that are commonly described as
pain in patient reports. Because pain has subtly complex characteristics, we believe that
its examination will have heuristic value for ontological accounts of symptoms (such as
feelings of nausea, fatigue, depression) more generally.
2 Barry Smith, Werner Ceusters, Louis J. Goldberg and Richard Ohrbach.

Our strategy is to pursue a view of pain as resting in every case on some physical ba-
sis that is perhaps as yet unknown. This is part of a more general strategy, defended in
[Scheuermann et al. 2009], which views all clinically relevant phenomena on the side of
the patient as having some physical basis within the organism. By ‘physical basis’ we
understand any configuration of one or more physical components within the organism
at any level of granularity, from a single nucleotide to an arthritically deformed joint.
Where they are non-disordered – which means: such as to reflect the coordinated ex-
pression of the corresponding structural genes for an organism of the given type [Rosse
et al. 2007] – such configurations support those dispositions in the organism which are
realized in normal, ordered functioning. Where they are disordered, such configurations
support dispositions to abnormal functioning, one family of which is manifested in
experiences of pain. We then use the term ‘disorder’ to refer to the physical basis of
such a disposition to abnormal functioning. Thus a disorder is some physical part of the
organism that gives rise to a potential for a clinically significant departure from normal
functioning of one or other kind.As we shall see, abnormal functioning may have bene-
ficial characteristics for the patient, above all – in the case of pain with concordant
tissue damage (PCT) – in signaling the presence of tissue damage. When, for example,
there is a persistent pain in a patient’s left temporomandibular joint (TMJ), then this is
because some physical structure or substance in the organism is disordered (for example,
the TMJ is deformed because of arthritis so that it serves as a direct source of nocicep-
tion). As a result of this disorder, the organism acts in a certain way that is consistent
with the presence of pain. When pain is acute, the behavioral manifestations of the
organism follow a generally predictable pattern that is oriented towards self-protection
and help-seeking: the pain excites the organism to some action to relieve the suffering
involved in the experience. In the case of chronic pain, multiple components of the
central nervous system are affected; behavioral manifestations will vary from patient to
patient and will fluctuate over time in a pattern difficult to predict on the basis of only
the reported pain but generally understandable when the multiple affected CNS compo-
nents are identified.
‘Symptom’, as we here use this term, covers a restricted family of phenomena (in-
cluding pain, anxiety, nausea, anger, drowsiness, itchiness), which are of their nature
such as to have features which can be experienced only in the first person. (In Table 1
below these are labeled as ‘Subjectively Observable Features’.) Symptoms can be re-
ported to, and associated behaviors and bodily qualities can be observed by, the clini-
cian; but symptoms themselves cannot be observed (except by the patient) and they
cannot be objectively measured. The absence of objective measurement does not, how-
ever, imply that symptoms cannot be reported reliably and so have objective validity.
To say of pain that it has features that can be observed only subjectively is therefore
consistent with the fact that given pain reports can be objectively true or false.

2 The IASP Definition of Pain

Pain is defined by the International Association for the Study of Pain (IASP) as follows:
pain (IASP) =def. an unpleasant sensory and emotional experience associated
with actual or potential tissue damage, or described in terms of such damage
[Merskey et al 1979].
This definition has proved to be of considerable value, having led to 50 years of highly
productive fundamental research on pain. On the other hand it has certain problems, as

Keio University Press Inc. 2011

 Ontology and Logic Conference 2011 3

recently reflected by significant discussion by an IASP task force (http://www.iasp-

pain.org/source/eforums). The definition ascribes a common phenomenology (‘unplea-
sant sensory and emotional experience’) to all instances of pain, together with the rec-
ognition of three distinct subtypes of pain involving, respectively:
1. actual tissue damage,
2. what is called ‘potential tissue damage’, or
3. a description (in some patient report) involving reference to tissue damage.
Clause 3. may be interpreted to mean that a mere description of a certain sort provides
sufficient evidence that pain is present. The intent, as we understand it, is to assign
those patient reports of pain that are not sufficiently grounded in observable
manifestations of tissue damage to some other (for example psychological) realm albeit
while still requiring reference to an experience anchored within the body. This causes
problems in the classification of cases for example of the following sort: A patient
presenting with pain and associated tissue damage was prescribed pain relief medication.
While moderate tissue damage remains, the medication is effective, so that there is no
longer pain. But because the patient has become addicted, he claims that there is still
pain in order to obtain more medication. Such cases are not pain; yet as we shall see
they will often be so classified by the clinician. Until recently the headache classifica-
tion system would not permit a headache diagnosis if the patient was effective in use of
medication for treating the headache. While this problem has been fixed, clinicians still
face problems because of the ambiguous use of ‘pain’ and cognate terms to mean either
pain as experienced or the underlying physical basis in the body of the patient. The
latter may give rise to a disposition to pain which may survive even though it is not
being realized as a result of effective medication.

3 Strategy for Defining Pain

In providing a modified version of the IASP definition in what follows we define, first,
what we shall call ‘pain with concordant tissue damage’ (PCT), which we hold to be
the canonical (normal, prototypical) and evolutionarily most basic case of pain. We then
describe a number of variant phenomena which are defined in terms of, and involve
specific kinds of departures from, this canonical case. Our strategy is thus comparable
to the way in which the results of genetic mutations or injuries affecting, for example,
the human hand, are most effectively described in terms of specific kinds of departures
from the anatomical structure of the normal human hand (with its 5 fingers, 10 metacar-
pal bones, etc.). This strategy has been pioneered by the Foundational Model of Anat-
omy (FMA) Ontology, a scientifically well-established reference ontology of human
(and more generally of mammalian) anatomy [Rosse et al. 2007] which describes the
‘canonical’ structure of the human organism and various ‘variant’ structures arising, for
instance, through amputation or genetic defect. Its implementation here, which draws
on an approach to mental functioning-related anatomical structures advanced in [Ceust-
ers et al. 2010], sees the nociceptive system as involving multiple components, inclu-
deng a reflex protective component, a sensory component, and an emotional component,
the latter involving many non-cortical brain systems (e.g. amygdala) that together pro-
duce in the organism a sense of crisis, activate the autonomic nervous system, and or-
ganize the behavioral response to the pain stimulus. (Compare also [Melzack 2001].)
4 Barry Smith, Werner Ceusters, Louis J. Goldberg and Richard Ohrbach.

Against this background we can now distinguish the following five different sorts of
cases of pain and of pain-related phenomena (see Table 1 and Figure 1):
PCT: pain with concordant tissue damage: the patient experiences pain of the
evolutionarily most basic sort, which is to say: pain in response to and in concordance
with simultaneously existing tissue damage. This tissue damage is both the experienced
target and the cause of the pain experience. Here the nociceptive system and the
peripheral trauma are aligned; clear signals connect the one with the other.
PNT: pain without concordant tissue damage: the patient experiences pain that is
associated with some disorder outside the nociceptive system, a disorder existing either
now or in the past. In some cases (for example myofascial pain) this disorder is
predominantly associated with peripheral trauma; in other cases (for example tension
headache) predominantly not. But in every case PNT pain is marked by the fact that it
is discordant in terms of some attribute, such as intensity, spatial extent, or episode
frequency or duration, relative to the state of the underlying tissue. Here the nociceptive
system is intact, but signals associated with a stimulus at a peripheral site are either (1)
ramped up relative to what would be the normal (PCT) case, or (2) ramped down, for
example in individuals with coping styles which have the effect of diminishing either
nociceptive transmission or some feature of experienced pain, or in cases of pain
asymbolia, a condition in which pain is experienced (perhaps in a manner otherwise
consistent with the PCT case) but does not cause suffering.
NN: neuropathic nociception: the patient experiences pain, as in the above, but here
the pain is caused by some disorder in the nociceptive system. The latter is, as it were,
derailed from its normal functioning and is firing on its own. Thus the patient is
experiencing pain, which may be experienced as having a peripheral cause, but the pain
is the result of a neuropathic disorder of the nociceptive system. An example is
phantom limb pain, where pain-system components in the brain which had been laid
down through the PCT experiences activated earlier by tissue damage in the once
present limb are re-activated. Some types of cancer patients fall under the NN heading,
where the nociceptive system has been derailed by a peripheral cancer disorder.
In addition, we distinguish two related cases of non-pain-phenomena:
PBWP: pain behavior without pain: there is, for example, a mere report in a medical
record attributed to the statements of a patient, but no pain is being experienced by the
patient (a fact which may or may not be detectable by an external observer).
TWP: Tissue-damage without pain: tissue damage normally of the sort to cause pain
does not activate the pain system (as contrasted with pain asymbolia, where pain is
experienced, so that the pain system is activated, but the patient does not experience the
pain as something that is suffered).
TWP: Tissue-damage without pain: tissue damage normally of the sort to cause pain
does not result in the subjective feeling of pain., for example because of stress asso-
ciated with sudden emergencies or because of physiological damping of the pain
process caused by endorphins. This can also be caused by the inherited disorder, con-
genital insensitivity to pain with anhidrosis.,
In a full account, we would need to distinguish also various combination cases, for
example where the patient experiences canonical (PCT) pain in conjunction with neu-
ropathic nociception In particular, we would need to take account of the fact that pain is
divided into two broad subtypes along the temporal dimension (as previously intro-
duced): Subtype 1. consists of pains of short duration (that is, acute pain): such as in the

Keio University Press Inc. 2011

 Ontology and Logic Conference 2011 5

case of a cut, a local burn, an abrasion; each involves a brief duration stimulus and
evokes a brief, intense experience of pain with accompanying reflex withdrawal that
moves the body away from the stimulus. Following the injury there is a continued expe-
rience of usually less intense pain associated with inflammation that gradually recedes
as healing occurs. Subtype 2. is chronic pain, a long-lasting sequence of experiences of
pain that emerge from acute pain that did not resolve, which may extend over many
years without relief, and which may involve the patient visiting many specialists (e.g.,
ENT, headache specialist, neurologist, dentist, psychologist) with no positive results.

4 Canonical Pain as an Evolutionarily Basic Mechanism

The canonical pain phenomenon reflects the fact that mammals have central nervous
system components that are associated with signals to the organism indicating that some
part of their peripheral structure is damaged or is in danger of being damaged. Such
signals result in various consistent outcomes on the side of the organism. This is the
sensory signalling system for pain.
It is canonical to have pain in a joint when the joint is inflamed. Coordination be-
tween tissue damage and pain is then part of the core orienting function of pain, which
is to protect the organism from harm. A patient can thus usually direct the clinician to a
particular site on or in the body where the pain is experienced.
The resultant definition of this evolutionary most basic, ‘canonical’ type of pain, in-
volving an aversive reaction towards that which is causing tissue damage, reads:
pain with concordant tissue damage (PCT) =def.
(1) a bodily process in an organism S involving two integrated levels:
(1a) activation of the nociceptive system including the pain-associated emotion-
generating brain components of S, and
(1b) a simultaneous sensory and aversive experience on the part of S
that is
(2) caused by damage to tissue located in a region R of the body of the subject S,
(3) experienced by S as being caused by this damage,
(4) such as to involve an aversive reaction on the part of S directed towards that which is
presumed by S to be causing this damage,
(5) concordant with the tissue damage on both levels (1a) and (1b),
and also
(6) such that the sensory experience is sufficiently intense to communicate the presence
of tissue damage to the subject.
Pain is thus a bodily process that has both a subjective or mental level, consisting of a
sensory and emotional experience, and an objective or physical level, comprising the
workings of the nociceptive system and of those parts of the sensory and emotional
systems that are narrowly involved in the realization of this pain; each can exist without
the other. But in the case of pain both must exist together, and they are then mutually
dependent parts within a single whole [Smith et al. 1982]. In the case of canonical PCT
pain both must further be concordant with the peripheral tissue damage that causes
them. ‘Experience of pain’ in its primary meaning refers to the sensory and emotional
6 Barry Smith, Werner Ceusters, Louis J. Goldberg and Richard Ohrbach.

experience that is an integral part of the pain process itself; in a secondary meaning it
may refer to optional cognitive representations of pain which may in certain circums-
tances be associated therewith. Animals and infants may experience pain in the former
sense but it is not clear whether they can do so also in the latter.
The canonical pain process will involve activity in many components of the central
nervous system. Part of the physical basis for this process is localized in the pain sen-
sory system and associated emotional centers. In addition, PCT pain has a physical
basis in simultaneously existing peripheral tissue damage. The tissue damage is loca-
lized in some part of the body, and the sensation is a sensation of processes in that part
of the body. The definition of PCT pain is ‘canonical’ also in the sense that it reflects
the default understanding brought to each new case by the clinician, who first assumes,
on the basis of medical necessity, that the experience of pain reported by the patient is
the result of simultaneous tissue damage.
PNT cases such as allodynia, in contrast, occur not only with tissue damage but also
often occur in a site where there was an injury that has healed. A non-noxious stimulus
to the site or an area surrounding the site produces pain. The mechanism for this could
be the local sprouting, during the healing process, of excess nerve terminals, and/or
permanent changes in the sensory system leading to the cortex that have nothing to do
with cognitive mechanisms associated with threat that are activated in the case of PCT.

5 A General Definition of Pain

We can now advance the following general definition of pain, which comprehends both
canonical pain and the distinguished variant phenomena:
pain =def. a bodily process in an organism S involving two integrated levels:
(a) activation of the nociceptive system and associated emotion generating brain
components of S, and
(b) a simultaneous aversive sensory and emotional experience on the part of S,
where (b) is phenomenologically similar to the sort of aversive experience involved in
PCT pain.
Here ‘phenomenologically similar’ means inter alia: (1) that the experience is ‘of’ or is
‘targeted towards’ some region R of the body of S, so that all pain is in this sense (and
however diffusely) localized; (2) that the experience involves a dimension of unplea-
santness which – as is shown by the case of pain asymbolia – is not necessarily of the
sort that involves suffering or aversion on the part of the subject S.
This definition is formulated in such a way that small children and even some ani-
mals can experience canonical pain, even though they do not have the cognitive re-
sources to represent their experience as one that is caused in this or that way. This ad-
dresses one recognized shortcoming of the IASP definition [Rollin 2006].

6 Problems of Diagnosis with PNT and NN Pains

Pain results in behavioral responses in animals similar to those observed in humans; but,
trivially, only humans experience pain in a way that is linked to the ability to speak of it
– the latter reflecting the contribution that cognition has in affecting our basic percepts.

Keio University Press Inc. 2011

 Ontology and Logic Conference 2011 7

It is for this reason that the IASP definition gives a prominent role to descriptions con-
taining reference to tissue damage of varying states. Such descriptions are central to the
clinician’s understanding of pain phenomena of both the PNT and NN types, neither of
which necessarily has apparent tissue damage at the putative locus of pain, because the
corresponding experiences are heavily influenced by processes independent of direct
stimulus transmission.
Clinicians have significant problems with evaluating (quantifying) pain intensity, or
the presence of pain itself, in those cases where no observable tissue damage or mal-
functions in any component of the patient’s body can be observed. Indeed, even if tis-
sue damage or malfunctions can be observed, pain intensity cannot be quantified in a
purely objective manner for the same reason that other feelings cannot be quantified.
While other attributes such as episode frequency or duration are more amenable to
quantification (because they are more readily observable in an objective manner), the
attribute of intensity is often the most salient one due to its greater susceptibility to
change as well as the suffering aspect of pain which appears to be heavily influenced by
intensity.
One such case is that of the well-known disorder referred to as ‘regional myofascial
pain’, in which the fibers in the muscles of the jaw, neck or lower back, affected by a
myofibril disorder that is putatively due to some form of trauma [Mense 1993], but
where biochemical exploration has largely failed to find signs of overt inflammation
suggestive of such trauma or of tissue damage. Many theories are associated with both
the phenomenon and why it is painful, but controversy is considerable. We believe that
this comes close to what IASP means by ‘potential tissue damage’. For us it is a case of
PNT, because while the pain is intense, the peripheral physical disorder – a disorder in
myofibrils – would not normally be of the sort to give rise to a pain of this intensity
Another example is provided by PNT pains with the dominant characteristic of ‘al-
lodynia’. The clinician applies non-noxious pressure to a tooth; the patient senses the
increase in pressure and reports ‘pain’. Examination of the tooth, via direct observation
and radiographs, fails to disclose any evidence of pathology (tissue damage), and yet
the patent clearly localizes the pain of complaint in the tooth; moreover, the pain
evoked by the examination replicates the pain of complaint.
This disorder was, for many years and in particular prior to the IASP definition, at-
tributed to psychiatric cause; but when better knowledge emerged regarding an underly-
ing neuroscience for allodynia, the psychiatric causation was transformed into a diag-
nostic entity currently referred to as ‘atypical odontalgia’ – pain that feels like a too-
thache but that is not due to peripheral damage of the tooth. It is one implication of our
discussion above, that atypical odontalgia must be divided into two distinct kinds,
representing pain, respectively, of the PNT or of the NN sort. In the first case, for ex-
ample, a tooth has been treated for caries and the process affected the nociceptive sys-
tem of the patient in such a way that the patient feels discordant pain that is (for the
patient) localized in the corresponding tooth. In the second case, the relevant portions of
the patient’s nociceptive system have become disordered for example in virtue of some
virus infection, in such a way that the patient feels pain that is localized in just the same
way as in the PNT case, but which involves no tissue damage at the corresponding
locus.
8 Barry Smith, Werner Ceusters, Louis J. Goldberg and Richard Ohrbach.

7 Ontology of Pain and Ethics of Pain Diagnosis: Problems

of Diagnosis with PBWP
Certainly there are behavior-based measures of pain, for example in terms of loudness
and frequency of overt expressions. Given the PBWP phenomenon, however, these are
in fact measuring two different things, since in the case of PBWP there is by definition
no pain to measure. Science based on comparing the two sets of data appears empirical-
ly ungrounded.
Such cases make diagnosis in matters of pain especially difficult, and although the
types of pain-related phenomena described here can be clearly distinguished in general
ontological terms, identifying which type is exemplified in any given instance is by no
means trivial.
For good, diagnostic reasons the IASP definition is standardly interpreted in such a
way as to allow even mere descriptions of pain to warrant pain classification; there are
also essential ethical reasons for the degree of permissiveness allowed by the IASP
definition with regard to clinician action. When an experience is described by the
patient in terms of peripheral tissue damage, the clinician will standardly not be in a
position to assert that the reported experience is not correctly so described. It is
precisely due to this feature of the IASP definition that pain research has progressed so
significantly in the past 50 years. Data could be collected at the margins of pain without
the clinician (investigator) having to be judge and jury regarding the patient’s reports of
his experience. But factitious pain poses significant problems, and if research into the
gigantic problem of pain that is experienced as being localized but is without localized
tissue damage is to be successful, then some supplement to the IASP definition is
needed, of the sort which, we believe, ontology can provide.
Our goal here is to initiate the development of an approach which allows the clini-
cian or researcher to better understand the physical basis underlying a report of pain and
not just to stay at the level of reports and of the assumption that, if the patient says that
it is pain (within the limits of language relating to tissue damage of one sort or another),
then therefore it is pain (or, as pain-clinicians will often say, for the benefit of patients,
‘all pain is real’).
If the clinician expects concordance between stated intensity (the symptom as re-
ported) and the clinical findings (the signs), but concordance does not occur, then sig-
nificant problems will ensue, either in the form of dismissing the disorder, or in labeling
the patient as ‘psychiatric’. If, in contrast, the clinician understands the neuropathic and
other non-peripherally localized contributions to pain experience, then this may serve a
more adequate diagnosis.

8 Practical Applications of the Pain Ontology

Increasingly, progress in pain research and in diagnosis of pain will require analysis of
new types of data, including:
(1) PET or fMRI data, ;
(2) data linking genotype risk to present pain in the interest of making prognostic
statements, including predicting likelihood of future relapse on the basis of identified
pain genes;

Keio University Press Inc. 2011

 Ontology and Logic Conference 2011 9

(3) classification structures that will incorporate multiple axes (including data al-
ready available concerning pain history, hard and soft tissue imaging, psychological
data) so as to generate a multi-dimensional classification, producing clarification of
pain subtypes as for example in the case of pains strongly characterized by allodynia,
which we believe may be applied also in the troubled field of cancer pain [Hjermstad et
al. 2009].
We believe that ontology-based research has already shown its value in supporting
the integration that is required for full exploitation of such bodies of multi-dimensional
data, and ontologies in a range of biomedical domains are now being developed in ways
designed to serve such integration [Smith et al. 2007]. In the pain domain the ontologi-
cal approach will in addition provide considerable value in allowing us to differentiate
the categories involved even where we do not know to which categories given patients
belong, because on the basis of the available clinical evidence we do not know the de-
gree of match – or mismatch – between reported experience and the underlying neural
processes. Knowing what these categories are then allows us to analyze the different
types of data in ways which are unavailable on a more diffuse approach by allowing
statistical analysis on the basis of alternative hypotheses as to the proportions of differ-
ent sorts of mismatch in any given batch of patient reports.

9 Conclusion
A number of pain-related disorders found in specific, localized regions of the body and
currently classified, for example, separately as temporomandibular disorders, orofacial
pain, mucosal pain, odontogenic pain, regional neuropathic pain, and headache span the
types of pain described in the above. Many of these phenomena are marked by a similar
array of medically unexplained symptoms. The primary goal of the work described in
the foregoing is to support research directed towards a better understanding of such
phenomena and thereby also a better classification of patients according to
susceptibility of responding positively to different kinds of therapy. Barriers to
improved classification and subsequent research have thus far been due, in part, to the
difficulty associated with going beyond traditional clinical perspectives and
assumptions. At issue is the emerging recognition that many body regions have
associated chronic pain disorders that are more alike than they are different, whereby
one of the primary characteristics is: presence of persistent dysfunctional pain
disproportionate to the observed pathology (here labeled PNT). Repeated observation
has indicated that many of these disorders appear to co-occur at a higher rate than
chance would suggest; yet research into this complex domain has, we believe, had
trouble moving forward in part due to the inadequacy of the classification of pain and
related phenomena that is implied by the IASP definition.
We believe that our ontological account of pain, and of those phenomena closely
related to pain commonly described as pain in patient pain reports, can significantly
contribute to advancing our ability to more successfully understand, diagnose and treat
pain and related phenomena.
10 Barry Smith, Werner Ceusters, Louis J. Goldberg and Richard Ohrbach.

Acknowledgements
We acknowledge support of the Oishei Foundation, OPPERA: Orofacial Pain Prospective Eval-
uation and Risk Assessment (NIDCR/NIH DE017018), and NCBO (NIH Roadmap 1 U 54
HG004028). We also thank Olivier Massin and Kevin Mulligan of the University of Geneva for
helpful comments.

References
[Bodenreider 2008] Bodenreider, O. Ontologies and data integration in biomedicine: Success stories and
challenging issues. In Data Integration in the Life Sciences (Lecture Notes in Computer Science),
New York: Springer, 2008; pp. 1-4.

[Ceusters et al 2010] Werner Ceusters and Barry Smith, “Foundations for a realist ontology of mental dis-
ease”, Journal of Biomedical Semantics, 2010, 1:10.
[Hjermstad et al. 2009] Hjermstad MJ, Fainsinger R, Kaasa S. Assessment and classification of cancer pain,
Current Opinion in Supportive & Palliative Care, 3; 2009: 24-30.

[Kanovich et al. 2003] Max I. Kanovich, Mitsuhiro Okada and Andre Scedrov, Phase Semantics for Light
Linear Logic, Theoretical Computer Science 294(2003), pp. 525-549.
[Knuth 1984] Donald E. Knuth. The TeXbook. Addison-Weskey, 1984.
[Lamport 1994] Leslie Lamport. LaTeX : A Document Preparation System. Addison-Wesley Professional, 2nd
edition, 1994

[Melzack 2001] Melzack R. Pain and the neuromatrix in the brain. Journal of Dental Education. 2001; 65(12):
1378-1382.
[Mense 1993] S. Mense. Nociception from skeletal muscle in relation to clinical muscle pain. Pain 54:241-291,
1993.

[Merskey et al 1979] Merskey H, Albe-Fessard DG, Bonica JJ, et al: Pain terms: A list with definitions and
notes on usage. Recommended by the IASP Subcommittee on Taxonomy. Pain 1979; 6:249-252.
http://www.iasp-pain.org/source/eforums
[Rollin 2006] Rollin BE. Science and Ethics, Cambridge : Cambridge University Press, 2006.

[Rosse et al. 2007] Rosse C and Mejino JLV. The Foundational Model of Anatomy ontology. In A. Burger, D.
Davidson, and R. Baldock, eds., Anatomy Ontologies for Bioinformatics: Principles and Practice,
London, 2007; pp. 59-117.
[Scheuermann et al. 2009] Scheuermann R, Ceusters W and Smith B. Toward an ontological treatment of
disease and diagnosis. AMIA Translational Bioinformatics Summit 2009, pp. 116-120.

[Smith et al. 1982] Smith B. and Mulligan, K. Pieces of a Theory, in B. Smith (ed.), Parts and Moments.
Studies in Logic and Formal Ontology, Munich: Philosophia, 1982, 15-109.

[Smith et al. 2007] Smith B, Ashburner M, Rosse C, et al. The OBO Foundry: Coordinated evolution of
ontologies to support biomedical data integration, Nature Biotechnology 2007; 25 (11): 1251-1255.

Keio University Press Inc. 2011

 Ontology and Logic Conference 2011 11

Table 1. Ontology of Pain and of Pain-Related Phenomena

Subjectively Objectively
Observable Observable Physical Basis Examples
Features Features

Canonical Pain

PCT: Pain Pain Manifestation of Peripheral tissue Primary sunburn

with tissue damage damage Pain from strained
concordant Signals sent to Intact muscle
tissue nociceptive nociceptive
damage Pain from fracture
system system
Pulpitis
Activation of
emotion-
generating brain
centers, which
can produce
increased heart
rate, blood
pressure,
galvanic skin
response.

Variant Pain

PNT: pain Pain Manifestation of Physical Myofascial pain

without some disorder in disorder of disorder
concordant the patient amplitude Tension-type
tissue Signals sent to control headache
damage nociceptive mechanisms
associated with Chronic back pain
system
the nociceptive
Patient reports of system
pain are either
exaggerated or Intact
muted relative to nociceptive
disorder system
Activation of
emotion
generating brain
centers
 12 Barry Smith, Werner Ceusters, Louis J. Goldberg and Richard Ohrbach.

NN: neuro- Pain Neurological test Disorder in the Trigeminal neuralgia

pathic confirming nociceptive Post-herpetic
nociception nerve damage system neuralgia
Diabetic neuropathy
Central pain

PRP: Pain-Related Phenomena Without Pain

PBWP: Unknown Report of pain Mental states Factitious pain

pain Sick role such as anxiety, Malingering
behavior behaviors rather than
without peripheral tissue Anxiety-induced pain
accompanied by report
pain normal clinical locus
examination Disordered
Grossly emotional or
exaggerated pain cognitive
behaviors systems
misinterpreting
Identified sensory signals
external
incentives

TWP: No pain Manifestation of Suppression of Stress associated with

tissue- tissue damage pain system by sudden emergencies
damage normally of the one or other Physiological damp-
without sort to cause mechanism ing of the pain process
pain pain caused by endorphins
Placebo-induced opio-
id analgesia
Genetic insensitivity
to pain

Keio University Press Inc. 2011

 Ontology and Logic Conference 2011 13

Figure 1: Top-Level Ontology of Pain and Pain-Related Phenomena