Referencing as Cooperation or Competition

Full text

Henry Small
Referencing as Cooperation or Competition
 The Citation Process Revisited
Blaise Cronin’s book The Citation Process (1984) gave the information science and
bibliometrics communities their rst major statement on citation theory. It framed
the theoretical discussion around the two principal approaches to science studies
at the time, the Mertonian normative view and the post-modernist social construc-
tivist view. The book saw the conict in sharp terms: between those who espouse a
positivist, normative, and aggregationist view of citation, and those who see it as
subtle, individualistic, and a product of research-in-practice, where norms carry
little or no weight.
Rereading The Citation Process after a hiatus of 30 years recalled feelings of
both excitement and depression. I was energized by these tentative theoretical
steps, but disheartened by his suggestion that we should follow the lead of the
social constructivist. As someone trained in science and the history of science,
the constructivist view did not ring true. Perhaps I was stuck in my story-book
version of science. In any event, the bibliometrics community ignored the new
sociology and remained largely empirical and atheoretical.
In a later paper, Cronin (1998) revisited the problem of citation theory in a
way more congenial to me, proposing a middle ground between the normative
and constructivist approaches, and then later even seemed to return to a more
normative position in the context of semiotic theory (2000).
This long interlude of three decades saw many shifts in the positions of major
players in the contending camps, among these, the split between the bibliometri-
cians and constructivists with the formation of ISSI in 1993 and the burgeoning
of the 4S society. My purpose in revisiting this debate is to better understand the
fundamental issues it raised regarding the nature of science, the shortcomings
of both constructivist and normative theories, and to suggest some possible ways
forward based on recent theories in evolutionary biology which give a rationale
for cooperative behavior.
DOI 10.1515/9783110308464-006, © 2020 Henry Small, published by De Gruyter.
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 License.

50 HenrySmall
 Social Construction and Its Problems
One of the fundamental tenants of social construction is that science does not
have direct access to the “reality” of the external world. Rather, that access is so
thoroughly mediated by devices, presuppositions, and social constraints that our
knowledge is not about the external world at all. Furthermore, what knowledge
we do have is relative to the framework and circumstance in which it is created,
and no one framework is more valid or closer to reality than any other. In Latour
and Woolgar’s (1979) view, scientists in the lab are engaged in processes of inscrip-
tions that lead to the creation of a “mythology.” Inscription devices in the lab act
as black boxes allowing social interactions to dictate their output: “There thus
occurs a transformation of the simple end product of inscription into the terms
of the mythology which informs participants’ activities” (Latour & Woolgar, 1979,
p. 63).
From the “strong program” of the Edinburgh school, we have the view that
scientic knowledge is no dierent from any other belief system, including those
of primitive cultures, and the notion that “belief systems cannot be objectively
ranked in terms of their proximity to reality or their rationality” (Barnes, 1974,
p. 154). Knorr-Cetina, in The Manufacture of Knowledge asks (1981, p.2), asks:
“Why should our interest-geared, instrumentally-generated world order mirror
some inherent structure in nature?” She claims that the lab is an articial and
constructed environment, and this social environment and its instruments manu-
facture what we call science, which bears no necessary relationship to the world.
According to Latour and Woolgar, the ultimate aim of the lab is the production
of papers. In writing papers the task is to persuade fellow scientists and the out-
side world that the mythology they have created in the lab is valid. It is impossible
for facts and theories to be convincing on their own because, as myth, they can
only be made convincing by rhetorical means.
Knorr-Cetina sees the scientic paper as yet a further construction with its
own unique arguments and end products which bear little or no resemblance to
what went on in the lab. In fact: “Compared with the work observed in the lab-
oratory, the written paper is … a rst complete perversion” (Knorr-Cetina, 1981,
p. 132). One reason for this is that the paper does not reveal the more or less irra-
tional and haphazard path that went on in the lab. The paper constructs a dierent
“logic” and sequence of events for the benet of various audiences. Knorr-Cetina’s
view is echoed by the rhetorician Ken Hyland who states: “Texts … can never be
regarded as accurate representations of what the world is like … Reality is con-
structed through processes that are essentially social and involve crafting texts in
ways which will be persuasive to readers” (2009, p.12).

Referencing as Cooperation or Competition  51
The message is that we need to analyze the various rhetorical devices that
scientists use to convince (i.e., mislead) readers into thinking that the writer is
describing the physical world. As Swales puts it: “The art of the matter, as far as
the creation of facts is concerned, lies in deceiving the reader into thinking that
there is no rhetoric … that the facts are indeed speaking for themselves” (1990,
p. 112).
To show how “facts” are created in the lab, Latour introduces the idea that
there is a gradation in the so-called “facticity” of knowledge-claims. This is em-
bedded in statements scientists make about these claims and is marked by the use
of “modalities” which, to varying degrees, cast doubt on or boost the credibility of
given assertions. Modalities are expressed by words or phrases such as “prema-
ture,” “suggested,” “reported,” “rst described,” “assumed,” “conrmed,” etc.
The goal of the lab is to move statements from low to high facticity by rhetori-
cal persuasion such that all modalities disappear from the language. “Facticity”
does not mean that some fact about the real world has been established, only that
actors view it as such.
Latour introduces a theory of citation in Science in Action. In the citation con-
text, the text surrounding the reference, modality words and phrases are deployed
either to strengthen those who support the author or weaken the opposition. Ref-
erences are massed as opposing armies on a battleeld. In citation warfare all
distortions of the meaning of the prior text is fair game: “…do whatever you need
to the former literature to render it as helpful as possible …” (1987, p. 37) and “… all
deformations are fair” (1987, p. 40). Referencing becomes part of the process of
persuasion in which one mythical version of the world is promoted over another.
Constructivist theory is now under attack from within, in part, because it has
been hijacked by reactionary groups that invent their own versions of science (La-
tour 2004; Collins, 2014). Nevertheless, it is useful to review some of the objections
that have a bearing on a theory of citation.
One problem is that fraud or error in science would be indistinguishable from
any other type of science. If facts are socially constructed, then error and fraud
are as well. Failure to replicate or corroborate an author’s result would carry no
weight because these activities are not anchored in the real world. Replication or
reproducibility is, however, the basis for the detection of fraud or error in science
and is fundamental to the normative operation of science (Zuckerman, 1977; Hull,
1988, p. 435).
For constructivists there is no way of evaluating or comparingthe relative mer-
its of one group’s scientic views over another group’s because neither is based on
reality. Each group can provide persuasive arguments for the validity of its beliefs.
Kuhn (1962), however, rejected this extremerelativism and conjectured that there
could be progress through scientic revolutions. Because of the increasing growth

52 HenrySmall
and specialization of science through a revolution “… both the list of problems
solved by science and the precision of individual problem-solutions will grow and
grow” ( p. 169).
Constructivists rely on rhetorical persuasion (Gilbert, 1977) to convert non-
believers to believers presumably because the evidence is not suciently com-
pelling on its own. Rhetorical devices include hedging and purging personal mo-
tives (Swales, 1990, p.112). However, if rhetorical persuasion is all that it takes
to win converts, it is unlikely that Einstein’s theory of relativity or Heisenberg’s
matrix mechanics would have gained many converts. Persuasion can also come
about by presenting a coherent framework of theory and observations. In some
cases, a new nding is compelling enough to bring about the emergence of a new
research area (Small, Boyack, & Klavans, 2014). This is because rhetoric is weak
compared to the perception of a new ordering of facts and theory (Cole, 1992,
p. 47).
It could also be argued that the rigid form of the modern scientic paper
(Swales, 1990, p. 134) militates against persuasive presentation. The conventional
IMRD format (introduction, methods, results, discussion) forces papers to present
their ndings in a uniform manner which may facilitate their browsing, compar-
ison, and registration in databases, but does not facilitate discursive argument
(Zuckerman, 1977, p. 125). Unlike the earliest scientic reports published in jour-
nals such as the Philosophical Transactions of the Royal Society, the modern
scientic paper is not designed to be a rst-hand account of an author’s obser-
vations in the lab and does not allow the “virtual witnessing” that Robert Boyle
practiced (Shapin & Schaer, 1985, p. 55).
Regarding Latour’s views on referencing as a strengthening of allies and a
weakening of enemies, there is little evidence that such a nuanced deploying of
individual references has much eect on a paper’s impact. Much more depends on
the actual ndings reported in the paper, on the data or theory it presents. Watson
and Crick’s 1953 paper on the structure of DNA had only six references, Darwin’s
Origin of Species relatively few, while Einstein’s 1905 paper on special relativity
contained no references at all. A careful deploying of modalized references is not
enough to make a trivial paper convincing.
 The Realist Alternative
The idealist position espoused by the constructivists can be contrasted with the re-
alist position that we are observers of a world which we have access to through our
senses (Leplin, 1984). We extend the power of our senses by inventing instruments

Referencing as Cooperation or Competition  53
and analytical methods that magnify or probe our environment. The instruments
are not the black boxes of the constructivists because they obey the same physical
laws as the thing being observed. To understand the world, we make systematic
observations, and invent hypotheses and theories to explain our observations.
When our observations and theories agree, we claim to have discovered some-
thing about the world, but discoveries are always tentative and subject to revision
by more accurate observations or dierent theories that have better agreement
with the observations. We can think of this tentativeness of knowledge as a norm
which resulted from the numerous theories in history which turned out to be false.
Despite this rosy picture, the signals from nature are not always clear. They
can be noisy, contaminated, ambiguous, and aected by our prejudices, precon-
ceptions, biases, and expectations. When an experiment is successful, it is not
always apparent that it is until it can be replicated or corroborated by other ex-
periments or with theory. Scientists can disagree about the meaning and inter-
pretation of these signals. The realist response is to call for more research, and
scientists are always thinking about their next experiment.
Most scientists committed to a theory are aware of the possibility that new
evidence might prove them wrong. But the amount of eort required to develop
and test a theory requires a level of commitment that makes it dicult to keep an
open mind to alternative theories (Mitro, 1974). However, what scientists work
toward is a plausible basis for further research, and this may involve abandoning
one hypothesis in favor a more plausible one. Latour’s levels of facticity are also
relevant in a realist approach as a means of sorting the plausible from the implau-
sible. Thus, an important question is what makes some theories more plausible
than others?
The colloquialism “the facts speak for themselves” expresses the common
sense notion that the plausibility of hypotheses depends on their degree of t with
the existing body of facts and theories, like the tting together of pieces of a puz-
zle. For example, quantum mechanics ts with the ionization potentials of simple
atoms. The double helix model of DNA ts with the concept of genetic replication.
Human-caused global warming is consistent with an increase in the burning of
fossil fuels and more carbon dioxide in the atmosphere.
Each of these examples of consistent theory and observation can be expanded
into networks comprised of many elements that t together in a larger puzzle. For
example, the quantum mechanical theory of the atom is also consistent with the
electromagnetic spectrum of the atom and the quantum nature of light. Anthro-
pogenic global warming is consistent not only with rising carbon dioxide levels in
the atmosphere, but also with the decrease in energy being re-radiated back into
space, and the wavelength of the main component of re-radiated energy is con-
sistent with the energy spectrum of carbon dioxide. A similar point was made by

54 HenrySmall
Gingras and Schweber in their critique of a social constructivist account of quarks
(1986, p. 379). In theory choice, scientists consider a network of multiple facts and
predictions, not single facts in isolation.
The general approach to tting facts and theories together is called con-
silience and derives from the work of the 19th century philosopher and historian
William Whewell (1847) recently made popular by E. O. Wilson (1998). Whewell
wrote: “The consilience of induction takes place when an induction, obtained
from one class of facts, coincides with an induction obtained from another
dierent class” (1847, p. 469). A related network approach called “explanatory
coherence” has been proposed by Paul Thagard (1992; 2007). In this approach
various types of coherence relations are treated as constraints and the network
having the highest constraint satisfaction is considered the most likely to be
true, or at least, the best one given the currently available facts and theories.
Explanatory coherence provides an approximate guide to theory choice and a
way of understanding the history of science.
In writing scientic papers, authors are also engaged in tting together the
pieces of a large puzzle that represents the problem space of their research area.
Some of this tting together (but not all) is evident in the references authors cite
which associate facts, theories, or methods with prior papers. In fact, Thagard
applies his explanatory coherence to the process by which scientists arrive at a
consensus which, he notes, requires communication among researchers (2000,
p. 223).
Referencing is a passive form of communication between the cited and citing
authors so we can use this to illustrate consilience. Citation contexts, that is, the
portions of text where the papers are referenced, can reveal how authors see the
prior literature. For example, a citation might represent the linking of a cause and
eect, where the eect is an observation and the cause is a deduction from theory
(Hanson, 1972).
To illustrate how a scientic paper builds a network of coherent facts and the-
ories through its references, we use a paper that was part of a co-citation cluster
from 2007 on the water pollution by estrogens (Small & Klavans, 2011). Focus-
ing on a single randomly selected citing paper (Thorpe, Benstead, Hutchinson, &
Tyler, 2007) from this cluster, all the contexts were extracted in which references
were made (see Table 1).
The 23 contexts are arranged sequentially by the section of the paper. Con-
tent words have been removed, leaving the non-technical or general words. This
allows us to see more clearly the function of the reference in the authors’ presen-
tation. The cited references within each context are denoted by integers in square
brackets. Some contexts reference multiple items, so-called redundant references,
and some items are cited repeatedly throughout the paper, the op. cit. references.

Referencing as Cooperation or Competition  55
Tab. 1: Reference structure of a scientic paper: citation contexts stripped of content words.
Introduction
. It is now well-established that… may impair… with potential detrimental consequences
[, , , , , ]
. … implicated as causative []… were not previously subject to routine monitoring due
to …
. In this regard …has been widely employed… that induce… response []
. … that is produced … in response to … [, , ]
. … but exposure to… has been shown to result in… [, ]
. There is some evidence that …production …as a consequence of … [, , , , , ,
] but the implications of …are less clear.
. Even less is known on the consequences of … although an association between … has
been reported [] …
. … provide the majority … [] and therefore … could potentially impact on …
. … but is of considerable… relevance and was therefore used in …to expand on the ear-
lier work of [] who investigated for association between …
Materials and Methods
. … basic design … same … replicate treatment in experiment … [ self]
. … samples were assayed … using []
Results and discussion
. … is consistent with reports from earlier investigations using… [ self, ] and indi-
cates that … signicantly exceeds …
. This supports the results of an earlier investigation reporting … []
. … it has been hypothesized that… is thought to result in … [, ]
. Indeed, a number of investigators have reported …eects … which are hypothesised to
result from … [, , , , ]
. The eect … [ self] shows …
. This compares with previous observations where … were linked with … eects … but not
in … [, , , ,]
. It has previously been demonstrated that … are associated with … results in … leading
to … [].
. The observed …here is consistent with a previous investigation [] …
. This may reflect alterations …due to … and supports the earlier work of [] who
showed … through … examination of …
Conclusions
. The collective results from these investigations support an earlier investigation [] in
demonstrating that … signals … adverse … impact …
. The poor ability … however could lead to … that have adverse health eects during … as
can occur …and should be considered further [, ]
. This further supports the work of [, ] in demonstrating that …could potentially be
used to signal for adverse …health eects.

56 HenrySmall
For example, the rst context in the Introduction refers to “well established” facts
that potentially lead to detrimental consequences. In other words, the established
facts form a coherent picture and together with theory predict certain undesirable
eects. In the ninth context earlier work was expanded on because an association
was found. Associations suggest possible causes. The contexts in the introductory
section emphasize causation using words such as “causative”, “induce”, “in re-
sponse to”, “result in”, etc. Words like “may”, “could”, and “potentially” act as
hypothetical connections that require further investigation.
Contexts in the Results and Discussion section focus mainly on documenting
the consistency of the paper’s results with empirical ndings and hypotheses of
earlier papers. Words such as “consistent”, “supports”, and “compares” build a
web of relationships between the authors’ current results and the previous liter-
ature. These relationships dier from the cause-eect relationships of the Intro-
duction in that they are mainly about similarities and parallels between results
by dierent investigators.
Instead of focusing on all contexts from a single paper, we can also look across
all contexts at a particular cited reference. In the selected paper, reference #11
had the highest op. cit. rate, suggesting that it was of particular importance to the
authors. There were a total of 16 contexts for reference #11 across the 868 con-
texts from all papers in the sample. We then selected the most characteristic of
these contexts by computing the cosine vector similarity of each of the 16 contexts
against a composite of all 16. The context with the highest cosine vector (0.61) was
the seventh context in Table 1, and is thus the most representative of the 16:
Even less is known on the consequences of disruptions in VTG dynamics in females and
although an association between VTG induction and reduced egg production has been re-
ported [11], the eects on egg production were only observed at concentrations that were
toxic to males.
(Thorpe et al., 2007, p.177)
Thus, this most typical context reports an association, or consilience, between
three observations: VTG induction, egg production, and toxicity to males. These
examples show that references serve to link observations with theory and eects
with varying degrees of certainty.
 The Problem of Norms in Science
What is missing in the previous discussion of the coherence model of plausibility
is what establishes the acceptable forms of coherence between the pieces of the

Referencing as Cooperation or Competition  57
puzzle. A few of these noted already are the degree of t between theory and ex-
periment, the accuracy of theoretical predictions, and the qualitative agreement
of a model and the evidence. Thagard mentions explanation, deduction, and as-
sociation (2000, p. 17). But where do these criteria come from?
One answer is that these are technical or epistemic norms analogous to the
norms that govern the social conduct of scientists proposed by Robert Merton,
such as, universalism, communality, disinterestedness, skepticism, and origi-
nality. Merton considered social norms necessary for the “extension of certied
knowledge” (1973 p. 270). Among the technical methods that allowed the creation
of certied knowledge were empirical conrmation and logical consistency. The
social norms supported the goal of certied knowledge.
Merton, however, does not spell out the “technical methods” that should gov-
ern scientic practice, or where they came from. These are discussed in greater
detail in a long article by Zuckerman (1977). She called them “cognitive norms
and methodological canons” (1977, p. 87), and related them to the philosophical
concept of demarcation, that is, the rules that dene what it is to be scientic.
The problem with both technical and social norms is that they appear to come
out of nowhere, although Merton’s early writings (1938) suggest a possible link to
the Puritan values held by many members of the Royal Society in the 17th cen-
tury. Zuckerman hints at the possible dynamic nature of norms when she dis-
cusses how the norm of disinterestedness, which encourages unrestrained pure
research, comes into conict with the social hazards of certain scientic ndings
(1977, p. 122). This suggests that under some circumstances restraints on pure re-
search may be necessary. Thus, a dynamic and evolutionary theory of social and
cognitive norms seems to be called for.
Mulkay rejects the idea that norms aect or control behavior. He sees moral
precepts as embodied in Mertonian norms as exible vocabularies that are in-
voked rhetorically to rationalize scientists’ interests (Mulkay, 1991, p. 69). Yet, just
paying lip service to norms does not seem adequate to explain why so many sci-
entists adhere to formal conventions and rules in their work, and how these con-
ventions and rules arose and become sustained.
Technical norms could be collectively considered part of the “scientic
method” (Gower, 1997) for a given historical period. Francis Bacon provided
many examples of technical prescriptions for science including the method of
induction, the gathering of systematic observations, and the conducting of ex-
periments. Philosophers of science have proposed criteria on which to judge the
adequacy of theories called criteria for theory choice. Kuhn provided a short
list of what he considered key criteria or “values”: the accuracy of theoretical
predictions, the consistency of the theory with other accepted knowledge, the
ability of the theory to expand its scope to predict other phenomena, the need for

58 HenrySmall
a theory to be simple or parsimonious, and fruitful in generating new problems
and solutions (1977, p. 322). Rather than seeing these as givens, however, it should
be possible to trace their origins in history.
Technical norms clearly change over time and are subject to selection and
extinction: what is acceptable or required in today’s science is not what was ac-
ceptable or required in earlier historical periods. But norms probably change in a
reactive rather than a proactive manner. As Hull puts it, “The nature of science is
constantly under negotiation, and the currency of these negotiations is success”
(1988, p. 297). The availability of a successful or highly visible social or technical
practice might rst become fashionable then later on required. That “accounts”
should “save appearances” is one ofthe oldest technical norms perhaps having its
origin in ancient Egyptian and Mesopotamian creation myths (Frankfort & Frank-
fort, 1949, p. 11).
Examples of technical norms that have gone extinct are the requirement that
theories be consistent with the writings of the ancients, or with the teaching of
the Church. A norm that emerged in the scientic revolution was that predictive
theories should be mathematical in form. This norm was spurred by the success
of Newton in predicting the motions of celestial bodies (and perhaps earlier ac-
cording to Crombie [1959]). Likewise, Lavoisier’s theory of combustion resulted
in adoption of quantitative criteria in the explanation of chemical change (Kuhn,
1977, p. 336).
A technical norm of more recent origin is the notion of symmetry which was
introduced in particle physics and relativity theory. Medicine has introduced its
own norms such as double-blind clinical trials and evidence based medicine. Be-
sides being induced by scientic discoveries or successes, it is likely that norms
migrate from one branch of science to another. An example is the diusion of
statistical methods to various disciplines. Practical innovations can also be exem-
plars for new epistemic norms. For example, as new and more accurate scientic
instruments are introduced, the standards of measurement increase and higher
precision becomes required.
A potential diculty in theory selection arises when a new discovery stimu-
lates the adoption of a new norm which is then used to rationalize the discovery
(Kuhn, 1977). It is not clear how often this situation arises, but it may account
for the delays in acceptance of some theories, such as relativity or string theory
where radical new ways to understand the world are proposed which are not eas-
ily testable. However, if the theory has multiple conrmatory paths, some of which
rely on traditional criteria, this circularity is less problematic.
Mulkay argues that scientists often justify their behavior using a wide range of
rules—some of which are contradictory—and their behavior can contradict their
own stated rules (1980; 1991). He also notes, referencing Kuhn (1977), that when

Referencing as Cooperation or Competition  59
technical norms are applied to theory selection, dierent scientists can arrive at
dierent choices. It is not dicult to envision situations where norms come into
conict, for example, when an author attempts to publish a paper that violates
technical norms. Merton (1963) recognized that conicting norms create ambiva-
lence towards them, for example when originality conicts with humility. The
complex and contradictory nature of norms does not, however, invalidate their
importance.
The social norms discussed by Merton under broad categories probably also
evolved from exemplars of good practice or as reactions to new social realities. For
example, the norm of universalism may be related to the rise of distinctive styles
of national science in Europe (Ben-David, 1984), and the need to assert that scien-
tic ndings are valid across national boundaries. The invention of the scientic
journal in the 1600s may have crystallized the norm of communality as well as
numerous publishing conventions as the medium evolved.
Norms also carry dierent weights and are associated with varying degrees
of sanctions, and the importance of a norm and the sanction that accompanies
its violation would likely change over time. For example, the norm of openness is
probably more important today than it was the in 17th century when many scien-
tists kept their discoveries secret for fear of not receiving proper credit.
The norm of honesty, which falls under Merton’s category of disinterested-
ness, however, carries a more severe sanction. Fraud, if proved, can jeopardize the
scientist’s career. Without adherence to the norm of honesty the scientic commu-
nity would probably ceaseto function. Scientists could no longer trust one another
and would lose the support of the larger society (Zuckerman, 1977). The norm of
honesty in reporting scientic results may derive, in part, from the impracticality
of eye witnessing experiments (Shapin & Schaer, 1985) and was a necessity if
scientists were to work independently.
 Norms and the Scientic Paper
Central to the integrity of science is the connection between what is done in the
lab and the nal written scientic paper. Knorr-Cetina has shown that the relation-
ship between what happens in the lab and the nal report for publication is com-
plex (1981, p. 94). Results are selected, not reported as they actually happened,
and the argument may be reframed for various audiences. However, through these
transformations the author must take care not to misrepresent his or her results.
In some labs this norm is enforced by requiring the maintenance of lab notebooks
and diaries which can be reviewed in cases of suspected misconduct (Gaulton,

60 HenrySmall
2004). What is missing from Knorr-Cetina’s account is the powerful eect exerted
on the authors by both social and technical norms. The social norm of honesty is
the most important, but many other norms and conventions govern the form and
content of the paper, its style, sections, and references. Failure to follow generally
accepted technical norms and conventions may jeopardize publication.
We could speculate about what the scientic “paper” would look like if we
stripped away the norms and editorial conventions. Loss of a standardized format
or organization would make science appear more “literary” or perhaps autobio-
graphical. The personalization of the paper would probably result in fewer col-
laborators and co-authors. The paper would likely not start with a review of the
state of knowledge with references to prior literature. The most dramatic eect
would be loss of trust. Authors would have no need to tell the truth, and would
not be held accountable. Readers would no longer have condence that the author
actually observed what was observed or did what was said was done.
In 1988, Merton explicitly discussed referencing as a normative constraint in
science as part of the “composite cognitive and moral framework” (p. 622) which
had historically evolved. He pointed to its main function as a “moral obligation to
acknowledge one’s sources” and explained its origin as a response to the social
problem of plagiary in the 17th century. In 1965, Kaplan noted that there were few
if any normative guides for citation practices in the available style handbooks.
Nowadays we nd many such guides and prescriptive texts (e.g., see Kamat &
Schatz, 2014).
So how do we know that a norm of citation is operating and has an eect on
behavior? Referee reports and letters to the editor are lled with complaints that
author X has failed to cite author Y (Retraction Watch, 2014; Hagstrom, 1974). One
kind of evidence is psychological discomfort: an author’s real or imagined em-
barrassment on failing to cite an obvious precursor (Wilson & O’Gorman, 2003).
Perhaps worse than the guilt the author may feel is concern that the omission will
be found out by colleagues. And there is always the nagging feeling that some-
where in the literature another author has made the same point and the feeling of
relief when a literature search fails to nd anything of relevance. Sanctions can
be psychological as well as social.
As Merton suggests, the social norm of referencing perhaps began with schol-
ars wanting to lay claim to their ideas and avoid priority disputes. At rst, the
only option was to keep their ideas secret, deposit sealed notes or anagrams, and
refrain from publication as Newton and others did. With the advent of the scien-
tic journal, scientists were able to disseminate and date their ideas. Thus, the
journal acted as a registry of their contributions. The author could then point to
this registry if a question of priority arose and this function gradually evolved into
the formal bibliographic reference. Thus, we would expect that the early referenc-

Referencing as Cooperation or Competition  61
ing would be skewed toward self-citations, and that self-citations would be more
complete in terms of specifying the cited item than citations to others, and there
is some evidence that this is the case (Small, 2010).
The reference format has evolved over time as shown by studies of journals
such as the Philosophical Transactions of the Royal Society of London which began
publication in 1665 (Allen, Qin & Lancaster, 1994). Early references were usually
embedded in the text, and often consisted only of an author name (italicized or
bolded), and, occasionally, a source. Later, references became more complete, giv-
ing pages, years, etc., and moved from embedded text to side notes, footnotes, and
nally endnotes.
Normative expectations probably evolved along with these changes in print-
ing and format. In addition to allowing the ownership of ideas and discouraging
plagiarism, referencing became a tool to carve out a niche for your idea by demon-
strating that it was dierent from those of other authors (Gilbert, 1977)—eectively
an extension of knowledge claiming.
Another evolutionary thread developed around summarizing the current
state of knowledge on a topic, what we would call a review of the literature.
The tradition of reviewing prior opinions on a topic goes back to the writings of
Aristotle (Small, 2010), and many examples of such proto-reviews can be found
in the Philosophical Transactions. The review, while not a novel knowledge claim,
can be a new synthesis and useful to others.
In referencing others either for dierentiation or for review, we can speculate
that the norm of generosity of referencing came into being. What was originally a
defensive practice could also be used in a generous way to credit other authors for
their ideas. From this point the practice evolved from being customary, to one that
is expected and eventually required. Readers would then expect certain authors
to be credited if a topic was reviewed or a related knowledge claim was made.
Authors who failed to reference would be suspected of intellectual theft or, at best,
ignorance, and referencing others became a norm of scholarly practice.
In contrast to this normative account, in the constructivist approach, refer-
ences are only made for persuasive reasons motivated by self-interest. In this
world authors would be less likely to cite prior work closely related to their own
claim since it might jeopardize their own priority. Authors would be more likely to
distort or misrepresent prior work to support their own point of view (Nicolaisen,
2007). Authors would be less likely to cite items that serve only to provide the
reader with background information, and they would be more likely to self-cite.
In addition, as White has argued (2004), constructivist authors would tend to
cite leading gures in order to convince readers, but this was not empirically
supported.

62 HenrySmall
In constructivist citation contexts we would expect to frequently encounter
modality terms that weaken or cast doubt on the cited work. However, studies
of citation contexts have found the rate of negative citations to be relatively low,
about 6 % over seven separate studies (Small, 1982). In addition, in a random sam-
ple of 265 citation contexts containing the word “not”, it was found that in about
85 % of cases the citing authors were supporting a negative nding of an earlier
author, and were not themselves directly negating a cited work, in eect a nega-
tion by indirection.
As we have seen, Latour’s theory calls for a no-holds-barred approach to refer-
encing. However, in a norm-governed publication world, misquoting or distorting
a prior author’s work would not be regarded with equanimity. These instances
could be classied as “constructivist” (Small, 2004) and are relatively rare. Most
references are normative in the sense of adhering to some literal message in the
cited text. This is supported by the word similarity of citing and cited texts (Peters,
Braam & van Raan, 1995).
This does not mean that a range of interpretations of the cited work is not
possible. In fact, capsulizing, summarizing, and pigeonholing a prior text is part
of the compacting of knowledge, the process of creating symbols for ideas (Small,
1978), and a step toward Merton’s obliteration by incorporation (1968, p. 35). There
is, in addition, a gray area between distortion and legitimate interpretation. This
provides some room for reconciliation between normative and constructivist po-
sitions (Luukkonen, 1997) because dierences of interpretation and debate are
expected in cases where the signals from nature are ambiguous, or there is ambi-
guity in the cited text. Cozzens shows that interpretations of specic papers can
dier within a eld of science (1982). Cole (1992) also sees the lack of consensus
at the research front as an area of potential agreement between realists and con-
structivists. Riviera (2013) uses normative theory to describe the phenomenon of
high citation rate and constructivist theory to explain low rates. However, inter-
pretations can also converge—as seen in the emergence of regularized language
in citation contexts of highly cited papers indicating the formation of a consensus
(Small, 1978). In such cases, the signicance of the paper for a majority of citing
authors is shared.
In the previous discussion we have shifted the focus from the citing side to
the cited side and the formation of consensus. Here a reconciliation of normative
and constructivist theories is less likely. For example, Mulkay (1980) argues that
scientists do not apply technical norms in a consistent way, and the meaning of
rules varies depending on the situation and who is applying them. In this view
it is dicult to see how a consensus could emerge, and yet, citation studies have
shown that consensus formation can be rapid and dramatic (Cole, 1982, p.48).

Referencing as Cooperation or Competition  63
A citing theory deals with individual decisions on what to cite and cited theory
with aggregate citation phenomena and the perspective of a community. Since
the sum of all the citing acts results in what we see on the cited side, it might
appear that a theory of citing is all we need. The resulting distributions of citations
are typical of cumulative advantage processes, or, to use current nomenclature,
preferential attachment networks, where the number of future cites depends on
the number already accumulated (Newman, 2010). To get such distributions there
must be some kind of coordination of action among citing authors, an awareness
of the references of other authors or a shared reaction to the cited work. Seeing
that an author has referenced a particular paper may motivate other authors to
read and cite the paper, but there also needs to be recognition of the paper’s value
or relevance (White, 2011). Of course, many social and intellectual factors could
contribute to citation inequality. Following Thagard’s (2007) theory, value may
derive from better alignment of theory and observation or, following sociological
theory, a higher degree of utility (Cole, 1982, p.47; Hull, 1988, p. 301).
 Generosity in Referencing
We tend to think of science as a competitive activity with scientists striving for
priority, recognition, and funding (Hagstrom, 1974). But scientists also act gener-
ously and cooperatively by sharing work and collaborating. The Mertonian norms,
of course, embody generosity in the norm of “communalism.” In constructivism,
by contrast, scientists are driven by self-interest. Mertonian referencing is gener-
ous in giving credit to others, but could such behavior also be motivated by self-
interest? In biology, Richard Dawkins (2006) is known for his rejection of altruistic
behavior, favoring selsh behavior at the level of the gene. However, others have
argued that altruistic behavior—beneting others at a personal cost—is reason-
able both biologically and psychologically (Sober & Wilson, 1998). Evolutionary
biologists and philosophers have long struggled with how cooperative behavior
could have existed at all in the face of erce evolutionary competition. Yet both
cooperation and competition seem ubiquitous in human and animal societies.
E. O. Wilson recently described the inherent conict between wanting to behave
competitively and cooperatively which has been hardwired in our genes by evo-
lutionary forces (2014, p. 24).
The decision to cite or not to cite a nding similar to our own is a dicult
one for authors who want to claim as much credit as possible. In some instances
authors may not go out of their way to nd others who have expressed similar
ideas. When we reference others, we are giving up credit to others that could have,

64 HenrySmall
hypothetically at least, come to us (Small, 2004; Hull, 1988, p. 319). This is espe-
cially true when the cited work is intellectually close to our own. At rst glance,
this seems to be an act of generosity, a sacricing of a portion of our originality to
others. On the other hand, we are motivated to cite others whose work is similar
to ours in order to demonstrate that our contribution is distinctive, and to avoid
negative sanctions for failing to cite related work. Thus, whether referencing is a
selsh or generous act is ambiguous. Sober and Wilson point out the hypothesis
of generosity is dicult to prove because, regardless of the apparent seless act,
we can always think of some way the actor could have beneted.
Nicolaisen (2007) makes a related point inspired by a theory from evolution-
ary biology called the “handicap principle” or “costly signaling” (Zahavi, 1975).
In nature, animals engage in behaviors such as ostentatious displays, blus, and
mock threats which serve to enhance the tness of the performer and protect the
herd from predators. Hence, costly signaling is behavior that risks our own well-
being for the apparent benet of others and might be interpreted as generous or
altruistic. Nicolaisen sees referencing as costly signaling because the author is tak-
ing a risk and going out on a limb which could easily be cut o if a diligent reader
discovers that the reference is irrelevant or fallacious. Here he sees a connection
to Latour’s theory of citation which is based on self-interested manipulation of
the prior literature. Hence, referencing is a handicap and a gamble in the interest
of gaining advantage. Thus, although referencing may appear to benet others,
it is actually done out of self-interest, to advance our own interests. The handi-
cap principle does not accord well with normative theory because the behavior is
based on trickery and deception.
The handicap principle is one of a number of theories now current in evolu-
tionary biology which may serve to stimulate further theorizing on citations and
other issues in science studies. Two of the most relevant theories are “reciprocal
altruism” and “strong reciprocity”. “Reciprocal altruism” is the tendency to help
those who are likely to return a favor (Arrow, 2007). But this form of altruism still
has a selsh motivation. Reciprocal altruism would work for referencing only if
the cited author is likely or capable of returning the citation. If the cited author is
incapable of reciprocating, as is often the case, this mechanism fails.
In another approach called “strong reciprocity” (Fehr, Fischbacker, & Gach-
ter, 2002) cooperators are rewarded and non-cooperators are punished but at a
cost to those who punish (the strong reciprocators). Under certain conditions this
model has been shown to lead to sustained cooperation in social groups. Unlike
reciprocal altruism or costly signaling, strong reciprocity is consistent with the
existence of norms which dene what it means to cooperate, but it requires that
someone is willing to sanction the norm violators.

Referencing as Cooperation or Competition  65
One version of this model begins with a population of three types of agents:
cooperators who obey the norms but do not punish, selsh agents who violate
the norms, and strong reciprocators who obey the norms and punish violators
(Bowles & Gintis, 2003). In referencing, we can imagine that the selsh agents do
notciteothers,thecooperatorsciteothers,and the strong reciprocators cite others
and “punish” the non-referencing violators. Of course interactions of these agents
would run over years, not generations, as in evolutionary models.
Cozzens (1989) has studied the degree to which scientists involved in a prior-
ity dispute over the discovery of the opiate receptor behaved generously in their
referencing. She looked at each of the main co-discoverers and how they credited
the other co-discoverers. Over time there was a tendency for some co-discoverers
to be more generous in their referencing, and less inclined to claim the credit ex-
clusively for themselves. Also, the trend over time is toward a more standardized
and less specic or qualied citation of the competitor’s work as evidenced in ci-
tation contexts. The initial divergence of views at the time of discovery is followed
by a convergence over the next few years as distinctions and qualications are
dropped, suggesting greater generosity over time. In addition, one co-discoverer
played the role of enforcer by complaining to the others about the excessive pri-
ority claiming of one of the co-discoverers.
Another well known “multiple” is the discovery of oxygen involving Lavoisier,
Priestley, and Scheele. Lavoisier had received information from the others that
he was able to put to good use in making his discovery. However, initially he
claimed credit for himself and failed to acknowledge the contributions of the oth-
ers. Priestley, however, played the role of strong reciprocator and wrote a letter to
Lavoisier complaining about his failure to credit others. Subsequently Lavoisier
did acknowledge his fellow co-discoverers but not without pointing out how their
discoveries diered from his (Small, 2010).
Einstein’s special relativity paper, although not a multiple discovery, is also an
example of delayed generosity. His celebrated 1905 paper contains no references
to other papers. Notable for its absence was the work ofMichelson and Morley on
the constancy of the speed of light relative to the ether. However, two years later
Einstein wrote a longer article on relativity which contained a number of refer-
ences among them the Michelson-Morley paper (Small, 2010). Many years later,
Einstein claimed that he had been unaware of the earlier work and would have
cited it had he known about it (Holton, 1973, p.282). It is not known if pressure
was brought to bear on Einstein to acknowledge others.
A possible hypothesis is that over time there is an increase in generosity of
referencing which is marked by a spreading of credit to multiple individuals, a
lessening of the tendency toward the dierentiation of the contributions, and an
increasing standardization of language in the citation contexts. This can occur in

66 HenrySmall
the evolution of a single author’s work, in the work of independent co-discoverers,
or within an invisible college. There is some evidence of strong reciprocators act-
ing to enforce normative compliance.
Other evolutionary models of cooperation use “multilevel” strategies, advo-
cated by Sober and Wilson (1998), where selection occurs at the level of both the
individual and the group. When the focus changes from individual to the group
level success, it turns out that groups with many altruists are favored over groups
with more selsh individuals (Arrow, 2007). Taking this perspective requires that
we can think of science in functional terms at the group level. Because scientists
comprise a relatively distinct social group, it is possible that behaviors could have
evolved that enhanced group success. Historically we know that scientists often
had to defend their views from attack by various outside authorities, whether po-
litical, economic, religious, or scientic. At the same time, scientists depended
on these authorities for their support. Under these external threats it is not unex-
pected to nd that cooperative behaviors and norms evolved that increased the
tness and success of the community. Such cooperative mechanisms could have
included norms of behavior, the punishing of deviant behavior, and the mecha-
nisms for recognition. These mutual support mechanisms would be magnied by
the sub-structure of invisible colleges where individuals come face to face. Evo-
lution at the sub-group level might also give rise to other specialized technical
norms.
The next step might be to apply game theory to the process of writing papers
and making references. For example, Chatterjee and Chowdhury (2012) have ap-
plied game theory to citation networks. Another possibility is to model the writing
process as a game between an author and an imagined reader or critic. Each move
in the writing process could be scored as cooperative or competitive in a balancing
act to maximize the paper’s strengths and minimize its weaknesses. For example,
not citing a precursor or citing an irrelevant paper would be scored as selsh,
while citing a review or rival would be seen as generous and enhance the score.
Obviously the paper’s tness is not just a matter of what references are cited, but
what connections are made to experiment and theory, that is, the paper’s explana-
tory coherence, which is as we have seen partially revealed in its citationcontexts.
 Conclusions
Social construction leads to an anti-realist position on scientic knowledge and
a community of scientists bent on self-interests. In this view, facts and theories
have no basis in reality, and the only means of convincing others of the “truth”

Referencing as Cooperation or Competition  67
of a knowledge claim is rhetorical persuasion—even if that involves deception or
fabrication. In constructivism it would be impossible to detect fraud or error in
science or make a rational choice between theories. Deception, fabrication, and
distortion would become institutionalized norms of behavior.
The realist view, on the other hand, arms that “eternal and immutable reg-
ularities” exist in nature (Hull, 1988, p. 476). Because science relies on the arbiter
of our senses, it does not require rhetorical persuasion or deception. The princi-
ple of consilience, favoring theories that have multiple empirical conrmations,
oers an approximate guide to theory selection and how science evolves. Citation
contexts in papers are shown to be a rich source of connections between theory
and observations, and can be used for the construction of consilience networks.
Both technical and social norms are pervasive in science and are critical in
regulating behavior. But their origin and evolution are little understood. Technical
norms govern what counts as a consilience and the general procedures we call the
“scientic method”. Changes in technical norms are perhaps spurred on by major
scientic successes or technical innovations that employ novel methods. If a new
method becomes popular and incorporated into general practice, it will eventu-
ally be seen as a rule. Social norms may have evolved from general cultural values,
but also from new social realities such as the rise of national styles of science and
the need to insure the integrity of independent researchers.
The invention of the scientic journal gave rise to numerous norms and con-
ventions, perhaps the most important of which was the communalism of scientic
knowledge. Numerous other norms pertain to the acceptable style and structure
of scientic papers. The norm of referencing may have originated as a means to
claim priority, to show how your work diers from others, and to review what is
known on a topic. Once specic concepts became associated with prior works,
normative expectations were raised that these works would be cited when these
concepts were used.
Competition and cooperation are pervasive in science as they are in all hu-
man endeavor, but cooperation is dicult to account for in an evolutionary view
stressing individual survival. Strong reciprocity, where cooperators are rewarded
and non-cooperators punished, is one viable mechanism for the emergence of co-
operation. Here norms play a critical role in dening what it means to cooperate.
An alternative explanation is multi-level selection, where individual selshness
is counteracted by society level norms. Here norms act as group level adaptations
to maximize the tness and success of the group. It seems obvious that science is
a mix of selsh and altruistic individuals, but, perhaps, each individual is also a
mix of these tendencies.
Referencing appears to t the model of strong reciprocity where generous
citation is rewarded and non-citers are sanctioned. Historical examples of co-

68 HenrySmall
discoveries oer a preliminary conrmation. Referencing may trend toward a
sharing of credit and symbolic consensus in cases of multiple discovery or priority
disputes.
The tools of game theory and computer simulation now being used by evo-
lutionary biologists and economists to study cooperation and competition may
oer a promising new avenue for research into citation practice and social norms
in science. While referencing decisions are undoubtedly situationally complex,
we can expect that both competitive and cooperative motives are at work.
Acknowledgment
Citation context data for the publication year 2007 was generously provided by
Elsevier under an agreement with SciTech Strategies, Inc. I would like to thank
Mike Patek for programming support, and David Pendlebury and an anonymous
reviewer for helpful suggestions.
Cited References
Allen, B., Qin, J., & Lancaster, F. W. (1994). Persuasive communities: a longitudinal analysis
of references in the Philosophical Transactions of the Royal Society, 1665–1990. Social
Studies of Science,24, 279–310.
Arrow, H. (2007). The sharp end of altruism. Science,318 (Oct. 26), 581–582.
Barnes, B. (1974). Scientic knowledge and sociological theory. London: Routledge & Kegan
Paul.
Ben-David, J. (1984). The scientist’s role in society: a comparative study. Chicago: University of
Chicago Press.
Bowles, S. & Gintis, H. (2003). Evolution of strong reciprocity: cooperation in heterogeneous
populations. Theoretical Population Biology,65(1), 17–28.
Chatterjee, K. & Chowdhury, A. (2012). Formation of citation networks by rational players and
the diusion of ideas. Review of Network Economics,11(3), article number 6.
Cole, S. (1992). Making science: between nature and society. Cambridge, Mass.: Harvard Uni-
versity Press.
Collins, H. (2014). Are we all scientic experts now? Cambridge, England: Polity Press.
Cozzens, S. E. (1982). Split citation identity: a case study from economics. Journal of the Ameri-
can Society for Information Science,33(4), 233–236.
Cozzens, S. E. (1989). Social control and multiple discovery in science: the opiate receptor case.
Albany, New York: State University of New York Press.
Crombie, A. C. (1959). The signicance of medieval discussions of scientic method for the sci-
entic revolution. In M. Clagett (Ed.), Critical problems in the history of science. (pp. 79–
101). Madison, Wisconsin: University of Wisconsin Press.

Referencing as Cooperation or Competition  69
Cronin, B. (1984). The citation process: the role and signicance of citation in scientic commu-
nication. London: Taylor Graham.
Cronin, B. (1998). Metatheorizing citation. Scientometrics.43(1), 45–55.
Cronin, B. (2000). Semiotics and evaluative bibliometrics. Journal of Documentation,56(4),
440–453.
Dawkins, R. (2006). The selsh gene. 30th anniversary edition. . Oxford: Oxford University
Press.
Einstein, A. (1905). Zur Elektrodynamik bewegter Körper. Annalen der Physik, 17, 891–921.
Reprinted in The Collected Papers of Albert Einstein, volume 2. Translated by Anna Beck,
consultant Peter Havas, pp. 140–171. Princeton: Princeton University Press.
Fehr, E., Fischbacker, U., & Gachter, S. (2002). Strong reciprocity, human cooperation, and the
enforcement of social norms. Human Nature and Interdisciplinary Biosocial Perspectives,
13(1), 1–25.
Frankfort, H. & Frankfort, H. A. (1949). Myth and reality. In H. Frankfort, H. A. Frankfort, J. A. Wil-
son, & T. Jacobsen (Eds.), Before philosophy: the intellectual adventure of ancient man.
(pp. 11–36). Baltimore, Maryland: Penguin Books.
Gaulton, G. N. (2004). Ethical conduct in biomedical research: a handbook for biomedical grad-
uate studies students and research fellows,3
rd edition. Biomedical graduate studies
program, the University of Pennsylvania. Retrieved July 9, 2014, from https://www.med.
upenn.edu/bgs/docs/BIOETHICSHANDBOOK4-04.pdf
Gilbert, G. N. (1977). Referencing as persuasion. Social Studies of Science,7(1), 113–122.
Gingras, Y. & Schweber, S.S. (1986). Constraints on construction. Social Studies of Science,
16(2), 372–383.
Gower, B. (1997). Scientic method: an historical and philosophical introduction. London: Rout-
ledge.
Hagstrom, W.O. (1974). Competition in science. American Sociological Review,39(1), 1–18.
Hanson, N. R. (1972). Patterns of discovery: an inquiry into the conceptual foundations of sci-
ence. Cambridge: Cambridge University Press.
Holton, G. (1973). Thematic origins of scientic thought: Kepler to Einstein. Cambridge, Mass.:
Harvard University Press.
Hull, D. (1988). Science as a process. Chicago: University of Chicago Press.
Hyland, K. (2009). Academic discourse: English in a global context. London: Continuum.
Kamat, P. & Schatz, G.C. (2014). Cite with sight. Journal of Physical Chemistry Letters,5(7),
1241–1242.
Kaplan, N. (1965). The norms of citation behavior: prolegomena to the footnote. American Doc-
umentation,10(3), 179–184.
Knorr-Cetina, K. D. (1981). The manufacture of knowledge: an essay on the constructivist and
contextual nature of science. Oxford, England: Pergamon Press.
Kuhn, T. S. (1962). The structure of scientic revolutions. Chicago: University of Chicago Press.
Kuhn, T.S. (1977). Objectivity, value judgment and theory choice. In The Essential Tension.
Chicago: University of Chicago Press.
Latour, B. & Woolgar, S. (1979). Laboratory life: the social construction of scientic facts. Bev-
erly Hills: Sage Publications.
Latour, B. (1987). Science in action: how to follow scientists and engineers through society.
Cambridge, Mass.: Harvard University Press.
Latour, B. (2004). Why has critique run out of steam? From matters of fact to matters of con-
cern. Critical Inquiry,39, 225–248.

70 HenrySmall
Leplin,J.(1984).Introduction.InJ.Leplin(Ed.),Scientic realism. (pp. 1–7). Berkeley, Califor-
nia: University of California Press.
Luukkonen, T. (1997). Why has Latour’s theory of citations been ignored by the bibliometric
community? discussion of sociological interpretations of citation analysis. Scientometrics,
38(1), 27–37.
Merton, R. K. (1938). Science, technology and society in seventeenth century England. Bruges,
Belgium: Saint Catherine Press.
Merton, R. K. (1942). Science and technology in a democratic order. Journal of Legal and Politi-
cal Sociology, 1, 115–126. Reprinted in: The sociology of science: theoretical and empirical
investigations. Chicago: University of Chicago Press, 1973, 267–278.
Merton, R. K. (1963). Resistance to the systematic study of multiple discoveries in science.
European Journal of Sociology,4, 250–282.
Merton, R. K. (1968). Social theory and social structure. New York: Free Press.
Merton, R. K. (1988). The Matthew eect in science. 2. cumulative advantage and the symbol-
ism of intellectual property. ISIS,79,606–623.
Mitro, I. I. (1974). The subjective side of science. Amsterdam: Elsevier.
Mulkay, M. (1980). Interpretation and the use of rules: the case of the norms of science. In
T. Gieryn (Ed.), Science and social structure: a festschrift for Robert K. Merton. Transac-
tions of the New York Academy of Sciences. Series 2, 39, 111–125.
Mulkay, M. (1991). Sociology of science: a sociological pilgrimage. Bloomington, Indiana: Indi-
ana University Press.
Newman, M. E. J. (2010). Networks: an introduction. Oxford, England: Oxford University Press.
Nicolaisen, J. (2007). Citation analysis. Annual Review of Information Science and Technol-
ogy,41,609–641.
Peters, H. P. F., Braam, R.R. & van Raan, A. F. J. (1995). Cognitive resemblance and citation rela-
tions in chemical engineering publications. Journal of the American Society for Information
Science,46(1), 9–21.
Retraction Watch: Lack of citation prompts correction in Nature journal. April 10, 2014. Re-
trieved July 9, 2014, from http://retractionwatch.com/2014/04/10/lack-of-citation-
prompts-correction-in-nature-journal/#more-19753
Riviera, E. (2013). Scientic communities as autopoietic systems: the reproductive function of
citations. Journal of the American Society for Information Science and Technology,64(7),
1442–1453.
Shapin, S. & Schaer, S. (1985). Leviathan and the air-pump: Hobbes, Boyle and the experi-
mental life. Princeton, New Jersey: Princeton University Press.
Small, H. (1978). Cited documents as concept symbols. Social Studies of Science,8, 327–340.
Small, H. (1982). Citation context analysis. In B. Dervin & M.J. Voigt, (Eds.), Progress in Com-
munication Sciences, 3 (pp. 287–310). Norwood, N.J.: Ablex Publishing Corp.
Small, H. (2004). On the shoulders of Robert Merton: towards a normative theory of citation.
Scientometrics,60(1), 71–79.
Small, H. (2010). Referencing through history: how the analysis of landmark scholarly texts can
inform citation theory. Research Evaluation,19(3), 185–193.
Small, H. & Klavans, R. (2011). Identifying scientic breakthroughs by combining co-citation
analysis and citation context. Proceedings of the 13th International Conference of the Inter-
national Society for Scientometrics and Informetrics, Durban, South Africa.
Small, H., Boyack, K.W. & Klavans, R. (2014). Identifying emerging topics in science and tech-
nology. Research Policy,43, 1450–1467.

Referencing as Cooperation or Competition  71
Sober, E. & Wilson, D. S. (1998). Unto others: the evolution and psychology of unselsh behav-
ior. Cambridge, Mass.: Harvard University Press.
Swales, J. M. (1990). Genre analysis: English in academic and research settings. Cambridge,
England: Cambridge University Press.
Thagard, P. (1992). Conceptual revolutions. Princeton, New Jersey: Princeton University Press.
Thagard, P. (2000). Coherence in thought and action. Cambridge, Mass.: MIT Press.
Thagard, P. (2007). Coherence, truth and the development of scientic knowledge. Philosophy
of Science,74,26–47.
Thorpe, K.L, Benstead, R., Hutchinson, T.H., & Tyler, C.R. (2007). Associations between altered
vitellogenin concentrations and adverse health eects in fathead minnow (Pimephales
promelas). Aquatic Toxicology,85, 176–183.
Watson, J. D. & Crick, F.H. C. (1953). A structure for deoxyribose nucleic acid. Nature,171, 737–
738.
Whewell, W. (1847). Philosophy of the inductive sciences, founded upon their history.Vol.I.
London: John W. Parker.
White, H. (2004). Reward, persuasion and the Sokal hoax. Scientometrics,60(1), 93–120.
White, H. (2011). Relevance theory and citations. Journal of Pragmatics,43, 3345–3361.
Wilson, D. S. & O’Gorman, R. (2003). Emotions and actions associated with norm-breaking
events. Human Nature: An Interdisciplinary Biosocial Perspective,14(3), 277–304.
Wilson, E. O. (1998). Consilience: the unity of knowledge. New York: Alfred A. Knopf.
Wilson, E. O. (2014). The meaning of human existence. New York: Liveright Publishing Corp.
Zahavi, A. (1975). Mate selection: selection for a handicap. Journal of Theoretical Biology,53(1),
205–214.
Zuckerman, H. (1977). Deviant behavior and social control in science. In E. Sagarin (Ed.), De-
viance and social change. (pp. 87–138). Beverley Hills, California: Sage Publications.