url to share this paper:


Sci-Hub is a project to make knowledge free.

International Studies in the Philosophy of Science


ISSN: 0269-8595 (Print) 1469-9281 (Online) Journal homepage: https://www.tandfonline.com/loi/cisp20

Published online: 22 May 2020.

Are Pseudosciences Like Seagulls? A Discriminant Metacriterion Facilitates the Solution of the Demarcation Problem

Angelo Fasce

Department of Philosophy, University of Valencia

© 2020 Open Society Foundation

CONTACT Angelo Fasce[email protected]

Department of Philosophy, University of Valencia, Avd. BlascoIbañez, 30, 46010 Valencia, Spain INTERNATIONAL STUDIES IN THE PHILOSOPHY OF SCIENCE



In this article, I develop a philosophical framework, ormetacriterion, for the demarcation of pseudoscience.

Firstly,´gradualist demarcation is discussed in depth, considering an ap-proach to the demarcation problem that presupposes the existence of a spect-rum between science and pseudoscience; six general problems are found by means of this analysis.

Secondly, based on the subsequent discussion of these problems, a discrimi-nant metacriterion composed of four philosophical requirements is proposed. Lastly, it is shown that this metacriterion is able to guide the development of a workable and well-founded demarcation criterion for pseudoscience.

KEYWORDS Pseudoscience; demarcationproblem; demarcation criterion; family resemblance

Interest in the demarcation problem is undergoing a boom after being shelved and evengiven up for dead. Nevertheless,despite cur- rent philosophical discussions, there are nosubstantial advances in the development of a functional demarcation criterion capable of achieving academic consensus. In my view, this is due to two reasons.

On one hand,some scholars still engage in time-consuming, unpro-ductive discussions on already dis-carded demarcation criteria, such as falsifiability.

On the other hand, the, currently, most disseminated metacriterion the general philosophical requirements for a demarcation crite-rion of pseudoscience to be deemed satisfactory, which I call here gradualist demarcation, does not solve the shortcomings of prior approaches.

Therefore, in thisarticle I will develop an alternative, discriminant metacriterion that is able to deal withthe traditional problems of the philosophical definition of pseudoscience.

Firstly, gradualist demarcation will be defined based on the analy-sis of two relevant proposals: the first, raised by Fred Gruenberger in 1962, and a more recent one, by Massimo Pigliucci in 2013. This analysis is pertinent given that it will show the conceptual evolu-tion between both authors and, furthermore, it will bring out the most remarkable general problems of demarcation projects.

Secondly,these general problems will be discussed ingreater depth, taking account of the underlying conceptual issues that explain their persistence. Finally, a discriminant metacriterion will be pre-sented: this metacriterion, comprising four requirements, is able to avoid, or at least to palliate, the aforementioned flaws, setting the theoretical basis for developing a well-grounded and workable demarcation criterion. Lastly, the current philosophical and scien-tific outcomes of this metacriterion arelisted in order to show its fruitfulness.


Gradualist Demarcation Through two Instances

Beyond the two examples to be analysed here, gradualist demar-cation is defended by several scholars (e.g. Mahner 2013; Dawes 2018), so there are many demarcation projectsthatfit, in general terms, with this philosophical framework (e.g. Ruse 1982; Beyerstein 1995; Lilienfeld, Ammirati, and David 2012).

This approach assumes that the demarcation of pseudoscience is a matter of degree, a continuum that goes from the most reliable science to the most unreliable pseudoscience, with soft sci-ence, proto-science, and bad science within the spectrum.

In this regard, it aims to offer a fuzzy answer to the demarcation problem: something could be pseudoscience to a certain extent, depending on how it fits with the universe of variables selected as a demarcation criterion. This strategy, according to its defenders, finds its main advantage in leaving aside necessary and sufficient conditionsas was demanded by, among others, Laudan (1983). Accordingly, gradualist conceptions consider that the denial of the vagueness of scientificity, and the consequent adoption of burden-some requirements such as necessary and sufficient conditions, would largely explain why philosophers of science have not been able to offer a consensual and operational demarcation criterion for pseudoscience.

Gruenberger ([1962](1964)) was sceptical about the philosophical viability of the demarcation problem, arguing that the best we can do is offer a checklist of some of the attributes of science and of the crackpots to help in making this decision (Gruenberger1962, 2). Following this path, he chose thirteen variables, so the best of the sciences would meet all of them whereas pure foolishness would not meet any. Additionally, an in-between score makes diagnosis dependent on subjective assessment based on a context-dependant threshold.

The complete list of characteristics comprises:public veri
fiability (12 points), predictability (12), controlled experiments (13), Occams razor (5), fruitfulness (10), authority (10), ability to communicate (8), humility (5), open-mindedness (5), the Ful-ton non sequitur (underrated genius; 5), paranoia (5), the dollar complex (excessive esteem of ones ideas; 12), and statistics compulsion (5).

When considering Gruenbergers list, there is one thing, in particu-lar,that should draw the attention of philosophers:theoretical argu- ments in order to justify it are explicitly relativised — ‘these are some significant items which we think are among the main attri-butes of the scientist (Gruenberger1962,4; italics are mine). More- over, the scores of the variables are also arbitrary and he, again, shows no qualms over this:

the scores are personal, arbitrary, and biased. The reader is urged to fill in his own values, rather than to waste time quibbling over mine. I cannot de-fend any precise values (indeed, if I were to fill out the sheet again, I would probably have different values). (Gruenberger1962, 12)

Henceforth, I will call this issue the problem of the weight of the variables. Similarly, his implementation of the criterion follows a corresponding logic: dowsinggets 28 points, extrasensory percep-tion gets 38, and physics gets 97. But Gruenberger assessed the field of physics as a whole.


This is problematic because physics is heterogeneous regarding Gruenbergers criteria; for example, some of its branches, such a sparticle physics, have strong limitations in conducting controlled experiments. Also, philosophers can choose to demarcate several dimensions of science, such as logical-methodological issues (Wil-son 2000), propositions (Popper 1963), fields of knowledge (Bunge 1982; Thagard 1988), theories (Kitcher 1982; Lugg 1987), and re-search programmes (Lakatos 1978a)henceforth, I will refer to these dimensions using the overall concept of units of demarca-tion. Additionally, we dont know on what basis he measured such vague concepts as humility, paranoia, parsimony, and the ability to communicate.

To make matters worse, a detailed analysis of the structure of the check-list reveals that a unit of demarcation that is authoritarian, fruitless, not able to predict, and does not conduct controlled ex-periments, could get as questionable a score as 55 points of scien-tificity, depending on how we define some terms of the criteriain fact, Gruenberger considers that extrasensory perception (38 points) would belong to the middle group, stillopen to debate(Gruenberger 1962, 12). The final shortcoming is of a practical nature: what to do if a unit of demarcation gets 55 or 38 points of scientificity? Do we fund research projects based on it? Do we in-clude it in the education and judicial systems or, instead, we must discredit it? Gruenberger does not offer us any recommendation inthis regard, which defines his criterion not only as easy to dodge but also as problematicfor intersubjective decision-making.

Pigliucci (2013) has developed a much more contemporary ap-proach to gradualist demarcation. Again,the conceptual basis from which he departs is the spectrum that would make up science, soft science,proto-science,and pseudoscience a spectrum that would shape fuzzy conceptual clusters instead of clearly delineated con-cepts. Pigliucci rejects necessary and sufficient conditions for de-marcation by considering themold-fashioned(Pigliucci 2013, 19). Instead, he asserts that these assumptionsought to givepause at least since Ludwig Wittgensteins talk of family resemblance con-cepts (Pigliucci 2013,19).Accordingly,as in Toulmin (1972), Dupré (1993), and Irzik and Nola (2011),Pigliucci explicitly interprets the demarcation problem through the Wittgensteinian notion of family resemblance:Demarcation should not be attempted on the basis of a small set of individually necessary and jointly sufficient con-ditions becausescience andpseudoscienceare inherently Wittgensteinian family resemblance concepts (Pigliucci 2013, 25).

The analogy between the science/pseudoscience distinction and species is also of greatrelevance within Pigliuccis conception of demarcation although analogies are not, stricto sensu, argu-ments. He considers that the answer for the philosophical eluci-dation of pseudoscience should be similar to the one he offered in the past regarding the concept of species, which, following Hull (1965) and Templeton (1992),he characterises as a Wittgensteinian cluster-type concept (Pigliucci 2003). One classical example of this idea is the controversial characterisation of herring gulls as a ring species (Irwin,Irwin,and Price 2001):as these seagulls usually hyb-ridise between nearby populations, it would be possible to observe a chain of hybridizations that ends up in two populations with greater problems in producing offspring. So, even though ring spe-cies show two clearly recognisable extremes (science and pseudo-science in Pigliuccis analogy), they are unified by a continuum of hybrid individuals that aremore or lesso f the same species. Hence, if in order to assess whether two individuals are or are not of the same species, we should consider a complex universe of variablessuch as reproductive compatibility, evolutionary proximity, mor-phological similarity, and ecological behaviour , the same should be done inorder to distinguish between science and pseudoscience.


Pigliucci does not extract complete results from his ideas that can be considered a comprehensive demarcation criterion. Instead, he selects two initial variables theoretical understanding and empi-rical knowledge in order to show the way forward to future au-thors willing tofinish his demarcation project. Despite this, he con-siders these variablesas already able to offer substantially nuanced demarcation of science, proto-science, softscience, and pseudo-science (Figure 1). An in-depth analysis of Figure 1 reveals several philosophical shortcomings1:

(1) The relevance criterionused by Pigliucci in order to justify his variables selection is, in his words, science attempts to give an empirically based theoretical understandingof the world (Pigliucci 2013, 22). On why he specifically selects these two featuresinstead of others i.e. why he does not select variables based on some other relevancecriterion such asscience attempts to develop effective technology based on predictionand manipulation2, Pigliucci argues thatwe have to start from somewhere(Pigliucci  2013, 22). Again, wefind the same issue of a lack of theoretical foundationsto justify variables selection, as with Gruenbergers cri-terion. Hereafter,I will call thisissue,referring to which characteris- tics are relevant and which are not for demar-cation purposes, as the problem of the relevance criterion.

(2) On what basis has Pigliucci drawn the borderlines of the clus-ters? In a cluster analysiscarried out in a satisfying way - that is, in which clear clusters are observed despitethe existence of a spect-rum -,we must cut offthe clusters somewhere.In this case, Pigliuc- ci places the cluster of pseudoscience in the lower left corner, that of science inthe upper right, that of soft science in the upper left, and that of proto-science in the lower right.



Nevertheless, considering that these borderlines are the crucial outputs ofdemarcation, Pigliucci offers little argumentation aimed at justifying them. Accordingly, gradualist demarcation does not prevent the proliferation of relativised borderlines for pseudo-sciencealthough defining economics or string theory as pseudo-sciences, or astrology as soft science, are momentous decisions. The scope of pseudoscience should not be left to a borderline that can be moved from one place to another without theoretical resistance, as it would entail a lack of normative power.

The normative power of a given demarcation criterion is defined in terms of justifications, standards, and norms directly derived from its acceptance. Normative force emergesfrom its capacity to offer an intersubjective definition of pseudoscience, acceptance of which compels individuals to acknowledge and put into practice specific rational-basedbehavioural prescriptions, thus, setting con-straints during philosophical analysis andpragmatic decision-making .3 Hence, within a normative framework the demarcation criterion of pseudoscience should fit with the general definition ofpublic reason (Torcello 2020), and contradictory outcomes of the same criterion should be regarded as inconsistent. However, when a demarcation criterion lacks normative power, contradictory con-ceptions and decisions can be consistently and justifiably derived from iti.e. mutually contradictory propositions could be legiti-mately derived from the same criterionbecause that criterion allows, or is based on,subjective assessment. Consequently, the borderline between science and pseudoscience would be greatly influenced by contingentbeliefs and values, either idiosyncratic or shared by a certain group, society, or culture.4From here on, I will refer to this issue as theproblem of normative power.

(1) What does Pigliucci mean bytheoretical understanding, and why string theory hasthe most? He brieflydefines the concept asinternal coherence and logic(Pigliucci2013, 22), in that, although astrologers certainly can producetheoreticalfoundations for their claims,these quickly turn out to be both internally incoherent and,more damning,entirely detached from,or in contradiction with, very establishednotions from a variety of other sciences (particu-larly physics and astronomy, but alsobiology)(Pigliucci 2013, 24).

Accordingly, one could criticise a particular pseudoscience because, even if internally coherent, it contradicts robust scientific theoriesa classic but controversial demarcation strategy (Toul-min 1984). Nevertheless, if internal and external coherence are to be regarded as independent variables when assessing the quality of theoretical understanding, they should be independently included in the demarcation criterion of pseudoscience.

Additionally,what does Pigliucci mean by understanding? He con- siders, as the bestexample, a hypothesis that is not supported by empirical evidence in fact,it is controversial if some of the basic assumptions of string theory are testable (Smolin 2007; Woit 2007).Pigliucci does not mention classical characteristics of scien- tific explanations,such as causal links,models,and mechanisms, whose inclusion would substantially increase the number of inde-pendent variables. Furthermore, the concept of understanding (de Regt 2017) is not elucidated enough to grasp the differences bet-ween real understanding and overconfidence bias (e.g. Kruger and Dunning 1999; Rozenblit and Keil 2002). Something similar can be said on the other variable,empirical knowledge, also briefly de-fined as empirical support (Pigliucci 2013,22).I will refer to these issues astheproblem of defining variables and the problem of measuring variables. 5


(1) What type of units of demarcation are within the scope of this criterion? Pigliucci refers to his sample of units as disciplines and notions(Pigliucci 2013, 23), but these lection is not convincing.

Firstly, the demarcation of economics, physics and psychology as a whole is philosophically problematic. For example, the theoretical understanding provided by personality psychology, which works al-most entirely on thebasis of constructs (Fried 2017), could be con-sidered as lower than that of psychobiology, which includes biolo-gical mechanisms that underlie behaviours this criticismalso applies to other authors who demarcate entirefields of knowledge, such as Bunge (1982) and Thagard (1988).

Secondly, Pigliuccis sample includes a unitusually considered as a branch of psychology,namely evolutionary psychology, that resides in a different cluster, as well as an uncommon concept such as scientific history, whose demarcation seems to be begging the question besides, scientific history is classified as protoscience. Henceforth, I will call this issuethe problem of the scope’ — that is,the problem regarding what can be assessed by the demarcation criterion of pseudoscience and what is beyond its reach. All the aforementioned problems can be found in gradualist demarcation projects, albeit they are not exclusive of this approach.

In sum, these problems are:

(1) The problem of the scope.

(2) The problem of defining the variables.

(3) The problem of the relevance criterion.

(4) The problem of the weight of the variables.(5) The problem of measuring the variables.

(6) The problem of normative power.

Three Misconceptions of Gradualist Demarcation

In this section, I will discuss three misconceptions that lie at the root of the aforementioned problems: science as an improved version of pseudoscience, Wittgensteinian family resemblance as a philosophical background for demarcation, and gradual demar-cation as an approach that facilitates pragmatic decision-making.

Science is not an Improved Version of Pseudoscience

The basic problem underlying gradualist demarcation is a misun-derstanding of the objectives of the demarcation problem as a phi-losophical endeavour, as it consists of delimiting the borderlines of science as a set of fields and research programmes (Hansson 2017), thus separating science from what is not.

Nevertheless, gradualist demarcationists often over-state the scope of the demarcation problem by including the recognition of limi-tations andstrengths among legitimatefields of knowledge which is not to demarcate but to establish quality parameters.


In this regard, Boudry (2013) has made an interesting distinction between unfeasible and uninterestingterritorial demarcationbet-ween science and other epistemic activities,andnormative demar- cation between science and pseudoscience, eminently worthy of philosophical attention. So, the existing differences between science, proto-science, soft science, philosophy, humanities and the like do not belong in the demarcation of pseudoscience.

Contrastingly, gradualist projects hold, as an implici tassumption, that science and pseudoscience are of the same class - in that they are each a set of entities defined by a shared set of variables. That is to say, that they are clusters defined according to the same vari-ables but with different quality levels. So,the demarcation problem would be in the general recognition of such quality levels, instead of the philosophical justification of why science and pseudoscience belong to different classes.

Furthermore,the arguments used in order to justify a gradualist ap-proach to demarcation often assume that science and pseudosci-ence would be separated by a direct fuzzy borderline. Neverthe-less, this latter assumption does not justify either Gruenbergers scepticism about the demarcation problem or Pigliuccis Figure 1, in which science and pseudoscience have no direct frontier as border disputes are not transitive relations 6, the prototypical terri-tories of proto-science and soft science generate a sharp distinc-tion between the extremes of the spectrum. In other words, even though both extremes could have ambiguous limits with these in-between clusters, pseudoscience can be right-fully considered as not an overlapping class when compared to science. This non sequitur derived from their basic assumptions has been overlooked by gradualist projects; scienceand pseudoscience can be deemed different classes even if accepting a spectrum betweenthem. 7

In fact, the same can be said about many other cultural concepts. For example, there is no fuzzy borderline between science and py-ramid schemes, phishing websites, and embezzlement - taking into account that pseudoscience can be conceptualised as an intellec-tual misconduct (Blancke,Boudry,and Pigliucci 2017;Fasce 2017). 8

Resemblant characteristics such as the inability to predict or to offer proper explanations should not be interpreted in the same way among scientists and pseudoscientists. Experts in pancreatic cancer are almost always incapable of curing it and experts in fib-romyalgia do not understand the etiology of the disease, but these limitations do not make them partially pseudoscientificoncolo-gists and rheumatologists, like many economists, psychologists, and the SETI team, use the most reliable methods and base their ideas on the current corpus of scientific evidence.

This leads to the question: if, for example, pseudoscientists make medical or psychological statements without controlling confirma-tion bias at all or appealing to tradition, whereas scientists are go-verned by completely different standards, how can a pseudoscien-tist improve these radically negligent behaviours without changing their nature?

For instance:a complete lack of evidence or a priori untestable con- tent cannot be improved in order to achieve scientific standards. A given hypothesis is or is not within the spectrum of evidence and refers or does not refer to potentially testable phenomena, yet me-taphysical content cannot be improved to become natural pheno-mena and remain as metaphysical content. So,for demarcation pur- poses it would be more reasonable to compare between classes and not between individuals classified within a spectrum. This situ-ation could explain why there are no historical cases of pseudosci-ences that evolved to become science, not even the oldest ones such as phrenology and homeopathy. 9


In addition, if gradualist demarcation does not consider that pseu-doscientists would need to change the nature of their actions in or-der to achieve scientific standards,but merely improve their former behaviour, what solution do they offer for the problem ofthe scope?

As science mimicry has been traditionally regarded as the corner-stone of pseudoscience (Hansson 2009; Blancke, Boudry, and Pig-liucci 2017), the most promising option would be to consider only units that are publicly presented as scientific withinthe scope of the demarcation criterion of pseudoscience.

Nevertheless, all mono-criterial approaches, and only two of the twenty-one demarcation criteria analysed in Autor, consider this to be one of the variables that define pseudoscience. In fact, science fiction or Greek mythology could be classified as pseudoscience by Pigliuccis criterion because itsscope is not restrictive enoughit does not require something to be publicly presentedas scientific in order for it to be considered as pseudoscientific. Thus, an undesi-rable practical implication of fuzzy definitions by family resemb-lance is the confusion between pseudoscience and other types of unwarranted beliefs, such as paranormal and conspiracy theories (for a discussion on this psychometric issue see Fasce and Picó, 2019a). 10

Wittgensteinian Family Resemblance Outlaws Normative Elucidation

The use of the Wittgensteinian conception of family resemblance (Wittgenstein 1958; Pompa 1967) as a solution to the demarcation problem is substantially more problematicthan often suggested. Wittgenstein theorised that entities that are named using concepts ofnatural language are linked together by acomplicated network of similarities overlappingand criss-crossing: sometimes overall similarities, sometimes similarities of detail (Witt-genstein 1958, 32). A classic exampleone that Wittgenstein usually uses in his arguments is that of games (e.g. Wittgenstein1958, 31). We do not have a clear and outright definition of what a game is and what is not: in some games there is no winner, some are played alone and others by teams, some are played with balls, some are played with cards, and sometimes we use the term for activities that are not even related to funfor example, when someone is playing with your feelings. For this reason, Wittgensteindid not consider it appropriate to carry out closed, sharp definitions of concepts, characterising them as a form of impoverished artificial language (Givón 1986). Instead, concepts would be an intricate network of language-games whose meanings are contextual tools to perform regulated pragmatic interactions (Wittgenstein 1958, 2627).

There is a classic problem about this conception,known aswide-open texture (Richman 1962; Andersen 2000). It is not easy to dis-card explicit and exhaustive definitions of concepts based on suffi-cient and necessary characteristics, because somehow everything is similar to anything. For example, unlike athletics, pickpocketing is not a sporting discipline; nevertheless, carrying out this distinc-tion forces us to go beyond family resemblance. It can be said that pickpocketing: demands complex techniques which the pickpocket must practice to remain skilled in; is practiced as a means of ear-ning a living; and can even be considered to have a winner and a loser. Moreover, this is also a problem for positive delimitations of kinds. For example, cousins are defined as the offspring of parents siblings, so this specific kinship not physical resemblance is the characteristic with normative power in order to decide if two persons arecousins or not. 11 Hence, to use that kind of distinction we are compelled to prioritise certain relationships among the entities potentially described by a certain concept. 8


Otherwise, the use of language would be unfeasible: the world, including the cultural one,has sharp categories that language must grasp to be functional. It is worth examining Wittgensteins solu-tion to the wide-open texture problem so as toreveal the counter-intuitive implications of family resemblance within the context of demarcation. In order to develop his philosophical conception of meaning, he chose t ocombine definitions by family resemblance and a pragmatist theory of truth whose potential relativist nature has been widely discussed authors such as Williams (2007) and Coliva (2010) argue that his contextualism is not relativistic, although the relativistic interpretation of Wittgenstein has been particularly influential (Crayford 1997; Boghossian 2006). 12

The concept form of life is particularly relevant in this regard. It is defined as the set of language-games of a community of speakers that generates a shared worldview (Wittgenstein 1958, 11). So, in this context, a form of life implies thatthere must be agreement notonly in definitions but also (queer as this may sound) in judg-ments (Wittgenstein 1958, 88). Language-games would not only constrain our range of potential interactions andexpressive possi-bilities,but also the perception and consequent acceptance of facts:

So you are saying that human agreement decides what is true and what is false?

It is what human beings say that is true and false, and they agree in the language they use. That is not agreement in opinions but in form of life. (Wittgenstein 1958, 88)

Wittgensteins relativism is only clearly limited by the rules that define language-games,which he considers to be almost immutable - in fact,the evolution of these rules isone of the most problematic issues of his philosophical system (Kripke 1982; Baker and Hacker 1984). Hence, according to the relativist interpretation, and in a more restrained way for the contextualist one, an individual does not know facts, but understands a form of life defined by socially inculcated rules.

Taking all these characteristics of Wittgensteins philosophical conceptions intoaccount, it is difficult to understand how Pigliucci and other authors directly influenced by him intend to base a de-marcation criterion on ideas from which otherreaders, and not pre-cisely marginal ones,have derived a rather radical relativismfor example, Rortys neopragmatism (Boghossian 2006).That is to say, pseudosciencecan not be a Wittgensteinian family resemblance conceptand, at the same time, be subjected to normative demarca-tion and decision-making. In fact, from this standpoint, the demar-cation problem would not be a philosophical problem. Instead, it wouldbe a descriptive issuethe ethnographic study of a form of life , so the normative use of such demarcation criterion would commit a kind of semantic elitism already cri-ticised by Lakatos (1978b).

Perhaps the authors who seek to carry out the demarcation of pseudoscience by meansof family resemblance definitions do not follow Wittgenstein in all his philosophical commitments. So, they could be considering this strategy as more promising merely due to thealleged fuzzy borderline between science and pseudoscience, thus leaving aside the futilenature of Wittgensteinian family resem-blance for demarcation purposes although thereis no convincing argument to support this interpretation. However, even accepting an interpretation of family resemblance detached from strong Witt-gensteinian relativism, gradualist demarcation does not get rid of the challenge of establishing a criterion with normative power. The only difference is that it would be called relevance criterion and used to specify a set of variables among the overwhelming amount of resemblant characteristics of science and pseudoscience.


In sum, Wittgensteinian family resemblance banishes the demarca-tion problem from philosophy, whereas a weak interpretation of family resemblance has no deep implications it just moves the age-old demarcation problem from one place to another.

Gradualist Demarcation Hinders Decision-making

The lack of normative power of gradualist demarcation is also a problem for pragmatic decision-making about pseudoscience as a social affair. The demarcation criterion of pseudoscience has a pro-minent role in the implementation of policies and, consequently, justifying decision-making within the framework of public reason e.g. policies related to-disinformation, social media algorithms, public health, professional ethics, education, etc. Nevertheless, even though a demarcation criterion is a useful conceptual tool for the assessment and implementation of such policies, it is not pre-scriptive on its ownwithin this pragmatic dimension in other words, policies are not directly derivedfrom the philosophical elu-cidation of pseudoscience. Instead, policy makers should engage in expert deliberation,taking account of issues that are beyond phi- losophy, such as public/ private distinction, the available resources to fund and deploy interventions, the legal framework, economic concerns, etc.

The public acceptance of pseudoscience could involve harmful so-cial consequences, as was the case of Lysenkoism (Kolchinsky et al.2017), scientific racism (Paludi and Haley 2014), and social Darwinism (Paul 2003). Therefore,it is a well-documented danger, which threatens key issues within the public sphere, such as food (Mulet 2018), education (Forrest and Gross 2004), health (e.g. Lilienfeld, Lynn, and Lohr 2003; Ernst, Lee, and Choi 2011), and justice (Snook 2008). Accordingly, being characterised as pseudo-science has dire implications, particularly for researchers. 13

In attemptsto avoid the risk of promoting these kind of misleading ideas and to rationalise limitedresources, researchers will not be given access to the funding 14 and publication systemsof science, they will not receive official professional accreditation,they will be banished from the educational system at primary,secondary, and university levels, and their practices and ideas will be considered as a threat to critical thinking and public deliberation. Because of all this, demarcation is a very important decision, and pseudoscientists, like everyone, deserve a fair trial.

Nevertheless, what should we do with a verdict that states that the accused is more or less guilty? If a unit of demarcation turns out to be quite, or not too much, or 7/10 pseudoscientific, there would still be a substantial chance for our verdict to be wrong. 15 As happens with Gruenbergers criterion and as could also happen with Pigliuccis, depending on interpretation it may be easy for a unit of demarcation to achieve acertain degree of perceived sci-entificity through the exploitation of the less demandingvariables of the criterion, such as internal coherence, open-mindedness, ad-hoc predictability, and controlled experiments. Because pseudo-scientists fake dispensable, superficial characteristics of science, the inclusion of degrees of these characteristics in a demarcation criterion lends a helping hand by facilitating the production of an illusory level of scientificity. Therefore, gradualist demarcation tends to introduce uncertainty in a class of decisions that show, in most cases, the greatest certainty.'


How to Cope with These Problems?

There is a tacit consensus about what is scientific and what is pseudoscienti-fic, sopeople with the adequate motivational state can normally differentiate betweenboth (e.g. van der Linden et al. 2015; Tabacchi and Cardaci 2016;Garrett and Cutting 2017). Hans- son describes this state of affairs in his assertion: distinguishing between science and pseudoscience is much like riding a bicycle. Most people can ride a bicycle,but only a few can account for  how they do it (Hansson 2013, 61). There are numerous examples of consensual demarcation in encyclopaedic publications comprehen-sively focused on pseudoscience (e.g. Shermer 2002) these, in turn, greatly converge with more informal lists of topics characte-rised as pseudoscience, published by online encyclopaedias (e.g. Rational Wiki 2019; Wikipedia 2019) and official reports commis-sioned by governments (e.g. MSPSI 2011; Clarke et al. 2015). Accordingly,the demarcation problem can be defined as the project to justify and optimise this already existing consensus. Hence, it should not necessarilybe a fuzzy task: a demarcation criterion could be restricted to discriminating between classes that are known in advance, thus offering philosophical justification to decisions that have already been made.

To properly understand the kind of discriminant demarcation I will develop in thissection, it is important to be aware thatpseudosci-ence will be regarded here as an extreme category. Therefore, it must be used only for radical instances of epistemic misconduct. Consensual pseudosciences, such as homeopathy, neurolinguis-tic programming, and climate change denial show radically flawed epistemic dimensions: there are no studies supporting the theoretical framework and the clinical efficacy of homeopathy and neuro-linguistic programming, whereas anthropogenic global warming has been broadly confirmed by scientific evidence. Thus, discriminant demarcation is contrary to overstating the scope of pseudoscience, instead, considering it a narrow philosophical cate-gory. There are cases in which pseudoscience has been weaponised to perform ideology-driven social criticism - for example, socio-biology (Thompson 1980) and neoclassical economics (Bunge 2016) have been labelled as pseudosciences. Even worse, radical instances of overstated scepticism, in which scientific theories and fields are accused of being pseudoscientific, constitute a popular rhetoric strategy among science deniers, known aspseudo-scepti-cism (Torcello 2016).  Tosummarise, within discriminant demarcation, pseudoscience is regarded as:

(1) A sharp, greatly restricted category defined by radical epistemic negligences.

(2) An independent class that must be defined by its distinctive characteristics.

Consequently, due to (1) and (2), pseudoscience cannot be negati-vely defined solely as a by-product of the definition of science. In-stead, it must be defined by its own distinctivecharacteristics, re-cognised by means of the following question: what characteristics does pseudoscience have that science and other types of non-sci-ence do not? 16 Therefore,a definition ofsciencecannot be extrac- ted from a discriminant demarcation criterion of pseudoscience, composed of its exclusive features science, including formal and applied fields, involves other relevant and greatly nuanced non-dis-criminant multi-dimensional characteristics that are still discussed among philosophers of science, such as models, measurement, theories, prediction, and replication.


A Discriminant Metacriterion

A metacriterion capable of coping with thepreviously discussed problems should bemade up of four requirements (hereby referred to as R1, R2, R3, and R4), classifiedinto two groups. On one hand, R1 and R2 are desirable procedural requirements - that is, they are general values that should guide the development of the demarca-tion criterion of pseudoscience. The fulfilment of R1 and R2 is not necessary, but they give rise to general conditions of plausibility: a demarcation criterion that does not satisfy these axiological stan-dards would have serious problems achieving an optimalthreshold of acceptance among experts. On the other hand, R3 and R4 are criterion requirements whose satisfaction is mandatory in order to provide a proper criterion of relevance for selecting the necessary and sufficient characteristics of pseudoscience.

Procedural requirements:

R1: The demarcation criterion of pseudoscience should entail the least amount of philosophical commitments. This is desirable in order to provide the tool with elegance, parsimony, and thecapability to achieve widespread acceptance. In this regard, it is important to incorporate well-elucidated variables, as well as to avoid the competition between multiple demarca-tion criteria producing disparate results according to the philosophical commitments they entail.

R2: The demarcation criterion of pseudoscience should explain and optimise current consensus.

Firstly, it must be able to explain all the units of demarcation that are well-foundedly and unanimously characterised as pseudoscience.

Secondly,as pseudoscience often performs pro-cesses of cultural evolution, the criterion must be able not only to demarcate future andunknown forms of pseudoscience but also to reach current levels of consensus withrespect to them.

Criterion requirements:

R3: Mimicry of science is a necessary requirement to be pseudoscience, given that this is its distinctive feature as a subclass of non-science.

R4: All the items of the demarcation criterion must be discriminant with respect to science.Discriminantrefers to characteristics not included in any basic definition of science, but which, instead,can be found among cer- tain instances of pseudoscience. Hence, a sufficient definition of pseudosci-ence must entail fulfilment of R3 and of at least one R4-type variable. 17

The internal structure for a demarcation criterion that satisfy this metacriterion wouldbe as follows:


R3 and at least one R4-type item. Where, R3 is necessary.

At least one R4-type item is necessary.

The conjunction of R3 and at least one R4-type item is sufficient.

Given that science justifiably shows the trappings of science, it satisfies R3, so this requirement discerns between the set that in-cludes both science and pseudoscience, on the one hand, and other forms of non-science, on the other. 12


Therefore, what is truly pro-blematic about the demarcation crite-rion of pseudoscience is distinguishing it from science. That is the task of R4-type items, selected by consensual negative answers to the following question:can something with this feature be science?.

Furthermore, both the philosophical requirements and the internal structure established by this metacriterion are ahistorical. This means that even though the exhaustivedefinition of R4-type vari-ables may vary over time, according to new forms and strategies of pseudoscience, their inclusion in the demarcation criterion of pseudoscience should notchange through time.This is because dra- matically regressive changes in the definition ofscience are not ex-pectedat least, not dramatically enough to turn the distinctive radicalepistemicflaws of pseudoscience into shared characteristics.

Regarding the problems detected in the analysis of gradualist demarcation:

The axiology expressed by R1 and R2 helps solve the problem of normative power andtheproblemofthedefinition of the variables by establishing as neutral as possiblephilosophical foundations, from which practical consensus could be derived. Accordingly, philoso-phical agendas should not be a burden for the demarcation tool.

For example,this kind of theoretical burden can be found in the de- marcation criterion proposed by Bunge (1982), which demands so-cial support, the use ofnomological reasoning, and commitment to strong scientific realism, highly controversial requirements that do not have relevant implications for the definition of pseudoscience.

Contrarily, the problem of measuring the variables still exists, although its severity hasbeen substantially restricted.

Firstly, it has been quantitatively palliated by account for onlya few discriminant items, which reduces the problem of measuring an excessive number ofvariables.

Secondly, given that the conjunction of R4-type items characterises all current instances of pseudoscience,but any science,the problem is also qualitatively reduced. As these variables are meaningful as discriminant characteristics as long as they constituteradical mis-conducts,it is unnecessary to measure a spectrum of degrees of de- ficiency. Forexample, if, for gradualist demarcation, it is necessary to measure the continuum of meth-odological reliability, it would now be enough to define the absolute lack of methodological reliability that takes place in some instances of pseudoscience.

R3 neutralises the problem of the scope: as pseudoscience is cha-racterised by sciencemimicry,its demarcation criterion should only be used over units that are publiclypresented as science. Additio-nally, the conjunction of R3 and at least one R4-typecharacteristic facilitates the distinction of pseudoscience from other types of ra-dical non-science - as soon as conspiracy theorists and proponents of the paranormalengage in science mimicry they become pseudo-scientific. As ensuring a narrow and well-elucidated conceptual scope for pseudoscience constitutes a relevant purpose of discrimi-nant demarcation,mimicry of science accounted by R3 solely refers to the fields of knowledge usually included within the conceptual scope of the Englishwordscience.

Accordingly,R3 does not denote mimicry of humanities and art. Incontrast, Hansson (2013) argues that it is not crucial whether something is called science but whether it is claimed to have the function of science, namely to provide the most reliable informa-tion available about its subject-matter; thus, appealing to the Ger-man concept Wissenschaft to delimit the scope of pseudoscience.


Consequently, within Hanssons framework, pseudophilosophy should also be deemed pseudoscience. This conception of science mimicry is too wide to fulfil the purposes of discriminant demarca- tion, as R4-type characteristics of pseudoscience and pseudophilo- sophy greatly differ, for example, regarding methods and empirical evidence (related remarks on Hanssons use of the concept Wissen- schaft can be foundin Fasce 2018 for a discussion of Hanssons demarcation criterion as an unfavourablywide concept see Letrud 2019).

R4 solves the problem of the criterion of relevance, the problem of the weight of the variables, and the problem of normative power. Regarding the problem of the criterion ofrelevance, R4 states that we must only take into account variables that can be found in pseudoscience, but not in science. Hence, shared characteristics, even in differentdegrees, must be regarded as irrelevant for demar-cation purposes.18 Moreover, as these variables are radical, exclu-sive, and not subjected to controversies, they entailnormative po-wer, thus establishing an intersubjective and robust borderline bet-ween science and pseudoscience.19All R4-type items have the same weight byshowing the same discriminant potential. Hence, it is worth wondering why it isnecessary to include all R4-type vari-ables in the demarcation criterion of pseudoscience instead of just one.This is because different pseudosciences are definedby distinc- tive R4-type variables - that is to say,the reasons why they are con- sideredradical instances of non-science are diverse. Accordingly, a certain unit of demarcation may be considered pseudoscience due to problems relating to its domain, whereas another may be due to lack of evidentiary support.Lastly,R4-type variables also contribute in distinguishing pseudoscience from non-pseudoscientific non-sci-ence that publicly presents as scientific: as they do no tfit with any R4-type criteria, these units of demarcation do not involve radical epistemic misconducts. Therefore, these are the kind of ambiguous cases that discriminant demarcationexcludes as instances of pseu-doscience - even if close to being pseudoscience, they are not devi- ant enough. Accordingly,these units must be classified as instances of proto-science, bad science, soft science, and the like.

Empirical Outcomes and Concluding Remarks

There are two related tasks within the demarcation problem:

(A) the demarcation of current consensual instances of pseudo- science and (B) the demarcation of ambiguousand future forms of pseudoscience. (A) is the basic aim of the demarcation problemand its solution should justify the tacit demarcation already made.

However, although (A) is a feasible task for both approaches, it is substantially more problematic withinthe gradualist framework. Particularly, due to the profuse inclusion of non-discriminant vari-ables lacking normative power,that are hard to select, define, rank, measure, and tobe proven as shared with science. Contrastingly, as the discriminant approach isfocused on the distinctive traits of pseudoscience, it leaves aside these problematic resem-blant features of science and pseudoscience. 14


This discriminant metacriterion strongly restricts (B), as it should optimise, not extend,(A). Hence, a unit of demarcation not accoun-ted by (A) should be characterised as pseudoscience only if it shows the same level of accordance with a demarcation criterion asconsensual instances of pseudoscience.In contrast, (B) is greatly problematic for the gradualist approach, particularly due to its ten-dency to unnecessarily stretch the scope of theanalysis of pseudo-science. In sum,discriminant demarcation is less demanding regar- ding (A) and more demanding regarding (B) than gradualist demarcation.

In Fasce (2018) I developed a demarcation criterion that fulfils the requirements of thisdiscriminant metacriterion, where pseudosci-ence radically differs from science regardingdomain, method, and evidence. So, it can be identified by being uncontroversially out-side the domain of science, particularly due to untestable content - such as reiki, morphic fields, acupunctures qi, and vertebral sub-luxationsand through the use of radicallyflawed methods for example research on EMDR without controlling exposure (Herbert et al.2000), the use of projective tests (Lilienfeld, Wood, and Garb 2000), pendulum diagnosis, and unspecific cases, such as illusions of causality (Torres, Barberia, and Rodríguez-Ferreiro 2020). Some others are characterised by a complete lack of evidencee.g. cli-mate change denial, orgone (Klee 2005), Bachflowers (Ernst 2010), facilitated communication (Hemsley et al.2018), electromagnetic hypersensitivity (Rubin et al. 2011), and the wide range of pseudo-phenomena considered by parapsychology, such astelepathy, clair-voyance, and precognition. Lastly,some pseudosciences are defined by a mix of these three R4-type variables, such as anthroposophy (Hansson 1991), pseudoarchaeology (Fagan 2006), and intelligent design (Boudry, Blancke, and Braeckman2010).

Moreover, the characterisation of certain units of demarcation as instances of pseudoscience using this discriminant demarcation criterion is the cornerstone of the Pseudoscientific Belief Scale (Fasce and Picó 2019a) the only currently validated scale of ge-neralpseudoscientific beliefs with tested psychometric soundnessand of a recent operationaldefinition of Complementary and Alternative Psychotherapies (Fasce and Adrián-Ventura 2020).

Accordingly, the outcomes of discriminant demarcation show high internal consistency, as well as strong convergent and discriminant validity (Fasce and Picó 2019b; Fasce, Adrián-Ventura, and Avenda-ño 2020). Lastly, the conjunction of these results and a literature review of prior research outcomes on pseudoscience has led to a new explanatorymodel for the epidemiology of pseudoscientific beliefs: the Explanation-Polarisation Model (Fasce, forthcoming). Hence, this discriminant demarcation is not only philosophically satisfactory, but scientificallyfruitful.

Ive developed a discriminant metacriterion based on the critical analysis of the gradualist approach and able to deal with some of the persistent shortcomings of demarcationprojects. This metacri-terion regards that of pseudoscienceas an extreme label; charac-terised by the conjunction of radical epistemic flaws and science mimicry. This philosophical framework has been shown to be fruit-ful, through the definition of a validated psychological construct that has led to novelfindings on the epidemiology and the psychological profile of pseudoscientific beliefs.



1.  Similar criticism over Pigliuccis approach to demarcation can be found in Schindler (Forthcoming).

2. This definition of science is linked to applied science, so pure science is beyond its scope. Im using this example to show that Pigliuccis definition involve a similar limitation regarding applied science, such as SETI, and studies focused on the detection of correlational patternsand prediction.

3. For example: if the code of ethics of clinical psychologists states that clinicians must reject pseudoscientific practices, and a particular clinician accepts a demarcation criterion with normative power that defines primal therapy, bioenergetic analysis, thought field therapy, and orgone therapy as instances of pseudoscience, then the practitioner is ethically compelled to avoid the use of these techniques. Otherwise, she must either reject the code of ethics or thedemarcation criterion of pseudoscience.

4. For example:two philosophers agree on Gruenbergers criteria - Pigliuc- cis can also beapplied to the same case. Further, they agree on the degree of authoritarianism, humility, open-mindedness, and paranoia of flat-ear-thers, orgone therapists, or graphologists by means of some reliable and intersubjective measurement tools - a scenario regarded as practically im-possible by Gruenberger, but lets consider it as feasible. Insofar as there are no constraints on how, when, and where to establish the borderline, they could choose different tolerance thresholds. That is to say, one can consider a unit of demarcation as pseudoscientific starting from forty points, whereas the other can do so starting from fifteen points. Under these conditions, demarcation is not an intersubjective and externally consistenttask. These philosophers do not need to reject their shared non-normative demarcation criterion in order to justify their endorsement of practices and beliefs that the other regards aspseudoscientific.

5. Im using the termmeasurementin a wide sense, also covering classification into categories not designated by numbers.

6. (1) Canada and the United Stated have border disputes. (2) Mexico and the United Stateshave border disputes. (1) and (2) do not entail that Canada and Mexico have border disputes.

7. Personally, I consider it very likely that science has blurred boundaries with, among others, proto-science and soft science. Moreover, I also consi-der it likely that pseudoscience could have blurred boundaries with other types of radical non-science, as some forms of science mimicry are ambi-guous. Nevertheless, this situation is not as problematic as these authors state. If pseudoscience is accounted as an extreme category denoting radical epistemic flaws, it does not directly overlap with science. This conception entails an explicit rejection of ambiguous cases as instances of pseudoscience (e.g. string theory, sociobiology, evolutionary psychology, neoclassical economics, astrobiology, and the like), thus ensuring a narrower conceptual scope.

8. As happens with the promotion of pseudoscientific beliefs, one can par-ticipate in misconducts unwittingly deliberate intentions to deceive are a necessary condition for sciencefraud, but not for pseudoscience. By means of self-deception and motivated reasoning, pseu-doscientific believers are often disconnected from the search for truth. Hence,pseudoscienceis to in- tentional fraud as bullshit is to lies (for an in-depth discussion on this issue see Ladyman 2013).

9. Alchemy is better defined as hermeticism, or even as a proto-science, not as a form of pseudoscience in contemporary terms.

10. Conspiracy theories are defined aslay beliefs that attribute the ulti-mate cause of an event, orthe concealment of an event from public know-ledge, to a secret, unlawful, and malevolentplot by multiple actors working together(Swami, Chamorro-Premuzic, and Furnham2010, 749). Conspira-cies exist and there are well-known historical examples, althoughthese beliefs are unwarranted when they are an unnecessary assumption when other expla-nations are more likely.

11. It should not be assumed that the Wittgensteinian family resemblance only refers to behavioural or observable aspects. Wittgenstein could have admitted that genetic, unobservablekinship is part of the family resemb-lance of cousins.Nevertheless,he might not have acceptedan explanation in objective terms on why this specific feature is more relevant than others.16

Instead, he would have possibly stated that its pre-eminence is based on its role withinlanguage-games that involve cousins.


12. Other authors, such as Kuhn, have provided another potential solution for the wide-open texture problem by including not only similarities between members of the same class, butalso dissimilarities to members of other classes (Andersen2000). So, as contrasting conceptsmay mutually limit the extensions of each other, it would be better to concentrate on the dissimilarity between contrast classes. It is my intention to defend a simi-lar solution within thecontext of demarcation: pseudoscience would be bet-ter defined as a contrast class by means ofits distinctive dissimilarities ordiscriminantcharacteristics.

13. It is worth mentioning thatto study pseudoscienceis not the same as - to produce pseudoscience. Indeed,unwarranted beliefs constitute a valid and worthy study domain, ascan be seen from clinical research on alterna-tive medicine, anomalistic psychology, andother related scientific research on conspiracy ideation and the cognitive roots of sciencedenial. This scien-tific approach to pseudoscience is based on the acceptance of a normative approach to demarcation, as well as on some methodological standards that are beyond thedemarcation problemsuch as validated measurement, meaningful comparisons, andproper data analysis.

14. Of course, there are risky and potentially fruitless projects that may be funded to increase research variability, but projects based on pseudoscien-tific assumptions are known before-hand to be unsuitable of funding particularly, unworthy of public funding.

15. As decision theory shows, we can perform decision-making under con-ditions of uncertainty through the assessment of probabilities. Nevertheless, probabilistic reasoning may be preferable or not, and surely all sources of uncertainty reduction are more than welcome. In thecontext of demarcation, a workable and well-founded cut off is preferable in gene-ral termsthan probabilistic uncertainty. Gradualist demarcationists would agree on this general principle, although they do not consider sharp cate-gorical assessment of pseudoscience as feasible. Discriminant demarca-tion, to be developed in the next section, is more optimistic inthis regard.

16.From my point of view,legitimate criticism oversilver bullets (Poppers falsifiability, Lakatosprogressivity, Kuhns puzzle solving, etc.) led philoso-phers to favourmachine guns of diffuse (often gradualist) criteria. As such, the problem that has impeded demarcation for the last century was a counterproductive dismissal of the discriminant characteristics of pseudo-science.Despite there being no such silver bullet,it is feasible to demarcate pseudoscience by means of few precise, discriminant criteria - most being already well-known, e.g. untestability, and complete lack of both reliability and confirmatory evidence.

17. This discriminant metacriterion must be interpreted as a whole. For example, without R4, R1would be more about achieving consensus than about getting it right.

18. This does not mean that they are irrelevant when assessing the level of quality of science andwhen assessing how deviant pseudoscientific doctrines are.

19. As I mentioned previously, the borderline between pseudoscience and other types of unwarranted beliefs could be blurred. Discriminant demar-cation only sets a framework for a clearcut offbetween science and pseudoscience.


Angelo Fasce