Pseudoscience is a claim, practice, or system of ideas that is presented as scientific while failing to meet relevant standards of scientific inquiry. Typical features include unreliable evidence, resistance to meaningful testing, selective treatment of results, and the continued assertion of conclusions despite substantial contrary findings. The term concerns the relationship between claims and their scientific credentials: something is not pseudoscientific merely because it is false, unconventional, unpopular, or outside science.
The distinction matters because scientific authority influences medical treatment, education, commercial decisions, and public policy. A scientific appearance can lend unwarranted credibility to ineffective products, unsupported historical narratives, or misleading explanations of natural phenomena. At the same time, the label can be misused to dismiss legitimate disagreement or emerging research. Identifying pseudoscience therefore requires examining particular claims, methods, and responses to criticism rather than simply identifying outsiders to an established profession.
In philosophy of science, pseudoscience forms part of the demarcation problem: how to distinguish scientific inquiry from other intellectual activities and, more specifically, from practices that misleadingly claim scientific standing. No single criterion resolves every case. Philosophical analysis, historical investigation, and scrutiny of evidence instead reveal recurring differences between productive inquiry and systems organized to protect predetermined conclusions.
Meaning, scope, and neighboring categories
The prefix “pseudo-” derives from a Greek word meaning false. In ordinary usage, “pseudoscience” can refer to an individual assertion, a technique, an explanatory system, or a community of practice. These are different units of assessment. A particular claim may be scientifically testable even when the organization promoting it behaves pseudoscientifically; conversely, an otherwise legitimate research field can contain poorly supported claims. Careful analysis specifies what is being evaluated and which scientific standards are relevant.
Pseudoscience is narrower than nonscience. Mathematics, for example, establishes results primarily through proof rather than empirical experimentation, but this does not make it pseudoscientific. Historical scholarship, artistic interpretation, moral reasoning, and many forms of religion have purposes and standards that are not identical to those of the natural sciences. They become relevant to scientific demarcation when they make empirical assertions or claim scientific validation. A theological statement about ultimate meaning and a laboratory claim about the age of a material object do not require the same kind of assessment.
Error is also insufficient. Science contains mistaken measurements, unsuccessful theories, and abandoned explanations. What matters is partly how errors are handled: whether investigators expose their claims to criticism, revise their conclusions, and allow independent checks. The view known as fallibilism holds that knowledge claims can remain open to correction without becoming arbitrary. Pseudoscientific systems often imitate the certainty popularly associated with science while neglecting the organized possibility of being wrong.
Fraud and pseudoscience overlap but are not identical. Scientific fraud involves deliberate deception, such as fabricating data. A fraudulent result may employ an otherwise sound experimental design, while a sincerely held pseudoscientific belief may involve no conscious dishonesty. Likewise, weak research, commercial exaggeration, and premature publicity are not automatically pseudoscience. They become stronger candidates for that designation when they form a persistent pattern of claiming scientific authority while evading the evidential responsibilities accompanying it.
The term also differs from “protoscience,” used for inquiry that has not yet developed mature theories or reliable methods but may do so. This distinction cannot depend solely on eventual success, since that would make classification possible only retrospectively. More useful questions concern whether a developing field identifies its uncertainties, produces increasingly discriminating tests, engages with relevant established knowledge, and changes in response to results. An unconventional proposal can be scientifically legitimate even when its probability of success is low.
Philosophical accounts of scientific demarcation
Twentieth-century debates gave scientific demarcation a central place in epistemology. Philosophers associated with logical empiricism examined the relation between meaningful empirical statements and possible observation. Versions of verificationism connected empirical significance with verification or confirmation. These projects were not simply lists of pseudosciences: they also addressed language, logic, and the status of metaphysics. Their difficulties included explaining universal laws, which cannot normally be conclusively verified by a finite number of observations.
Karl Popper proposed falsifiability as a criterion of scientific status. On his account, a scientific theory excludes some possible observations and therefore risks conflict with experience. A theory compatible with every conceivable outcome lacks that kind of empirical content. Popper developed this approach in Logik der Forschung, published in 1934 with a 1935 imprint, and subsequently elaborated it in English. His distinction was not between meaningful and meaningless statements, nor between true and false theories: a false theory can still be scientific if it makes testable claims.
Falsifiability does not provide an automatic procedure for rejecting theories. An apparent failed prediction may result from faulty equipment, incorrect assumptions about initial conditions, or errors in auxiliary hypotheses. Confirmation holism emphasizes that observations usually test networks of assumptions rather than isolated propositions. Responsible researchers sometimes retain a productive theory while investigating anomalies. The difficulty is distinguishing a justified adjustment from an improvised maneuver that protects a claim against every possible objection.
Thomas Kuhn, in The Structure of Scientific Revolutions (1962), emphasized the historically organized character of research. Established scientific communities typically work within shared frameworks, solving problems and refining measurements rather than repeatedly testing their most basic assumptions. His account complicated portrayals of scientists as constantly attempting to falsify foundational theories. It also suggested that a field’s capacity for sustained, technically constrained problem-solving matters alongside the testability of individual statements.
Imre Lakatos shifted attention to research programmes developing over time. A progressive programme generates theoretical advances and novel empirical successes; a degenerating one increasingly accommodates known difficulties without comparable gains. In 1983, Larry Laudan criticized efforts to find a universal demarcation criterion and argued for focusing on the evidential credentials of specific claims. Later approaches have often combined several considerations: testability, evidential support, methodological reliability, explanatory progress, and responsiveness to criticism. Their shared implication is that one failed experiment or one superficial checklist rarely settles a complex case.
Historical development and changing boundaries
Practices now classified as pseudoscientific often arose before modern disciplinary boundaries existed. Astrology, for example, was historically intertwined with astronomical observation, calendrical calculation, medicine, and political advice. Its presence in learned institutions does not establish the truth of its predictive claims, but it explains why its earlier status cannot be understood simply as an attempt to imitate an already separate modern science. Over time, astronomy developed increasingly successful mathematical models and measurements, while astrological interpretations did not establish comparable predictive reliability.
Alchemy presents a different historical problem. It combined practical work with substances, theories of matter, medical aims, and religious or symbolic interpretations. Some activities contributed to experimental techniques later important in chemistry, while projects such as metallic transmutation relied on assumptions that subsequently changed. Historians therefore avoid treating the entire alchemical tradition as merely fraudulent chemistry. Present-day assertions that an alchemical procedure has scientifically demonstrated an extraordinary effect must nevertheless be evaluated using present evidence.
In the late eighteenth century, Franz Anton Mesmer attributed therapeutic effects to “animal magnetism,” conceived as a subtle influence or fluid. In 1784, a French royal commission including Benjamin Franklin and Antoine Lavoisier investigated the supposed agent. Its procedures helped separate subjects’ expectations from the alleged magnetic influence, and it found no evidence for the proposed fluid. The episode is important not because it explained every reported symptom, but because it demonstrated how a controlled comparison could challenge a seemingly persuasive therapeutic performance.
Phrenology, developed around the turn of the nineteenth century by Franz Joseph Gall and later popularized by Johann Gaspar Spurzheim, associated mental faculties with localized brain regions and attempted to infer their development from skull shape. The broad proposition that the brain has differentiated functions was not itself the central problem. The distinctive phrenological mapping of character onto external cranial features lacked reliable validation. Its use in judgments about personality and social hierarchy illustrates how an appealing classificatory system can outrun its evidence.
Retrospective judgment must also allow for legitimate ideas that initially lacked a complete explanation. Alfred Wegener proposed continental drift in 1912, but his mechanisms were inadequate and his case faced substantial criticism. Later evidence from seafloor spreading, paleomagnetism, and global seismic patterns contributed to plate tectonics in the 1960s. This history does not show that rejected proposals are generally correct. It shows that controversial ideas gain scientific standing through improved explanations and convergent evidence, not through rejection itself. Being opposed by established authorities is neither evidence of truth nor a sufficient mark of pseudoscience.
Recurring argumentative and methodological patterns
One common pattern is the insulation of a claim from unfavorable evidence. A prediction may initially be precise, but after failure its conditions are reinterpreted so that no result can count against it. An unsuccessful treatment is attributed to insufficient belief; an absent paranormal effect is blamed on skeptical observers; a failed forecast is assigned a new symbolic meaning. Such explanations are not necessarily illegitimate merely because they introduce additional conditions. The problem arises when those conditions have no independent support and are invoked only to preserve the original conclusion.
Selective evidence is another recurring feature. Advocates may collect striking successes while omitting failures, quote favorable passages while ignoring an author’s qualifications, or treat isolated anomalies as decisive against a well-supported theory. This reverses the normal burden of comparative explanation. An anomaly may justify investigation, but it does not automatically support a preferred alternative. A competing explanation must account for the relevant observations, including those already successfully explained by established research.
Pseudoscientific arguments frequently confuse unexplained events with evidence for a particular cause. A person recovering after an intervention establishes a temporal sequence, not necessarily causation. A strange light in the sky does not by itself identify its source. The inferential gap matters because several explanations may fit the same observation. Abductive reasoning, or inference to an explanatory hypothesis, requires comparing alternatives rather than treating the inability to disprove one favored explanation as confirmation of it.
Technical language can create an appearance of precision without supplying measurable content. Terms borrowed from quantum mechanics, energy, or genetics sometimes appear in claims that specify neither a physical quantity nor a mechanism connecting it to the alleged effect. “Energy” has defined meanings in physics; attaching the word to a product does not establish that any relevant transfer has been measured. Analogies can generate hypotheses, but they do not substitute for operational definitions and tests.
Arguments from authority also take distinctive forms. Testimonials, prestigious titles, patents, institutional-looking certificates, and claims of suppression may be presented in place of evidence. None is decisive: a patent is not a general certification of scientific effectiveness, and professional credentials do not make an assertion true. Conversely, lack of credentials does not make a claim false. A related error, sometimes called the Galileo gambit, treats persecution or ridicule as evidence of correctness. Historical cases of vindicated dissent do not establish that a contemporary dissenter’s claim survives scientific scrutiny.
Testing, statistical evidence, and reproducibility
A scientifically useful hypothesis must connect its terms to observations in a way that allows meaningful discrimination among possibilities. Testing an alleged diagnostic device, for example, requires specifying what condition it detects, how outputs are classified, and what independent standard determines whether the condition is present. Without these definitions, almost any output can receive a favorable interpretation afterward. Operational precision limits the freedom to reinterpret an unsuccessful result as success.
Controlled experiments address the fact that outcomes often change for reasons unrelated to the proposed cause. In a randomized controlled trial, random allocation helps distribute known and unknown influences across comparison groups. Blinding reduces the possibility that expectations affect treatment, observation, or reporting. These methods are particularly useful when outcomes are subjective, but they are not universal requirements for every science. Astronomy, geology, and evolutionary research often test explanations through observations, natural comparisons, and converging predictions rather than randomized interventions.
Statistics distinguishes observed differences from the variation expected under specified assumptions. A small -value is not the probability that a hypothesis is false, nor does it establish practical importance. Testing many outcomes and reporting only favorable ones can produce apparently impressive findings even when no genuine effect exists. Small samples, flexible stopping rules, undisclosed exclusions, and publication preferences can further exaggerate evidence. These problems occur within mainstream research as well as in pseudoscience; the distinction concerns whether they are acknowledged and systematically addressed.
Bayesian inference makes explicit the relationship between prior plausibility and new evidence. In odds form,
The equation does not authorize dismissing unfamiliar ideas without examination. It shows why a noisy, weakly controlled observation may provide little reason to accept a claim that conflicts with many independently successful measurements. Conversely, strong and repeatable evidence can overturn a low initial probability. The evidential weight depends on how much more expected the observations are under one explanation than under competing explanations.
Reproducibility and independent replication constrain personal bias, hidden procedural choices, and accidental results. Terminology varies across disciplines, but the central requirement is that other investigators can examine the reasoning and determine whether the effect persists under appropriate conditions. A systematic review can evaluate the whole body of relevant studies rather than a selection of favorable examples. Neither one positive study nor one failed replication normally settles a complicated question. Persistent failure under well-designed tests, combined with an absence of theoretical or empirical progress, is more informative than isolated successes promoted without their broader context.
Representative cases in nature, origins, and history
Astrology offers a clear distinction between a scientifically studied influence and an unsupported interpretive system. The gravitational effects of celestial bodies, seasonal variation, and solar radiation are legitimate subjects of research. These do not establish that zodiacal signs or birth-chart configurations reliably determine personality or individual destiny. Modern tests of astrological claims have not established dependable predictive performance beyond relevant chance expectations and ordinary sources of information. Vague descriptions and retrospective matching can nevertheless make a reading appear highly individualized.
Claims for perpetual-motion machines illustrate the importance of distinguishing a conceptual proposal from a demonstrated apparatus. A machine that continuously delivers net work without the required energy input conflicts with energy conservation; a device that cyclically converts heat from a single reservoir entirely into work conflicts with the second law of thermodynamics. Such laws are supported by extensive, diverse observations. A proposed exception therefore requires careful accounting of electrical supplies, chemical changes, stored mechanical energy, environmental heat, and measurement error. An impressive demonstration with inaccessible components is not an adequate energy balance.
Creationism encompasses religious positions as well as attempts to replace scientific accounts of biological and geological history. Its religious forms are not automatically scientific claims. Young-Earth creationism, however, makes empirical assertions about the chronology of Earth and life that conflict with convergent evidence from radiometric dating, geology, astronomy, and genetics. Intelligent design argues that some features of organisms are best explained by an intelligent cause. Its scientific critics emphasize the lack of an independently testable account of the designer, mechanism, and historical process, rather than treating every philosophical argument about purpose as a laboratory hypothesis.
In archaeology, pseudoscientific narratives frequently infer lost global civilizations, extraordinary ancient technologies, or extraterrestrial intervention from selected monuments and apparent similarities. Such proposals may neglect excavation context, securely dated sequences, local technological development, and the accumulated evidence of ordinary manufacture. A large stone structure can present genuine engineering questions without establishing an unknown civilization. Competing explanations must be assessed against material traces, not merely the modern observer’s surprise.
These cases differ substantially. Astrology is primarily a set of interpretive and predictive practices; perpetual-motion claims concern measurable physical performance; creationist controversies involve both empirical and theological issues; pseudoarchaeology concerns evidence about the human past. Their classification does not rest on one shared implausible belief. It rests on recurring failures to connect claims with adequate tests, incorporate contrary evidence, and develop explanations that outperform available alternatives. Particular assertions can therefore be evaluated without assigning identical motives to everyone who holds them.
Health claims and therapeutic pseudoscience
Health claims require especially careful distinctions because symptoms fluctuate, illnesses have different natural courses, and patients often use several interventions simultaneously. Improvement after treatment can reflect spontaneous recovery, concurrent medical care, changes in reporting, or regression to the mean: unusually severe symptoms may be followed by more typical levels even without an effective intervention. These possibilities do not make a person’s experience unreal. They show why an experience alone cannot establish which component caused the improvement.
The placebo effect is also frequently misunderstood. Expectations, conditioning, clinical interaction, and the context of treatment can influence some outcomes, particularly subjective symptoms. Improvement in a placebo group, however, includes more than a placebo effect: it can also include natural history, measurement variation, and other changes. Contextual benefits do not demonstrate the claimed specific mechanism of a treatment, nor do they establish that a disease process has been reversed. Reliable evaluation separates symptom reports, functional outcomes, biological measurements, and long-term clinical outcomes.
Homeopathy, developed by Samuel Hahnemann from the late eighteenth century, is based on principles including “like cures like” and repeated dilution with shaking. A centesimal dilution reduces concentration by a factor of 100 at each step. After 30 such steps, the nominal dilution factor is ; under ordinary preparation assumptions, a typical dose would not be expected to contain a molecule of the starting substance. Clinical research has not established reliable effectiveness of homeopathic remedies for treating specific conditions beyond placebo effects. Proposals that water retains a therapeutically specific memory of removed substances have not supplied a reproducible mechanism that resolves this evidential problem.
A treatment’s geographic origin, age, or cultural association does not determine whether it works. Traditional practices can contain pharmacologically active substances, ineffective procedures, and interventions whose effects depend on the condition and preparation. Conversely, an industrially manufactured product can be ineffective or harmful. Scientific assessment concerns the specific formulation, dose, indication, comparator, and outcome. An ingredient’s demonstrated biochemical activity also does not by itself establish clinical benefit in humans.
The ethical stakes include direct toxicity, contamination, interactions, financial loss, and delayed effective care. Informed consent requires truthful communication about what is known and uncertain; unsupported certainty undermines that process. In public health, inaccurate claims can also affect others through decisions about communicable disease and collective prevention. These consequences explain the importance of rigorous evaluation without requiring the assumption that every practitioner is deceptive or that every person reporting benefit has been mistaken about the experience itself.
Psychological and social mechanisms of persistence
Pseudoscientific beliefs can be compelling because ordinary cognition is adapted to finding patterns, identifying causes, and learning from experience under uncertainty. These capacities are indispensable but imperfect. Confirmation bias favors information consistent with existing expectations, while memorable events may dominate judgments more than their statistical frequency warrants. A striking coincidence can therefore outweigh many uneventful occasions. Reasoning from a vivid personal case is especially persuasive when the alternative is an abstract comparison across large groups.
The Barnum effect, also called the Forer effect, concerns the acceptance of broad personality descriptions as uniquely applicable to oneself. In a classroom demonstration published in 1949, psychologist Bertram Forer gave students the same general personality sketch while presenting it as individualized feedback. Their favorable assessments showed how apparent personal accuracy can arise without genuine discrimination between individuals. Statements that combine flattering traits with common insecurities are particularly easy to match to one’s experience.
Social interaction can reinforce these effects. In cold reading, a practitioner uses general statements, observation, questions, and the subject’s responses to produce an appearance of unusual personal knowledge. The exchange may involve deliberate technique, ordinary interpersonal inference, or a mixture of both. Subjects often remember accurate statements better than misses and may supply details that later seem to have originated with the reader. An apparently uncanny encounter therefore needs to be assessed as an interactive process, not merely as a final list of successful impressions.
Beliefs also serve social functions. A system may provide identity, belonging, a vocabulary for suffering, or a sense of control during uncertainty. Communities can preserve testimonials, reward expressions of commitment, and interpret outside criticism as confirmation of exclusion. Commercial incentives may favor dramatic promises over qualified explanations. These mechanisms help explain persistence, but they do not by themselves refute a claim: explaining why someone believes something and determining whether it is true are separate tasks.
Scientific training offers some protection through domain knowledge and methodological practice, but expertise is specialized. A highly accomplished scientist can make unsupported claims outside a familiar field, and an expert’s commitment to a favored explanation can distort judgment within it. Institutional safeguards matter because individual intelligence and sincerity do not eliminate bias. Social epistemology examines how trust, expertise, testimony, and collective criticism affect knowledge. From this perspective, the reliability of a community depends partly on whether disagreement can expose errors rather than merely threaten group membership.
Institutions, education, and the limits of the label
Scientific institutions employ peer review, editorial scrutiny, data inspection, replication, and professional criticism to improve reliability. None guarantees correctness. Journals can publish flawed work, reviewers can overlook errors, and reputational incentives can discourage correction. A journal article is therefore evidence to examine, not an unquestionable certification. Retractions and revisions may reveal serious failures, but they can also demonstrate that mechanisms for correction are operating. The relevant comparison is not between flawless science and fallible alternatives, but between systems with different capacities to discover and repair mistakes.
Commercial and legal institutions face related problems. Advertising may imply scientific proof through technical vocabulary, laboratory imagery, or endorsements without providing adequate studies. A conflict of interest can create incentives for selective reporting, although its presence does not prove that a finding is false. Regulatory authorization, product registration, patent protection, and professional licensing serve different purposes and vary across jurisdictions. Treating them as interchangeable can mislead consumers about what has actually been evaluated.
Courts sometimes need to assess purported scientific expertise. In the United States, Daubert v. Merrell Dow Pharmaceuticals, Inc. (1993) identified considerations including testability, peer review and publication, error rates, standards, and general acceptance for evaluating scientific evidence under federal rules. These considerations were not presented as a universal philosophical definition of science. In the 2005 federal district court case Kitzmiller v. Dover Area School District, the court held that the school district’s intelligent-design policy violated the constitutional prohibition on government establishment of religion and concluded that intelligent design was not science. Such rulings address particular legal records and institutional responsibilities.
Education about pseudoscience is more informative when it examines how evidence works than when it supplies only lists of discredited beliefs. Comparing a testimonial with a controlled study, distinguishing prediction from retrospective interpretation, and tracing how a measurement could be independently checked make the underlying reasoning visible. Scientific method is not one inflexible sequence followed identically in every discipline. Its more general commitments include public accountability to evidence, explicit reasoning, and opportunities for consequential criticism.
The label “pseudoscience” itself remains open to scrutiny. It can identify a genuine pattern of misleading scientific claims, but it can also function as a rhetorical device in disputes over professional authority, funding, or cultural legitimacy. A defensible application specifies the proposition or practice, identifies the relevant evidential standards, and explains the persistent failures involved. In borderline cases, narrower descriptions—untested, contradicted, methodologically weak, or unsupported by available evidence—can convey the epistemic situation more precisely than a categorical label.