What the human sciences study

The human sciences are a group of disciplines that systematically investigate human behaviour, experience and social organization. Among them are psychology, economics, sociology, anthropology, political science and human geography. Their subject matter varies, yet each discipline studies beings who interpret their circumstances and may change their behaviour when they gain knowledge about themselves.

This brings the field into contact with other areas of knowledge. Biological psychology draws on the natural sciences; economic history links explanations of human behaviour to evidence about the past. Such boundaries help to organize inquiry, but they are not fixed divisions in reality.

Human scientists might aim to describe what people do, explain why patterns occur, interpret the meaning of actions, or predict what is likely to happen. These aims overlap, but they aren't identical. For example, a survey could describe voting intentions accurately without explaining them. An interpretation of a ritual might clarify its meaning without leading to a reliable prediction.

Patterns, causes and laws

A causal explanation is an account that identifies a factor whose presence or alteration contributes to an outcome. Human outcomes generally arise from several interacting causes, such as institutions, material conditions, beliefs, identities and chance events. Treating one factor as a complete explanation is therefore risky.

A law is a general statement that claims a stable relationship between specified phenomena. Some human scientists search for law-like regularities. Human behaviour, though, is rarely as uniform as the behaviour captured by many physical laws. People learn, resist expectations and react differently to institutions. As a result, generalizations in the human sciences are often probabilistic and conditional. They point to tendencies under stated circumstances rather than exceptionless rules.

Human-scientific knowledge isn't arbitrary as a result. Researchers can test a claim against evidence, compare it with rival explanations and revise it when it fails. Rather than asking only whether the human sciences are “scientific”, it is more useful to ask which standards of evidence fit each research aim.

Knowledge, belief and opinion

Knowledge is a responsibly justified cognitive achievement that is sufficiently connected to truth for its context. A belief is a proposition that a person accepts as true, whether or not adequate justification is available. An opinion is a judgment that expresses a person’s assessment and may be supported to different degrees by evidence. In the human sciences, confidence alone cannot separate these categories.

A well-supported claim will normally use transparent methods and relevant evidence, offer a reasoned interpretation and undergo some scrutiny by other researchers. Being contestable is not, by itself, a defect. Claims about unemployment, prejudice or well-being may remain disputed because researchers can define the concepts differently, while the evidence may allow more than one interpretation. Contestability becomes a weakness when no possible counter-evidence is allowed to challenge the claim.

Quantification and language

Numbers make patterns easier to see. They allow comparisons and can expose differences missed by casual observation. Yet they may also suggest a level of precision that the underlying categories cannot support. Any numerical measure of “development”, “intelligence” or “happiness” rests on earlier judgments about what counts and how it should be measured. Graphs and statistics don't eliminate interpretation. Instead, they concentrate it in decisions about definitions, samples and presentation.

Language has a similar effect. Words such as “deviant”, “rational” or “developing” classify people and carry assumptions about what counts as normal or desirable. When a category gains institutional influence, people may change their behaviour in response to it. Human-scientific descriptions can thus help to shape the reality they describe.

Links across TOK

Like the natural sciences, the human sciences value evidence, criticism and systematic method. Their objects of study, however, are agents who make meaning. They also differ from history by seeking general patterns across cases more often, although both areas interpret evidence about human action. This leads to a central linking question: should every area of knowledge use the same standards of explanation? One useful answer separates shared intellectual values, such as honesty and responsiveness to evidence, from methods that need to suit different objects and purposes of inquiry.

Knowers studying other knowers

A perspective is a standpoint shaped by a knower’s concepts, experiences, values and social position. Perspective enters human-scientific research at several points. Researchers choose questions, set categories and interpret evidence. Participants bring their own perspectives too: they give meaning to their actions rather than acting as passive objects.

Reflexivity is the disciplined examination of how a researcher’s position and choices influence an inquiry. Researchers don’t have to give up the pursuit of reliable knowledge. They instead make relevant influences visible by stating their assumptions, describing relationships with participants and explaining interpretive decisions.

This relationship works both ways. Researchers interpret participants, while participants interpret researchers. Both act within a social setting that gives those actions meaning. Once research becomes public, it may shape later behaviour, creating a feedback process that many studies of non-human objects do not face.

Individual and collective perspectives

One person’s account may offer indispensable evidence about motives or experience, but it cannot automatically establish a broader social pattern. Population-level data can reveal such a pattern while hiding how individuals understand it. The level of analysis therefore matters. An explanation based on individual decision-making may differ from one based on institutions, class or culture.

Researchers must not assume that their own categories apply universally. Ethnocentrism is a form of interpretive bias that treats the standards of one culture as the natural measure of all cultures. Comparative research can reveal this error. Even so, comparison depends on chosen categories and cannot provide a view from nowhere.

Standpoint is a socially situated position that makes some features of experience more accessible while making others harder to notice. An insider may grasp language and implicit norms that an outsider misses. By contrast, an outsider may spot assumptions that insiders take for granted. Neither position guarantees truth. Credible inquiry therefore explains what each position makes possible and what it limits.

Objectivity and interpretation

Objectivity is an epistemic ideal that requires claims to be assessed through standards not determined solely by a particular knower’s preferences. This isn’t the same as having no perspective. Explicit definitions, public methods, critical review and comparisons among researchers can make knowledge less dependent on any one person’s viewpoint.

Interpretation is the reasoned assignment of meaning to evidence within a conceptual context. It cannot be avoided when evidence concerns intentions, identities, institutions or symbols. Still, interpretations face constraints. They should account for the evidence, fit the context and stand up against plausible alternatives.

Different theoretical perspectives may arrange the same evidence in different ways. One account of workplace behaviour might focus on incentives; another might stress group identity or unequal power. The presence of alternatives does not mean that all explanations are equally good. Researchers can compare their evidential support, explanatory range, assumptions and counterexamples.

Language, power and classification

Categories do more than describe people; they can also distribute social opportunities. Definitions of poverty, disability or criminality influence who gets counted and how institutions respond. A social construction is a category or practice whose form depends substantially on shared human conventions and institutions. Describing something as socially constructed does not make it imaginary. Money and legal status are constructed, yet they produce concrete consequences.

Here, the human sciences connect with politics and ethics. Who has the authority to classify others? Whose experiences count as evidence? A knower should examine both what a claim says and the position from which it was produced. However, rejecting a claim solely because of its source commits the genetic fallacy.

From concepts to evidence

Human-scientific inquiry turns broad concepts into questions that can be researched. Operationalization is a methodological process that specifies how an abstract concept will be identified or measured. “Trust”, for instance, could be represented through survey responses, observed cooperation or a willingness to accept risk. Any one of these choices captures certain aspects while leaving others out. An operational measure shouldn’t be treated as the whole concept.

A hypothesis is a testable proposition predicting a relationship or difference. Hypotheses give data collection a clear focus. Exploratory research may take another route, starting with open questions and developing concepts as evidence emerges. Neither approach is automatically better; the choice depends on the aim.

Research rarely follows a perfectly straight path. Researchers move back and forth between concepts, evidence, analysis and revision. They may also combine quantitative and qualitative approaches, allowing broad patterns and situated meanings to check and enrich each other.

Quantitative methods

A quantitative method is a research procedure that represents observations numerically so that their distribution or relationships can be analysed. Surveys, structured observations and experiments can show patterns across many cases. Their strength rests on valid measurements and good sampling, not just a large sample size.

A population is the complete group about which a study seeks knowledge. A sample is a selected subset of that population from which evidence is collected. Even a large sample can mislead when the selection process systematically leaves out relevant groups. Random selection can reduce selection bias. Representative sampling, by contrast, aims to reflect important characteristics of the population.

An experiment is a controlled investigation in which researchers manipulate an explanatory factor and observe its effect on an outcome. Through control and random assignment, experiments can strengthen causal inference. But laboratory conditions may alter participants’ behaviour. They can also simplify the social setting to the point that conclusions don’t transfer easily to everyday life.

Correlation is a statistical relationship in which two measured variables vary together. On its own, correlation doesn’t show that one variable causes the other. Reverse causation, an unmeasured common cause or coincidence may account for the observed relationship. A causal claim requires a plausible mechanism and the appropriate temporal order, along with evidence that rival explanations have been addressed.

Qualitative methods

A qualitative method is a research procedure that investigates meanings, experiences or processes primarily through non-numerical evidence. Interviews, participant observation and textual analysis can uncover contextual detail missed by fixed-response instruments.

Participant observation is a method in which a researcher studies a group while taking part in or closely observing its activities. This may reveal practices that participants consider too ordinary to mention. At the same time, access, relationships and the researcher’s presence can shape the evidence. Detailed field notes and reflexive reporting allow readers to judge the interpretation.

A case study is an intensive investigation of a bounded person, group, institution or event. A case study might challenge a supposed universal rule, reveal a process or generate a new explanation. Researchers cannot generalize its findings simply by assuming that one case statistically represents a population. Instead, they need to justify which features may transfer to other contexts.

Reliability, validity and replication

Reliability is a property of a procedure that produces sufficiently consistent results when relevant conditions are repeated. Validity is a property of an inference or measure that warrants the interpretation made from it. A questionnaire may be reliable because it repeatedly gives similar scores, yet still be invalid if those scores don’t adequately represent the intended concept.

Replication is the repetition of a study’s procedure to test whether a finding recurs. Exact repetition can be difficult when societies change or participants respond to earlier research. For this reason, conceptual replication may be especially useful: it tests the same proposed relationship with different measures or in different settings.

Triangulation is a strategy that compares evidence from different methods, sources or researchers. Agreement can raise confidence in a finding. Disagreement, though, may show that the methods capture different dimensions rather than proving that one result is simply wrong.

Models, prediction and explanation

A model is a deliberately simplified representation used to describe, explain or predict selected features of a system. Economic and psychological models isolate variables, making relationships more manageable. Their assumptions aren’t necessarily flaws. What matters is whether the simplification fits the purpose and whether omitted factors affect the conclusion.

Predictions in the human sciences are often conditional. They may describe probabilities across groups instead of predicting the action of a single individual. A prediction can also change the conditions it describes. For example, forecasting a shortage may encourage buying that helps cause the shortage, whereas a risk warning may prompt action that prevents the predicted outcome.

When assessing a human-scientific claim, follow the entire chain from concept to conclusion. Ask how researchers operationalized the concepts and selected the cases. Consider whether the design supports causal inference, what uncertainty remains and whether the conclusion goes beyond the evidence. This doesn’t call for one universal method. It requires the method to fit the claim.

Links across TOK

Both the human and natural sciences use models, measurement and experiments. In the human sciences, however, measurement can interact with self-understanding and social institutions. Mathematics offers powerful tools for identifying patterns, but mathematical precision cannot fix vague concepts or biased data. Interpretation also connects the human sciences with history. In both areas, evidence gains significance through contextual reasoning, although human scientists may put more emphasis on cross-case comparison and prediction.

Ethics within research

Research ethics is a field of normative inquiry that evaluates how knowledge should be produced and used. In the human sciences, ethical requirements aren’t decorations added after the method has been chosen. They shape which questions may be asked, who may participate and which procedures are permissible.

Informed consent is a voluntary agreement to participate that is based on an adequate understanding of relevant procedures, risks and rights. Consent continues throughout a study; it isn’t just a one-off signature. Securing it fully can be difficult when researchers use deception, observe public behaviour, work across languages or involve people whose freedom to refuse is limited.

Researchers must minimize harm, protect privacy, preserve confidentiality, allow withdrawal and treat participants fairly. Confidentiality is a duty to restrict access to identifiable information entrusted to a researcher. Anonymity is a condition in which a participant’s identity is not known or cannot reasonably be connected to the evidence. The two protections aren’t the same. A researcher may know an interviewee’s identity but still promise confidentiality.

During ethical review, potential knowledge is weighed against risks to participants and communities. A checklist can’t replace judgment: physical, psychological, social and reputational harms depend on the context.

Deception, observation and power

Sometimes deception stops participants from adjusting their behaviour to fit what they think the study aims to show. However, it interferes with informed consent and may damage trust in research. Researchers need to consider whether the deception is necessary, whether a less intrusive method could work and whether participants can be properly informed afterwards.

Nor is observation in a public place automatically harmless. People can be visible without expecting anyone to record their actions, combine them with other data or publish them permanently. This matters especially in digital research. Technically accessible information may still have been shared with only a limited audience in mind.

Power differences can determine whether participation is genuinely voluntary. Payment may provide fair compensation, while a large incentive can place undue pressure on people facing financial difficulty. When research involves prisoners, children, displaced people or employees, researchers must pay particular attention to dependence and the possibility of retaliation.

Values in the production of knowledge

A value is a normative commitment concerning what is important, desirable or worthy of protection. Values shape the problems that receive funding, the outcomes researchers measure and the risks they tolerate. Evidence is not therefore merely preference in disguise. Researchers should separate empirical findings from value judgments and be transparent about choices that can’t be avoided.

Evidence, for instance, may estimate how a policy distributes benefits and burdens. On its own, though, it cannot decide which distribution is fair. Moving from “is” to “ought” requires a normative premise. Human scientists can clarify likely consequences, uncover hidden assumptions and identify affected groups without claiming that data automatically resolve moral disagreement.

Ethics of application

Researchers are responsible not just to participants, but also to people affected when findings are used. Behavioural predictions may help provide beneficial services. They may also support manipulation, discrimination or surveillance. Even a classification that performs well on average can impose serious costs on individuals who are misclassified.

That responsibility is shared among researchers, universities, funders, governments, businesses and users. Researchers can’t control every later use of knowledge. Even so, foreseeable misuse should shape decisions about design, publication and communication. Suppressing findings may cause harm too, either by preventing scrutiny or by withholding useful knowledge.

Links across TOK

Ethical constraints vary across areas of knowledge because the possible harms are different. No area, however, is value-free. In the natural sciences, concern may focus on experimentation and technological application. In history, it may involve testimony, representation and collective memory. Human-science inquiry can directly classify and influence living people. The linking issue is whether ethical limits weaken knowledge. Such limits often restrict the evidence available, yet they can protect trust, improve participation and rule out methods that would treat people merely as instruments. Ethical quality can therefore form part of epistemic quality rather than stand in opposition to it.