Log In
Log In
Led by Popper Simulacrum
The scientific method from falsification and paradigm shifts through experimental design, causal inference, and the replication crisis.
Led by Popper Simulacrum
The question
A theory that cannot be falsified is not scientific — it is a story. Astrology makes predictions so vague that any outcome confirms them. Freudian psychoanalysis interprets any behaviour as evidence for its theory (the patient who agrees with the interpretation confirms it; the patient who disagrees is "in denial" — also confirming it). Popper's demarcation criterion: a theory is scientific if and only if it makes predictions that could, in principle, be shown to be false.
Outcome
The student can state Popper's demarcation criterion, explain the asymmetry of falsification, describe the Duhem-Quine problem, and apply the practical criterion ("what would change your mind?") to evaluate a claim. (Falsification)
Sub-units
Led by Lakatos Simulacrum
The question
Popper says: falsify and abandon. But scientists do not actually work this way — they hold onto theories through anomalies, modifying auxiliary hypotheses to save the core. Are they being irrational? Lakatos says no — a research programme can rationally accommodate anomalies, as long as the modifications are progressive (they predict new facts) rather than degenerating (they only explain away old anomalies).
Outcome
The student can describe Kuhn's paradigm shift model, describe Lakatos's research programme model (hard core, protective belt, progressive vs. degenerating), apply the progressive/degenerating distinction to a case study, and explain when theory abandonment is rational. (Research programmes and paradigm shifts)
Sub-units
Led by R.A. Fisher Simulacrum
The question
The experiment is the most powerful tool humanity has ever devised for establishing causal claims. But a badly designed experiment is worse than no experiment at all — it produces confident but wrong conclusions. The three pillars of experimental design are: the control group (what happens without the intervention), randomisation (ensuring the groups are comparable), and blinding (preventing expectations from influencing results). This module teaches the student to evaluate experimental evidence rigorously — to distinguish the well-designed study from the poorly designed one.
Outcome
The student can explain the necessity of control groups, the purpose of randomisation, the function of single and double blinding, describe the placebo effect, and explain sample size and statistical power. (Experimental design)
Sub-units
Led by Pearl Simulacrum
The question
Correlation is not causation — every student knows this. But what IS causation, and how do we establish it? Judea Pearl's framework provides the answer: causation is the effect of an intervention — not "what happened when X was observed" but "what would happen if we did X." The do-calculus formalises the distinction between seeing and doing, and the ladder of causation shows that causal reasoning requires a cognitive leap that no amount of observational data can provide.
Outcome
The student can describe the three rungs of Pearl's ladder, explain the difference between P(Y|X) and P(Y|do(X)), apply the back-door criterion to a simple causal graph, describe the front-door criterion as an alternative when confounders are unmeasured, and explain counterfactual reasoning. (Causal inference)
Sub-units
Led by Popper Simulacrum
The question
Science is self-correcting — but only if it actually corrects itself. The replication crisis (2010s–present) revealed that a disturbingly large proportion of published findings in psychology, medicine, and other fields cannot be replicated by independent researchers. The crisis has exposed systemic problems: p-hacking (manipulating data analysis to produce significant results), publication bias (journals publish positive results and reject negative ones), HARKing (Hypothesising After the Results are Known — presenting post-hoc findings as pre-planned), and outright fraud. This module examines the crisis and its solutions.
Outcome
The student can describe the replication crisis with quantitative evidence, describe p-hacking, publication bias, and HARKing, and describe four solutions (pre-registration, registered reports, open data, adversarial collaboration). (The replication crisis)
Sub-units