24. Historical Science & Causal Reasoning

 

Summary

Moving on from experimental science and Mill's methods, this episode turns to historical science and then examines the most common mistakes in causal reasoning.

Experimental vs. Historical Science

The scientific search for causes splits into two categories. Experimental science (which Di Donato prefers to "empirical," since all science rests on empiricism) deals with present observation, experimentation, and recurring events governed by natural laws. Historical science deals with unobserved events of the past for which no direct experimental data exists: disciplines such as forensic science, paleontology, archaeology, historical geology, and cosmology.

Crucially, experimentation still matters for historical science. Scientists and historians stage experimental reenactments to test hypotheses about the past, for example how the Egyptians moved pyramid blocks, how the Greeks produced red- and black-figure vases, or how far a man could run wearing 70 pounds of hoplite armor. Observation remains the key to all science, even when the objects themselves (subatomic particles, very large or very old things) cannot be seen directly.

Two Core Principles: Causality and Uniformity

Historical reconstruction rests on two principles. The principle of causality holds that every event has a cause: "everything that begins has a cause," and "nothing comes from nothing." It does not claim that everything whatsoever has a cause, and we must never confuse ignorance of a cause with the absence of one.

Causes come in two kinds: natural causes (the ordinary, non-intelligent patterns of nature, like erosion) and agent causes (goal-directed, intelligent causes, like a sculptor). Correspondingly, explanations are either physical (natural sciences, explaining material background) or behavioral (psychology, sociology, economics, explaining behavior via reasons, motives, goals, and habits).

The principle of uniformity states that the present is the key to the past: the same causes operating now operated then. This is an inductive argument by analogy, comparing present with past, and it can also project from past and present toward the future. Di Donato reviews the analogical argument form (ground, basis, objective extension, problematic extension, with conclusions only ever probable) and the six criteria for evaluating analogies: more cases compared, more similarities, relevance over sheer number, the weakening effect of disanalogies, the strengthening effect of variety within a strong ground, and the rule that stronger conclusions place a heavier burden on the premises.

Mistakes in Causal Reasoning

The bulk of the episode catalogs causal fallacies, errors of presumption that all share the theme of misidentifying or confusing causes:

1. Post hoc ergo propter hoc ("after this, therefore because of this") — the false cause fallacy, mistaking a temporal sequence for causation (the rooster's crow "causing" the sunrise). Its sibling, cum hoc ergo propter hoc, makes the same error with concomitant factors.
2. Emphasizing irrelevant factors — oversimplifying by overstating an insufficient or irrelevant factor, often from failing to use the joint method.
3. Neglecting negative evidence — overlooking cases where the supposed cause fails to produce the effect, driven by confirmation bias and loss aversion; Mill's negative method of difference corrects this.
4. Neglecting differences — treating mere similarity as identity (claiming Jesus must have been an Essene because of overlapping concerns).
5. Reversing cause and effect — confusing which factor has priority; cause and effect are not strictly temporal but often simultaneous (the brick and the breaking window). The "Latin students have the highest GPAs, so Latin makes students smarter" example illustrates the reversal.
6. Ignoring reciprocal causality — assuming causation is one-directional when it may be bidirectional (video-game violence and societal violence).
7. Confusing cause and condition — conflating necessary and sufficient conditions; a condition is merely needed, while a cause is necessary and sufficient (high LDL cholesterol is a contributing condition, not an outright cause, of heart disease).
8. Ignoring a possible common cause — treating X as the cause of Y when a third factor Z causes both (ice-cream sales and swimsuits both driven by hot weather). Correlation does not equal causation.
9. Discounting coincidence — assuming two proximate events share a cause without evidence (Kennedy, C.S. Lewis, and Aldous Huxley all dying on the same day implying Soviet involvement).
10. Confusing conditional probabilities — a positive test is an effect of a condition, not proof of it. Even a 90%-accurate bladder-cancer test, given false positives and a low base rate, yields only an 8% chance of actually having the disease.
11. Overlooking statistical regression (regression to the mean) — extreme measurements tend to be followed by ones nearer the average. This explains why very tall fathers have somewhat shorter children, why casinos profit, and why one-hit wonders exist; it stems from sampling error, underscoring the need for randomization and control groups in trials.
12. Proof by absence of disproof (argumentum ad ignorantiam) — treating the failure to disprove a hypothesis as evidence for it (nobody has disproven homeopathy).
13. Confusing explanations with excuses — assuming that explaining the cause of an atrocity (e.g., 9/11) amounts to justifying it; an explanation carries no moral endorsement.
14. Confusing kinds of causality — Di Donato's "Aristotelian" favorite. Aristotle distinguished four causes, expanded here to six, illustrated by making soup: the efficient cause (the cook, the agent that makes it happen), the material cause (water, chicken, pasta), the formal cause (the essence, "soupness"), the final cause (the goal: to satisfy hunger), the exemplar cause (the recipe or model), and the instrumental cause (the means: pot, stove, spoon).

Wrap-Up

This closes the course's treatment of induction and science. The next episode shifts gears entirely to deductive logic, which Di Donato promises is among the most practical skills critical thinking has to offer.