Psychological Science: What It Is and How It Differs From Common Sense

Ask ten different people why a teenager is acting a certain way, and you’ll get ten confident explanations. Ask a psychologist, and the first response is often another question: how would we check that? That’s the difference that matters, more than the subject matter itself.

A test, not an argument

The philosopher Karl Popper proposed a precise way to separate scientific claims from ones that aren’t: an idea is scientific if some observation, even a merely conceivable one, could prove it wrong. The Stanford Encyclopedia of Philosophy entry “Science and Pseudo-Science” reports Popper’s original statement of the demarcation problem: the need for a criterion to differentiate between “empirical-scientific” and “metaphysical” assertions. In a later text, Popper sharpened it: to count as scientific, statements must be capable of conflicting with possible or conceivable observations.

Applied to psychology, this means a hypothesis about human behavior only counts as scientific if it predicts something that might not happen. “People act in their own self-interest” doesn’t clear that bar, as long as it stays vague enough to explain any behavior, including one that looks selfless. “Under conditions of anonymity, the share of people who return a lost item drops below some measurable threshold” does, because an experiment could disprove it.

Real-world example: the Stanford Encyclopedia of Philosophy entry makes a point that surprises anyone hearing it for the first time: psychoanalysis, often trotted out as the textbook pseudoscience, doesn’t have the problem most people assume. According to the entry, plenty of its claims have gone through testing, on more than one front, and haven’t survived the scrutiny — that’s the real difficulty, not some inability to test them in the first place. Astrology fits the same pattern: Popper himself pointed to it as pseudoscience’s clearest textbook case, and it earned that label because researchers checked its predictions and they came up short, not because nobody could check them.

Why common sense doesn’t count as proof

A 2022 philosophy-of-science paper in Science & Education (Blancke & Boudry) zeroes in on the underlying issue: intuition and “common sense” are extremely unreliable once you step outside the environment our minds evolved to handle. Common sense is fine for crossing a street or reading a friend’s mood. It does a much worse job when the question is whether a drug works, whether a teaching method improves learning, or whether a vivid memory is necessarily an accurate one.

The same authors point out that science compensates for these built-in cognitive limits with purpose-built tools: telescopes and scanners to extend what we can observe, mathematics, logic and statistics to sharpen how we reason. They add a point about people rather than instruments: the most important error-correcting mechanism in science is the scrutiny of fellow scientists, which over time converges on the best available approximation of what’s true.

One concrete case of ordinary intuition going wrong once it’s pushed past its comfort zone comes straight from psychology: hindsight bias, the “I knew it all along” effect. A 2007 study in Child Development tested preschoolers with a simple task: after recognizing an object shown in progressively sharper detail, they had to guess when a make-believe peer (the Sesame Street puppet Ernie) would be able to identify it. Across 144 children aged 3 to 5, the effect showed up clearly: once they knew what the object was, the children systematically underrated how hard it would have been to spot it earlier on. The study also found a correlation between this bias and the children’s theory of mind (r = -0.31).

Correlation is not causation

The third ingredient separating psychological science from off-the-cuff reasoning is how data gets handled. Finding that two variables move together is a starting point, not a conclusion. A psychologist writing for the Association for Psychological Science made the point in a 2006 article: a purely correlational design rarely pins down a cause on its own, because too many possible third variables could be behind the same pattern.

Real-world example: the article walks through a concrete case, often used for teaching purposes: a positive correlation between babies’ birth weight and their maternal grandfathers’ income. Taken on its own, that might look like proof of a direct link between family wealth and health at birth. The author instead proposes an alternative explanation involving a third variable: residence in a severely low-income neighborhood marked by poor prenatal nutrition. That neighborhood can account for both the grandfathers’ low income and the grandchildren’s lower birth weight, with neither causing the other.

That’s why, whenever it’s ethically possible, psychological science turns to experimental designs with random assignment of participants to conditions: spreading every other difference between people evenly across groups isolates the effect of the variable under study more cleanly. It’s slower and more expensive than an eyeballed observation, and it leans on the same toolkit covered in /en/science/probability-and-statistics/: that extra effort is exactly what separates a science from a well-argued opinion.

The same principle holds outside psychology too. Evolutionary biology, for instance, remains scientific by this same falsifiability standard: Darwin’s theory of natural selection, covered in /en/science/darwin-natural-selection/, makes testable predictions about the fossil record and population genetics, which is why it has stood up to the evidence for more than a century and a half.

What this changes, in practice

Knowing that psychology is an empirical science doesn’t mean demanding absolute certainty about every piece of human behavior: it means knowing which questions to ask a claim before believing it. Could this be shown false by something that might happen? Does the data come from a controlled observation, or from a memory revised after the fact? Has the correlation found ruled out the more obvious alternative explanations? Researchers in psychology ask themselves these questions before writing a conclusion; common sense, by its nature, tends to skip all three.