What Is Priming? A Real Effect Surrounded by Discredited Claims

Repetition priming and semantic priming are among the most robust findings in cognitive psychology. Presenting a word makes that word faster to recognise later, which is repetition priming, and presenting a related word makes recognition faster too, so somebody shown nurse identifies doctor more quickly than an unrelated word. The effects are measured in tens of milliseconds, replicate consistently across laboratories and languages, and are used as a standard tool for investigating how meaning is organised in memory rather than being studied for their own sake. Related forms include perceptual priming, where a partially degraded image is easier to identify after previous exposure, and masked priming, where the prime is presented too briefly to be consciously perceived and still produces measurable facilitation. These are effects on processing speed and accuracy, they are small, and they are exactly what a network-based account of memory predicts, which is why they are theoretically interesting.

A separate body of work claimed much larger effects on behaviour and most of it has not held up:

• •A widely cited study reporting that participants exposed to words associated with old age subsequently walked more slowly, which failed to replicate in large well-controlled attempts
• •Studies reporting that priming intelligence-related concepts improved quiz performance, which likewise failed
• •Claims about physical warmth influencing judgements of social warmth, whose replication record is poor
• •Money priming, cleanliness priming and numerous variants reported through the 2000s, most of which have weak or failed replications
• •A pattern of small samples, flexible analysis and publication of positive results only, which inflates apparent effects across a literature
• •The response from within the field, including a widely discussed open letter in 2012 urging researchers to check their methods, which marked the start of a broader reckoning

The collapse of behavioural priming was a substantial part of what is now called the replication crisis, and the lessons extend far beyond this topic. Small samples produce unstable results and exaggerate whatever effect reaches significance, which is the winner's curse and which explains how a literature of positive findings can describe effects that do not exist. Flexibility in analysis, including choosing which measures to report and when to stop collecting data, inflates false positives enormously, and the practice was widespread and not regarded as misconduct. Publication bias removes the failures from the record. Direct replications were undervalued and hard to publish, so nobody checked. The remedies adopted since include preregistration of hypotheses and analysis plans, larger samples, multi-laboratory replication projects and registered reports where a journal accepts a study on its design before results exist. Those changes are real and their adoption is partial, and psychology has moved further on them than most fields.

The practical question for anyone encountering a psychological finding is which category it falls into, and a few checks help. Effect size matters more than significance, since a statistically significant result from a large sample may describe a change too small to matter. Sample size and whether the study was preregistered are usually stated and are informative. Whether independent replication exists is the single strongest signal, and a finding repeated by other laboratories with larger samples is in a different class from one reported once. Extraordinary claims about subtle cues producing large behavioural changes deserve particular scepticism, since that is exactly the pattern that failed. Media reporting systematically overstates, and the original paper frequently claims less than the coverage. None of this requires dismissing psychology, since the field identified its own problem and has changed its practices substantially, which is more than most disciplines facing the same issues have done.

Seeing a word makes a related word faster to recognise, measured in tens of milliseconds, and that effect replicates consistently and is used as a research tool. Claims that exposure to concepts changes behaviour substantially mostly failed replication, and that failure drove the wider reckoning over small samples, flexible analysis and publication bias. Independent replication is the strongest signal a finding is real.