A Failure of Metacognition: Why the Dunning-Kruger Effect Is Actually Stronger for Experts Than for Novices

On April 18, 2022, astrophysicist and Nobel laureate Adam Riess publicly detailed a phenomenon that plagues the highest echelons of science: receiving half a dozen proposals every month from individuals with zero physics background who believe they have solved the mysteries of the universe. This interaction perfectly encapsulates the traditional understanding of the Dunning-Kruger effect. The individuals emailing Riess possessed so little knowledge of astrophysics that they lacked the capacity to recognize their own errors.[3]

David Dunning and Justin Kruger formalized this cognitive bias in 1999, demonstrating that individuals scoring in the 12th percentile on logic and grammar tests routinely estimated their performance to be in the 62nd percentile. The bottom quartile overestimated their raw scores by a massive 45.5 percentage points, guessing they answered 68.6% of questions correctly when they actually managed only 23.1%. The researchers attributed this to a deficit in metacognitive skill—the ability to think about one's own thinking.[1]

The same knowledge required to produce a correct answer is required to recognize a correct answer. Without it, novices suffer a dual burden: they are incompetent, and their incompetence robs them of the ability to realize it. The Decision Lab notes that this manifests clearly in corporate environments. In one software engineering firm, 42% of employees assessed their own performance as belonging in the top 5%, a mathematical impossibility that prevents nearly half the workforce from seeking necessary mentorship.[1][5]

On the road, the consequences of this metacognitive gap are physical. Drivers with less than six months of experience are eight times more likely to crash than the baseline average. This elevated risk is driven not just by a lack of mechanical skill, but by an overconfidence that outpaces their actual vehicle control. Because they do not know what they do not know, they push boundaries they cannot handle.[5]

However, a structural re-evaluation of the 1999 data reveals a secondary, often ignored metacognitive failure that occurs at the opposite end of the spectrum. The top quartile of performers in the original study actually underestimated their performance by 14.1 percentage points. They estimated a score of 72.1% against an actual 86.2%. Because the tasks felt easy to them, they incorrectly assumed the tasks were universally easy for everyone else.[1]

While the absolute numerical gap is smaller for experts—14.1 points compared to 45.5 points—the systemic consequence is arguably more severe. Experts suffer from the 'curse of knowledge,' projecting their comprehension onto others and failing to accurately gauge the difficulty of a task for a true beginner. This creates the 'Expert Blind Spot,' a structural failure in knowledge transfer.[5][6]

A novice's overestimation isolates their own learning. If a beginner believes they are an expert, they simply stop improving themselves. But when an expert underestimates their relative skill and assumes their baseline is the average, they actively degrade the learning environment for everyone else. They design unusable software products, write incomprehensible public policies, and teach highly ineffective classes that leave cohorts of students behind.[6]

The scientific method demands constant hypothesis testing, which breeds inherent doubt among the highly skilled. As Joe Pierre M.D. noted in Psychology Today, this creates a stark contrast in public discourse. Riess observed the disparity firsthand: 'The scientists I know are always worried about whether their understanding is right. These guys aren't worried about that at all.'[3]

The true danger of the Dunning-Kruger effect does not lie solely at the bottom of the competence curve. The metacognitive failure of the expert—the inability to remember what ignorance feels like—creates a bottleneck that prevents the 80% of people who think they are above-average drivers or thinkers from ever actually getting there. The gap at the top does more structural damage than the noise at the bottom.[5][6]