Definition
Anchoring bias is a systematic departure from rational estimation in which an initial reference point — the anchor — exerts disproportionate influence on subsequent judgments, even when the anchor is arbitrary, irrelevant, or explicitly identified as such. First documented experimentally by Daniel Kahneman and Amos Tversky in the 1970s, anchoring is one of the most replicable findings in behavioral economics and decision psychology.
The classic demonstration involves two groups: one is asked whether the percentage of African nations in the United Nations is above or below 65, the other whether it is above or below 10. Both groups then give their best estimate. The group anchored at 65 produces estimates averaging around 45; the group anchored at 10 produces estimates averaging around 25 — despite both groups having access to the same knowledge and being asked an identical subsequent question. The anchor, arbitrary and transparently so, shifts the distribution of responses by twenty percentage points.
What makes anchoring remarkable is its robustness. It persists even when participants are warned about it, even when the anchor is patently absurd (a spinning wheel of fortune producing a number), and even when participants are financial professionals estimating quantities in their area of expertise. Partial adjustment — moving away from an anchor but not far enough — is the default mode of human estimation.
Why it matters
How it works
The adjustment heuristic
The dominant mechanism proposed for anchoring is the anchor-and-adjust heuristic: people take the initial value as a starting point and adjust from it in what seems like the right direction, stopping when they reach a value that feels plausible. The adjustment is typically insufficient because the process terminates as soon as the estimate enters a range that no longer seems clearly wrong — not when it reaches the most accurate estimate. The anchor thus functions as a gravitational attractor that the final estimate orbits too closely.
A secondary mechanism is selective accessibility: exposure to an anchor activates anchor-consistent information in memory, making evidence compatible with the anchor more cognitively salient. This means even estimates that don't follow the classic adjust-from-anchor path may be subtly shaped by the anchor through the distortion of what evidence comes to mind.
Anchoring in social and institutional contexts
Anchors are not always accidental. Skilled negotiators, salespeople, and policymakers deliberately deploy anchors to shape outcomes. A high opening price in a negotiation anchors the entire price discussion. A bold policy proposal anchors the range of what counts as 'moderate.' A diagnostic hypothesis voiced early in a clinical case conference anchors the team's interpretation of subsequent evidence.
Institutions can also create structural anchors through default options, reference prices, and standard templates. A salary band shown to candidates anchors their expectations; a default subscription price anchors willingness to pay. Designing these anchors — or de-biasing them — is a form of choice architecture with real distributional consequences.
Where it goes next
Anchoring sits within the broader ecology of cognitive heuristics alongside hyperbolic discounting and other departures from rational-agent models. Understanding how anchors interact with reasoning processes — why adjustment terminates early, what debiasing actually works — connects to the psychology of judgment under uncertainty. In practical application, the field of negotiation has built systematic strategies around both exploiting and neutralizing anchors.