Anchors bias estimates by activating anchor-consistent memories

Correct: B — Considering the anchor selectively activates anchor-consistent information from memory, biasing the subsequent estimate

In a series of experiments, Strack and Mussweiler (1997) showed that the classic anchoring effect (estimates pulled toward an arbitrary anchor) is largest when the anchor and target judgement share a feature dimension and when participants test the anchor as a possible answer. Their selective-accessibility model holds that the comparative judgement triggered by the anchor activates anchor-consistent information in memory, which is then disproportionately used to construct the absolute estimate. The account explains why anchoring is robust even when participants are warned about the bias and even when anchors are explicitly random.