Abstract: There are two main models accounting for anchoring effects: the anchoring and adjustment heuristic and selective accessibility model. Further, anchors can be externally provided or self-generated. Considering prior studies researchers of the prior studies found that the adjustment heuristic model accounted more for anchoring effects in the case of responding to self-generated anchors, while the SA model accounted more for responses to externally provided anchors. In the present study, a mental scale was first introduced to distinguish the effortful adjustment process from effortless accessibility, using an event-related brain potential (ERP) technique. We hypothesized that the coarseness of the scale would influence the process related to the response to self-generated anchors but not that related to experimenter provided anchors.
Two ERP experiments on young normal subjects examined the impact of a mental scale on responses to self-generated anchors and externally provided anchors, respectively. We also designed a new paradigm for ERP recording. In Experiment 1, higher or lower numbers as externally provided anchors were presented as cues on the screen; the participants’ task was to estimate the quantity of dots in a target circle, analogous to the two-stage standard anchoring paradigm. In Experiment 2, the cues serving as self-generated anchors were the circles for which the participants had already estimated the number of the dots, analogous to the one-stage anchoring paradigm. Behavioral results and ERP results were collected for statistical analysis.
Behavioral results validated the effectiveness of the mental scale in diminishing the anchoring bias in the case of responding to self-generated anchors. Adjustment on a coarser scale with fewer categories resulted in a greater skewness from the anchors. Further, ERP results showed a scale effect in the anchoring process even in the case of responding to self-generated anchors too. A positive deflection appeared at 250–800 ms after target onset, and the amplitude of this deflection was more positive in the finer scale condition. Source localization found two free dipoles: one was at the lingual gyrus; the other at the parietal lobe. In the anchoring process, an N300 component was the only component that responded to externally provided anchors and whose amplitude was not influenced by the mental scales. The results of dipole analysis showed that N300 was localized at the cingulate gyrus.
In conclusion, the two experiments in the present research showed that the manipulation of the mental scale was proficient in anchoring effects only in the case of responding to self-generated anchors, when the true adjustment process was invoked. ERP results provided more evidence of the neural correlate of anchoring effects. An N300 component was related to the accessibility process responding to externally provided anchors, and the positive deflection at 250–800 ms poststimulus, which was modulated by the mental scale, was related to the adjustment process responding to self-generated anchors. In all, our findings provided electrophysiological evidence for the dual nature of anchoring effects. Whether anchoring serves the purpose of adjustment or accessibility is subject to the relevant context, which is consistent with the two-system review. Although anchoring leads to judgment bias, anchoring effects reflect the flexible intelligence of human beings

Key words: anchoring, mental scale, adjustment, accessibility, event-related potentials