Metacognitive Hindsight Bias Across Age Groups
Year: 2026
Type: other
Abstract: This project is an addition to two previously preregistered experiments: Exp. 1 [https://osf.io/7u862/overview] and Exp 2 [https://osf.io/4w9gm/overview]. We will use multinomial processing tree (MPT) modelling to reanalyse data from these two completed experiments examining visual hindsight bias (HB) and metacognitive hindsight bias (MCHB) across age groups. Hindsight bias refers to the tendency to remember an earlier judgment as being closer to a known outcome than it actually was. MCHB extends this idea to confidence: after learning the identity of an object, people may remember having been more confident in their earlier identification judgments than they originally were. Although the original analyses showed HB and MCHB across the age groups, conventional difference scores cannot identify the cognitive processes responsible for these biases. The purpose of the present project is to determine whether age groups may show similar overall levels of bias while differing in recollection, biased reconstruction, and confusion between current and prior knowledge. We will apply the VHB3 model developed by Bernstein et al. (2011) to the visual-HB data from both experiments. We will also extend the same framework to MCHB by modelling participants’ foresight and hindsight confidence judgments. For visual HB, the model will estimate r, the probability of recollecting and using the original foresight judgment; b, the probability that outcome knowledge biases reconstruction when the foresight judgment is not used; and c, the probability of confusing current outcome knowledge with one’s earlier knowledge. The corresponding MCHB parameters will be labelled r_m, b_m, and c_m. For the MCHB analysis, only corresponding foresight–hindsight confidence pairs treated as correct in both phases will be included. Experiment 1 included Grade 3 students, Grade 7 students, and young adults who completed a visual object-identification task involving six objects presented across 10 increasingly clear blur levels. During the foresight phase, participants identified each object and rated their confidence at every level. After learning each object’s identity, they attempted to recall their earlier responses and confidence ratings. Experiment 2 used a similar task with younger and older adults. No new data will be collected. The response categories required for the MPT models have been constructed, but no MPT parameter estimates or model comparisons have been examined. Based on the real-object results reported by Bernstein et al. (2011), we expect recollection parameters rand r_mto be relatively stable from older childhood through young adulthood and lower among older adults than younger adults. Reconstruction parameters band b_mare expected to be relatively stable across the younger age groups, with a tendency toward higher estimates among older adults. Knowledge-confusion parameters c and c_mare expected to be close to zero across all age groups. The expected contribution of this project is to clarify whether age affects the processes underlying HB and MCHB even when age groups do not differ significantly in their overall bias scores. For example, two groups could show similar average bias while differing in whether it resulted from poorer recollection, more outcome-biased reconstruction, or greater confusion between current and prior knowledge. The analyses will apply VHB3 to visual hindsight bias assessed using a memory design and will test whether the same modelling framework can also be extended to biased memory for prior confidence. To our knowledge, VHB3, originally applied by Bernstein et al. (2011), has previously been used only with hypothetical hindsight designs, in which participants make judgments on behalf of a naïve peer rather than attempt to recall their own earlier judgments.
Source: Open Science Framework
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7171732224