Memory-based and online strategies in retrospective evaluations

Abstract

The “construction” view of preference holds that we sample information from memory in the moment to make judgments and decisions. Building on previous work examining the relationship between memory performance and judgment, we used incentive-compatible methods to investigate both the extent to which memory supports evaluation and the potential trade-off between online and memory-based retrospective evaluation. In a series of pre-registered experiments, we presented people with a sequence of numerical values and asked them to complete either a free recall task, a willingness to-pay (WTP) task, or both. Using Bayesian mixed effects modeling, we predicted WTP (for the values in the remembered sequence) from both recalled items and presented items, and examined how the position of a value in a sequence determined its contribution to overall evaluation. Uniquely, we compared models of overall performance in memory across trials (e.g., primacy and recency) with those that predict evaluation from specific items recalled on individual trials. Previous studies are likely to have underestimated the strength of the relationship by focusing only on general trends in memory performances. Our findings indicated that memory predicts evaluation when examined at the level of individual people and individual sequences. We replicate the findings that when participants anticipate a judgment task, they show less accurate recall from memory and show a preference for online updating. We demonstrated that even under these conditions, relying on memory is an effective strategy for performing evaluation, and discuss implications for understanding how people resolve the trade-off between online and retrospective evaluation.

Biased confabulation in risky choice

“…There is a large literature examining memory for numbers including number representation (Thevenot & Barrouillet, 2006), recognition memory for decision outcomes (Sobkow, Olszewska, & Traczyk, 2020), and memory for probability distributions (Goldstein & Rothschild, 2014). There is, however, scant evidence on free recall of numeric stimuli (but see Dale and Baddeley (1966) and Mason, Brown, Ward, and Farrell (2019)). In similar word-learning tasks, people are more likely to recall semantically related items from lists other than the target list (Miller, Weidemann, & Kahana, 2012).…”

Valuation and estimation from experience

“…Hence, memory effects can mediate behavioral characteristics as, for example, the overweighting of extreme events (Kahneman et al 1993;Madan et al 2014). The process of sequential number integration can also be explicitly modeled through online-updating or memory-based individual number recall (Erev et al 2008;Mason et al 2019;Gonzalez et al 2003). Future research is needed to link these memory models to the compressed mental number line and to examine whether compression differs depending on the number of memory processes that are necessary to integrate numeric information.…”