Testing error-management predictions in forgiveness decisions with cognitive modeling and process-tracing tools.

Abstract

We investigated the forgiveness decision as an error-management task and demonstrated how tools from decision science can facilitate testing precise predictions about bias and its cognitive implementation. We combined decision modeling (using a weighting-and-adding model and a lexicographic heuristic) with process-tracing tools that track response times as well as the pattern of information acquisition. Our modeling results indicate that individuals adopted a decision bias commensurate with the relative cost of errors and that they also adjusted their bias after the perceived costs of errors were experimentally manipulated. Even though the 2 decision models were accurate in fitting the decisions (accuracies of around 85%), they were less successful in fitting the process measures. Our process-tracing results do not support either model-response times were in favor of the heuristic, whereas information-acquisition patterns favored the linear model, albeit slightly. Nevertheless, our methodology used to investigate the forgiveness decision can be a seen as a "blueprint" of how the cognitive processes of other error-management tasks can be investigated and how a more detailed mapping of the adapted mind can be achieved.

A framework for building cognitive process models

“…The response-time prediction in the example has been tested in studies on information search, failing to support the process predictions despite support for the output predictions (Glöckner & Betsch, 2008;Johnson et al, 2008). Another example comes from the domain of forgiveness, where both Franklin's rule and fast-and-frugal trees predicted the output (choices) well, but the (nested) information acquisition process poorly (Tan, Luan, Gonzalez, & Jablonskis, 2018). As these examples show, careful experimental design that aims to discriminate models based on their output-level predictions optimally (e.g., Myung & Pitt, 2009;Westfall, Kenny, & Judd, 2014) may not suffice to discriminate models.…”

Responding to risk in evolutionarily important domains.

“…The three papers by Tan, Luan, Gonzalez, and Jablonskis (2018), Frankenhuis, Roelofs, and de Vries (2018), and Jarecki and Wilke (2018) have a central underlying theme: optimization under constraints, a principle that is important in a number of fields, including my own, behavioral ecology. When studying other species, we usually have little or no knowledge of underlying biases or possible thought processes.…”

“…All three papers also accomplish something I wish were more feasible in behavioral studies of nonhumans: They go beyond seeking to measure the fitness of responses, to considering how differing processes contribute, or do not, to the fitness of a response. Tan et al (2018) examine the decision to forgive as a tool, asking which is more costly: to forgive too much or too little? I am reminded of the tit-for-two-tats strategy (forgive one potentially mistaken hostile move by your opponent but not two), which is sometimes successful in Prisoner's Dilemma (see Axelrod, 1984).…”

“…Both Frankenhuis et al (2018) and Tan et al (2018) examine aspects of error avoidance and management. In both papers I appreciated their careful reporting of complicated tradeoffs.…”

“…Tan et al (2018) examine the decision to forgive as a tool, asking which is more costly: to forgive too much or too little? I am reminded of the tit-for-two-tats strategy (forgive one potentially mistaken hostile move by your opponent but not two), which is sometimes successful in Prisoner’s Dilemma (see Axelrod, 1984).…”