The role of decision confidence in advice-taking and trust formation

TL;DR

This study investigates how internal confidence judgments influence advice perception and trust formation, especially when objective feedback is unavailable, revealing that metacognitive evaluations help discriminate advice quality.

Contribution

It introduces a computational model linking confidence to trustworthiness estimation and demonstrates its relevance in social decision-making contexts.

Findings

People can discriminate advice confidence calibration without feedback.

Correlated judgments lead to distortions in advice perception.

A simple model explains advice reliability estimation based on confidence.

Abstract

In a world where ideas flow freely between people across multiple platforms, we often find ourselves relying on others' information without an objective standard to judge whether those opinions are accurate. The present study tests an agreement-in-confidence hypothesis of advice perception, which holds that internal metacognitive evaluations of decision confidence play an important functional role in the perception and use of social information, such as peers' advice. We propose that confidence can be used, computationally, to estimate advisors' trustworthiness and advice reliability. Specifically, these processes are hypothesized to be particularly important in situations where objective feedback is absent or difficult to acquire. Here, we use a judge-advisor system paradigm to precisely manipulate the profiles of virtual advisors whose opinions are provided to participants performing a perceptual decision making task. We find that when advisors' and participants' judgments are independent, people are able to discriminate subtle advice features, like confidence calibration, whether or not objective feedback is available. However, when observers' judgments (and judgment errors) are correlated - as is the case in many social contexts - predictable distortions can be observed between feedback and feedback-free scenarios. A simple model of advice reliability estimation, endowed with metacognitive insight, is able to explain key patterns of results observed in the human data. We use agent-based modeling to explore implications of these individual-level decision strategies for network-level patterns of trust and belief formation.