IEEE VIS 2024 Content: Unmasking Dunning-Kruger Effect in Visual Reasoning and Visual Data Analysis

Unmasking Dunning-Kruger Effect in Visual Reasoning and Visual Data Analysis



Mengyu Chen - Emory University, Atlanta, United States

 Yijun Liu - Emory University, Atlanta, United States

 Emily Wall - Emory University, Atlanta, United States

 Download Supplemental Material

 Room: Bayshore II

2024-10-16T14:27:00ZGMT-0600Change your timezone on the schedule page
2024-10-16T14:27:00Z

Exemplar figure, described by caption below — We replicated the Dunning-Kruger Effect (DKE) across tasks involving visual reasoning and judgment. We observed a typical DKE pattern, where highly skilled people tend to underestimate their performance, while those with lower skills often overestimate it. Additionally, we explored potential indicators of DKE, including participants’ interactions, personality traits, and domain familiarity, and identified several factors related to DKE.

Fast forward

Full Video

Keywords

Cognitive Bias, Dunning Kruger Effect, Metacognition, Personality Traits, Interactions, Visual Reasoning

Abstract

The Dunning-Kruger Effect (DKE) is a metacognitive phenomenon where low-skilled individuals tend to overestimate their competence while high-skilled individuals tend to underestimate their competence. This effect has been observed in a number of domains including humor, grammar, and logic. In this paper, we explore if and how DKE manifests in visual reasoning and judgment tasks. Across two online user studies involving (1) a sliding puzzle game and (2) a scatterplot-based categorization task, we demonstrate that individuals are susceptible to DKE in visual reasoning and judgment tasks: those who performed best underestimated their performance, while bottom performers overestimated their performance. In addition, we contribute novel analyses that correlate susceptibility of DKE with personality traits and user interactions. Our findings pave the way for novel modes of bias detection via interaction patterns and establish promising directions towards interventions tailored to an individual’s personality traits. All materials and analyses are in supplemental materials: https://github.com/CAV-Lab/DKE_supplemental.git.