Left, Right, and Gender: Exploring Interaction Traces to Mitigate Human Biases.

Area of Science:

• Human-computer interaction
• Cognitive science
• Data visualization

Background:

• Human decision-making is susceptible to cognitive biases.
• Data visualization can support cognitive processes and mitigate these biases.
• Understanding user interaction with data is key to improving decision support.

Purpose of the Study:

• To explore how visualizing user interaction history can mitigate cognitive biases.
• To investigate real-time and summative visualizations of interaction history.
• To assess the impact of interaction history on bias in decision-making tasks.

Main Methods:

• Developed an interactive scatterplot tool displaying user interaction history.
• Conducted in-lab and crowd-sourced experiments with political and movie selection scenarios.
• Evaluated bias mitigation through real-time and summative visualization formats.

Main Results:

• Inconclusive evidence for interaction history directly mitigating biased decision-making.
• Mixed support found for interaction traces increasing awareness of unconscious biases.
• Summative visualization format showed a potential trend towards increased bias awareness.

Conclusions:

• Visualizing interaction history shows potential for enhancing decision-making awareness.
• Further research is needed to confirm the effectiveness of interaction traces in bias mitigation.
• Interaction history visualization may serve as a tool for metacognitive reflection during data analysis.