Paper Type

Complete

Paper Number

PACIS2026-1459

Description

Explainable AI (XAI) can enhance human-AI collaboration, yet how explanation strategies align with different task contexts remains unclear. Grounded in task-technology fit theory, we conducted two behavioral experiments (N₁ = 293; N₂ = 323) and two EEG experiments (N₃ = 51; N₄ = 47) examining how explanation strategy (example-based vs. feature-based) affects collaboration performance across decision-making and creative tasks. Behaviorally, example-based explanations improve performance in decision-making tasks, while feature-based explanations benefit creative tasks. EEG findings replicate these results and reveal underlying mechanisms: in decision-making tasks, example-based explanations support systematic processing by reducing cognitive burden (theta-band ERS), while in creative tasks, feature-based explanations promote heuristic processing by stimulating divergent thinking (alpha-band ERD). These findings suggest that AI explanations enhance collaboration when achieving cognitive fit between explanation-induced and task-demanded processing. This work advances XAI and human-AI collaboration research, extends task-technology fit theory, highlights NeuroIS methods, and recommends task-contingent XAI deployment.

Comments

13-Design

Share

COinS
 
Jul 5th, 12:00 AM

Task-Driven Explanation Strategies: Enhance Human-AI Collaboration Performance by Achieving Cognitive Fit

Explainable AI (XAI) can enhance human-AI collaboration, yet how explanation strategies align with different task contexts remains unclear. Grounded in task-technology fit theory, we conducted two behavioral experiments (N₁ = 293; N₂ = 323) and two EEG experiments (N₃ = 51; N₄ = 47) examining how explanation strategy (example-based vs. feature-based) affects collaboration performance across decision-making and creative tasks. Behaviorally, example-based explanations improve performance in decision-making tasks, while feature-based explanations benefit creative tasks. EEG findings replicate these results and reveal underlying mechanisms: in decision-making tasks, example-based explanations support systematic processing by reducing cognitive burden (theta-band ERS), while in creative tasks, feature-based explanations promote heuristic processing by stimulating divergent thinking (alpha-band ERD). These findings suggest that AI explanations enhance collaboration when achieving cognitive fit between explanation-induced and task-demanded processing. This work advances XAI and human-AI collaboration research, extends task-technology fit theory, highlights NeuroIS methods, and recommends task-contingent XAI deployment.