Research map / Large language models / AI and society
Human Cognitive Biases in Explanation-based Interaction: The Case of Within and Between Session Order Effect
Dario Pesenti, Alessandro Bogani, Katya Tentori, Stefano Teso
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Early model errors (increasing-accuracy order vs constant or decreasing order) (ablation) | Within-session debugging of a fictitious face-box model · Exp1 agreement on incorrect boxes, difficult images | agreement with model: Inc 0.19 vs Const 0.24 and Dec 0.24 | Hurt |
| Increasing vs decreasing vs constant model accuracy order within a session (ablation) | Within-session debugging · Exp1 overall feedback accuracy | feedback accuracy: Inc 0.76, Const 0.75, Dec 0.76 | No clear effect |
| Decreasing accuracy order (errors late) vs constant order (ablation) | Within-session debugging · Exp1 accuracy on incorrect vs correct images | feedback accuracy: Dec gap 0.79 vs 0.70; Const gap 0.80 vs 0.67 | Mixed |
| Order condition (inc/const/dec) within a session (ablation) | Within-session debugging · Exp1 questionnaire | perceived accuracy and trust: 3.06 vs 3.11 vs 3.13 (p = .909) | No clear effect |
| Increasing model accuracy between two sessions (ablation) | Between-session debugging · Exp2 session 2 agreement | agreement with model: Inc 0.63, Const 0.62, Dec 0.61 | No clear effect |
| Increasing model accuracy order across sessions (ablation) | Between-session debugging · Exp2 session 2 accuracy on incorrect images | feedback accuracy: Inc 0.68 vs Const 0.69 and Dec 0.70 | Mixed |
| Order condition across sessions (ablation) | Between-session debugging · Exp2 questionnaire | perceived accuracy and trust: 3.38 vs 3.43 vs 3.34 (p = .821) | No clear effect |
| Vary model accuracy in session 1 (40%, 60%, 80%) (ablation) | Between-session manipulation check · Exp2 session 1 agreement | agreement with model: 0.42 -> 0.57 -> 0.70 | Improved |
Closest papers by meaning
- iLLuMinaTE: An LLM-XAI Framework Leveraging Social Science Explanation Theories Towards Actionable Student Performance Feedback (AAAI 2025 · similarity 0.86)
- Comparing Explanations is not Enough,Explain the Change: New Standards are Needed to Explain Behavioral Shifts in Large Language Models (NeurIPS 2026 · similarity 0.86)
- Failing to Falsify: Evaluating and Mitigating Confirmation Bias in Language Models (NeurIPS 2026 · similarity 0.85)
- LLM Agents Can Be Choice-Supportive Biased Evaluators: An Empirical Study (AAAI 2025 · similarity 0.85)
- InterMT: Multi-Turn Interleaved Preference Alignment with Human Feedback (NeurIPS 2025 · similarity 0.85)
- Beyond correlation: The impact of human uncertainty in measuring the effectiveness of automatic evaluation and LLM-as-a-judge (ICLR 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
