Saved in:
Bibliographic Details
Main Authors: Hong, Pingjun, Roth, Benjamin
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2606.01148
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914620829073408
author Hong, Pingjun
Roth, Benjamin
author_facet Hong, Pingjun
Roth, Benjamin
contents Natural-language explanations are often treated as a unified interface for understanding model behavior, but different explanation sources may support simulation in different ways. This paper compares two families of explanations for question answering models: verbalized feature attributions and self-generated rationales. We evaluate them under a shared counterfactual simulation setting, using an LLM judge as predictor and measuring whether it can better predict a model's answers to follow-up questions when given its explanation. Across multiple instruction-tuned models, we analyze how explanation source, verbalization strategy, and feature granularity affect the simulatability of explanations. Our results show that explanation format and granularity affect simulatability: attribution-based explanations and self-generated rationales differ in how much they improve counterfactual prediction, with effects that vary across models and formats.
format Preprint
id arxiv_https___arxiv_org_abs_2606_01148
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Not All Explanations Simulate Equally: Comparing Verbalized Feature Attributions and Self-Generated Rationales
Hong, Pingjun
Roth, Benjamin
Computation and Language
Natural-language explanations are often treated as a unified interface for understanding model behavior, but different explanation sources may support simulation in different ways. This paper compares two families of explanations for question answering models: verbalized feature attributions and self-generated rationales. We evaluate them under a shared counterfactual simulation setting, using an LLM judge as predictor and measuring whether it can better predict a model's answers to follow-up questions when given its explanation. Across multiple instruction-tuned models, we analyze how explanation source, verbalization strategy, and feature granularity affect the simulatability of explanations. Our results show that explanation format and granularity affect simulatability: attribution-based explanations and self-generated rationales differ in how much they improve counterfactual prediction, with effects that vary across models and formats.
title Not All Explanations Simulate Equally: Comparing Verbalized Feature Attributions and Self-Generated Rationales
topic Computation and Language
url https://arxiv.org/abs/2606.01148