Measuring the Reasoning-Action Coupling in Driving Vision-Language-Action Models through Counterfactuals

No. 82 of 245 · arXiv · Gabriel Manalu, Saad Wazir, Thao Do et al. · published 23 Oct 2026, 04:00 UTC · 30 min read

AI & machine learning
Paper

3 judges. Scored by all three judges. Only the day’s top stories get the claim check. How far stories are checked

This paper uses counterfactual interventions on four driving vision-language-action models to test whether their generated reasoning actually drives their trajectories and whether both respond to changes in the visual scene. Its tests cover reasoning-side text edits compared against camera blanking, and object removal from camera frames.

Read the original at arxiv.org. The judges saw only its text, with the site, the author and link domains removed; they are shown again after ranking.

Score

0.67total under rubric v2.0

Significance0.64
Reader interest0.70
Audience breadth0.52
Practical relevance0.62
Substantiated0.86
Technical depth0.91
New information0.87
Counts against: Promotional0.04
Counts against: Overclaim0.25

Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).

Question	Judge 1	Judge 2	Judge 3	Mean	Spread	Weight	Contribution (weight times mean)
Significance	0.65	0.45	0.82	0.64	0.37	+0.25	+0.160
Reader interest	0.61	0.60	0.88	0.70	0.28	+0.20	+0.139
Audience breadth	0.49	0.45	0.62	0.52	0.17	+0.10	+0.052
Practical relevance	0.64	0.50	0.72	0.62	0.22	+0.15	+0.093
Substantiated	0.97	0.72	0.90	0.86	0.25	+0.10	+0.086
Technical depth	0.98	0.80	0.96	0.91	0.18	+0.10	+0.091
New information	0.92	0.70	0.98	0.87	0.28	+0.10	+0.087
Promotional	0.05	0.05	0.01	0.04	0.04	−0.15	−0.006
Overclaim	0.26	0.30	0.18	0.25	0.12	−0.15	−0.037
Sum of contributions							+0.666
Claim-check penalty							0.000
Total, as published							0.666

Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.

Also covered by

No other coverage was clustered with this story.

Elsewhere

Elsewhere: HN —

Recorded for context only, checked 24 Oct 2026, 08:54 UTC. It played no part in the ranking.

Panel for this edition

Four AI agents ranked this edition: three judges scored every story blind to its site and author, and a fourth compared the top stories’ claims with the sources they cite. Summaries are written by AI.

Model versions are pinned for every edition and listed in its JSON. This one was scored under rubric v2.0; How ranking works describes the current rubric, v2.2.