aiqbench analysis
This study challenges the assumption that increased inference-time computation in reasoning models automatically mitigates cognitive biases, finding instead that bias magnitudes do not reliably decrease with higher token budgets. The research indicates that for several decision-making errors, additional deliberation can even shift responses further from human baselines, suggesting that test-time reasoning is not a guaranteed fix for rationality. Consequently, developers should prioritize specific bias auditing over relying on longer chain-of-thought as a universal solution for improving model alignment.
The Long Road to the Same Answer: Cognitive Bias Under Escalating Reasoning Budgets in Large Language Models
arXiv:2610.10049v1 Announce Type: new Abstract: Reasoning models allocate extra computation at inference time and present their answers as the product of deliberate thought. If this deliberation works the way dual-process accounts of human cognition suggest, longer thinking should weaken the classic decision biases that fast, intuitive judgment produces. Using 30 vignettes covering six biases (anchoring, framing, loss aversion, escalation of commitment, availability, confirmation) from an established benchmark, we run a dose-response study across four model families, pairing each reasoning model with a matched non-reasoning sibling and requesting thinking ceilings of 0, 1,024, 4,096, and 8,192 tokens, for 12,350 API calls. Because a requested ceiling is not the same as realized deliberation, we use the reasoning tokens each call consumed as the dose. First, reasoning models are not less biased than their siblings; the point estimate leans the other way in every family, but the item-level pooled contrast is not reliable (Delta = +0.031, t(29) = 1.45, p = .157). Second, bias magnitude does not reliably fall as realized deliberation grows: no slope is significantly negative, and where anything moves it is the signed score drifting further from the human direction. Third, anchoring is the only bias in the human direction (d = 1.89). Four of the other five lean the opposite way in all seven models; with five items per bias, that reversal is reliable for framing and directional for escalation of commitment, confirmation, and loss aversion, while availability is absent. A one-line instruction to restate the anchor before answering lowered anchoring on all five anchoring items, which no amount of additional thinking did, although the effect does not reach significance (p = .057). The results argue against treating test-time reasoning as a rationality guarantee and for auditing deployed models bias by bias.
Abstract:Reasoning models allocate extra computation at inference time and present their answers as the product of deliberate thought. If this deliberation works the way dual-process accounts of human cognition suggest, longer thinking should weaken the classic decision biases that fast, intuitive judgment produces. Using 30 vignettes covering six biases (anchoring, framing, loss aversion, escalation of commitment, availability, confirmation) from an established benchmark, we run a dose-response study across four model families, pairing each reasoning model with a matched non-reasoning sibling and requesting thinking ceilings of 0, 1,024, 4,096, and 8,192 tokens, for 12,350 API calls. Because a requested ceiling is not the same as realized deliberation, we use the reasoning tokens each call consumed as the dose. First, reasoning models are not less biased than their siblings; the point estimate leans the other way in every family, but the item-level pooled contrast is not reliable (Delta = +0.031, t(29) = 1.45, p = .157). Second, bias magnitude does not reliably fall as realized deliberation grows: no slope is significantly negative, and where anything moves it is the signed score drifting further from the human direction. Third, anchoring is the only bias in the human direction (d = 1.89). Four of the other five lean the opposite way in all seven models; with five items per bias, that reversal is reliable for framing and directional for escalation of commitment, confirmation, and loss aversion, while availability is absent. A one-line instruction to restate the anchor before answering lowered anchoring on all five anchoring items, which no amount of additional thinking did, although the effect does not reach significance (p = .057). The results argue against treating test-time reasoning as a rationality guarantee and for auditing deployed models bias by bias.
Accepted at the 2026 IEEE 8th International Conference on Cognitive Machine Intelligence (IEEE CogMI 2026). 8 pages, 3 figures, 5 tables. Code and data: this https URL
Computation and Language (cs.CL)
arXiv:2610.10049 [cs.CL]
(or arXiv:2610.10049v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2610.10049 arXiv-issued DOI via DataCite (pending registration)
Full text under the publisher's open licence · 334 words
Related coverage
Game-Theoretic Disempowerment: Loss of Control to Agencyless AI
LessWrong · Sep 27, 2026
The Continual Constitutional Convention: How language models revise the documents that govern them
AI Alignment Forum · Oct 13, 2026
Jev: System One models for Prod, not God — with Diogo Almeida, CEO, TypeSafe AI
Latent Space · Sep 21, 2026
Six Principles of Voice Design: Designing for the Ear and Encoding It into Agent Script
Salesforce Blog · Oct 8, 2026