# Reasoning in machine vision by learning fast and slow thinking

arXiv: https://arxiv.org/abs/2506.22075 · HTML: https://mlsanity.com/paper/2506.22075/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 25 of its 114 references that are in the map.
- First posted on arXiv 27 Jun 2025.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)

## Builds on (most cited first, 25 of 25)

- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [mixup: Beyond Empirical Risk Minimization](https://mlsanity.com/paper/1710.09412/) (2017)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://mlsanity.com/paper/2408.03314/) (2024)
- [On First-Order Meta-Learning Algorithms](https://mlsanity.com/paper/1803.02999/) (2018)
- [Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations](https://mlsanity.com/paper/2312.08935/) (2023)
- [Solving math word problems with process- and outcome-based feedback](https://mlsanity.com/paper/2211.14275/) (2022)
- [Training Language Models to Self-Correct via Reinforcement Learning](https://mlsanity.com/paper/2409.12917/) (2024)
- [Alphazero-like Tree-Search can Guide Large Language Model Decoding and Training](https://mlsanity.com/paper/2309.17179/) (2023)
- [V-STaR: Training Verifiers for Self-Taught Reasoners](https://mlsanity.com/paper/2402.06457/) (2024)
- [DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search](https://mlsanity.com/paper/2408.08152/) (2024)
- [Hierarchical Reasoning Model](https://mlsanity.com/paper/2506.21734/) (2025)
- [The KiTS19 Challenge Data: 300 Kidney Tumor Cases with Clinical Context, CT Semantic Segmentations, and Surgical Outcomes](https://mlsanity.com/paper/1904.00445/) (2019)
- [Can 1B LLM Surpass 405B LLM? Rethinking Compute-Optimal Test-Time Scaling](https://mlsanity.com/paper/2502.06703/) (2025)
- [System 2 Attention (is something you might need too)](https://mlsanity.com/paper/2311.11829/) (2023)
- [Massively Parallel Methods for Deep Reinforcement Learning](https://mlsanity.com/paper/1507.04296/) (2015)
- [Dualformer: Controllable Fast and Slow Thinking by Learning with Randomized Reasoning Traces](https://mlsanity.com/paper/2410.09918/) (2024)
- [MM-IQ: Benchmarking Human-Like Abstraction and Reasoning in Multimodal Models](https://mlsanity.com/paper/2502.00698/) (2025)
- [Image quality assessment by overlapping task-specific and task-agnostic measures: application to prostate multiparametric MR images for cancer segmentation](https://mlsanity.com/paper/2202.09798/) (2022)
- [Towards Sustainable Self-supervised Learning](https://mlsanity.com/paper/2210.11016/) (2022)
- [Learning to ignore: rethinking attention in CNNs](https://mlsanity.com/paper/2111.05684/) (2021)
- [Weakly supervised localisation of prostate cancer using reinforcement learning for bi-parametric MR images](https://mlsanity.com/paper/2402.13778/) (2024)
- [Fast and Slow Planning](https://mlsanity.com/paper/2303.04283/) (2023)
- [Biochemical Prostate Cancer Recurrence Prediction: Thinking Fast & Slow](https://mlsanity.com/paper/2409.02284/) (2024)

## Built on by (most cited first, 1 of 1)

- [Active few-shot segmentation by reinforcing data selection](https://mlsanity.com/paper/2607.22371/) (2026)

## Abstract (arXiv)

Reasoning is a hallmark of human intelligence, enabling adaptive decision-making in complex unfamiliar scenarios. In contrast, machine intelligence remains bound to training data, unable to dynamically refine solutions at inference. While recent advances have explored machine reasoning - trading inference-time compute for improved performance - they focus on verbal domains such as mathematical problem-solving where explicit rules govern step-by-step solution generation. Many tasks lack sufficient labelled data and require alternative performance improvement mechanisms, such as inference-time compute. Here we present a paradigm for machine reasoning in vision, enabling performance improvements with increasing thinking time (inference-time compute), even with limited labelled data. Our approach is inspired by dual-process theories of human cognition, integrating a fast-thinking System I module for generating and verifying solutions in familiar tasks, with a slow-thinking System II module that iteratively refines predictions using self-play reinforcement learning, even when task-specific data is limited. This paradigm involves proposing, competing over, and refining solutions until convergence. We demonstrate that extended inference-time compute yields superior performance compared to large-scale supervised learning, foundation models, and human experts in vision tasks. These include computer-vision benchmarks and cancer localisation across five organs, highlighting the potential of inference-time compute for data-scarce problems.