# Neuro-Symbolic Reasoning Shortcuts: Mitigation Strategies and their Limitations

arXiv: https://arxiv.org/abs/2303.12578 · HTML: https://mlsanity.com/paper/2303.12578/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning.
- Builds on 7 of its 10 references that are in the map.
- First posted on arXiv 22 Mar 2023.

## Placed in

- LLM reasoning / [Reasoning benchmarks and evaluation](https://mlsanity.com/lr/reasoning-benchmarks-and-evaluation/)

## Builds on (most cited first, 7 of 7)

- [Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations](https://mlsanity.com/paper/1811.12359/) (2018)
- [Variational Autoencoders and Nonlinear ICA: A Unifying Framework](https://mlsanity.com/paper/1907.04809/) (2019)
- [Neural Probabilistic Logic Programming in DeepProbLog](https://mlsanity.com/paper/1907.08194/) (2019)
- [Self-Supervised Learning with Data Augmentations Provably Isolates Content from Style](https://mlsanity.com/paper/2106.04619/) (2021)
- [Robustly Disentangled Causal Mechanisms: Validating Deep Representations for Interventional Robustness](https://mlsanity.com/paper/1811.00007/) (2018)
- [Neuro-Symbolic Continual Learning: Knowledge, Reasoning Shortcuts and Concept Rehearsal](https://mlsanity.com/paper/2302.01242/) (2023)
- [Towards efficient representation identification in supervised learning](https://mlsanity.com/paper/2204.04606/) (2022)

## Built on by (most cited first, 7 of 7)

- [Learning to Intervene on Concept Bottlenecks](https://mlsanity.com/paper/2308.13453/) (2023)
- [Interpretable Concept Bottlenecks to Align Reinforcement Learning Agents](https://mlsanity.com/paper/2401.05821/) (2024)
- [On the Independence Assumption in Neurosymbolic Learning](https://mlsanity.com/paper/2404.08458/) (2024)
- [Counterfactual Concept Bottleneck Models](https://mlsanity.com/paper/2402.01408/) (2024)
- [Unifying Self-Supervised Clustering and Energy-Based Models](https://mlsanity.com/paper/2401.00873/) (2023)
- [Simple and Effective Transfer Learning for Neuro-Symbolic Integration](https://mlsanity.com/paper/2402.14047/) (2024)
- [Neurosymbolic Deep Learning Semantics](https://mlsanity.com/paper/2511.02825/) (2025)

## Abstract (arXiv)

Neuro-symbolic predictors learn a mapping from sub-symbolic inputs to higher-level concepts and then carry out (probabilistic) logical inference on this intermediate representation. This setup offers clear advantages in terms of consistency to symbolic prior knowledge, and is often believed to provide interpretability benefits in that - by virtue of complying with the knowledge - the learned concepts can be better understood by human stakeholders. However, it was recently shown that this setup is affected by reasoning shortcuts whereby predictions attain high accuracy by leveraging concepts with unintended semantics, yielding poor out-of-distribution performance and compromising interpretability. In this short paper, we establish a formal link between reasoning shortcuts and the optima of the loss function, and identify situations in which reasoning shortcuts can arise. Based on this, we discuss limitations of natural mitigation strategies such as reconstruction and concept supervision.