# Grounding vs. Compositionality: On the Non-Complementarity of Reasoning in Neuro-Symbolic Systems

arXiv: https://arxiv.org/abs/2604.26521 · HTML: https://mlsanity.com/paper/2604.26521/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 23 of its 39 references that are in the map.
- First posted on arXiv 29 Apr 2026.

## Builds on (most cited first, 23 of 23)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [VQA: Visual Question Answering](https://mlsanity.com/paper/1505.00468/) (2015)
- [CLEVR: A Diagnostic Dataset for Compositional Language and Elementary Visual Reasoning](https://mlsanity.com/paper/1612.06890/) (2016)
- [Neural Module Networks](https://mlsanity.com/paper/1511.02799/) (2015)
- [The Neuro-Symbolic Concept Learner: Interpreting Scenes, Words, and Sentences From Natural Supervision](https://mlsanity.com/paper/1904.12584/) (2019)
- [Neural-Symbolic VQA: Disentangling Reasoning from Vision and Language Understanding](https://mlsanity.com/paper/1810.02338/) (2018)
- [COGS: A Compositional Generalization Challenge Based on Semantic Interpretation](https://mlsanity.com/paper/2010.05465/) (2020)
- [Compositional Attention Networks for Machine Reasoning](https://mlsanity.com/paper/1803.03067/) (2018)
- [A Benchmark for Systematic Generalization in Grounded Language Understanding](https://mlsanity.com/paper/2003.05161/) (2020)
- [Learning to Infer Graphics Programs from Hand-Drawn Images](https://mlsanity.com/paper/1707.09627/) (2017)
- [Meta-Learning to Compositionally Generalize](https://mlsanity.com/paper/2106.04252/) (2021)
- [NS3D: Neuro-Symbolic Grounding of 3D Objects and Relations](https://mlsanity.com/paper/2303.13483/) (2023)
- [Learning Differentiable Programs with Admissible Neural Heuristics](https://mlsanity.com/paper/2007.12101/) (2020)
- [Techniques for Symbol Grounding with SATNet](https://mlsanity.com/paper/2106.11072/) (2021)
- [Softened Symbol Grounding for Neuro-symbolic Systems](https://mlsanity.com/paper/2403.00323/) (2024)
- [Neurosymbolic Grounding for Compositional World Models](https://mlsanity.com/paper/2310.12690/) (2023)
- [From Perception to Programs: Regularize, Overparameterize, and Amortize](https://mlsanity.com/paper/2206.05922/) (2022)
- [Mapping the Neuro-Symbolic AI Landscape by Architectures: A Handbook on Augmenting Deep Learning Through Symbolic Reasoning](https://mlsanity.com/paper/2410.22077/) (2024)
- [Program-Guided Image Manipulators](https://mlsanity.com/paper/1909.02116/) (2019)
- [A Survey on Compositional Generalization in Applications](https://mlsanity.com/paper/2302.01067/) (2023)
- [Unsupervised Learning of Neurosymbolic Encoders](https://mlsanity.com/paper/2107.13132/) (2021)
- [Fodor and Pylyshyn's Legacy: Still No Human-like Systematic Compositionality in Neural Networks](https://mlsanity.com/paper/2506.01820/) (2025)

## Abstract (arXiv)

Compositional generalization remains a foundational weakness of modern neural networks, limiting their robustness and applicability in domains requiring out-of-distribution reasoning. A central, yet unverified, assumption in neuro-symbolic AI is that compositional reasoning will emerge as a byproduct of successful symbol grounding. This work presents the first systematic empirical analysis to challenge this assumption by disentangling the contributions of grounding and reasoning. To operationalize this investigation, we introduce the Iterative Logic Tensor Network (iLTN), a fully differentiable architecture designed for multi-step deduction. Using a formal taxonomy of generalization -- probing for novel entities, unseen relations, and complex rule compositions -- we demonstrate that a model trained solely on a grounding objective fails to generalize. In contrast, our full iLTN, trained jointly on perceptual grounding and multi-step reasoning, achieves high zero-shot accuracy across all tasks. Our findings provide conclusive evidence that symbol grounding, while necessary, is insufficient for generalization, establishing that reasoning is not an emergent property but a distinct capability that requires an explicit learning objective.