Right for the Right Reasons: Avoiding Reasoning Shortcuts via Prototypical Neurosymbolic AI

2025-10-29

AI Summary

Neural-symbolic AI often suffers from shortcut reasoning due to spurious correlations—learning erroneous neural predicates that satisfy symbolic constraints under sparse supervision, rather than modeling true underlying concepts. To address this, we propose the Prototype Neural-Symbolic Architecture (P-NSA), which tightly integrates prototype learning with symbolic knowledge: neural modules are explicitly guided to learn interpretable, verifiable primitive concepts via semantic similarity between inputs and a small set of labeled examples, while strictly adhering to background knowledge during inference. This architectural design inherently suppresses shortcut pathways, enabling reliable concept-level grounding. Evaluated on the RSBench benchmark—including high-risk, sparse-supervision tasks such as MNIST-EvenOdd, Kand-Logic, and BDD-OIA—P-NSA achieves substantial gains in concept accuracy and out-of-distribution generalization robustness. Our approach establishes a new paradigm for trustworthy AI under low-resource settings.

Technology Category

• Machine Learning: Neuro-Symbolic Learning
• Cognitive Modeling & Cognitive Systems: Symbolic Representations
• Computer Vision: Visual Reasoning & Symbolic Representations

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Neurosymbolic AI is growing in popularity thanks to its ability to combine neural perception and symbolic reasoning in end-to-end trainable models. However, recent findings reveal these are prone to shortcut reasoning, i.e., to learning unindented concepts--or neural predicates--which exploit spurious correlations to satisfy the symbolic constraints. In this paper, we address reasoning shortcuts at their root cause and we introduce prototypical neurosymbolic architectures. These models are able to satisfy the symbolic constraints (be right) because they have learnt the correct basic concepts (for the right reasons) and not because of spurious correlations, even in extremely low data regimes. Leveraging the theory of prototypical learning, we demonstrate that we can effectively avoid reasoning shortcuts by training the models to satisfy the background knowledge while taking into account the similarity of the input with respect to the handful of labelled datapoints. We extensively validate our approach on the recently proposed rsbench benchmark suite in a variety of settings and tasks with very scarce supervision: we show significant improvements in learning the right concepts both in synthetic tasks (MNIST-EvenOdd and Kand-Logic) and real-world, high-stake ones (BDD-OIA). Our findings pave the way to prototype grounding as an effective, annotation-efficient strategy for safe and reliable neurosymbolic learning.

Problem

• Preventing neurosymbolic AI from learning spurious correlations
• Addressing reasoning shortcuts in low-data training regimes
• Ensuring models learn correct concepts for reliable decisions

Innovation

• Prototypical neurosymbolic architectures prevent reasoning shortcuts
• Training uses similarity to labeled prototypes and background knowledge
• Effective in low-data regimes for reliable concept learning