# Think before you act: A simple baseline for compositional generalization

arXiv: https://arxiv.org/abs/2009.13962 · HTML: https://mlsanity.com/paper/2009.13962/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 29 Sep 2020.

## Built on by (most cited first, 11 of 11)

- [Improving Coherence and Consistency in Neural Sequence Models with Dual-System, Neuro-Symbolic Reasoning](https://mlsanity.com/paper/2107.02794/) (2021)
- [ReaSCAN: Compositional Reasoning in Language Grounding](https://mlsanity.com/paper/2109.08994/) (2021)
- [Systematic Generalization on gSCAN: What is Nearly Solved and What is Next?](https://mlsanity.com/paper/2109.12243/) (2021)
- [Improving Systematic Generalization Through Modularity and Augmentation](https://mlsanity.com/paper/2202.10745/) (2022)
- [When Can Transformers Ground and Compose: Insights from Compositional Generalization Benchmarks](https://mlsanity.com/paper/2210.12786/) (2022)
- [Compositional Generalization in Grounded Language Learning via Induced Model Sparsity](https://mlsanity.com/paper/2207.02518/) (2022)
- [Recursive Decoding: A Situated Cognition Approach to Compositional Generation in Grounded Language Understanding](https://mlsanity.com/paper/2201.11766/) (2022)
- [Is Feedback All You Need? Leveraging Natural Language Feedback in Goal-Conditioned Reinforcement Learning](https://mlsanity.com/paper/2312.04736/) (2023)
- [Compositional Generalization in Spoken Language Understanding](https://mlsanity.com/paper/2312.15815/) (2023)
- [A Short Survey of Systematic Generalization](https://mlsanity.com/paper/2211.11956/) (2022)
- [Data Distributional Properties As Inductive Bias for Systematic Generalization](https://mlsanity.com/paper/2502.20499/) (2025)

## Abstract (arXiv)

Contrarily to humans who have the ability to recombine familiar expressions to create novel ones, modern neural networks struggle to do so. This has been emphasized recently with the introduction of the benchmark dataset "gSCAN" (Ruis et al. 2020), aiming to evaluate models' performance at compositional generalization in grounded language understanding. In this work, we challenge the gSCAN benchmark by proposing a simple model that achieves surprisingly good performance on two of the gSCAN test splits. Our model is based on the observation that, to succeed on gSCAN tasks, the agent must (i) identify the target object (think) before (ii) navigating to it successfully (act). Concretely, we propose an attention-inspired modification of the baseline model from (Ruis et al. 2020), together with an auxiliary loss, that takes into account the sequential nature of steps (i) and (ii). While two compositional tasks are trivially solved with our approach, we also find that the other tasks remain unsolved, validating the relevance of gSCAN as a benchmark for evaluating models' compositional abilities.