# Removing Timing Shortcuts Improves Non-Invasive Brain-to-Text

arXiv: https://arxiv.org/abs/2609.40359 · HTML: https://mlsanity.com/paper/2609.40359/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 21 of its 32 references that are in the map.
- First posted on arXiv 30 Sep 2026.

## Builds on (most cited first, 21 of 21)

- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [SGDR: Stochastic Gradient Descent with Warm Restarts](https://mlsanity.com/paper/1608.03983/) (2016)
- [Sigmoid Loss for Language Image Pre-Training](https://mlsanity.com/paper/2303.15343/) (2023)
- [Shortcut Learning in Deep Neural Networks](https://mlsanity.com/paper/2004.07780/) (2020)
- [Decoding speech perception from non-invasive brain recordings](https://mlsanity.com/paper/2208.12266/) (2022)
- [The Brain's Bitter Lesson: Scaling Speech Decoding With Self-Supervised Learning](https://mlsanity.com/paper/2406.04328/) (2024)
- [LibriBrain: Over 50 Hours of Within-Subject MEG to Improve Speech Decoding Methods at Scale](https://mlsanity.com/paper/2506.02098/) (2025)
- [The 2025 PNPL Competition: Speech Detection and Phoneme Classification in the LibriBrain Dataset](https://mlsanity.com/paper/2506.10165/) (2025)
- [NeuralBench: A Unifying Framework to Benchmark NeuroAI Models](https://mlsanity.com/paper/2605.08495/) (2026)
- [Unlocking Non-Invasive Brain-to-Text](https://mlsanity.com/paper/2505.13446/) (2025)
- [From Thought to Action: How a Hierarchy of Neural Dynamics Supports Language Production](https://mlsanity.com/paper/2502.07429/) (2025)
- [MEG-XL: Data-Efficient Brain-to-Text via Long-Context Pre-Training](https://mlsanity.com/paper/2602.02494/) (2026)
- [LibriBrain100: One Hundred Hours of Broad and Deep MEG Data for Neural Speech Decoding at Scale](https://mlsanity.com/paper/2608.25204/) (2026)
- [Accurate Decoding of Natural Sentences from Non-Invasive Brain Recordings](https://mlsanity.com/paper/2608.18114/) (2026)
- [A Common Measure of Communication for Speech Brain-Computer Interfaces](https://mlsanity.com/paper/2609.02887/) (2026)
- [The 2026 PNPL Competition: Word Classification and Efficient Cross-Subject Generalisation in LibriBrain100](https://mlsanity.com/paper/2609.03231/) (2026)
- [The Semantic Bottleneck: Leveraging Semantic Representations for Non-Invasive Speech Decoding](https://mlsanity.com/paper/2609.10296/) (2026)
- [Margin-Regularized Structured Semantic Alignment for Brain-Language Correspondence](https://mlsanity.com/paper/2608.16975/) (2026)
- [HDND: Hierarchical Dynamic Neural Decoding for Multilingual Word/Character Retrieval from Non-Invasive Brain Recordings](https://mlsanity.com/paper/2609.24095/) (2026)
- [Brain-to-Language Decoding: Tasks, Signals, Methods, Evaluation, Practical Use and Beyond](https://mlsanity.com/paper/2609.27650/) (2026)

## Abstract (arXiv)

We find that major reported improvements in decoding words from non-invasive brain recordings are largely reproducible without any brain data. In the influential work of d'Ascoli et al. (2025), time series of brain activity from subjects perceiving continuous speech are segmented into fixed-length windows starting at each word. A neural network then generates predictions for all of the words in a sentence together. Neighbouring windows partially overlap, implicitly revealing the interval between words. Since these intervals indicate the duration of the words spoken, and different words tend to have different durations - for example, "the" is much shorter than "supercalifragilisticexpialidocious" - the neural network can improve its predictions of words without relying on the underlying brain activity. Consistent with this, the method reaches 22.0% balanced accuracy on synthetic signals containing no brain information, compared with 22.3% on real brain recordings. To prevent the network from learning this shortcut, we make a single, simple change. Instead of jointly encoding all windows in a sentence, we process each independently. As a result, the neural network achieves better performance by learning underlying word-specific information from brain recordings. This makes two existing strategies become much more effective than before. Both aggregating predictions from distinct neural responses to the same word and using a pretrained LLM as a linguistic prior now substantially improve results. On our perceived speech benchmark, this simple recipe (SimpleB2T) achieves a word error rate of 36.6% with five observations per word, approaching past invasive speech decoding performance, albeit under different conditions. The results in this work expose an important shortcut in brain-to-text decoding and show that removing it leads to a simple and considerably more effective strategy.