Removing Timing Shortcuts Improves Non-Invasive Brain-to-Text
A word-timing shortcut inflated non-invasive brain-to-text results; removing it cuts error to 36.6%.
The authors show that a reported non-invasive brain-to-text improvement largely exploits word timing leaked by overlapping fixed-length windows rather than brain activity. The earlier method reaches 22.0% balanced accuracy on synthetic signals containing no brain information, compared with 22.3% on real recordings. Processing each window independently blocks that shortcut, after which aggregating repeated neural responses and using a pretrained language-model prior become much more effective. Their SimpleB2T recipe achieves a 36.6% word error rate with five observations per word on a perceived-speech benchmark, approaching some invasive results under different conditions.
- Overlapping windows leak word duration independent of brain signals.
- No-brain synthetic signals score 22.0% versus 22.3% on real data.
- Encoding each window independently removes the timing shortcut.
- SimpleB2T reaches 36.6% word error rate with five observations.
Full article276 words · extracted from arxiv.org · click to collapse
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.
Text extracted automatically; images, tables and formatting may be missing. Original: https://arxiv.org/abs/2609.40359