Surprisal contributes little beyond contextual embeddings in high-gamma ECoG encoding
Sakuma, T.
Show abstract
Surprisal and contextual embeddings are both derived from large language models and are widely used to predict neural responses during language comprehension, but it is unclear whether surprisal adds information beyond embeddings. We test this directly: does word surprisal improve out-of-sample prediction of high-gamma ECoG responses after GPT-2 XL contextual embeddings are included? Using public ECoG recordings from natural speech, we fit word-aligned ridge encoding models with baseline stimulus features, GPT-2 XL embeddings, and GPT-2 XL surprisal. Adding one surprisal predictor left held-out correlation essentially unchanged at the center lag, and the effect remained within a prespecified equivalence margin across alternative lags and sensitivity analyses. This near-zero increment suggests that surprisal does not act as an independent predictor. It is better understood as a compressed readout of the same broader predictive state that the embeddings already capture.
Matching journals
The top 6 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Direct Speech Reconstruction from Sensorimotor Brain Activity with Optimized Deep Learning Models 95%
- Robust assessment of the cortical encoding of word-level expectations using the temporal response function 95%
- Implications of shared motor and perceptual activations on the sensorimotor cortex for neuroprosthetic decoding 94%
Similar papers in this journal
- Recurrent neural networks as neuro-computational models of human speech recognition 96%
- Convolutional neural networks can identify brain interactions involved in decoding spatial auditory attention 95%
- Time-resolved dynamic computational modeling of human EEG recordings reveals gradients of generative mechanisms for the MMN response 94%
Similar papers in this journal
- Dynamic Network Analysis of Electrophysiological Task Data 94%
- Alignment massive of auditory individual artificial networks with fMRI brain data leads to generalizable improvements in brain encoding and downstream tasks 94%
- Representational similarity learning reveals a graded multi-dimensional semantic space in the human anterior temporal cortex 94%
Similar papers in this journal
Similar papers in this journal
- Online speech synthesis using a chronically implanted brain-computer interface in an individual with ALS 94%
- Brain-Guided Convolutional Neural Networks Reveal Task-Specific Representations in Scene Processing 94%
- Audio-visual combination of syllables involves time-sensitive dynamics following from fusion failure 93%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.