The Language Neuroscience Podcast Encoding and decoding semantic representations with Alexander Huth
May 4, 2023
Alexander Huth, Assistant Professor at the University of Texas, Austin, specializes in fMRI and computational models studying language and meaning. He discusses using natural stimuli for semantic research and how he transitioned from AI to neuroscience. Huth shares insights on creating rich semantic models, mapping semantic responses in the brain, and decoding stories from brain activity. He highlights advancements in neural language models and ethical considerations in brain decoding, revealing how our thoughts can be reconstructed from brain data.
AI Snips
Chapters
Transcript
Episode notes
Visualize Heavily And Share Interactive Maps
- Invest in high-quality visualization and interactive tools to convey complex, multidimensional brain maps.
- Build or use shareable web-based viewers (like PyCortex) to make results explorable and transparent.
Model Context, Not Just Single Words
- Move beyond single-word embeddings to contextual language models to capture real language processing.
- Use neural language models (RNNs/transformers) that integrate multi-word context to improve encoding performance.
Next-Word Prediction Yields Rich Linguistic Features
- Predicting the next word trains models to learn syntax, semantics, and long-range dependencies implicitly.
- Word embeddings and hidden states are learned jointly through prediction objectives.