Big-data approaches to increasing amounts of raw brain data will enable researchers to decode cognitive and emotional states and to elucidate fundamental principles of cognition and memory.¹⁰ Machine learning has implications for neural decoding, including determination of the neural correlates of memory encoding, retention and retrieval.¹¹ Speech decoding is now possible at near-normal speeds, with some systems boasting vocabularies of 125,000 words and with extremely low error rates, allowing near-naturalistic communication. ¹²¹³

The rise of predictive large language models means semantic decoding becomes possible even without brain implants; one has decoded continuous language from fMRI.¹⁴ Brain monitoring for interfacing with external technology is seeing rapid progress. Neuralink, for instance, has implanted its Telepathy chip in a human brain, enabling wireless, thought-enabled operation of a computer cursor. Many other companies are operating in this space, hoping to respond to brain signals to restore a person’s movement or speech. Synchron, for example, has implanted its Stentrodes in 10 people’s brains, and is making preparations to conduct a much larger trial of its technology.^{15 16}