intro’d to Jack via these tweets from wef16
There is activity in many parts of our brain that cannot be expressed in #language: @gallantlab https://t.co/2KM3nEpZKb #wef #whatif
Original Tweet: https://twitter.com/Davos/status/690545273030971392
taking him in first here:
Mapping language-related information across the human cerebral cortex
on problem being not having enough data
document everything ness
(youtube down.. i guess.. can’t play any videos.. have to come back to later)
i can read your mind
Even though he’s not working on one, Gallant knows what kind of brain decoder he might build, should he chose to. “My personal opinion is that if you wanted to build the best one, you would decode covert internal speech. If you could build something that takes internal speech and translates into external speech,” he says, “then you could use it to control a *car. It could be a universal translator.”
*car – or.. perhaps.. help humans grok what matters .. ie: augmenting (listening deeper to) self-talk as data
Some groups are edging closer to this goal; a team in the Netherlands, for instance, scanned the brains of bilingual speakers to detect the concepts each participant were forming – such as the idea of a horse or cow, correctly identifying the meaning whether the subjects were thinking in English or Dutch. Like the dream decoder, however, the system needed to be trained on each individual, so it is a far cry from a universal translator.
idiosyncratic jargon ness
If nothing else, the brain reader has sparked more widespread interest in Gallant’s work. “If I go up to someone on the street and tell them how their brains work their eyes glaze over,” he says. When he shows them a video of their brains actually at work, they start to pay attention.
his page at uc berkeley
Computational encoding models that accurately predict brain activity have many practical uses. First, they provide a critical foundation for other work aimed at rehabilitation of visual function; after all, one needs to understand how a system functions before one can hope to repair it. Second, these models provide a new tool for neurological evaluation and diagnosis. Third, the models can be inverted in order to decode brain activity, providing a direct and principled way to do brain reading and to build brain-machine interfaces (BMI) and neural prosthetic