4 ms·
> a way to exchange information across millions of people without using keyboards or voice recognition devices Brain-to-brain communication will be convenient,
by 6ren 14y ago
> a way to exchange information across millions of people without using keyboards or voice recognition devices
Brain-to-brain communication will be convenient, because we won't need physical I/O devices. However, I wonder if there will be any advantage to interfacing any deeper into the brain than those corresponding nerves... that is, perhaps direct-brain interfaces will be pretty much like conventional I/O, just without the physical I/O device (like the cochlear implants/bionic eyes/bionic arms of today). So we can't improve fundamentally over things like google glass/MYO.
The reason I say this is because we already have brain-to-brain communication in the form of language. This involves reducing our internal representation (whatever it is) into a serial format that follows grammatical rules - not necessarily "proper" grammar, but sufficient to enable the recipient, using shared knowledge, to decode the meaning.
This has striking parallels to network communication: we have the internal representation of data structures (e.g. objects), which are serialized to a low-level protocol (e.g. xml, json, yaml, protobuf, asn.1) which follows a higher-level grammar (explicit like xsd or implicit as in json). Well, not entirely surprising, since Chomsky's grammar hierarchy was based on human language.
The interesting question is whether brain-to-brain communication can be any more "direct" than the serial representation (language) we already have...
That is, are our internal representations similar enough that they can be "directly" communicated; or are they so different between individuals that converting between them will amount to going through a common grammar in order to interface between them?
Favouring direct interface is the argument that serial representation was only needed to communicate over the distance to another human - i.e. because we didn't have direct brain-to-brain contact - and it is not needed apart from that.
Also in favour of direct communication is that if individuals are reasoning about the same things in the same ways, our internal representations must be equivalent in some sense. Other hidden aspects of ourselves also turn out to have strong similarities (such as internal organs); and where they vary, they often vary in strong groups (such as blood types), so perhaps our hidden mental features will also be similar.
Against this is the observation that everyone's perspective is unique - literally, where their eyes are - but also of course their figurative perspective, resulting from cumulative experiences, genetic background, internal thoughts and randomly different wiring/interpretations. Therefore, we don't have the same internal models. We don't even have different models of the same reality, because we perceive reality differently - our models are of a different reality. And this diversity is one of the great creative strengths of our collective intelligence, that different people have different insights into the same problem.
Another argument against this is that grammar is not only used in communicating to others, but is also used to communicate to ourselves - when we talk to ourselves, debate a course of action internally, "marshal our thoughts" into words. There's even an argument that the full power of human thought requires language - it's not just a serialization format, but is thought itself. By this view, even if we can communicate directly, it would only be associations, senses, emotions, recollections etc. We couldn't communicate an argument, for example.
SUMMARY our internal representations will differ at least to the extent that we interpret things differently (i.e. that we live in a different reality). We will probably fall into groups, like blood types; also by profession/training/personality/social group. i.e. between persons "on the same wavelength". This is a limit on how "directly" communicate, i.e. without requiring a shared intermediate general-purpose language.
tl;dr qualia