Fascinating writing … but I do not think the current neurological research can support the author’s claims. For instance, it can be easily shown there is no one to one correspondence between perceived images and places in the brain, whether one is using a machine or not to decode a perception. But leaving this argument to the side, let us suppose this one to one brain-image correspondence is possible, and not just a folkloric supposition of psychology, then the problem of how a machine could read a mind/brain is still speculation.

At the most fundamental level, the author’s argument supposes not that machines can read/decode what people think, dream, etc., but more importantly that people must think like machines.

This is the hidden premiss: people dream, perceive, think, etc. like machines. Once this hidden premiss is accepted, then what is actually being proposed is that one can use another machine to decode the first machine (the model of the human). There are two machines involved in a parallel program, where one is supposed to read/decode what the other writes/codes. A completely normal problem of AI.

Second, since the author is really trying to put forward the argument of how a person would think, dream, or perceive like a machine if they were a machine, then the model of a human as a parallel program is possible. But note once one tries to describe how only one machine thinks and perceives, he will be forced at some point to admit that no machine can determine the semantics of its own proof procedures (aka Godel’s incompleteness theorem and Church’s Thesis), much less where images are in its own system. It is only by putting two machines together and running them in a parallel program can one machine decode what the other is doing, thinking, etc. as a coding of the semantics of the first system. It still leaves open the question of where the act of reading the images/perceptions occurs in the second machine is, since this is where the real dream interpretation is occurring. The interpreted dream, i.e., the actual read dream that is relayed between the two systems can never itself be given a place in the parallel program, unless one calls in a third machine or a human to read the second. Thus, one falls into infinite regress.

Needless to say, it is important to know how machines think, dream, perceive, etc.as digital coding/decoding. Admittedly, some people not only think like machines, but can become animals. But it is an altogether a different question to know how and why the human subject does not think, dream, or perceive like a machine or an animal.

Cheers,

$