I’ve just finished reading Consciousness and the Brain by Stanislas Dehaene, neuroscientist, experimental psychologist. This work is illuminating: experimental results are detailed and explained, with critics reviewed and countered by more experiments. The main hypothesis, the global neural workspace, is convincing by its abilities to explain experimental data as well as some clinical behavior.
But maybe the main interest of this book is the questions it raises.
We learn for example that language is processed at a high level unconsciously, but not math, even not basic arithmetic. Could human evolve an arithmetic / mathematics module in the brain, the way specialized language areas have evolved? Do humans who are incredible calculators have managed to dedicate specific brain areas for this purpose, which would be different in people who don’t have these abilities? Could bio-technology / bio-hacking achieve this one day? (considering that biological evolution is slow, and environmental, cultural, social pressures might not be aligned to select such capacities)
It doesn’t seem impossible, when you learn that unconscious processing of sensory data is performing advanced statistical calculations to produce distribution of probabilities (Bayesian brain). Unfortunately, these remain inaccessible to consciousness. These unconscious systems perform much better in these tasks than machine learning algorithms, which needs thousands or millions of trials to recognize a cat from a dog.
Why then try to interface our brain with machine and computers, if the brain circuitry is so much more efficient in processing data?
First, obvious advantage would be calculations that even a dumb computer can do effortlessly, and a human brain can’t.
But I don’t see how a direct interface with AI would really improve our capacities. Stanislas Dehaene mention an example in his book of trained satellite image analysts, who are shown images far below the level of consciousness, while their brain is monitored to see if they unconsciously detect anomalies. This allow them to review more images in less time. When you think about this example, it is a bit dystopian. The brain of these people is used more or less like a machine, inputs / outputs, no matter what these persons are conscious of. However, it’s no just any brain that can perform this task unconsciously. They needed hours of conscious training to tune their brain first, so dedicated neurons and pathways could code the normal characteristics of satellite imagery, so that any deviation would automatically send a signal.
The main problem of consciousness, as is apparent in this example, is its limited bandwidth, and slowness. Our conscience is capable, by nature and structure, of only holding one thought at a time. There are already multiple preconscious ideas floating around in our brains, which all compete for attention. The process of sorting and selecting through these ideas is another area where I would like to learn a lot more, if any research has been done. Adding an additional source of ideas to the global neural network means that this source will be in competition with all others. How will we able to regulate our conscious access to this external AI, if we don’t know exactly how some things find their way into consciousness, and not others?
Worse: if we understand eventually how this process works, how will we able to prevent it from being hijacked? This is a very high possibility, considering the existing attempts to capture people’s attention through marketing and media strategies, and increase engagement at all costs for businesses. This is literally at all costs, as Facebook demonstrated, with democracy being a potential casualty of their addiction to clicks.
I’m also wondering how this theory of consciousness resonates or not with zen meditation practice. Or, how to understand / explain meditation in terms of the global neural workspace.
One experience typically relates how deep meditative state looks like: when you cease to see a tree for a tree, but see things (matter / colors) for themselves, without the labelling and distanciation. In this case, meditation goal would be to allow sensory signal of lower order to make it into consciousness, and bypass the higher level order processing areas.
We also learn that fixed stimuli tend to disappear from consciousness, as they are discarded as non-relevant signals. Visual processing areas are trained to discard all fixed signals, which in nature are more likely to be related to imperfections in our eyes, rather the world itself (blood vessels, fovea).
So a perfectly still meditating person (immobile body and eyes) would likely see the world vanish in front of their eyes. Whether such a perfect absence of motion is achievable is difficult to know, but we shouldn’t dismiss what training can do. If this is the case, one can imagine how powerful an experience it would be, and how it relates with the teachings of void and emptiness.
Finally, this book made me realized how finely tuned our brains are, since our birth, and possibly before, to the world we live in, our mother tongue(s), human society, etc. I was aware of this regarding our consciousness, but, by demonstrating how much of our thoughts are really born in unconscious processes, which are built early on and throughout our life, this book shows how much more immersed and imprinted by our environment we are. Even though there are still a number of processes, that arose early in evolution, which we share with other animal species (spatial awareness, numeration).