Tuesday, October 07, 2014

Brain as a computer

The big question is – is the brain a computer? (not only like a computer)

Turing et al set up the computational theory of computers. Using those ideas, it seems highly unlikely that the brain too is one computer – however complicated. But then there are others like Dennett that argue that the word computer has been used in a narrow context of a top down machine made of cooperative algorithms. In the wider context that allows computational capability embedded in competing modules, the brain, according to this line of thinking, can be safely thought of as a computer.

The other attack on the idea of brain as a computer comes from the ‘thinking’ side. We as human beings think. Computers don’t think, they ‘mechanically’ carry out instructions. Deep down in the software and hardware of the computer, there are only electrons moving about – as per a pre-determined circuitry and well defined rules of logic. This set up, however complex it gets, remains at heart a deterministic machine that at best simulates the idea that it carries out complicated procedures to solve problems – and these problems are actually solved by the ingenious programming designed by humans from outside.

This is partly correct. The computer is ultimately a collection of electrons moving about and the overall impression it gives of immense computing power is simply the effect of the miniaturization of its circuitry and the consequent space efficiency in managing all that in a small box sized CPU of it.
Where this analogy loses track is in forgetting that the brain being ‘more’ than the sum total of its computational/cognitive circuitry is simply a claim, a matter of faith, an unverified hypothesis if that. So while it’s entirely correct that computer is sum total of its billions of logic gates, the brain can also be thought of – in absence of any compelling evidence to the contrary – as sum total of its billions of synapses. There is of course the matter of being ‘inside’ the brain and being able to ‘experience’ this purposeful behavior of human thinking as against the mere dance of electrons through the logic gates. But this thinking commits the usual twin-sins of anthropocentrism and lack of imagination. 

Anthropocentrism in the sense that when seen from inside the brain, of course it is going to seem magical, purposive and more sensible than the computer. This sense, this feeling of purpose and so on is part of its programming. There is nothing magical about it. But then computers don’t think do they? This is where the lack of imagination comes into picture. We are unable to imagine that the much revered thinking of human beings can ultimately be broken down – with a lot of work sure – into smaller computations that individually are simply signals. Being ‘inside’ this, our imagination does not generally extend enough to allow us to see the trees in the woods of our thoughts.
I imagine thoughts to be computations being carried out inside the brain using the signal processing mechanisms built on the infrastructure of neurons and synapses. Clearly the modern computers differ significantly in their architecture and their very organization from the human brain. However, the fact that individual signals are processed in a huge amount to carry out an overall computational or cognitive task is the fundamental common thread between the two.

Another attack is generally from the qualia camp. We ‘perceive’ things – some are red, some are pungent and some are symmetrical. We do not merely compute and measure these things, we actually ‘experience’ or ‘sense’ or ‘perceive’ them. The image of a desk with a phone, remote and small box is ‘real’ in my mind. It might have been arrived using computation by my brain. But the final product is this distinct image that cannot be explained using computational terms. This is the summary of the qualia argument. I do realize there is some unexplained phenomenological account that is needed of this experience. However to me, when seen from an alien’s point of view and from outside, this qualia problem is more curiosity than a fundamental premise of human mind being non-computational. Clearly human mind’s computational architecture is not fully understood by – well human minds! There could be several things that we do not know yet about the details of the perception process that can explain the presence of qualia. It is a sub-problem in my view.

Stating succinctly, my current thoughts on the computational view – the brain is vast collection of neurons and synapses – which act as the logic gates for computations. There are several systems or modules in the brain. Some are nearly autonomous systems (breathing, digestion etc) while others are learned but semi-autonomous – walking, cycling, language etc. Lastly, there are systems that are equipped to handle highly unspecific situations – which are located mostly in the neo-cortex and are most well developed in humans. These systems are an evolution driven feature to survive in the world using one’s wits – i.e. ability to think on the fly using the inputs from the surroundings and computations about a suitable course of action highly customized to that specific instance. This ability to deal with each situation as it turns out differently requires different mode of computation than say the one that deals with digestion or even locomotion.

Connecting my other thoughts about the self written elsewhere, this system also harbours the socially constructed self. The reflector module is inside this system. It is required by this system only – you don’t need reflection to walk or to digest food. The special place of the reflector module inside this system makes human beings believe that they are different from the ‘dumb’ systems of computation carried out in the silicon based computers. When told more about the modules of digestion and locomotion, most humans would grant that these modules are indeed like the silicon based computers. They will most likely still exclude higher thinking (the ‘self’) from this lowly description. Ask an alien though, and it would simply believe that the ‘higher’ thinking is different only in its details from ‘lower’ thinking and the ‘self’ created by the higher thinking is another module inside the brain of the being.

The way mind is computational is very different from the way the brain is computational. For the mind, the logical reasoning and cognitive processing comes at a very later stage of development. The brain has multiple modules – some mechanical (respiration), some purposive (problem solving). The mechanical modules might resemble the silicon based computer in the processing of signals and information. The higher (or those dealing with less deterministic tasks) modules are unlikely to be computational in this way.

For example, if I am solving 2+3 in a silicon based computer, I would use logic gates that help me solve this. That might have a few of those gates at best which will effectively throw the output. If I am solving it in my brain though, the problem is set up in the world of very high level concepts of 2,3 and +. This makes the neuronal support required for it several orders of magnitude larger than that needed for the silicon based computer. This all fine from evolutional point of view because the need to solve 2+3 came up much later for organisms (if at all it can be said to be a need.) The ability to deal with a complex and ever changing environment is their first priority. For that they need the complex modules dealing with ideas. That same module if pressed into the service of solving 2+3 will continue to use its established methods – which from the computing efficiency point of view are highly inefficient.

No comments: