The term cognitive refers to mental processes such as judgement, reason, memory and perception. These are processes that alter the brain cognitive structures, which in effect are neural patterns representing information retained in the long term memory. The study of these activities is very complex and it is being shared by multiple disciplines that adopt a variety of approaches in describing what is going on in there: physiology, psychology, neuroscience, etc. However, we could say in a nutshell that cognitive processes are calculations, crunching information bits that flow into the brain system through various sensorial channels.
We mostly associate cognitive processes with thinking in a conscious state, but they can occur in other states. The keyword here is “thinking”. In computer speak, that is the running of an algorithm, the execution of program routines that break down the input data into components to assemble a response.
Computer systems do exactly the same thing in their silicon world, with one little important difference: all their programming routines are created by humans. Their thinking is a result of our thinking, which is an entirely rational cognitive process. Sure, we can get emotional in our programming, but in the end the lines of code must obey very dry syntax rules. No matter how hard we try, we can only code software that fakes emotion very well at best, but never software that is actually emotional.
In my mind, this is why artificial intelligence will never match our intelligence: it lacks emotional capability. Emotion is what drives us. Emotion is our most precious attribute that allows us to create something from nothing (almost). We don’t know how emotion really works, but you have a sense that if cognition is a calculation on which reason is based, emotion must be a super-calculation that takes place at deeper and inaccessible levels from which inspiration and unexpected creativity is drawn from.
But, this is a big BUT: is it possible that when you have billions of computers linked through a network, when their state change at huge scale as result of their rapid interaction and massive parallel sensorial input, to have waves of computation patterns forming unexpectedly, creating a significant response that never existed before and that was never programmed to occur, which will change the global distributed long-term memory in ways that will change their future behaviour? That is learning and creating emotionally, a capability specific to living systems.
I suspect that is already occurring, but we just don’t see it clearly yet. If we would take a closer look at the financial trading systems as they are networked around the world, maybe we see a glimpse of that. The quant trading systems that have been programs to crunch huge chunks of market information and detect human psychological patterns could display such “emotional” behaviour. I am not referring to them being programmed to behave emotionally, because they don’t for the reason I mentioned above, but because as they are networked at a massive scale receiving large amounts of input data, they could display “emotion”. I suspect the “glitch” that caused almost 1000 points drop in Dow Jones Industrial Average in May 2010 was in fact an “emotional” behaviour of these networked systems (it was not a bug).
That computational ability is a type of artificial intelligence indistinguishable from the natural. This would pass an imaginary Turing test with flying colours..
PS: Financial markets are one of the most networked systems at a global scale. There is more to come, see here (NYT, 2011)and here (CBS report). Also, social networks such as Facebook will likely display, if not already do, independent intelligent behaviour.