Credit: Joe Sutliff

Science 314: 1093

Computers and Brains

A comment from a naïve neuroscientist
(OeGAI Journal 19/2, 2000)
A warning to begin with
This issue of the ÖGAI journal is dedicated to the neurosciences. Nevertheless, the AI in ÖGAI stands for artificial intelligence, so most readers will be familiar with the specific formalism of information processing by machines. Not so the author. If you expect that the following essay will try to explain the functions of the brain in terms of mechanistic data processing on a high expert level, you will be disappointed. Don't waste your time. Without any doubt, you are the expert and I'm not. I'm just a neuroscientist, and I use my computer to write down a text like this, for not much more.
The brain is a machine?
Without any doubt, the brain is a machine consisting of biological building blocks, and producing outputs in response to inputs. At some (modest) level of complexity, the output can be predicted from the input with certainty. We are now in the position to construct machines with properties reminiscent of simple animal behaviors. We do this to make our industrial life easier, or simply for distraction and amusement. And we are fascinated by a little man-made machine moving across the floor and avoiding obstacles, exactly like a living being would do it. We have equipped this little artefact with an electronic device we rightfully might call a little brain, if we call a brain also the handful of neurons governing the behavior of a worm.
The free will of a moving vacuum cleaner
It makes a big difference, whether we deal with the firing of a single neuron in a tissue culture dish, with the behavior of an Aplysia snail, or with the feelings of a human being. Although the basic mechanisms remain (almost) the same, the complexity of the systems varies dramatically. We may watch for hours with childish delight our new vacuum cleaner as it (he?) finds its way across the dusty floor of our apartment without our guiding hands. But after some time of accommodation, the first impression of magic will fade away. The first impression provoked the illusion of a little creature exhibiting goal directed behavior, as if an independent autonomic authority was to decide where to go and where not to go. However, we rapidly realize, that the movements of the machine are 100% predictable, and our fascination cools down.
What we are really interested in
The fascination of a man-made device exhibiting apparently goal directed behavior resides in the fact, that it is man-made. The fruitless attempts of a fly trying to get on the other side of an invisible window-pane are a much more spectacular view, but usually we are only disturbed by the annoying sound, not really interested in the event as such. But we immediately focus our attention to a little mouse, trying desperately to escape from a dustbin. There is something special in the behavior of the mouse, something reminiscent of our own behavioral drives. And finally, no normal human being can ignore the heart-breaking sobbing of a child, not in reality, and even not if he knows that it is just a recording. The closer it gets to ourselves, the more interested we are. The same is true for those information processing devices we call brains: the more similar they get to our own brains, the more we wonder, on what mechanisms they are based on.
A computer acting like a human being?
Modern days computers are calculating faster than Carl Friedrich Gauss, they offer automatic correction of orthography, and they memorize with 100% reliability more data than we can ever read (not to mention retain) in a whole lifetime. However, these remarkable qualities are not the typical attributes of an average human being. Why is it, then, that we dare to speculate that the functioning of our human brain could ever be explained in terms of the complex wiring of electronic circuits? Our admirable attempts to simulate brain function have not even reached the level of complexity of an insect's brain. Maybe sooner or later we will be in the position to understand the essence of what it means to be a fly. But will it ever be possible to simulate a human being, i.e. the information processing background responsible for our motivations and decisions?
What being human really means
Human beings arise from (nearly) nothing. In the course of their development, they are shaped and permanently acted upon by environmental influences. Once they enter as discernable actors into this world, they are engaged in continuous processes of interaction with entities similar to themselves, over narrow and over long distances, over short and over long times. They have to defend their integrity, always at the risk of loosing it, at the risk of being eliminated from the world. And most of them like their living and try to continue it as long as possible. In all these respects, a human being is not much different from a mouse. What makes us special as a human being is the quality and the quantity of inter-personal communication. We establish societies (like rats and honey bees too), but we also maintain communication far beyond a few generations, over thousands of years.
First steps to create a human-like machine
First of all: a human-like machine has not to look like a human. It's the brain that counts, so a computer would do the job, whatsoever it looks like. It's not even necessary that it stays associated with always the same material object (as we are associated with the object called our body). But it must adopt several essential features. (1) It must be in danger to disappear. (2) It must be able to influence the conditions it depends on. (3) Its decisions must never lead to its own disappearance. (4) It must be able to create other similar machines. (5) It must be able to recognize similar machines. (6) Its decisions must not lead to the disappearance of similar machines. (7) If rules 3 and 6 get in conflict, the solution with the greatest number of surviving similar machines (including, but not favoring the original machine) is to be preferred. And that would be just the most rudimentary beginning of a concept.
Cats and dogs?
Even if such an attempt would arrive at some information processing construct responsible for motivations and decisions leading on a long term to its own propagation, it is very unlikely that anything else than another devastating computer virus would emerge, i.e. a cybernetic creature at a very low organizational level. But why should we be interested in such a thing? As long as the organizational level does not reach ours, we better communicate with cats and dogs, that's by far more rewarding...
6/00 <         MB (7/00)         > 7/00