Networks of the Mind

Today's topic is something I thought of as I was just leaving work today. Recently I've been listening to two audiobooks as I bike to and from work. The first is "Linked: How Everything is Connected to Everything Else", by Albert-Laszlo Barabasi. It's about recent (1998-2001) discoveries in network topology, but written for the layperson. He talks about the difference between random networks and so-called "scale-free" structured networks centering around interconnected hubs, and he describes examples of these networks that can be seen in such wide-ranging systems as social and professional networks, protein interactions within a biological cell, food webs in an ecosystem, national power grids, business affiliations, marketing strategies, terrorist cells, and of course the internet and world wide web.

One key concept he presents involves the way networks form: while nodes in a randomly-linked system have an average of <1 connection with other nodes, the network is broken-up and disorganized. Once the critical threshold of 1 connection per node is exceeded, the system suddenly "gels" into a coherent network. (This is one connection on average, of course - many nodes will have no connections and a few will have many connections.) This behavior can be seen in ice crystals forming, for example. While the system is a liquid, the molecules are not coherently connected, but once a critical threshold is exceeded, the water crystallizes into ice.

The other book is "Memory Optimizer", by Vera Birkenbihl and Paul Scheele. It describes some techniques for improving memory... I haven't gotten too far into it, but they seem to be stressing the idea that memory is an active process, and that the formation of new memories is a process of constructing them based on existing connections. They talk about creating a "memory web", and continually expanding your potential connections by spending a little bit of time each week learning new things and reviewing connections you've made in the past.

Although I haven't read a formal study on this, it seems likely to me that creating a highly effective memory web involves building up a network of memories, concepts, and images. Once the interconnectivity in that network exceeds some sort of critical threshold, the connections your mind makes will suddenly bring very distantly-related and hard-to-reach ideas and memories close to each other, at which point your memory and creative facilities should dramatically increase in efficiency. That seems to be the gist of what the Memory Optimizer book is driving at with the memory web technique, and it seems to be consistent with the self-organization of real networks described in the other book.

The first thought I had today regarding networks was actually in regard to my performance at work. I've been thinking a lot about this lately - I work in sort of an in-between capacity at my company. On the one hand I'm in a support position, developing software to help the researchers with their experiments. On the other hand, I'm a little bit above the interns, and I think I'm expected also to do my own, original research. So far I haven't published anything, although I'll hear back about my first submitted paper at the end of next month. I feel like I have the potential to do lots of interesting research and write many papers, but something is holding me back.

Reflecting on what that might be, I've concluded that I feel like my background in the academic knowledge pertaining to the field of robotics isn't quite deep enough to make bold statements with confidence, and at the same time I feel like my knowledge of the workings and procedures within my company is slightly below the level where I'd feel confident rushing out and coordinating a major experiment. Finally, my experience doing pure research is somewhat limited, and my intuitions about what is necessary and useful to do to achieve publishable results have not been fully developed.

All this will come in time, of course. It's a natural part of the learning curve of any job. But what occurred to me is that this is a network as well. All of these bits of knowledge and knowhow are related, and when enough nodes and connections have been pulled together, the result is the intuitive ability to do something well. I think this is probably the case with any learned skill, but with complex job-related skills reaching into the social and scientific spheres, a particularly large number of connections need to be pulled together. Once this internal mental network is established, ideas can become reality just as smoothly as an email can travel around the world to its destination.

My other thought was that the complexity of networks is obvious in software design as well. If one considers function calls and dependencies on variables as links in a network, even a small piece of software can quickly exceed a critical threshold of intraconnectivity. At this point, the complexity of the system exceeds our mental ability to understand and control it, and the system has topologically transformed into a highly-interconnected configuration known as "spaghetti code". Managing this complexity is a primary function of the disciplines of abstraction and modular design. I had never thought of it as a network before, though.

So I found it interesting to consider that these seemingly unrelated concepts are all connected, not by actual networks, but by the concept of networks. And that's my thought for the day. :)