What are the characteristics of networks within the autistic brain?
Ways of looking at networks: What is a small world network?
The world that we live in is overflowing with networks. Close to home, relatives are one network; if you go out and ride the train, for example, the trains are a giant network comprising large and small stations with Tokyo Station as its starting point; and the SNS that you connect to with your smartphone also has a network structure that spans the globe.
Thinking about SNSs, the transfer of information across networks is extremely fast, but what characteristics of networks could this be due to?
Let’s use an air network as an example to think about this. Airplanes are convenient, but they do not connect every airport directly. There are innumerable domestic air networks centered on large airports like Haneda, Narita or Kansai International, and international air networks linking these domestic air networks criss-cross the globe.
Or we could think about workplaces. Workplaces are made up of various organizations, which are small networks, and the bosses of the organizations each collaborate to create one large network. Thanks to this, even if we do not know them directly, we pick up things said by a range of staff members and information relating to them.
To summarize, an effective network has a structure:
- that is formed from a collection of smaller networks,
- which are connected through the central point of the smaller networks (the international airport or the boss).
Ways of looking at network characteristics
As shown above, networks consist of points and lines and small blobs where they are connected, but how can we look at these points and lines?
One way is to look at how many lines come from each point. In a network where all points are joined to each other, the number of lines coming from each point will be greater, whereas in a network like a small world network, many points are connected indirectly through hubs, so the number of lines coming from each point will necessary be fewer. (Think of how regional airports usually only have routes linking them to Tokyo, Kansai, Sapporo or Fukuoka.) The number of lines is called the path length, and the shorter the average path length of the network, the more like a small world network it is.
Another way is an index that shows how well each point is connected to nearby points. In a village that has a long history, people will associate closely with their neighbors, but in an apartment building in a city, residents may rarely communicate with the people living next door to them. How closely points are linked to nearby points is shown by an index of the cohesiveness of the blobs called the clustering coefficient. In a small world network, it is important for each blob to have strong links, so the higher its clustering coefficient, the more like a small world network it is.
A further way of looking is an index that shows how much the blobs are broken up. In many organizations, smaller groups divide functions between themselves and they each have roles to play, which means that a certain level of work can be completed within each group. The nursing department and the rehabilitation department may do something together at times, but they each show their true skills when they are separated. In other words, in small world networks, the blobs must be broken down as shown in the diagram above, and the degree to which blobs are broken up is described by the term “modularity”.
Cerebral networks in autistic spectrum disorder
The paper I discuss today investigates cerebral networks in autistic spectrum disorder. In the experiment, ten participants with autistic spectrum disorder and ten participants in a control group had their resting brain activity measured, and the relationship between electrodes was subjected to network analysis focusing mainly on delta waves.
The results show that those with autistic spectrum disorder have a less small-world network than the control group, as anticipated, that specifically, they had many short connections, but few long ones—a tendency that grew stronger as the autistic spectrum disorder was more severe—and that those with autistic spectrum disorder had a small clustering coefficient (an index that shows how much the small networks form blobs) and a large path length, making a so-called “big world network” in contrast with the small world network of the control group. (Top is from the control group, and bottom is from the autistic spectrum disorder group)
Reference URL for the above diagram
At times, the slowness of information processing in autistic spectrum disorder creates difficulties in life in society, but I thought that this could be because of the unique structure of the networks in the brain.
