Brains working towards machine consciousness

Over at her blog, Frances Bell has a useful critique of Connectivism underway (I believe this is also being combined into a wiki somewhere...)

I posted a comment over there:
I would agree about the network of 'theories' - you can envisage the theories themselves as a Connectivist network, with adherents and practitioners being the connections between the Nodes, and learning being the shifts in position of the theories themselves. The more links there are between theories (in terms of people experienced with several of them) the faster the progress as there is a better chance of critiquing.
In this view, the 'reality' is just another node which interacts with the theories via the network. Because the social reality comprises the theory-network Actors (sorry, that isn't really an ANT reference), and is thus changed through the interactions, the reality can change just as much as the theories which attempt to describe it.
This is where Siemen's point about currency is important - we deal with a realm where the theory can act almost as an observer of reality and (quantum theory style) thus modify the reality. The links must then be able to re-adjust to take account of the changes, and in doing so change the underlying basis of the nature of the theories (which are, in a sense, an aggregation of the beliefs of the links - in this case, people)

And decided I needed to post more about the concepts here.

In mathematical terms, networks are 'graphs'. Which doesn't help people much, because in the UK 'graphs' are things like Stock market performance, and daily rainfall (the Americans tend to call these 'charts', something we use more often for maps of the sea, and the weather). Essentially, networks (and graphs) consist of nodes and links. In the case of Connectivism and neural networks, the links have attributes, such as 'strength'. It is quite reasonable to view them as little Input/Output mechanisms which take in data, and transform it in some way, and then output it again. This can be amplifying the signal, attenuating it, or filtering it (or, indeed, anything else you can think of doing to a signal). They are also capable of adapting their Transfer Function (TF) which allows the system as a whole to adjust its behaviour (because the changing TFs mean that data gets passed around differently).

Typically, people are used to thinking of the Nodes in a network as being a representation of Agents (people, machines) and the links as just being conduits for the data. In this way, the Nodes have to adjust the parameters of the links, apply filters, and generally be in charge of the learning process. But if we view the Actors as being the links between some form of conceptual Nodes, the learning which takes place in the adjustment of link parameters becomes easily understood.

In the general case, you can take a graph (network) and re-draw it as its complement - representing nodes as links and vice versa.

Here, though, I am suggesting that we view the Theories we develop and use as being the Nodes of a network - which means they are taking the place of what are typically thought of as links. The implication is that the Theories are the connections through which we communicate, which seems reasonable.

Of course, both the Nodes and the links are adaptive, and this suggests a view in which all of the people, machines and concepts are represented in the network as Nodes (or links, take your pick). The 'other class' of network elements is then relegated back to being a simple connection between things. If we take Agents (including theories, memes) as Nodes, then the connections are links; if Agents are links, the connections are nodes or nexi (is that the right plural? nexusses sounds horrid, but I don't know which languages nexus is derived from off hand).

I favour a view in which our concept of Agents (people, machines) is really just a boundary condition - the Network continues throughout, using a number of different linkages (for humans, not only each of the senses, but realistically each nerve which conveys information) - a continuous whole, which is only comprised of discrete 'systems' because we choose to view it that way.