Brains working towards machine consciousness

First, my model of mind and (therefore) of learning - we are pattern matching creatures by dint of the way out brains work, and look for the similarities and differences between things. We also create models in our minds of what we perceive, which are, by definition, abstractions and simplifications of the 'objective reality'. Our perceptions are, I believe, affected by the internal models. In order to deal with noise in the environment and from the stochastic nature of neuronal firing, I think we run perceptions through a 'reality check', comparing what is perceived now to our model of what has gone before, and the output of this process is used by other processes in the brain. This has the great advantage of meaning that we readily compare, contrast and classify stimuli with what has gone before, but the potential disadvantage that we can see patterns in systems because they were there in ones we have seen before, even though they may not be fully present in the system under scrutiny. However, given language, we can also build the internal models through communication which means that we can share a model with others (to a greater or lesser degree) - a handy trait for a social animal.
Why is this important to note? I think it is the underlying mechanism behind pre- and post-event suggestions - we see what we expect to see, or believe we saw what the subsequent context suggests we saw, rather than what our perceptions at the time actually registered. This means that if we are looking for networks everywhere, we will, indeed, tend to find them everywhere.
Having said that, I think networks are everywhere - after all, a network is just an interconnected bunch of things with relationships to other things, and almost any world view is going to see that entities have relationships to others.
*phew* got all that out of the way.
Mind and community.
Communities are made up of a number of similar-but-different entities able to communicate with one another, with 'something' which ties them together, such as an ideology, local proximity or need. In general each can take on different roles within the community (unless specialisation forces them to be in a fixed role, such as in many work places, and the social insects). The norms of the community are negotiated (informally or formally), even if they are just that set of behaviours and attitudes which are common to the majority of the members. They have a degree of cognitive centrality and tend to reach cognitive consensus (i.e. there are 'core beliefs' common to the majority, and the tendency is to agree rather than disagree). They also have mechanisms (oral, written) for retaining information, and either formal or informal ways of producing rules to delineate both acceptable behaviours and means of optimising activity (to get the best return for effort, the highest safety or greatest group cohesion, for example).
An individual's brain can be seen as being made up of a number of modules (functional or physical, though I believe functionalism is the better of the two) which have different roles. The modules can communicate with one another, through the massively interconnected architecture of the brain (and, probably, because each is distributed across various physical locations anyway, making inter-connectivity easier). They have mechanisms for retaining information and for producing rules (learning). In most cases, the brain has a degree of cognitive centrality and cognitive consensus - but sometimes it just cannot make up its mind! The modules are 'tied' together by their physical location, but also through their high level of connectivity and mutual reliance.
In many cases damaged areas of the brain cause significant impairment, of course, but thanks to its plasticity and underlying robustness of design (distributed) it can often adapt over time to take on the role of the damaged area - re-assigning other parts of the brain to do their job. This capacity is obviously important in the case of injury or disease, but also
highlights a particular issue - 'forcing' new behaviours on modules (or areas) of the brain takes time and energy.
Learning is the day to day equivalent of the massive re-training necessary to deal with problems in the brain. Although the effort involved is less, it takes energy to build and strengthen the relevant pathways in the brain to support knew knowledge and skills. It is reasonable to think that new 'patterns' which need to be acquired will be learned more quickly and with less energy expenditure if they are small modifications to existing structures. Deeper understanding will be formed if the new patterns form associative links with existing ones, and the internal modelling system of the brain can readily abstract rules from both the old and the new to cover both sets. Indeed, deep understanding is the production of these abstractions in a way that they can be applied to new situations based on prior experience.

Connectivism
So, what does all that mean in terms of connectivism? The structure and qualities of the brain/mind and community are remarkably similar. This should not be a surprise, after all, the way we form communities is governed by our brains, and in the interests of minimising energy costs we will form them in a way that naturally fits with our brains architecture to minimise the cost of translating from one structure to another. Similarly, in order to minimise the overheads invovled in learning, it makes sense to utilise the similarity of nature between community and mind.
According to Dennet's Multiple Drafts theory (which seems at least as good as any other model of the mind, in my view), and others, the mind does not have any single controlling 'thread' of consciousness. Different modules can take on the role of 'conscious thought' depending on which one is producing the right patterns to match the attentional mechanism (which seems to me is based on the modelling system and perception mechanism, matching the salience of the thought to the current situation). This does not mean, however, that the brain is only able to 'think' about one thing at once. Indeed, different modules within the brain have access to perceptions both before and after they are filtered by the internal contextualiser. Although we may only be consciously aware of a single train of thought, the subconscious strands are processing in the background. Providing these with sufficient information to continue working without having to context switch would seem to be a good way of utilising the parallel resources of the brain. Therefore, multiple sources of information, available (as near as the physics of the system allow) simultaneously seems like a good idea. The snag is that because certain modules of the brain have quite strong specialisation, and can be on the critical path for processing the inputs, it is important to support the mind in only passing one set of perceptions through them at one time. The costs of "overloading" an area of the brain are, at a minimum, a waste of energy as different perceptions must be filtered and distibguished from one another, and worse, the potential for cross-over of information resulting in misunderstanding.
Not that all misunderstanding is bad, mind you. It can provide creative insights, and the sense of mild confusion whilst trying to resolve a case of cross-over can provide motivation for further enquiry.

The brain is not a static hierarchy - the dominant thread of consciousness is prone to change with the environment.
Energy costs are minimised when you do not need to perform context switches, and when the pattern of incoming perception is similar to the architecture of the mind (not the brain, as the functional modules in the mind are distributed).
Having too many demands on individual areas of the brain at any one time increases costs and risks information cross-over and confusion.
Communities are organized in a similar way to the organization of the mind, with many similarities in their 'pattern'.
Minds are not hierarchical, but interconnected networks.

It would seem to make sense, therefore, to use the interconnected nature of communities to train incoming members (adult or child) about the norms of the community.
To produce adaptive, flexible problem solvers (which requires education on top of training), it would seem to make sense to use communities to provide a wide-band range of information across a number of topics and let the mind of the learner do what it does best - pattern matching and prioritising, and resolving confusions.
This was written in reply to George Siemens asking how the parallels between mind and community can be used to help the design and delivery of education. I am not entirely sure it answers the question, but hopefully it provides some talking points and generates more interest in the ideas.

This is best supported by providing motivating factors for the learner to acquire specific skills or knowledge so that their prioritisation and confusion-resolving mechanisms work in a specific context.
The cognitive centrality of the community provide this context.
Cognitive consensus provides motivation because it minimises the energy costs, which is desirable on biological grounds.
Elements within the community but on the cognitive periphery (those who do not always agree) provide the community with the possibility of adaptation when the environment changes, and provide the individual learner with challenges to their preconceptions (their internal models) - whilst they cause an energy drain on the community and the individual, they provide stimulus for change and this outweighs the energy cost (if it doesn't, the community is likely to refect the 'outlier').

In terms of my role taxonomy, Explorers within a community act to bring fresh resources to it, and will tend to be on the cognitive periphery because they experience different environments more frequently. Reviewers and Editors distill this information and 'sanitize' it, in a role similar to the internal modeling system. Mentors guide community members to resources, including resources which define learning skills (similar to the attentional mechanism). Translators bridge linguistic gaps between community members. Ambassadors negotiate exchanges of resources (potentially including community members) with other communities, helping to provide diversity and cognitive centrality.

How do we fit these learning communities into the education system? That is a good question - it may be that the hierarchical nature of the existing system will not accommodate them.