I attended a lecture from Herb J. Weingartner yesterday afternoon here at UMaine. The title of the lecture was "Cognitive Neuroscience and Education". Unfortunately I had to leave as soon as the presentation finished, and could not stay for the question and answer period that followed. However, the content of the presentation served a great function - it reminded me about how important I think it is to make sure that conceptual models follow the general principles of what we know about the brain.
One of the slides in the presentation could have been a basic map of my vision of the refined coordination class model for concepts. Of course, the labels were different, but the essential features were there: a read-out of information from the environment by the sensory apparatus, a complex processing network that analyzed and interpreted the information presented by the sensory apparatus, and an output from that complex network that was some sort of externally-observable (objective...?) behavior.
But you, reader, might know enough about the original coordination class model to say: hey, why revise? That sounds just like what diSessa and Sheren were talking about: read-out, causal net, and another read-out. Well, yes, but there was another part of Weingartner's diagram that I haven't mentioned yet -- a SECOND input of information into the causal network. That input, I suspect, is incredibly important in order for teachers and other educational researchers to understand the pseudo-random interjections and explanations that students exhibit in using concepts. The second input to the causal network comes from the activity of other nerves within the brain that may or may not be a stable component of the concept.
Various neuroscience experiments have shown that neurons don't just fire when stimulated by another neuron. At least some nerve cells, in various parts of the brain, fire spontaneously - independent of any external stimulus. Some of these spontaneously firing nerve cells act as oscillators, firing with regularity over variable temporal intervals. Within this set of neural oscillators, the degree of regularity is variable, and context-sensitive. In 1998, Prut et. al. published "Spatiotemporal Structure of Cortical Activity: Properties and Behavioral Relevance", which describes some ways in which neurons act in networks that can be related to specific behavior. The authors describe the complexity of neural networks, showing that single neurons can participate in multiple networks through variability in activation in time (a reinforcement of the notion that neural networks exhibit scale-free properties). Furthermore, the authors demonstrate that behavior is highly correlated with the temporal qualities of neural network activation, and not only with external stimulus. In essence, this research highlights the role of spontaneous neural activity in animal behavior. The reading list provided by the authors' citations is comprehensive and provides further sources for information on spontaneous neural activity.
There are multiple consequences from this neuroscience research to the development of cognitive models for concepts. Behavior is certainly linked to external stimulus, and neural activity is coordinated by that external stimulus. This supports the notion that concepts are constructed on-the-fly with a strong degree of context sensitivity. However, the research also demonstrates that there are some components of a particular conceptual construction at a particular instant in time that are not necessarily a stable component of that concept, or that are directly linked to the external environment of the learner. At this point I think it is possible to make a claim that is not yet a part of the knowledge-in-pieces, resources-based models for concepts: we really can think of cognitive resources as activated single nerves or activated networks of nerves. Furthermore, educational researchers should expand the possible behaviors of resources to include the possibility of spontaneous activation that may have significant variability in the regularity of activation.
So, the long and the short of this boils down to the following. Sometimes a student's concept is going to include resources that are not involved in expert concepts and that are not intended to be activated by the teacher when creating the learning context.