Despite the simplicity of the Ising model, we can use it to demonstrate the relevance of certain features responsible for changes in language, such as the propagation and the level of linguistic interaction and attenuation. However, there are several features that we have ignored in this model, such as long-range neighbour interactions and the aspect of competing uses of vocables that we described earlier (The dynamics of shareability) . We have sketched a dynamic in which a population of users are engaged in linguistic transactions. Each brings forth a unique use of vocables, which are then modified by a process comparable to a competition between individual effects generated from perceptual cues. The effects are, we assume, of a neural kind. However our population dynamics account is too coarse-grained to give a description of what they may be. We suggest that the dynamics of competition at a population level may be a reflection of what goes on at the synaptic level.
William H. Calvin has created a model that offers a detailed description of what neural effects may be. He describes a context of competing synaptic patterns that are responsible for our capacity to abstract including our capacity to schematize and possibly for our use of language. Calvin's refers to his model as a Darwinian machine. His inspiration is based on the structure of the superficial layers of the neocortex and their behavior. Once stimulated, the neural arrangement in these areas, generates synaptic patterns that are subsequently replicated across pairs of neurons that consequently form a whole structure. Calvin suggests that the replication process is not unlike putting down tiles on a floor as we try and fit the edges of the tiles to cover the surface as well as possible. Several features of the tile, such as its shape and color, perhaps its thickness, will define the overall pattern of the flooring. This metaphor illustrates how a perceptual structure occurs in synaptic functions. Calvin's model is not unlike the Ising model, except for its capacity for long range interactions (as opposed to nearest neighbour interactions) and a few other less crucial features.
In Calvin, linguistic activities are perceptual structures, that is, neural activations that occur from our interaction with our environment. These neural activations generate patterns. For Calvin, a complete pattern requires three pairs of neurons configured as a hexagon. The strength of activation of this pattern may activate an entire structure of many hexagons. These structure may be in competition with each other in a context of partial or degraded perceptual cues. The competition is usually resolved, often with inaccuracies that can give rise to several of our thinking processes such as generalized schemas.
The idiom of physics is not a large part of the elaboration of Calvin's model, but he does explain how attractors bias and promote the propagation of spatio-temporal neural structures. An attractor can be understood as a constituent locked temporarily in a particular state. Constituents locked in that state can potentially entrain neighbours into a similar state. This locked state can be a periodic vibration of sorts, a resonance. We have mentioned an example of entrainment from resonance with the synchronization in the firing of fireflies and the synchronization of women's periods. Resonance is what causes neurons to fire rhythmically for limited cycles. There are many types of periodicity in resonance patterns. Some can be compared to a simple melody while other are more akin to white noise.
For Calvin, multiple attractors can be in competition with each other for the propagation of their own pattern. A resolution is achieved through a first-order phase transition the result of which can be a merging of attractors. For example, the experience of the object apple will generate an attractor and so will the syntactic experience of the word. The merging of both attractors will promote the propagation of the specific neural pattern of combined perceptual and linguistic experience.
We are interested in Calvin's account because it offers a finer grained description of linguistic interactions, one that is compatible, and even complements, our more coarse-grained physical, population dynamical, account of linguistic interactions. Calvin's model is a good illustration of how a fine grained description of language does not have to rely on a semantic theory and how it can be compatible with a wider family of physical theories including physics, biology and neuroscience.