Dynamic Patterns: The Self-Organization of Brain and Behavior

This book was a great help in my quest to understand the Recognition Process. Very informative.

What language would It use to describe itself? Can something like mathematics, music, language or art, which It created contain the essence to even describe itself? Do modern models describe It any better than earlier models? In the context of the times does Non-linear Dynamics attractors describe (or for that matter predict) any better than C.Jung's archetypes? As far as that goes, since they both describe scales of magnitudes, why is their so much similarity between so many of each others concepts?(enantiomorphism vs. entrainment). One would think the biological scientific community would really embrace the concepts in this book for the simple reason to get them off the hook with one of their most difficult areas. If the human brain and thus cognitive and conscious processes are a product of limit cycles and memory simply a function of a hystersis curve the entire process becomes a function of Newtonian physics in support of biology's basic premise. Self-determination and teleological arguments become academic. Mind boggling. (I think the human brain is highly overrated). While Kelso's book Dynamic Patterns doesn't answer these questions either, it is still a pretty good book (overlooking not having any equations on the first page). Already in it's third printing it doesn't need me to critique it. For computer modelers: Mathcad contains the necessary (logistics) for scalar mapping and iterated bifurcations to provide the background to create the models used in the book. (including Lorenz models, Duffings, ODE solvers). Dynamics: Numerical Explorations, (Springer) with related program even goes one better. Matlab's signal processing toolbox, (creates any signal p.48, check out the Hofstader sequence relation to Kelso's basic signal), wavelets toolbox (creates Farey sequence, most converging fraction sequences paint this same picture). Because these equations can be set up as state-space models the same model can be used Matlab's system identification, control, and neural net toolboxes (use feedforward instead of feedback). For biologicals: Most applicable non-linear equations are much better handled by differential equations. While a unit circle torus may work for pictures it is does not give unique solutions and thus calculable answers. Assuming the time constant is the same in each differential equation and then cascading the solutions into each other to build matrices that can then be used to build the phase plots easily solves this. Check out Spikes, Decisions, and Actions (Wilson) probably the best as far as Matlab add-ons go (real action figures basic formulas and models can be modified to see the effect of gain i.e., use of amphetamines or inhibitors SSRI). Although Koch's (Biophysics of Computation provides good background in stochastic processes in the Markovian sense and Tri-diagonal matrices) startneuro Simulink from Methods in Neuronal Modeling (MIT Press) is a lot more fun!(Gabrianni and