On the Necessity of Mixed Models: Dynamical Frustrations in the Mind
This addresses the challenge of modeling complex mental processes in linguistics, but it is incremental as it builds on existing ideas about quantum computation and mixed models.
The paper tackles the problem of modeling linguistic derivations that appear to exceed Markovian or Turing-like computation, proposing that a quantum processor is necessary, and argues that mixed linear and non-linear models are required for a unified understanding of mental computations.
In the present work we will present and analyze some basic processes at the local and global level in linguistic derivations that seem to go beyond the limits of Markovian or Turing-like computation, and require, in our opinion, a quantum processor. We will first present briefly the working hypothesis and then focus on the empirical domain. At the same time, we will argue that a model appealing to only one kind of computation (be it quantum or not) is necessarily insufficient, and thus both linear and non-linear formal models are to be invoked in order to pursue a fuller understanding of mental computations within a unified framework.