Subcortical loops through the basal ganglia and cerebellum form computationally powerful distributed processing modules (DPMs). This chapter relates the computational features of a DPM's loop through the basal ganglia to experimental results for two kinds of natural action selection. First, data from both monkeys and humans in a step-tracking task were used to decipher the neural mechanisms that underlie the detection of movement errors leading to selection of corrective movements called submovements. Second, functional brain imaging of human subjects during a serial-order recall task was used to study brain activity associated with decoding a sequence of actions from information held in working memory. Our DPM-based model assists in the interpretation of puzzling data from both of these experiments. These analyses lead to a broad discussion of the DPM concept and how it relates to neuroscience, modularity, engineering, evolution, mathematical recursion, agent-based modelling, Bayesian computations, and brain disorders. The loops through basal ganglia and cerebellum profit from exceptional combinations of unique cellular properties together with advantageous neural circuitry. Their modular organisation means that DPMs regulate pattern formation in multiple areas of the cerebral cortex, thus initiating and refining different kinds of action (or thought), depending on the area of the brain. We then use our findings to formulate a novel model of the etiology of schizophrenia.
The higher order circuitry of the brain is comprised of a large-scale network of distributed processing modules (DPMs). Each of approximately 100 cerebral cortical areas is individually regulated by relatively private loops through subcortical structures, particularly through the basal ganglia and cerebellum (Houk, 2005; Houk and Wise, 1995; Kelly and Strick, 2003, 2004). These DPMs have powerful computational architectures as summarised in Figure 10.1. Each DPM receives cortico-cortical input vectors from approximately seven other DPMs (although only two are shown in Figure 10.1). (The estimate of seven derives from Felleman and Van Essen, 1991.) The final outcome of all of the computations in a given DPM is a spatiotemporal pattern of activity in the module's output vector, representing the activity in its set of cortical output neurons. This output is sent as input to other DPMs, or to the brainstem or spinal cord. In this manner, arrays of DPMs form large-scale networks that function in combination to control behaviour, or thought. The reader should consult Houk (2005) for a detailed description of this architecture and a justification of its capacity to control both actions and thoughts. The brief overview of functional operations in loops through basal ganglia (BG) and cerebellum (CB) given in the next two paragraphs is a summary that applies to the selection and initiation of movement commands that control discrete actions.