The prefrontal cortex has long been thought to subserve both working memory (the holding of information online for processing) and executive functions (deciding how to manipulate working memory and perform processing). Although many computational models of working memory have been developed, the mechanistic basis of executive function remains elusive, often amounting to a homunculus. This article presents an attempt to deconstruct this homunculus through powerful learning mechanisms that allow a computational model of the prefrontal cortex to control both itself and other brain areas in a strategic, task-appropriate manner. These learning mechanisms are based on subcortical structures in the midbrain, basal ganglia, and amygdala, which together form an actor-critic architecture. The critic system learns which prefrontal representations are task relevant and trains the actor, which in turn provides a dynamic gating mechanism for controlling working memory updating. Computationally, the learning mechanism is designed to simultaneously solve the temporal and structural credit assignment problems. The model’s performance compares favorably with standard backpropagation-based temporal learning mechanisms on the challenging 1-2-AX working memory task and other benchmark working memory tasks.

The frontal cortex and basal ganglia interact via a relatively well-understood and elaborate system of interconnections. In the context of motor function, these interconnections can be understood as disinhibiting or "releasing the brakes" on frontal motor action plans --- the basal ganglia detect appropriate contexts for performing motor actions, and enable the frontal cortex to execute such actions at the appropriate time. We build on this idea in the domain of working memory through the use of computational neural network models of this circuit. In our model, the frontal cortex exhibits robust active maintenance, while the basal ganglia contribute a selective, dynamic gating function that enables frontal memory representations to be rapidly updated in a task-relevant manner. We apply the model to a novel version of the continuous performance task (CPT) that requires subroutine-like selective working memory updating, and compare and contrast our model with other existing models and th...

A selection problem arises whenever two or more competing systems seek simultaneous access to a restricted resource. Consideration of several selection architectures suggests there are significant advantages for systems which incorporate a central switching mechanism. We propose that the vertebrate basal ganglia have evolved as a centralised selection device, specialised to resolve conflicts over access to limited motor and cognitive resources. Analysis of basal ganglia functional architecture and its position within a wider anatomical framework suggests it can satisfy many of the requirements expected of an efficient selection mechanism. Key words: behaviour, action, movement, switching, model, architecture, motor control Citation: Redgrave, P., Prescott, T.J. and Gurney, K. (1999). The basal ganglia: a vertebrate solution to the selection problem?, Neuroscience, 89, 1009--1023. INTRODUCTION Despite a prodigious volume of work in recent years there is still no consensus co...

A fundamental question is whether people have available one category learning system, or many. Most multiple systems advocates postulate one explicit and one implicit system. Although there is much agreement about the nature of the explicit system, there is less agreement about the nature of the implicit system. In this article, we review a dual systems theory of category learning called competition between verbal and implicit systems (COVIS) developed by Ashby et al. (1998). The explicit system dominates the learning of verbalizable, rule-based category structures and is mediated by frontal brain areas such as the anterior cingulate, prefrontal cortex (PFC), and head of the caudate nucleus. The implicit system, which uses procedural learning, dominates the learning of non-verbalizable, information-integration category structures, and is mediated by the tail of the caudate nucleus and a dopamine-mediated reward signal. We review nine studies that test six a priori predictions from COVIS, each of which is supported by the data.

rule-based but not information-integration

this article should be addressed to W. Todd Maddox, Department of Psychology, Mezes Hall 330 Mail Code B3800, University of Texas, Austin, Texas, 78712. E-mail: maddox@psy.utexas.edu

Twelve male listeners categorized 54 synthetic vowel stimuli that varied orthogonally in F2 and F3 on a BARK scale into the American English vowel categories /I/, /U/, and / ˛ /. A neuropsychological model of visual categorization, called the Striatal Pattern Classifier (SPC; [1]) is generalized to the auditory domain, and applied separately to the data from each observer. Performance of the SPC is compared with the successful Normal A Posteriori Probability model (NAPP; [2], [3]) of auditory categorization. Versions of the SPC and NAPP that assume linear response region partitions provided similar accounts of the data. Nonlinear versions of both models provided only small improvements in fit. 1.

tasks, whereas response position is learned by the procedural-learning system to solve informationintegration tasks. Accuracy rates were examined to isolate global performance deficits, and modelbased analyses were performed to identify the types of response strategies used by observers. A–B training (consistent mapping) led to more accurate responding relative to yes–no training (variable mapping) in the information-integration category learning task. Model-based analyses indicated that the yes–no accuracy decline was due to an increase in the use of rule-based strategies to solve the information-integration task. Yes–no training had no effect on the accuracy of responding or distribution of best-fitting models relative to A–B training in the rule-based category learning tasks. These results both provide support for a multiple-systems approach to category learning in which one system is procedural-learning–based and argue against the validity of single-system approaches.

The role of verbal and visuospatial working memory in rule-based and information-integration category learning was examined. Previously, Maddox, Ashby, Ing, and Pickering (2004) found that a sequentially presented verbal working memory task did not affect information-integration learning, but disrupted rule-based learning when the rule was on the spatial frequency of a Gabor stimulus. This pattern was replicated in Experiment 1, in which the same category structures were used, but in which the verbal working memory task was replaced with a visuospatial analog. Experiment 2A examined rule-based learning on an oblique orientation and also found both verbal and visuospatial working memory tasks disrupting learning. Experiment 2B examined rule-based learning on a cardinal orientation and found a minimal effect of the verbal working memory task, but a large effect of the visuospatial working memory task. The conceptual significance of cardinal orientations and the role of visuospatial and verbal working memory in category learning are discussed. Perceptual categorization is a systematic differentiation among classes (or categories) of objects based on their perceptual features. A large body of evidence suggests that humans have available several category learning systems (see Ashby & Maddox, 2005, or Keri, 2003, for a review). The classical approach to categorization (Bruner, Goodnow, & Austin, 1956), and most multiplesystems approaches, postulate one categorization system based on an extraction of a categorization rule via hypothesis testing during an explicit reasoning process

Physiological studies of the principal neurons of the neostriatum indicate that these cells have unusual membrane properties, which effectively give them two dynamically stable levels of membrane potential. One of these potential levels, the "up" state, allows the cell to fire in response to small changes in input spike frequency, while the other, the "down" state, leaves the cell relatively unresponsive and very resistant to firing. This chapter describes a biophysically-based model of this cell type, the medium spiny neuron, which demonstrates how the cell's ion channels interact to produce this bistable response pattern. The model includes observed modulatory influences of dopamine and acetylcholine on these ion channels, which lead to the surprising property that the medium spiny cell can effectively act as a programmable binary latch. A network of these model cells can support proposed functions of the neostriatum, and predicts some of the symptoms of basal ganglia disorders such ...