Network Modeling of Adult Neurogenesis: Shifting Rates of Neuronal Turnover Optimally Gears Network Learning according to Novelty Gradient

Abstract

Apoptotic and neurogenic events in the adult hippocampus are hypothesized to play a role in cognitive responses to new contexts. Corticosteroid-mediated stress responses and other neural processes invoked by substantially novel contextual changes may regulate these processes. Using elementary three-layer neural networks that learn by incremental synaptic plasticity, we explored whether the cognitive effects of differential regimens of neuronal turnover depend on the environmental context in terms of the degree of novelty in the new information to be learned. In "adult" networks that had achieved mature synaptic connectivity upon prior learning of the Roman alphabet, imposition of apoptosis/neurogenesis before learning increasingly novel information (alternate Roman < Russian < Hebrew) reveals optimality of informatic cost benefits when rates of turnover are geared in proportion to the degree of novelty. These findings predict that flexible control of rates of apoptosis-neurogenesis within plastic, mature neural systems optimizes learning attributes under varying degrees of contextual change, and that failures in this regulation may define a role for adult hippocampal neurogenesis in novelty- and stress-responsive psychiatric disorders.

Additive neurogenesis as a strategy for avoiding interference in a sparsely-coding dentate gyrus

“…Existing studies have explored neurogenesis in a variety of networks, using a variety of different learning rules (Gould et al, 1999;Cecchi et al, 2001;Chambers et al, 2004;Becker, 2005;Crick & Miranker, 2005;Aimone et al, 2006;Chambers & Conroy, 2007). These studies have been useful in illustrating some of the computational properties of neurogenesis, in particular when neurogenesis acts as a general or targeted turnover of neurons and new units are generated as replacements for dying units in the network.…”

“…A number of explanations of neurogenesis have been put forward in the literature. Many theoretical studies focus on the role of neurogenesis in learning and the computational advantages of neurogenesis have been illustrated in a variety of networks and learning tasks (Gould et al, 1999;Cecchi et al, 2001;Chambers et al, 2004;Becker, 2005;Crick & Miranker, 2005;Aimone et al, 2006;Chambers & Conroy, 2007). Typically neurogenesis is implemented as part of a replacement process for units which die and are subsequently removed from the network, so that the neurogenesis actually takes place as part of a neuronal turnover mechanism.…”

Working memory updating and the development of rule-guided behavior

“…Notably, the decisions at any level of policy are made more difficult by increasing ‘competition’, or the number of competing alternatives at a given branch point (i.e., increasing the width of the tree). Badre & D’Esposito (2007) developed a set of paradigms to consider behavioral effects of competition at different levels of policy abstraction in adults. Here, we adapt these paradigms to examine the developmental course of these processes.…”

“…As such, the developmental course of higher order rule-guided behavior could be a function of updating or gating information into working memory at the level of policy abstraction, at the level of resolving the competition among the increasing fan of options, or some combination. Our work sought to address this question using the paradigms established in Badre & D’Esposito (2007). The potential for specificity and mechanistic insight has broad implications for informing basic science on the topic of the development of rule-guided behavior in this age range, but also fills an important gap with respect to the mechanisms underlying perseverative behavior in a host of developmental disorders, both marked by inefficient cognitive control of thought and action.…”

Computational Modeling of Adult Neurogenesis

“…Subsequently, work by both Chambers and Miranker on this initial study further explored the effect of turnover in ANN networks with progressively more biologically realistic learning rules, the inclusion of inhibition, and targeted replacement strategies (Crick and Miranker 2006;Chambers and Conroy 2007). These studies combined illustrate that the effect described in the initial study is not simply a feature of ANNs, but is a general principle that is likely relevant to more biologically realistic networks.…”