BioSpaun: A large-scale behaving brain model with complex neurons

📝 Abstract

We describe a large-scale functional brain model that includes detailed, conductance-based, compartmental models of individual neurons. We call the model BioSpaun, to indicate the increased biological plausibility of these neurons, and because it is a direct extension of the Spaun model \cite{Eliasmith2012b}. We demonstrate that including these detailed compartmental models does not adversely affect performance across a variety of tasks, including digit recognition, serial working memory, and counting. We then explore the effects of applying TTX, a sodium channel blocking drug, to the model. We characterize the behavioral changes that result from this molecular level intervention. We believe this is the first demonstration of a large-scale brain model that clearly links low-level molecular interventions and high-level behavior.

💡 Analysis

We describe a large-scale functional brain model that includes detailed, conductance-based, compartmental models of individual neurons. We call the model BioSpaun, to indicate the increased biological plausibility of these neurons, and because it is a direct extension of the Spaun model \cite{Eliasmith2012b}. We demonstrate that including these detailed compartmental models does not adversely affect performance across a variety of tasks, including digit recognition, serial working memory, and counting. We then explore the effects of applying TTX, a sodium channel blocking drug, to the model. We characterize the behavioral changes that result from this molecular level intervention. We believe this is the first demonstration of a large-scale brain model that clearly links low-level molecular interventions and high-level behavior.

📄 Content

BioSpaun: A large-scale behaving brain model with complex neurons Chris Eliasmith, Jan Gosmann, Xuan Choo Centre for Theoretical Neuroscience, University of Waterloo, Waterloo, Ontario, Canada Abstract We describe a large-scale functional brain model that includes detailed, conductance- based, compartmental models of individual neurons. We call the model BioSpaun, to indicate the increased biological plausibility of these neurons, and because it is a direct extension of the Spaun model [1]. We demonstrate that including these detailed compartmental models does not adversely affect performance across a variety of tasks, including digit recognition, serial work- ing memory, and counting. We then explore the effects of applying TTX, a sodium channel blocking drug, to the model. We characterize the behav- ioral changes that result from this molecular level intervention. We believe this is the first demonstration of a large-scale brain model that clearly links low-level molecular interventions and high-level behavior. Keywords: Spaun, Neural Engineering Framework, Semantic Pointer Architecture, conductance neurons, biological cognition

1. Introduction Recently, several large-scale brain models have been described. These include a biophysically detailed model from Markram’s group in the Human Brain Project (HBP) [2], which includes about 31,000 compartmental neu- rons and 37 million synapses, modelled with many equations per cell. This model is large-scale because of the amount of computation required to sim- ulate its behavior at this level of biological detail. Another model reported earlier by the Synapse project has simulated 500 billion neurons – more than 5x the number in the human brain – although each neuron is much simpler than those in the HBP model, and the connectivity is far more limited [3, 4]. Preprint submitted to ArXiv February 18, 2016 arXiv:1602.05220v1 [q-bio.NC] 16 Feb 2016 We have also previously proposed a large-scale model that includes 2.5 mil- lion neurons, 8 billion connections, and, unlike these other large-scale models, exhibits a wide variety of cognitive behavior [1]. However, our model uses simple leaky integrate-and-fire neurons, and for this reason Markram has claimed “It’s not a brain model” [5]. In this paper we incorporate detailed compartmental models of the type used in the recent HBP model into different cortical areas of our large-scale, behaving brain model. We refer to this augmented model as “BioSpaun”. We show that the behavior of the original Spaun model is not adversely affected by changing the neuron model. We further show that the additional com- plexity can be used to test hypotheses not possible with the original model. Specifically, we demonstrate that BioSpaun can be used to simulate the ef- fects of adding the drug tetrodotoxin (TTX) to these areas of cortex. We perform this manipulation to both visual cortex and frontal cortex, demon- strating performance declines related to the dosage of drug applied, both within and across different tasks. While much remains to be done to verify the accuracy of these simulations in vivo, we believe this is the first demon- stration of a large-scale behaving neural model that includes a high degree of biophysical detail. Integrating these two aspects of brain modeling pro- vides a new method for testing low-level molecular and other physiological interventions on high-level behavior.
2. Methods 2.1. Modeling approach The Neural Engineering Framework (NEF) identifies three quantitatively specified principles that can be used to implement nonlinear dynamical sys- tems in a spiking neural substrate [6]. These methods have been used to propose novel models of a wide variety of neural systems including parts of the rodent navigation system [7], tactile working memory in monkeys [8], and simple decision making in humans [9] and rats [10]. These methods have also been used to better understand more general issues about neural function, such as how the variability of neural spike trains and the timing of individ- ual spikes relates to information that can be extracted from spike patterns [11], and how mixed weight neuron models can be transformed into models respecting Dales Principle, while preserving function [12]. Conceptually, the NEF can be thought of as a neural compiler which allows a researcher to specify a computation as a general nonlinear dynamical 2 system in some state space, which is then implemented in a spiking neural substrate using an efficient optimization method. There are several sources for detailed descriptions of these methods [6, 13, 14], so we do not describe them here. Centrally, the NEF answers questions about how neural systems might compute, but it does not address the issue of what, specifically, is computed by biological brains. In our more recent work, we address this second question by proposing a general neural architecture that includes specific functional hypotheses. We call this proposal the Semantic Pointer

This content is AI-processed based on ArXiv data.