Technical integration of hippocampus, Basal Ganglia and physical models for spatial navigation

Abstract

Computational neuroscience is increasingly moving beyond modeling individual neurons or neural systems to consider the integration of multiple models, often constructed by different research groups. We report on our preliminary technical integration of recent hippocampal formation, basal ganglia and physical environment models, together with visualisation tools, as a case study in the use of Python across the modelling tool-chain. We do not present new modeling results here. The architecture incorporates leaky-integrator and rate-coded neurons, a 3D environment with collision detection and tactile sensors, 3D graphics and 2D plots. We found Python to be a flexible platform, offering a significant reduction in development time, without a corresponding significant increase in execution time. We illustrate this by implementing a part of the model in various alternative languages and coding styles, and comparing their execution times. For very large-scale system integration, communication with other languages and parallel execution may be required, which we demonstrate using the BRAHMS framework's Python bindings.

Keywords: BRAHMS; Python; basal ganglia; hippocampus; place cells; plus-maze; spatial navigation.

Figure 1

Architecture of the basal ganglia model. The main circuit (centre) can be decomposed into two copies of an off-centre, on-surround network: a selection pathway (right) and a control pathway (left). Three parallel loops – channels – are shown in both pathways, with example activity levels in the bar charts to illustrate the relative contributions of the nuclei (the three channels are colour-coded black/grey/white, corresponding to the example bar charts). Note that, for clarity, full connectivity is only shown for the second channel. Briefly, the selection mechanism works as follows. Constant inhibitory output from substantia nigra pars reticulata (SNr) provides an “off” signal to its widespread targets in the thalamus and brainstem. Cortical inputs representing competing saliences are organised in separate channels (groups of co-active cortical neurons), which project to corresponding populations in striatum and STN. In the selection circuit, the balance of focussed (one-to-one) inhibition from striatum and diffuse (one-to-many) excitation from STN results in the most salient input suppressing the inhibitory output from SNr on that channel, signalling “on” to that SNr channel's targets. In the control circuit, a similar overlap of projections to GP exists, but the feedback from GP to the STN acts as a self-regulating mechanism for the activity in STN, which ensures that overall basal ganglia activity remains within operational limits as more and more channels become active. For quantitative demonstrations of this model, see Gurney et al. (2001b, 2004) and Humphries et al. (2006).

Figure 2

Basic structure of the hippocampal model. Different connections are active during learning and recall modes. Learning is performed on the asterisked connections only. Thin lines indicate connections which do not drive their targets, but perform learning only.

Figure 3

Grid cell receptive fields from the model, over physical 2D space. These are plotted with Pylab's Matlab-style imagesc command.

Figure 4

Receptive fields for nine CA3 place cells, superimposed on the robot's path around the plus maze. Crosses show locations where cells firing rates are in the top 5% of their activity throughout the path. Plotting was performed with Pylab's plot command, which has similar syntax to Matlab.

Figure 5

The simulated plus-maze environment. The hippocampus reports the current estimated location, shown by the cross on the floor. When this estimate is close to the center of the plus-maze, the basal ganglia is consulted for an action to turn. 3D physical simulation and visualisation uses PyODE and Pivy.

Figure 6

Real-time graphical neuron monitor, showing basal ganglia and hippocampus model populations. The monitor runs remotely from the simulation over TCP/IP using Pyro, and displays graphics using Pivy.