A computational model of parallel navigation systems in rodents

被引:35
作者
Chavarriaga, R
Strösslin, T
Sheynikhovich, D
Gerstner, W [1 ]
机构
[1] Ecole Polytech Fed Lausanne, Sch Comp & Commun Sci, CH-1015 Lausanne, Switzerland
[2] Ecole Polytech Fed Lausanne, Brain Mind Inst, CH-1015 Lausanne, Switzerland
关键词
action selection; biomimetic agents; navigation strategies; reinforcement learning;
D O I
10.1385/NI:3:3:223
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
Several studies in rats support the idea of multiple neural systems competing to select the best action for reaching a goal or food location. Locale navigation strategies, necessary for reaching invisible goals, seem to be mediated by the hippocampus and the ventral and dorsomedial striatum whereas taxon strategies, applied for approaching goals in the visual field, are believed to involve the dorsolateral striatum. A computational model of action selection is presented, in which different experts, implementing locale and taxon strategies, compete in order to select the appropriate behavior for the current task. The model was tested in a simulated robot using an experimental paradigm that dissociates the use of cue and spatial information.
引用
收藏
页码:223 / 241
页数:19
相关论文
共 40 条