Information Technology Reference
In-Depth Information
- hippocampus - cortex to realize episode memory and long-term memory, respec-
tively. For its characteristic structure of neuroanatomy and computational availability
of mathematics, we adopt this hippocampus-cortex model as a part of our model of
limbic system.
2) The structure of hippocampus is organized with stratified slices and chaotic
response is observed appearing among these neurons. We proposed a new hippocam-
pus-neocortex model to realize mutual association and long-term memory in our pre-
vious works [8, 9]. The main difference between our model and Ito
et al
's model is
that a model of CA3 layer of hippocampus was presented by multi-layered chaotic
neural networks (MCNN) [7]. MCNN is also used to realize dynamical memory proc-
ess in this study.
3) Amygdala plays important roles in emotional responses of brain, and neighbored
to hippocampus in location. We consider that emotion can promote the efficiency of
memory process, so adopt an emotion model proposed by Balkenius and Moren [3]
[13] into our model of limbic system. The main components of the emotion model are
thalamus, amygdala, orbitofrontal cortex and sensory cortex. The model has been
used successfully in an intelligent control system [14] and we also used it to improve
the chaos control process in hippocampus-neocortex model efficiently [10].
According to the knowledge of neuroanatomy, neurophysiology and physiological
psychology, we propose a higher function model of limbic system fusing hippocam-
pus-neocortex model, MCNN and emotion model described above in this paper.
2 A Model of Limbic System
A model of limbic system which is composed with a hippocampus-neocortex model
[8, 9] and an amygdala model [3] [13] is shown in Fig. 1. The hippocampus-neocortex
model consists of a memory circuit including cortex - dentate gyrus - hippocampus -
cortex given by Ito
et al
[4] [5]. Entorhinal cortex and CA2 of hippocampus are omit-
ted for their weak connections. A chaotic model of CA3 layer in hippocampus is
adopted in with multi-layered chaotic neural networks (MCNN) [7] which combines
two chaotic neural networks [1] instead of the single layer of Ito
et al
's model to serve
intermediate-memory processing. MCNN showed its effectiveness of mutual associa-
tion of plural time series patterns and long-term memory formation in our previous
simulations [7] [8] [9]. However, emotional effect is not considered in the memoriza-
tion process of MCNN, likewise almost of conventional association systems. In fact,
amygdala, which locates adjacent of hippocampus, plays important roles in the emo-
tional responses such as fear and aggression, and also in the memorization process
such as learning of Pavlovian fear conditioning and enhancing of long-term memory
formation [14]. We adopt an amygdala model given by Balkenius and Moren [3] [13]
here to evaluate and promote the efficiency of memory processing.
2.1 A Hippocampus-Neocortex Model
The dark dots and gray dots in Fig. 1 express neurons of neocortex and hippocampus,
and symbols of CX1, CX2, DG, CA3 and CA1 denote the first layer of cortex, the
