Fukushima's Selective Attention Model is a model of biological vision system. It has engineering merit of deformation and position shift tolerant recognition and recollection of the recognized object in an input image. Based on this feature, we have proposed a method for hybrid image understanding in which each object is recognized, recollected and segmented sequentially even when the objects are overlapped and occluded. However, parameter setting for the fine object recollection is difficult with the Selective Attention Model and parameters search is necessary to tune the given image. In this paper, we propose a method for the parameter search based on an evaluation of image recollection precision in the hybrid image understanding process. A computer simulation result demonstrates the validity of the proposed method.
It is believed that the hippocampus and the neocortex are involved in several aspects of learning and memory. However, the relationship of the activities between the hippocampus and the neocortex are not well investigated. In the present study, we recorded the response to a sound stimuli in the auditory cortex of the anesthetized guinea pig by voltage sensitive optical recording. And we examined the effect of the hippocampal stimulation on the auditory cortex response by changing the stimulation current intensity. The results showed that the weak stimulation inhibits the auditory responses but the strong stimulation facilitates the responses. It means that the hippocampal activity can produce excitatory or inhibitory potentials, suggesting that it is an important neuronal basis of the memory consolidation process.
In spatial alternation tasks, the same place in different behavioral sequences is represented by different neurons in the rat hippocampus. This indicates that associative memory in the brain solves the memory interference problem at each instance. By assuming that different behavioral sequences are encoded in different continuous attractors, we hypothesize that theta rhythm-dependent activity provides cooperation between external input and retrieval of the behavioral episode. In our computer experiments, these two activities with an appropriate phase difference enable the maintaining of a consistent attractor in accordance with behavior. It suggests neural dynamics for the real-time process of associate memory.
Classical conditioning induced map plasticity in the adult sensory cortex has not been well investigated. In this study, we used voltage sensitive optical recording method to study map changes in the auditory cortex induced by fear conditioning with parining of sound and electric shock. From pair-conditioned group and pseudoconditioned group, the responses were recorded both before and after the conditioning. The result shows that after conditioning, only the response area to CS sound in the pair-condtioned group was increased. This suggests that many neurons in the auditory cortex exhibit a rapid tuning to the CS sound by the fear conditioning.
The modular network SOM (mnSOM) proposed in this paper is a Self-Organizing Map (SOM) in function space, as opposed to the conventional SOM in vector space. Whereas each node of the conventional SOM represents a codebook vector, each unit of mnSOM represents a function (i.e., input-output relationship) which may be a dynamical one. In other words, all nodes of the competitive layer are replaced by some kind of neural networks which may be of a multi-layer perceptron type or a recurrent type. The performance of mnSOM is examined by simulation examples such as one dealing with geology-dependent meteorological changes in Japan, one involving musical scale and one simulating a mass-spring-dashpot system. These results show that the functions acquired by the winner modules are mapped into the 2D lattice with topological continuity, i.e., similar functions are close to each other and desimilar ones are allocated far apart. Moreover, “test functions” whose corresponding input-output data are not used during the training are mapped as “test winner modules” that appear at interpolated locations between “training winner modules”.
Many studies have emphasized simulation theory as a core concept for the mirror neurons that are found in the ventral premotor area (F5) and inferior parietal cortex (PF) of the macaque. Imitation, social cognition, empathy and language are linked to mirror neurons. However, function of mirror neurons in the macaque is still missing, because such cognitive functions are limited or lacking in these monkeys. It is important to discuss these neurons in the aspect of motor function. I would suggest that mirror neurons in area PF might have a roll of monitoring of self generated action. This idea may provide neural correlate of self other distinction.
Current conceptual models of mirror neurons attribute high level functions such as imitation, action understanding and (precursor of) language to mirror neurons. However the number of studies that take a computational viewpoint to realize these proposals is limited. In this article we first present current opinions (`conceptual models') on the mirror neuron function. Then we review the computational models available and explain the strengths and weakness of these models in addressing the mirror neurons. Our meta-analysis underlines the gap between conceptual and computational models and points out the required research effort at both sides to reduce this gap.