A successive period-doubling cascade (‘spd’) and chaotic solutions are numerically found as a function of leakage conductance in the space-clamped Hodgkin-Huxley equations (HH) for muscle fibres. They occur under the condition of inhibition of sodium inactivation, within a realistic parameter range. While it has been known that HH with a periodic current stimulus have a ‘spd’ root to chaos, the present result is the first report of ‘spd’ in an autonomous style of HH.
Incremental learning of knowledge and context dependency of recognition are important characteristics for an intelligent machine in the real world environment. Unknown objects may appear among known objects in such environment and the context requires change of the recognition result even if the input is the same. The system has to learn which object is unknown, what knowledge is necessary and how the context acts on the recognition process. Associative memory model PATON (Pattern+ton) has been proposed to realize such context dependency of recognition on the basis of the attention vectors. External inputs and candidates for the recognition can both be selected by an attention vector. In this paper, we propose a method of incremental learning involving the attention vectors and the knowledge, based on a reward through conversational interactions between PATON and the environment.
In our previous report (Tsukada et al. (1996): Neural Networks, Vol. 9, pp. 1357-1365), the temporal pattern sensitivity of long-term potentiation (LTP) in hippocampal CA1 neurons was estimated by using Markov chain stimuli with different values of the serial correlation coefficient ρ1 between successive interstimulus-intervals. In this paper, the effect of chaotic stimuli on induction of LTP in the hippocampal CA1 area was investigated in comparison with those of Markov chain stimuli and periodic stimuli. The chaotic stimuli were produced by a modified Bernoulli map, so that interstimulus sequences with various values of ρ1 can be generated by changing the parameter B. These stimuli had an identical first order statistics (mean interstimulus-interval and variance), but their higher order statistics were different in the serial correlation coefficient as well as in fluctuation regarding the dynamic property of stimulus sequences. The LTP induced by chaotic stimuli with B=2 were significantly larger in magnitude than those of periodic stimuli and Markov chain stimuli, and also depended on the initial value of chaotic stimulus at B=2 and 3. These results suggest that chaotic signals play an important role for memory coding in the hippocampal CA1 network, and are discussed in relation to findings from a model simulation.