The shape and excitability of neuronal dendrites are expected to be responsible for the functional characteristics of information processing in the brain. In the present study, we proposed that excitable media with branching patterns mimicked the multi-signal integration of neuronal computation. We initially examined the conditions of the coincidence detection of two inputs as the simplest form of signal integration. We considered a gate with two channels that was bound by a circular joint with uniform excitability and demonstrated that the time window for the coincidence detection was controlled by the geometry and excitability of the gate. The functions of the gate were due to the unique property of the excitation waves, known as the curvature effect. The expanded spatial spread diluted the incoming excitation signals to insufficient levels to sustain wave advancement. Next, we applied dendritic gates that were reminiscent of neuronal dendrites for multi-signal integration. The irregular dendritic patterns were produced by a cellular automaton model of self-organizing pattern formation that adopted the semi-random grid in numerical simulations. We demonstrated that the threshold operation for multiple inputs was conducted by the dendritic pattern. The thresholds varied among gates owing to their irregular patterns, and were adjusted by changing the excitability without changing the gate geometry. The materializable model may provide a novel biomimetic approach for developing fuzzy hardware with adjustable responsiveness.
All Science Journal Classification (ASJC) codes
- Cognitive Neuroscience
- Artificial Intelligence