Researchers study a simple model for spiking systems
A study compares a High-Pass Model to a conductance-based model to see how well it mimics biological neurons.
Researchers investigated how easily the High-Pass Model (HPM) can be tuned. They compared its behaviors to a Hodgkin-Huxley-type Conductance-Based Model (CBM).
The HPM consists of a static sigmoid with a high-pass filter in positive feedback. A high-pass filter is a frequency-selective element. The CBM is a conductance-based model.
Below rheobase, both models fire spikes from transient inputs. This includes post-inhibitory rebound spikes. Above rheobase, the HPM can fit the CBM frequency-current curve using an affine reparametrization of the input.
The models differ near their rheobases. The first-order HPM lacks bistability and spike latency. Modifying the frequency-selective element may overcome this limitation.
The HPM has potential as a fundamental computational primitive. This is relevant for neuromorphic and spike-based technologies.
Key facts
- The HPM can fit the CBM frequency-current curve through an affine reparametrization of the input above rheobase.
- The first-order HPM lacks bistability and spike latency near the rheobase.
- Modifying the frequency-selective element may overcome current limitations of the HPM.