Computational Advances in Taste Perception: From Ion Channels and Taste Receptors to Neural Coding

Area of Science:

• Computational neuroscience
• Biophysics
• Sensory systems modeling

Background:

• Understanding taste processing requires bridging molecular receptor events with network-level neural activity.
• Existing models often lack either biophysical detail or computational efficiency for end-to-end simulation.
• Developing efficient, accurate models is crucial for advancing neuromorphic engineering and sensory science.

Purpose of the Study:

• To create a multiscale computational model of taste perception.
• To achieve both biophysical fidelity and computational efficiency in taste simulation.
• To establish a foundation for developing real-time, energy-efficient neuromorphic gustatory sensors.

Main Methods:

• Coupling Hodgkin-Huxley taste receptor cells with Goldman-Hodgkin-Katz ion currents and specific taste receptors (T1R/T2R, ENaC).
• Integrating these receptor models with an Izhikevich spiking neural network featuring glutamatergic synapses and spike-timing-dependent plasticity.
• Employing a hybrid training approach combining spike synchrony and genetic algorithms for network optimization and synaptic plasticity.

Main Results:

• The multiscale model successfully simulates taste transduction to network-level coding.
• Distinct and sparse spiking neural activity patterns ('fingerprints') were observed for different taste qualities and mixtures.
• The model demonstrates computational efficiency suitable for real-time applications.

Conclusions:

• The developed model provides a biophysically faithful and computationally efficient framework for taste processing.
• The distinct neural fingerprints offer insights into taste coding mechanisms.
• This work lays the groundwork for practical neuromorphic gustatory sensors.