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

We present a multiscale model of taste that is both biophysically faithful and computationally efficient, enabling end-to-end simulation from receptor transduction to network-level coding. The novelty lies in coupling Hodgkin-Huxley taste receptor cells with Goldman-Hodgkin-Katz ion currents and modality-specific receptors (T1R/T2R, ENaC), to an Izhikevich spiking network equipped with realistic glutamatergic synapses and spike-timing-dependent plasticity. Training combines spike synchrony and a genetic approach in order to reach both globally optimized network structure and biomorphic synaptic plasticity. This hybrid design yields distinct, sparse spiking "fingerprints" for taste qualities and mixtures, and provides a practical foundation for neuromorphic gustatory sensors that require real-time, energy-efficient operation.