Most neural systems encode information through temporally varying sequences of action potentials, which are usually related to specific functions. However, their intrinsic variability poses a challenge to study these sequences. This paper analyzes the flexibility of a closed-loop stimulation protocol based on the Temporal Code-Driven Stimulation algorithm to study temporal coding in neural systems while accounting for variability. The protocol uses the Victor-Purpura distance to assess the similarity between sequences emitted by a neural system and a target sequence. When it detects in real-time a sequence whose temporal structure has an established degree of similarity to the target, it delivers a stimulus to the system to drive it to a target state. The adaptability of the protocol to variability in the information coding of neural systems was evaluated using the Hindmarsh-Rose neural model. In the validation experiments, the protocol stimulated the model after detecting a target spike sequence in the bursting activity. Gaussian noise was injected into the model to induce variability and evaluate the protocol’s adaptability. The results demonstrate that, even in the presence of considerable neural variability, the closed-loop stimulation protocol effectively conditions the model's activity, outperforming an open-loop approach to drive the system into the desired state. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.