Neural systems produce functionally equivalent sequential activity patterns with a high variability in their temporal structure. To study whether a set of temporally similar neural patterns has the same functional meaning it is convenient to use closed-loop stimulation. The patterns are studied in real-time and when a neural pattern with a specific temporal structure is detected, a stimulus is delivered to the system. If the system responds similarly to these stimuli, the set of neural patterns detected in real-time potentially represent an internal state of the system, i.e., they transmit the same information. However, due to their natural variability, detecting these patterns in real-time is a challenge. This work focuses especially on the real-time detection of neural patterns that can be functionally equivalent considering their variability. The Temporal Code-Driven Stimulation protocol with the Victor-Purpura distance allows the study of neural patterns that can be functionally equivalent in closed-loop experiments, but the parameterization of this protocol to match a set of patterns with a specific temporal structure in a certain research context is not an easy task. This work proposes a methodology to optimize the parameterization of the protocol for the robust detection of early accelerations in the weakly electric fish electroreception signaling, taking into account the variability with which these patterns are generated. As an example illustration of the protocol parameterization methodology, early accelerations have been selected as patterns of interest and separated from the rest. The values of the protocol parameters were configured through metrics and algorithms designed on purpose to robustly detect these early accelerations. Finally, the protocol was tested in real-time and the methodology was validated by showing that the detected patterns, in electrical signals that had not previously been seen, had a similar temporal structure to the patterns of interest. |
*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.