Model dependence in quantification of spike interdependence by joint peri-stimulus time histogram

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Multineuronal recordings allow studying context-dependent neural interactions.
• The joint peri-stimulus time histogram (JPSTH) is commonly used to assess spike event interdependence between neuronal pairs.
• JPSTH results are often interpreted as interaction strength, independent of firing rate changes.

Purpose of the Study:

• To evaluate the accuracy of JPSTH in estimating interaction strength, specifically after attempting to remove firing rate contributions.
• To investigate the mathematical basis of JPSTH's quantification of interaction strength.
• To determine if effective connectivity measures are independent of firing rate modulations.

Main Methods:

• Developed a probabilistic model of interacting point processes to simulate neuronal spike data.
• Generated simulated data to test the JPSTH method under controlled conditions.
• Analyzed the mathematical structure of JPSTH and compared it to an effective connectivity model.

Main Results:

• The normalized JPSTH incorrectly infers the temporal dynamics of interaction strength variations.
• The required normalization to correct for firing rates differs from that used in established effective connectivity models.
• Firing rate modulations cannot be corrected in a model-independent way.

Conclusions:

• Effective connectivity, as calculated by JPSTH, is not a universal characteristic independent of firing rate modulations.
• JPSTH provides one specific way to describe interdependence, not an absolute measure.
• Awareness of JPSTH's limitations is crucial when analyzing experimental data on neural interdependence.