Reconstructing ERP amplitude effects after compensating for trial-to-trial latency jitter: A solution based on a novel application of residue iteration decomposition

- The within person, trial-to-trial latency jitter ubiquitously exists in ERP data
- The trial-to-trial latency jitter blurs ERP waveforms and affects ERP amplitude effects between conditions
- Residue iteration decomposition (RIDE) was applied to correct the blurred amplitude effects caused by latency jitter

Stimulus-locked averaged event-related potentials (ERPs) are among the most frequently used signals in Cognitive Neuroscience. However, the late, cognitive or endogenous ERP components are often variable in latency from trial to trial in a component-specific way, compromising the stability assumption underlying the averaging scheme. Here we show that trial-to-trial latency variability of ERP components not only blurs the average ERP waveforms, but may also attenuate existing or artificially induce condition effects in amplitude. Hitherto this problem has not been well investigated. To tackle this problem, a method to measure and compensate component-specific trial-to-trial latency variability is required. Here we first systematically analyze the problem of single trial latency variability for condition effects based on simulation. Then, we introduce a solution by applying residue iteration decomposition (RIDE) to experimental data. RIDE separates different clusters of ERP components according to their time-locking to stimulus onsets, response times, or neither, based on an algorithm of iterative subtraction. We suggest to reconstruct ERPs by re-aligning the component clusters to their most probable single trial latencies. We demonstrate that RIDE-reconstructed ERPs may recover amplitude effects that are diminished or exaggerated in conventional averages by trial-to-trial latency jitter. Hence, RIDE-corrected ERPs may be a valuable tool in conditions where ERP effects may be compromised by latency variability.