4.4 Systematic Changes Demonstrated Across Single-Trial Amplitude and Latency Measures
Previous studies using ST analyses have examined systematic changes in latency across the duration of the ERP task to identify informative patterns that may represent learning over time (Jongsma et al., 2006, Quian Quiroga et al., 2007; Rey et al., 2015). In this current study, an exploratory curve fitting procedure was used to determine whether changes in amplitude, latency, or RT across trials in a single EEG/ERP session of the speeded version of the flanker task were random or systematic. Curve fitting revealed systematic changes that were similar across all components, for both amplitude and latency, as well as RT.
These systematic changes suggest that for the speeded flanker task participants require some time at the start of a session to develop a strategy for successful goal directed behavior. This is reflected by the increasing of the components’ amplitudes, increasing latencies, and longer RTs across the initial trials, peaking around 50 to 100 trials. Following these initial peaks, there was a systematic trend of successively smaller component amplitudes, decreasing component latencies, and shorter RTs. This suggests that once a strategy was developed, participants were able to sustain their behavioral performance more efficiently, more quickly, and with fewer allocated resources. At the end of the block, there were slight increases in component amplitudes and latencies with concomitant RT increases which may reflect fatigue or boredom. Similar systematic changes were repeated in the 2nd block (i.e., after the hashed line in Figures 2 and 3) although blunted, indicating that the strategy was already developed during block one and less time was needed to re-establish and less effort to sustain the behavioral pattern. These systematic changes were most prominent for peak-to-peak data from the STP approach and much less so in the data from the STW approach.
The observed non-linear trends in the peak-to-peak amplitude measures accounted for 12% to 25% of variance across all trials depending on the component. Curve fitting models for latency revealed non-linear systematic change similar to component amplitudes and response times. These findings are consistent with other ST analyses which also have examined systematic changes in latency across the duration of the ERP task to identify informative patterns that may represent learning over time (Jongsma et al., 2006, Quian Quiroga et al., 2007; Rey et al., 2015). Thus, the consistency of the trends in the trial-to-trial changes in amplitude and latency across components and RT suggest that the participants “learn” to implement a strategy for task performance which improves with the accumulation of experience, i.e., successive trials. Similarly, trends of increased amplitudes and latencies at the end of blocks suggests a time-dependent effect such as boredom or fatigue.