Empirical Analysis of EEG and ERPs for Psychophysiological Adaptive Task Allocation

The present study was designedto test the efficacy of using Electroencephalogram (EEG) and Event-Related Potentials (ERPs) for making task allocation decisions. Thirty-six participants were randomly assigned to an experimental, yoked, or control group condition. Under the experimental condition, a compensatory tracking task was switched between manual and automatic task modes based upon the participant's EEG. ERPs were also gathered to an auditory, oddball task. Participants in the yoked condition performed the same tasks under the exact sequence of task allocations that participants in the experimental group experienced. The control condition consisted of a random sequence of task allocations that was representative of each participant in the experimental group condition. Therefore, the design allowed a test of whether the performance and workload benefits seen in previous studies using the biocybernetic system were due to adaptive aiding or merely to the increase in task mode allocations. The results showed that the use of adaptive aiding improved performance and lowered subjective workload under negative feedback as predicted. Additionally, participants in the adaptive group had significantly lower tracking error scores and NASA-TLX ratings than participants in either the yoked or control group conditions. Furthermore, the amplitudes of the N 1 and P3 ERP components were significantly larger underthe experimental group condition than under either the yoked or control group conditions. These results are discussed in terms of their implications for adaptive automation design.