Human Resource Scheduling in Performing a Sequence of Discrete Responses

摘要

Behavior in military domains typically requires a sequence of decisions and actions. Yet, characteristics and limitation of cognitive processing are typically based on discrete-trial laboratory studies. The broad objective of this work was to bridge the gap between basic science and applications by: (1) exploring the suitability of central bottleneck models as the basis for computational models of sequence behavior, (2) identifying emergent properties in scheduling behavioral sequences; and (3) determining if and how sequence execution is optimized. Studies examined eye movements and manual responses to sequences of speeded choice response time tasks arrayed linearly on a visual display. A consistent emergent property was discovered in the deferral of the first response, which was shown to be a strategy only loosely linked to resource constraints. Given this strategy central bottleneck theory provided an accurate account of sequence execution. Deferring the first response may represent an optimal response to stochastic fluctuations in the duration of internal processes. The distribution of eye fixation was well fit by reinforcement learning models, evidence for optimality with respect to target probability.