Proactive and Reactive Cognitive Control Under Threat of Unpredictable Shock: A Combined Eye-Tracking and EEG Study Using Multilevel Modeling

Abstract

We are constantly bombarded by environmental distractors in daily life which interfere with internal, ongoing goals, thus cognitive control processes need to be in place to adapt to maintain these goals in light of the environmental demands. These cognitive processes (generally referred to cognitive control) are thought to be adjusted reactively or proactively to deal with distractors. There is little evidence on how state anxiety dynamically interacts with these two modes of cognitive control. Taking advantage of a multimodal methodology, through two experiments, we replicated existing findings of reactive and proactive control processes via utilizing a Flanker task in a laboratory setting, and acquired evidence of neurocognitive (N2, 200-350ms, and frontal slow-wave, 500-700ms, components) and eye-gaze (dwell-time) indices corresponding to these modes using a highly stringent, multilevel modeling approach. In the second experiment, we administered the threat of unpredictable shock and demonstrated that induced state anxiety, versus safe, had an overall enhancing effect on reaction time (RT) but no effect on accuracy, regardless of the mode of cognitive control. However, shock had a unique enhancing effects on reactive control by shifting its mode of operation from “reactiveness” toward “preemptiveness” while having a dampening effect on the proactive mode through using attentional resources and leaving limited capacity for proactivness in the face of highly frequent distractors. Unlike previous studies, we found a potentiation of N2 amplitude and longer eye-gaze fixations for proactive mode to further support the idea that the proactive mode might be associated with some compensatory activity under the threat of shock which might result in a better overall performance compared to reactive mode, however, this compensation could not outperform the proactive mode under the safe condition. Overall, the multilevel modeling along with the multimodal methodology adopted in this experiment provided strong supportive evidence of previous experiments in the context of induced state anxiety and suggested a replication of this finding with individuals with trait anxiety to further disentangle the differences observed in cognitive control between induced state anxiety and trait anxiety.