Date of Degree


Document Type


Degree Name



Cognitive Neuroscience


Elizabeth Chua

Subject Categories

Cognitive Neuroscience | Cognitive Psychology


Metamemory, Arousal, Bayesian, Uncertainty


Successful recognition often depends on probabilistic estimation of memory-signal. Arousal has been shown to offset the influence of heuristic evidence on memory prediction. We conducted three experiments that tested whether arousal also suppresses predictions relevant to memory confidence. Experiments employed associative face-name memory tasks that included retrospective (Experiments 1 and 2) or concurrent (Experiment 3) judgements of confidence. During test, subjects were presented with a masked-affective face on a subset of trials. By timing the masked-affective face to precede a recognition judgement (Experiment 1), we replicated the finding that unexpected arousal offsets the influence of heuristic evidence on expectations of memory. However, by delaying the masked-affective cue to the post-decisional epoch (Experiment 2) we observed that subliminal arousal suppressed retrospective estimates of decision-making confidence. Critically, the effect was shown to be specific to confidence as it was not confounded by differences in first-order memory-strength. Furthermore, since the only difference between Experiments 1 and 2 was the timing of the masked-affective cue, we demonstrated that unexpected arousal was attributed to either memory or confidence depending on its timing with respect to response-selection. Finally, by timing the masked-affective cue to precede an integrated judgement of recognition and confidence, we observed that subliminal arousal also modulated concurrent estimates of decision-confidence (Experiment 3). However, compared to its effect on retrospective confidence, subliminal arousal was shown to inflate rather than suppress predictions of decision-confidence. This difference likely reflects the extent to which internal performance-monitoring systems shift in bias over the course of decision-making. Data from all experiments help to extend Bayesian and embodied models of cognition into the domain of memory metacognition, or metamemory.