Conjunctive Representations in Dynamic Action Control
Loading...
Date
2020-09-24
Authors
Kikumoto, Atsushi
Journal Title
Journal ISSN
Volume Title
Publisher
University of Oregon
Abstract
In the present work, I examine the functional role of highly integrated, conjunctive representations of basic task features during dynamic action control. People can use abstract action rules to flexibly configure and select actions for specific situations. Yet, how exactly rules shape actions towards specific sensory and/or motor requirements remains unclear. One theoretical possibility is that rules become integrated with sensory/response features in a nonlinear, conjunctive manner during action selection (i.e., event files). Such conjunctive representations in turn are a precursor of successful action. To test this hypothesis, it is necessary to dynamically track neural representation of multiple action features that become active concurrently during action selection. We applied multivariate decoding analysis to the time-resolved EEG signal at the level of single-trials, while participants selected actions based on varying action rules. In Chapter II, we provide initial evidence that conjunctive representations can be tracked during action selection. Specifically, we show that these representations emerged throughout the entire response selection period and that they were robust and unique predictors of the variability in trial-to-trial performance. Moreover, they were related to a theoretically important, behavioral indicator of event files—the partial-overlap priming effects. In Chapter III, we tested how conjunctive representations contribute to stopping of planned actions. Because the formation of conjunctive representations is theorized as a necessary stage of successful actions, we hypothesized conjunctions should be the primary target of the stopping-related activity. Indeed, using the stop-signal paradigm, we found (a) that conjunctions were selectively suppressed on stop trials, and (b) that stopping became particularly difficult when the conjunctive representations of to-be-stopped actions prior to the stop-signal. In Chapter IV, we discuss the implications of the findings and propose ideas for future directions. Specifically, I propose further experiments to characterize key processes or properties associated with conjunctive representations, such how they are actively maintained in working memory, how they are functionally related to the dimensionality of neural responses, or how effects of actions are integrated into such representations. The work presented in this dissertation confirms that conjunctive representations are functionally independent of the constituent features and play a critical role in action control. It also provides broad insights to how we can study cognitive control functions in humans by directly decoding goal-relevant information.
Description
Keywords
action control, conjunctive representations, EEG, event files