- Title
- Quantifying contextual information for cognitive control
- Creator
- Barceló, Francisco; Cooper, Patrick S.
- Relation
- Frontiers in Psychology Vol. 9, Issue 10 September 2018, no. 1693
- Publisher Link
- http://dx.doi.org/10.3389/fpsyg.2018.01693
- Publisher
- Frontiers Research Foundation
- Resource Type
- journal article
- Date
- 2018
- Description
- In the tradition of serial processing models that link the stimuli, the organism and its responses (S → O → R) (Hull, 1943), cognitive control has long been defined from stimulus-centric views implicit in simple “target detection” tasks (e.g., flanker and Stroop-type tasks) (Posner and Petersen, 1990). According to these views, cognitive control is invoked for processing (i.e., perceiving, attending, memorizing) and, ultimately responding to, a task-relevant set of stimulus features. These task-relevant features occur in close spatiotemporal proximity to competing non-targets and distracters that need not be fully processed or responded to. As a consequence, cognitive demands are normally defined as a function of such sets of relevant and irrelevant stimulus features. These features compete for limited central resources at the cost of less efficient goal-directed behavior. However, in such stimulus-centric views, the actions themselves are largely disregarded. For example, a button press is simply considered the final output of stimulus-feature competition with no or little influence on earlier information processes. Such stimulus-centric views permeate interpretation of other executive function tasks like the Wisconsin card sorting test (WCST), and task-switching paradigms, inasmuch as these more complex tasks share common features with simpler target detection paradigms. Indeed, in all these tasks, targeted stimulus features appear within the spatiotemporal context of other nontarget features that may afford alternative and potentially conflicting courses of action. Here we embrace current enactivist views of cognition to propose an alternative framework to probe cognitive control. To do so, we contend that (1) cognitive control is essentially context-sensitive, and hence (2) cognitive demands should be defined and quantified in terms of contextual information or uncertainty. We suggest shifting the focus of context to specifically capture the uncertainty about upcoming actions, which may be represented in the brain through hierarchies of sensorimotor loops that evolve dynamically over time.
- Subject
- information theory; Bayesian inference; context-updating; context-learning; frontoparietal cortical dynamics; P300
- Identifier
- http://hdl.handle.net/1959.13/1440918
- Identifier
- uon:41262
- Identifier
- ISSN:1664-1078
- Language
- eng
- Reviewed
- Hits: 1036
- Visitors: 1036
- Downloads: 0