Cognitive Control

Cognitive Control in Humans

The human brain contains a vast amount of information such as perceptual, semantic, motoric, etc. During task execution, only a small set of these resources are needed. Human cognitive skills involve the ability to organize resources necessary for the task at hand, including selection and maintenance of required information and avoidance of disruption from other influences [1]. Disruptions can come in the form of sensorimotor-type actions which tend to be triggered in response to certain stimuli regardless of current goals. In order to accomplish particular goals, execution of actions must be controlled according to the context. Actions are selected based on information obtained from past experiences and knowledge about the task and environment. In some cases, sensorimotor coordination must be either inhibited or overridden. The ability to control actions in this manner to accomplish a task is called cognitive control [1][2][3][4][5]. Effective cognitive control requires a subtle manipulation of task switching that prevents disruption of ongoing tasks while providing flexibility to allow rapid execution of other tasks that arise in a given situation [6]. The concept of cognitive control is illustrated in the following figure.

According to Miller, executive functions of cognitive control include the ability of the brain to [4]:

  • Generate plans and monitor task progress
  • Focus on task related information
  • Maintain and update goal information
  • Inhibit distractions to current tasks
  • Shift between different levels of cognition ranging from routine actions to complex deliberation
  • Learn new responses in novel situations.
Cognitive Control in Robotics

In robotics, cognitive control is expected to help the robot intelligently and robustly select actions based on attention, emotion, and past experience that are suitable for the current situation. It gives flexibility to the robot to reason and act according to stimuli and under conflicting goals in dynamic environment. Generally, a fast response is preferred during a task execution which could be achieved if reactive sensorimotor actions are allowed, i.e. actions are carried out instantaneously upon sensing of stimuli. However, there are situations where this type of action is not suitable to the task execution since it distracts the robot from the task in hand. Cognitive control, together with attention, emotion, and past experience, will help the robot to exhibit robust sensorimotor intelligence to cope with these situations.

ISAC’s cognitive control is modeled and implemented based on Baddeley’s psychological human working memory model [7]. His working memory model consists of the “central executive,” which controls two working memory systems namely the phonological loop and the visuo-spatial sketchpad. Cognitive control in ISAC is implemented using the Central Executive Agent (CEA) which interfaces with the Working Memory System (WMS). Functions of the CEA include deciding which action to execute and monitoring the state of the system. In the current design, the CEA responds to the Focus of Attention (FOA) obtained from WMS by selecting and assigning actions. The CEA learns to make decisions using feedback from prior task execution. The operation of the CEA is illustrated in the following figure.

References

1. Monsell S. and J. Driver (eds), Control of Cognitive Processes: Attention and performance XVIII, Cambridge, MA: MIT Press, 2000.

2. Gazzaniga, M.S., R.B. Ivry, and G.R. Mangun, Cognitive Neuroscience: The biology of the mind. NY: Norton, 2002.

3. Baars, B.J., In the Theatre of Consciousness: The workspace of mind, Oxford: Oxford University Press, 1996.

4. Miller, E.K., Cognitive Control, Fundamental of Brain and Mind Lecture Series, Mass. Institute of Technology, June 11-13 2003.

5. Hommel, B., K.R. Ridderinkhof, and J. Theeuwes, “Cognitive control of attention and action: Issues and trends,” Psychological Research, 66, 215-219, 2002.

6. Monsell, S., “Task switching,” Trends in Cognitive Sciences, 7(3), 134-140, March 2003.

7. A. Baddeley, Working Memory, 11, Oxford Psychology Series, Oxford: Clarendon Press, 1986.