AUDREYSERNA
PhD in Cognitive Science and Computer Science
Post-Doctorat in HCI and Ergonomics

012 Cognitive modeling

Executive processes model

Background

Smart homes are technological environments aimed to support people who are loosing their autonomy, especially persons suffering from cognitive and executive disorders (older people and Alzheimer patients). To improve their efficiency and their responsiveness, cognitive remediation and assistive systems should be based on a better understanding of cognitive dysfunctions and their impact on people’s daily living. Observation and modelling of executive processes allows mechanisms involved in activities of daily living (ADL) performance to be observed and to be better understood.

Methodology

This thesis project is aimed to study executive processes, especially executive control, functions of the cognitive impairment due to aging and dementia. A three steps approach has been applied:

  1. experiment to observe executive mechanisms;
  2. analysis to qualify these mechanisms and their impairment; and
  3. theoretical and computational model of these results within an existent neuropsychological theory (Norman and Shallice’s model).

The experiment has been realized on three populations: young subjects, older subjects and MCI-AD subjects (Mild Cognitive Impairments and Alzheimer’s disease). Subjects were asked to perform an ADL, which was perturbed in order to observe and to qualify their abilities to control their actions, to adapt or correct their behaviour when abnormal or unexpected situations occur.

Results

Observations show an impairment of executive processes functions of cognitive impairment. Older people and MCI-AD people show difficulties to change their strategy and their plan. They need help from the experimenter and more time than young subjects to complete the task. With these observations, planning and control processes have been specified to elaborate a model of executive processes.

Conclusion

Such a model, with a computational implementation and a simulation part, could predict the most frequently made errors for specific ADL and could be used to improve the design of smart homes and cognitive assistance systems.