Entry Date:
January 25, 2017

Think - Inferring Cognitive State From Subtle Behaviors

Principal Investigator Randall Davis

Project Start Date September 2014

Project End Date
 August 2017


Cognitive disorders are a key health problem affecting millions of individuals, a problem that is increasing in importance as America's population ages. Just a single dementia, Alzheimer's disease, is estimated to afflict some 5.4 million Americans in 2012, projected to rise to 11-16 million by 2050, with the potential for staggering cost both in dollars and human costs for the afflicted and their caregivers. Yet cognitive health is not monitored effectively in the way that physical conditions and wellness are: physician visits routinely include screens for heart disease, diabetes, etc., but there is far less consistent screening for cognitive status. This project seeks to develop rapid, precise, and objective cognitive assessment tools that will provide operationally defined, quantitative metrics with sufficient sensitivity and specificity to aid in early detection and diagnosis of neurodegenerative disorders, as well as monitoring disease process and treatment efficacy. Its evidence-based approach offers the possibility of earlier treatment, which can slow the progress of cognitive disease, resulting in better quality of life.

The first research objective is to develop two new digital versions of two traditional tests. The digital Maze Completion (dMC) assesses frontal brain system associated cognitive functions by providing measures of tiered decision making and executive planning, and calibration of memory function by measuring the effect of priming. The digital Symbol-Digit Test (dSDT) assesses executive function by measuring information processing speed, and calibrates planning and cognitive shifting by detecting and measuring subtle behaviors. Both tests provide new measures of learning, assessing reaction time changes and short term memory retention. The second objective is to develop the algorithms required to detect subtle behaviors during the dMC and dSDT tests, which can serve as a platform to which additional tests can easily be added. The third objective is to make all the tests as self-scoring as possible, through advances in signal interpretation. The fourth objective involves administering the dCDT, dMC and dSDT to both healthy adults and subjects with known cognitive impairments, permitting the calibration of the power and value of the tests. The final objective is to establish the validity of selected measures provided by the tests, by comparing the measures to standard cognitive test results, to specific neuroanatomical structures, and to external and subcortical gray matter integrity relative to white matter integrity. These research activities are integrated with education through projects in courses, engagement of graduate and undergraduate students in the research, and annual workshops for students.