Age vs Alzheimer's: a computational model of changes in representation

Area of Science:

• Cognitive Neuroscience
• Psycholinguistics
• Computational Linguistics

Background:

• Previous research indicates that older adults' language yields denser semantic representations than younger adults' language.
• Density in the Hyperspace Analogue to Language (HAL) model may serve as a metric for memory comparison across age groups.

Purpose of the Study:

• To investigate the role of semantic representation density in adults with Alzheimer's disease.
• To compare semantic density in Alzheimer's patients with age-matched normal controls.

Main Methods:

• Analysis of semantic representations derived from spoken language.
• Utilizing the Hyperspace Analogue to Language (HAL) model to quantify representation density.
• Comparison of density metrics between younger adults, older adults, and adults with Alzheimer's disease.

Main Results:

• Adults with Alzheimer's disease exhibit denser semantic representations than normal older adults.
• Consistent with prior findings, older adults show denser representations than younger adults.
• Increased density in semantic space correlates with impaired retrieval in older adults and Alzheimer's patients.

Conclusions:

• Semantic representation density is a potential biomarker for cognitive decline in aging and Alzheimer's disease.
• Excessive density in semantic space may indicate a "saturation point" impacting memory retrieval.
• The HAL model's density metric offers insights into age-related memory changes and Alzheimer's pathology.