Growth and evolution of category fluency network graphs

Rajeet Shrestha,Thomas Fritsch,Barclay Shaw,Alan J Lerner,Peter J Thomas,Wojbor Woyczynski

doi:10.15761/jsin.1000103

Rajeet Shrestha, Thomas Fritsch + Show 4 more

Open Access

https://doi.org/10.15761/jsin.1000103

Copy DOI

Abstract

Background: Category fluency is a sensitive measure of cognitive integrity and is known to involve both frontal and temporal cortical areas. Network graph analysis is a technique used to analyze relationships between nodes and edges and calculate metrics such as path lengths between nodes and clustering coefficients. Objectives: To investigate network growth and preferential attachment in a network model of category fluency. Method: Category fluency results (“animals” recorded over 60 seconds) from subjects (N=374) contacted via telephone were converted to undirected network graphs of all unique neighbors and network parameters were calculated. Growth was also modeled using an extended cognitive network model. Random subsamples of people or of node pairs were used to model network growth and study preferential attachment. Results: The final network had 275 nodes and 2035 edges. The network showed scale free and small world properties, which change with network size. Both methods of modeling connectivity showed exponential growth of nodes and edges as increasing fractions of the complete network were sampled. Preferential attachment was demonstrated by using Newman’s method. Conclusions: Network growth patterns show a sharp transition to scale free and small world properties with early network growth. Networks based on category fluency show preferential attachment and appear to be a valid model for studying network dynamics based on cognitive output.

Highlights

Complex networks consisting of nodes and edges have received increasing attention as a means of understanding complex relationships in the natural world
Networks based on category fluency show preferential attachment and appear to be a valid model for studying network dynamics based on cognitive output
We examine the emergence of scale free properties with network growth and whether preferential attachment occurs in category fluency networks, in accordance with the Barabasi-Albert (BA) model [1]

Summary

Introduction

Complex networks consisting of nodes and edges have received increasing attention as a means of understanding complex relationships in the natural world. A form of verbal fluency, are a standard part of neuropsychological testing, and verbal fluency may be tested either as lexical fluency (such as words beginning with the letter “F”) or semantic fluency, such as naming of animals, vegetables or tools. Verbal fluency output reflects the accessibility of “semantic space,” representations of which are closely aligned with our understanding of how semantic information is organized at the neuronal level, in left hemisphere structures, and how this is modified in those at risk for dementia, with Mild Cognitive Impairment (MCI), or mild to moderate Alzheimer’s disease [8,9,10,11,12,13]. Network graph analysis is a technique used to analyze relationships between nodes and edges and calculate metrics such as path lengths between nodes and clustering coefficients

Objectives

Methods

Results

Conclusion