Neural network interactions related to auditory learning analyzed with structural equation modeling

Abstract

Abstract:If learning is an emergent property of interacting brain regions, understanding it requires a network analysis of the patterns of interaction between brain regions. Brain mapping techniques have the potential of providing information about functional interactions within entire neural systems, but computational methods are needed to make sense of the complex interactions that take place in the brain. This review paper illustrates, with examples from auditory learning studies, the application of a computational method that quantifies the interactions among neural regions in terms of networks, and what is gained from this approach. The method is Structural Equation Modeling, and it computes network interactions by combining anatomical circuitry with the covariation in the activity between auditory structures. Activity was assessed by fluorodeoxyglucose uptake, and the functional strengths of auditory pathways were quantified by computing path coefficients representing the strength of the functional influence through each anatomical path. Changes in these values were used as indices of how learned information was processed and modified within the auditory system. Structural models of the auditory system revealed the patterns of network interactions related to habituation to a sound, as well as the patterns related to the opposite learned associative properties of the same sound. They illustrated how the functional interactions among parallel auditory neural pathways were modified as a result of auditory learning. This analysis supports the hypothesis that auditory learning is an emergent network property, distributed among connected brain regions showing specific patterns of interaction.

Changes in regional activity are accompanied with changes in inter-regional connectivity during 4 weeks motor learning

“…A complete SEM model can be decomposed into a measurement model and a structure model (Gonzalez-Lima and McIntosh, 1994). The measurement model is expressed as x=LX+E.…”

Network activation during bimanual movements in humans

“…Generation of the structural equations: The interregional correlations of activity were used to assign initial numerical weights to the connections (path coefficients) in the anatomical model, leading to the functional model (Gonzalez-Lima and McIntosh 1994; McIntosh 1999; McIntosh and Gonzalez-Lima 1994). …”

Cocaine reduces cytochrome oxidase activity in the prefrontal cortex and modifies its functional connectivity with brainstem nuclei

“…These two statistical approaches have been embraced by functional neuroimaging studies because they are complementary, as the nervous system can be viewed as a complex network of interacting individual regions (Gonzalez-Lima and McIntosh, 1994). The analysis of regional means of cytochrome oxidase activity focuses on how the baseline metabolic activity of specific brain regions change after repeated cocaine, whereas the analysis of functional connectivity focuses on how baseline interactions among brain regions change after repeated cocaine.…”

“…4). But inter-regional correlations do not provide information on the direction of the influence, so they need to be interpreted based on other known anatomical and functional data (Gonzalez-Lima and McIntosh, 1994). The noradrenergic system contributes to control stress responses, arousal, mood and alters learning and memory (Huether, 1996; Sved et al, 2001) and plays a key role in mediating reward (Poschel and Ninteman, 1963; Stein, 1964, 1975; Wise, 1978).…”