1. The objective was to examine how opposite learned behavioral responses to the same physical tone were differentiated by the pattern of interactions between extraauditory neural regions. This was pursued using a new approach combining behavior, neuroimaging, and network analysis to integrate information about differences in regional activity with differences in the covariance relationships between brain areas. 2. A tone was used as either a Pavlovian conditioned excitor or inhibitor. Rats were conditioned with reinforced trials of a conditioned excitor (A+) intermixed with nonreinforced trials of a tone-light compound (AX-). The tone was the excitor (A+) for the tone-excitor group and was the inhibitor (X-) for the tone-inhibitor group. After conditioning, all rats were injected with [14C(U)]2-fluoro-2-deoxyglucose (FDG) and presented with the same tone. 3. FDG autoradiography was used to measure regional activity and to generate interregional correlations of activity resulting from the presentation of the tone. A stepwise discriminant analysis was used to select brain regions that differentiated the excitor from the inhibitor effects. 4. Network analysis consisted of constructing an anatomic model of the brain regions, selected by the discriminant analysis, linking the regions with their known anatomical connections. Then, functional models for the tone-excitor and -inhibitor groups were constructed using structural equation modeling. Correlations of activity between regions were decomposed to calculate numerical weights, or path coefficients, for each anatomic path. These path coefficients were used to compare the interactions for the tone-excitor and -inhibitor models. 5. Regional differences in FDG uptake were found in the sulcal frontal cortex (SFC), lateral septum (LS), medial septum/diagonal band (MS/DB), retrosplenial cortex (RS), and dentate-interpositus nuclei of the cerebellum (DEN). Discriminant analysis selected three other regions that significantly discriminated the tone-excitor and -inhibitor groups: perirhinal cortex (PRh), nucleus accumbens (ACB), and the anteroventral nucleus of the thalamus (AVN). 6. Structural equation modeling identified two functional circuits that differentiated the groups. One involved the basal forebrain regions (LS, MS/DB, ACB) and the other limbic thalamocortical structures (SFC, RS, PRh, AVN). Differences in the interactions within these circuits were mainly in sign of the covariance relationships between regions, from positive for the tone-excitor model to negative path coefficients for the tone-inhibitor model. The path coefficient between the basal forebrain circuit and the limbic thalamocortical circuit showed the largest magnitude difference. This quantitative difference was mediated by a path from the MS/DB to PRh.(ABSTRACT TRUNCATED AT 400 WORDS)
A Theoretical Investigation of the Relationship between Structural Equation Modeling and Partial Correlation in Functional MRI Effective Connectivity