If history is a guide, in the area of emotion, the patterns that appear to reliably distinguish emotion categories may vary from study to study (e.g., [24] vs. [25] ), making it difficult to identify generalizable models of specific emtion types.

For instance, Activation Likelihood Estimation (ALE), multilevel kernel density analysis (MKDA), and co-activation approaches are 1) not generative models of the emotion, and 2) not multivariate in brain space.

Because they are not generative models , standard analyses provide only descriptive, summary maps of activity or bivariate co-activation for different psychological states.

The generative model concerns the process by Which emotional instances of a given category produce observed peak activation foci, and the likelihood With Which they do so.

Because it is a generative model , the BSPP model of emotion categories is capable of making predictions about new instances.

Other methods—such as our previous MKDA analyses, ALE analyses, and bivariate co-activation analyses that we and others have developed—are not generative models , and would not be expected to be appropriate to or perform well in classification.

While our multi-step voltage clamp protocol alone is very useful for estimating many of the parameter values (Fig 4B), it tends to generate models that fail to predict novel stochastic pacing data well (Fig 4C), which is unsurprising given that it does not train the models according to membrane potential.

Furthermore, because our model enables validation on data from the same cell for which a model was optimized, we were able to demonstrate that the cell-specific models are markedly better at predicting the response to novel stimulation sequences than was the generic model .

Generic models have the advantage that direct comparisons can be made among different simulation studies.

However, when comparing a generic model such as the out-of-the-box FR model to our experimental data, there are substantial differences, which likely would cause inaccurate predictions if simulating, e.g., effects of pharmacological agents or genetic variations.

Whole-cell optimization approaches have focused on generating models that can match action potentials from different types of cardiomyocytes, using both simple models consisting of a few generic currents [13,18—20] and more physiologically detailed ionic models [21—25].

Considered together with the demonstration that the approach accurately identifies model parameters (Figs 2—4), these findings suggest that the approach significantly improves the fidelity of the model for cellular data, relative to the published generic model .

We confirm analytical results with computational simulations using general model networks and anatomical brain networks, as well as high-density electroencephalography collected from humans in the conscious and anesthetized states.

Our mathematical analysis allowed us to predict the directionality patterns in general model networks as well as human brain networks across different states of consciousness.

One can gain general insights about the behavior of more complex interacting oscillator models by analyzing such generalized models .

In this respect the Stuart-Landau model can be considered as the generalized model of the Kur-amoto model, with the inclusion of the amplitude equation.