[Left] Networks of coupled neural masses are fitted to 1 minute worth of data recorded from 4 minutes after the light exposure (to capture steady state responses). These fitted models capture key differences between the phenotypes.

[Right and Center Bottom] To assess whether the main deviations of the two mutant genotypes from wildtypes between conditions can be explained with subsets of parameters, we compared combinations of models across two model families - one (right) where parameters only vary in restricted parts of the anatomy; and one (bottom) where only some aspects of local inhibition are altered by the mutations.

[Center Top] This graph shows the Bayesian model comparison across reduced models that best explain the effects of mutation on steady state power spectra for the two mutations separately.
For gabra1 -/-, Bayesian model comparison indicates, that changes across all anatomical brain regions are required, but that those changes can be restricted to alterations in between-population inhibitory coupling. For gabrg2 -/-, on the other hand, changes in all types of inhibitory coupling parameters are required, but these can be restricted to only the brain stem nodes.
These findings indicate that there are differential impairments in inhibitory neurotransmission between fish, which we will explore further in future studies.