Behavioral modeling shows that increased postural variability following lesions emerges from combined impairments to swim timing and corrective capacity. A, A generative model of swimming adapted from previously published work (Ehrlich and Schoppik, 2017b) consists of four computations (tan boxes) determining swim timing, pitch angle, and AV. Each computation contains condition-specific parameters, shown by plots with a green (control) or blue (vestibulospinal lesion) line. B, Variations of the swimming model were created using parameters either entirely from control data (“full control model,” green), lesion data (“full lesion model,” blue), or combinations of parameters from control and lesion data (“hybrid models,” gray). C, Probability distributions of simulated inter-bout pitch angles in full control model (green), bout timing null model (magenta), and full null model (purple) are comparable for modeling data from 4 and 7 dpf, and are comparable to previous models of larval swimming (Ehrlich and Schoppik, 2017b). D, Probability distributions of observed inter-bout pitch angles at 4 and 7 dpf for empirical control data (black) and simulated control fish (green) and E, empirical lesion data (black) and simulated lesion fish (blue). For simulated data in panels C–E, lines are probability distribution across all 30 simulated model runs, shaded error bars ± 1 SD across simulated runs. For empirical data in panels C–E, lines are probability distribution across all observations per condition, shaded error bars represent bootstrapped estimates of 95% confidence intervals. F, Mean standard deviation of 50 simulated pitch probability distributions  (±1 SD) using different combinations of 4 dpf condition-specific parameters or G, with all control parameters except for a single parameter derived from 4 dpf lesion data. Parameters shown here calculate pitch bout correction (m_Θ, b_Θ, σ_ΔΘ, <mml:math display="inline"><mml:msubsup><mml:mi>R</mml:mi><mml:mrow><mml:mi mathvariant="normal">Θ</mml:mi></mml:mrow><mml:mn>2</mml:mn></mml:msubsup></mml:math>RΘ2 ) and AV bout correction (<mml:math display="inline"><mml:msub><mml:mi>m</mml:mi><mml:mrow><mml:mrow><mml:mover><mml:mi mathvariant="normal">Θ</mml:mi><mml:mo>˙</mml:mo></mml:mover></mml:mrow></mml:mrow></mml:msub></mml:math>mΘ˙ , <mml:math display="inline"><mml:msub><mml:mi>b</mml:mi><mml:mrow><mml:mrow><mml:mover><mml:mi mathvariant="normal">Θ</mml:mi><mml:mo>˙</mml:mo></mml:mover></mml:mrow></mml:mrow></mml:msub></mml:math>bΘ˙ , <mml:math display="inline"><mml:msub><mml:mi>σ</mml:mi><mml:mrow><mml:mrow><mml:mover><mml:mi mathvariant="normal">Θ</mml:mi><mml:mo>˙</mml:mo></mml:mover></mml:mrow></mml:mrow></mml:msub></mml:math>σΘ˙ , <mml:math display="inline"><mml:msubsup><mml:mi>R</mml:mi><mml:mrow><mml:mrow><mml:mover><mml:mi mathvariant="normal">Θ</mml:mi><mml:mo>˙</mml:mo></mml:mover></mml:mrow></mml:mrow><mml:mn>2</mml:mn></mml:msubsup></mml:math>RΘ˙2 ) (see Methods, “Behavioral modeling” for additional explanation of parameters). H, Mean standard deviation of 50 simulated pitch probability distributions (±1 SD) using different combinations of 7 dpf condition-specific parameters or I, with all control parameters except for a single parameter derived from 7 dpf lesion data. Asterisks within inset grids represent statistically significant differences between simulated standard deviations in Tukey’s post hoc significance tests, p < 0.05.