Figure 3.

Illustration of the benefits of a strong gravity prior. (A) The upper panel shows simulated tuning curves of neurons specialized for different rates of change of the elevation angle (light gray) together with neurons’ responses to the stimulus value of $\dot{\gamma }$ = 0.797 rad/s (red). The lower panel depicts the likelihood of different responses given this stimulation. (B) Maximum Likelihood Estimate of the stimulus value given the neural responses with a 10% error margin and the combinations of stimulus values that fall into this error margin. (C) Fraction of correct estimates of the horizontal velocity as a function of the gravity prior’s value for a Maximum Likelihood Estimation (thin blue line) and a Bayesian estimation based on the elevation angle (bold yellow line) or its rate of change (bold blue line) as decision variables. The gray area denotes chance level.