Skip to main content
. 2020 Mar 20;14:218. doi: 10.3389/fnins.2020.00218

FIGURE 7.

FIGURE 7

Analysis of reduced models and parameter sweeps. (A) The full model (CSM) included three components: (1) a concentration-dependent velocity term (V), (2) a concentration-dependent casting term (C); and, (3) a binaral-olfaction term (B). Removal of any of these components (–V), (–C), or (–B) may, in principle, reduce model performance. (B) Comparison of noise effects on CSM success rate shows that the CSM model is robust for a wide array of different noise conditions, but fails if random noise approaches 100% or Gaussian filtering approaches 0 mm. (C) We added a coefficient in front of the hyperbolic tangent term in Eq. 5 and set it equal to 3. With this additional term, (–B) had a larger effect on performance; however, the effect of removing (–V) or (–C) terms decreased.