<p>A) Comparison of model variants testing cueing vs. integration hypotheses.</p>

The Fully Shared Model (bottom) uses identical priors () and likelihoods (). The arrows illustrate the computational flow: prior and likelihood are combined via integration () to produce a posterior distribution ().

B) 2D Bayesian inference showing how spatial localization (posterior; green) emerges from combining prior expectations (blue) with sensory evidence (likelihood; red). As can be seen, the presence of a prior leads to a localization error (i.e., offset between likelihood and posterior).</p>