After completion of this work, we became aware of a paper by Yoshiyuki Sato, Taro Toyoizumi and Kazuyuki Aihara, who independently developed a similar model (Neural Computation, in press).
http://plosone.org/article/info:doi/10.1371/journal.pone.0000943#article1.back1.ack1.p1
This excellent paper has just appeared in NC. I highly recommend reading this paper as it exposes many additional issues and shows the strength of the causal inference approach in partially overlapping and partially non-overlapping areas.
I have done some informal comparisons of their results and ours and indeed the model predictions are very similar.
Sato, Y., Toyoizumi, T., & Aihara, K. (2007) Bayesian inference
explains perception of unity and ventriloquism aftereffect:
identification of common sources of audiovisual stimuli. Neural
Computation, 19 (12), 3335-3355.