Probability, mean, and variance of model parameters, derived from Bayesian parameter averaging
| Probability | Mean | Variance | |
|---|---|---|---|
| Bottom-up model (projectors, N = 10) | |||
| Modulatory connectivity LSA→V4 | 1.000 | 0.190 | 0.0011 |
| Modulatory connectivity V4→SPL | 1.000 | 0.358 | 0.0020 |
| Fixed connection LSA→V4 | 1.000 | 0.194 | 0.0011 |
| Fixed connection V4→SPL | 0.997 | 0.102 | 0.0014 |
| Top-down model (associators, N = 5) | |||
| Modulatory connectivity LSA→SPL | 1.000 | 0.360 | 0.0033 |
| Modulatory connectivity SPL→V4 | 1.000 | 0.362 | 0.0075 |
| Fixed connection LSA→SPL | 1.000 | 0.427 | 0.0030 |
| Fixed connection SPL→V4 | 0.910 | 0.116 | 0.0074 |
| Fixed connection V4→SPL | 0.992 | −0.178 | 0.0055 |
Table 1 lists the outcomes of BPA for all parameters of the DCMs, for the winning model within each group. See also Figure 1, C and E.