Table 1.
Model parameters. Light gray shading indicates that the parameter was fixed to a set value and not optimized during fitting. All times (i.e., time constants, latencies) are given in ms. Negative latencies for tVAOn mean that the voluntary control signal started before the target.
| Parameter | Description | Value | 95% CI |
|---|---|---|---|
| All layers | |||
| n | exponent | 1.5 | -- |
| Sensory layer 1 | |||
| τ S1 | time constant | 52 | [49 116] |
| σ S1 | semi-saturation constant | 1.4 | [1.2 2.0] |
| Sensory layer 2 | |||
| τ S2 | time constant | 100 | [69 120] |
| σ S2 | semi-saturation constant | 0.1 | -- |
| Decision layer | |||
| τ D | time constant | 1e5 | -- |
| σ D | semi-saturation constant | 0.7 | -- |
| Voluntary attention layer | |||
| τ VA | time constant | 50 | -- |
| σA (shared with IA) | semi-saturation constant | 20 | -- |
| b VA | amplitude of voluntary gain modulation | 40 | [21 50] |
| t VAOn | latency of voluntary control signal onset | −34 | [−223 −6] |
| t VADur | duration of voluntary control signal | 124 | [99 374] |
| t R | recovery time of voluntary gain | 918 | [600 1091] |
| w N | weight to treat neutral precue more like precue T1 (1) or precue T2 (0) | 0.28 | [0.01 0.53] |
| Involuntary attention layer | |||
| τ IA | time constant | 2 | -- |
| σA (shared with VA) | semi-saturation constant | 20 | -- |
| b IA | amplitude of involuntary gain modulation | 8.5 | [0.8 27.9] |
| h IA :p | shape parameter for involuntary temporal prefilter | 2.2 | [0.04 49.9] |
| h IA :q | scaling parameter for involuntary temporal prefilter | 0.023 | [0.01 0.09] |
| Fitting | |||
| s T1 | scaling constant to relate model output to d’ for T1 | 1 | -- |
| s T2 | scaling constant to relate model output to d’ for T2 | 0.80 | [0.77 0.84] |
| Number of parameters | |||
| Total | 20 | -- | |
| Fitted | 12 | -- | |