Table 1. Artificial neural networks with lowest E_rms for each trial.

Trial	1	2	3	4
No. units in 1st hidden layer	8	11	12	6
No. units in 2nd hidden layer	11	7	10	10
Erms ± s.e.	0.085±0.002	0.083±0.002	0.097±0.003	0.103±0.004

Searching the space of possible topologies seen in Figure 3B, the most accurate decoding ANNs had different topologies for each trial.