“…In a similar way, several researchers in neurobiology are using evolutionary algorithms, either for setting synaptic weights of neural network models (Eurich, Roth, Schwegler, & Wiggers, 1995;Eurich, Schwegler, & NVoesler, 1997), or for determining conductance parameters of biophysical models of single neurons (West & Wilcox, 1997). From a general point of view, note that, although the GA can be very useful for demonstrating that a model can produce a specific behavior (by finding efficient sets of unknown variables), it is less useful for invalidating a hypothetical model, as an inability to find successful variable instantiations may be due to failings of the model or to problems with the GA set up, or both.…”