Condition definition in spike_analysis is rather flexible. keys.condition_parameters defines your unique conditions (as in all unique combinations of these parameters. Of course, the parameter needs to be present in the data (see …). Additionally, you will have to define labels for each parameter value, and set colors for each condition in your settings.
Lables define a (ideally two characters long) string for each parameter value. At the moment, ph_get_condition_matrix takes care of correct assignments of values and labels, should actually just be a second input.