This panel will provide you access to female preference models.
First, a text box let you tune the strength of the preference. A value between -100 and 100 is required here. However, as we will see further on, this strength is dependent on the preference model. A general idea though is that it is the slope of a logit-like function that relate the phenotype of the male to the probability that a female picks up this male. Positive values tend to increase preference for MED phenotypes, whereas negative values increase preference for ATL phenotype. But this is subject to nuances depending on the preference model (see below).
Second, the user has to choose is the preference models apply to the whole range of phenotypic variation (which is [0;1] with 0 being the ATL pure phenotype and 1 the MED pure phenotype). This is an important choice: is the probability that a female picks up a male just dependent on the male phenotype on an universal gradient ([0;1]) or is this probability affected by what is available for the female (or else said, the phenotype range available in the OSR). In the first case, a female will not be very choosy between two males having close phenotypes (say 0.4 and 0.5). In the second case, if she can only sample these two males, then they will represent the boundaries of her preference expression, and she will be very choosy. The absolute preference is kind of fixed whatever the context, whereas a relative preference is highly context dependent.
Here, we are let with three general options. For each choice of preference model, as well as each value of preference strength, a synthetic graph is provided on the right: it pictures a preview of the preference curve.
A/ The first option assumes a female phenotype independent preference: female express a preference independently of who they are. The preference curve is therefore universal in the population. Basically, a logit model applies:
logit(preference) = preference strengh * (male phenotype - 0.5)
This ensures that preference for intermediates phenotypes will be 0.5, and that all other values are contained between 0 and 1. If preference strength is set to 0, then the curve will be flat and all phenotypes will have the same probability to be picked up. Values above zero will increase preference for MED phenotypes, and vice versa. However, there is a catch, if not several here: let enter a value of 2 or 3 for the preference strength, and you will see a relatively linear relationship where the female can discriminate between any phenotype. But if you enter a higher value, say 50, then the curve will present a steep threshold: females will be very efficient to discriminate between pure ATL or pure MED phenotype, but they won't be able to discriminate between phenotypes that would be all located on one side of the threshold. This mathematical approach of preference curve has some limit, the user should be careful of them.
B/ The second option accounts for the phenotype of the female. In fact, it ponders the approach described above by flattening the curve depending on the female phenotype. Three sub-options are available that illustrates the approach:
C/ The third option is a model not directly based on the value of female phenotype, but that simulates a preference behaviour based on similarity: if the preference strength value is positive, females will prefer males having a different phenotype (heterogamy), whereas if the value is negative, they will prefer males displaying a phenotype close to their own (homogamy). This system is acceptable for pure breed, but there is a problem with hybrids as shown by the curves: they tend to either be very choosy or not at all, irrespective of male's phenotype.