# The Basie Model

The model was constructed using a succession of algorithms to simulate how human offspring might process crude ecological information to derive an adaptive growth strategy (Box 2). In order to express our results in a way that facilitates comparison with empirical birth weight data, we assigned an average value for *B* of 3 kg (in other words, we treat this value as a population average that is present now because it maximized fitness in the recent past). The average value for *R* over the entire period was 79.7 mm; hence, we converted all R-data entered into the model into output B-data by dividing by (79.7/3), or 26.6.

Box 2: Algorithms Tested Using the Model

The basic model |
Matching birth weight to rainfall in the last month |

Simple smoothing |
Extracting a smoothed signal from rainfall using a rolling average |

Lengthening pregnancy |
Increasing the time period over which the rolled average is collected |

Minimal information processing |
Weighting birth weight toward a fixed value and reducing the contribution of the rolling average |

Generational effects |
Collecting rolled averages from the plastic periods of two or more successive generations |

Variability in *B* can be crudely assessed using the coefficient of variation (CV). Although a large CV provides one indication of the risk of excessive or insufficient birth weight values, it is also helpful to see how this risk is distributed. We therefore calculated “fitness penalties,” as the difference between the actual *B* value and 3 kg. The larger the difference, the more birth weight was inadequate or excessive. These fitness values could be expressed in absolute values or squared to make positive and negative penalties equivalent.

Using this model, we considered a range of information-processing algorithms whereby growth could be calibrated to environmental conditions so as to reduce fitness penalties.