Today's goal: definition of model and list of valid uses of models
List of models:
-
Hardy-Weinberg model (Ted)
-
IN: how many parents with various genotypes
-
OUT: fraction of offspring with various genotypes
-
parents mate randomly, so offspring genotypes are figured out by laws of probability.
-
ASSUMPTIONS: no mutation, no selection -- no evolution
-
UNREALISTIC: because some of the assumptions (like infinite pop size) are never met
-
Schrodinger's cat model (Holland)
-
SET-UP: there's radioactive decay, or not. Radioactive decay kills cat.
-
PREDICTION: cat has various quantum superpositions.
-
MODEL is bunch of equations about superpositions of states
-
UNREALISTIC because the eqns require mathematically tractable numbers of particles
-
Standard Model of Particle Physics (Hannah)
-
Can be tested, but not well
-
Descriptive model
-
Two overlapping worlds, joined by EM force
-
Quantum world: there are quarks, which (via Strong Force) make up larger objects, including matter.
-
Classical world: Force is gravity
-
PREDICTIONS: properties of particles seen in detectors
-
FICO credit-rating formula (Irina)
-
IN: payment history, level of debt, no. of credit cards, etc.
-
PREDICTION: estimate of a potential borrower's likelihood to repay a debt on time
-
UNREALISTIC: doesn't account for temporary things
-
Stick-and-ball theory of molecular structure (Carolyn)
-
Consists of interatomic distances, atomic sizes, angles
-
Allows prediction ofoverall shapes and something about interactions
-
UNREALISTIC: because ignores orbitals and motion
-
Central Dogma (Cindy)
-
information flows from DNA --> RNA --> proteins
-
PREDICTIONS: if you take a nucleus and put it into an enucleated cell, the resulting cell will look like the cell that originally had that nucleus.
-
falsified (by existence of RNA viruses
-
But for the most part, still useful
-
Computer simulations of HIV epidemiology (Arun)
-
IN: popn growth rates, poverty, transmission rates, mortality rates
-
OUT: PREDICTIONS of HIV infection rates in future
-
UNREALISTIC because ignores wars, unknown factors
-
Linear-programming model of behavior in games (Ananya)
-
ASSUMES: ability to estimate expected payoffs, rational behvior of players
-
PREDICTS game decisions of players based on rules and payoffs
-
UNREALISTIC because sometimes players have incomplete knowledge or are irrational
-
Optimal foraging theory (Danni)
-
IN: costs and benefits of foraging activities
-
PREDICTIONS: probability and deration of foraging
-
UNREALISTIC because ignores idiosyncratic behavior among indivs or populations
Characteristics of models
-
Reductive: take complicated reality and describe in terms of several or few actors or variables
-
Models make simplifying assumptions and therefore can't always be trusted
-
Models make good predictions often enough to be useful
-
Allow for predictions, which can be tested
-
sometimes quantitative
-
sometimes qualitative
-
Good models are consistent with mother good models, and often, incomplete models work together to form a complete picture
-
Because they're testable and subjected to attempts at falsification, good models agree with our observations of the world
Uses of models
-
Null models allow for detection of assumptions' violations
-
Some models make quantitative predictions (e.g., HIV sim, credit-rating model)
-
Models can sometimes be used in lieu of impossible or immoral experiments (e.g., virtual dissection, nuclear-explosions)
-
Models are useful as articulations of theory
-
Some models are fun to watch and play with (e.g., Sim City)
