Toward a mechanics of adaptive behavior: evolutionary dynamics and matching theory statics.

The team built a computer world filled with digital organisms.

Each organism lived or died based on how well it picked between two levers.

Over thousands of generations the survivors produced the same choice patterns that real pigeons and people show in matching-law experiments.

The model had no built-in math rules; only birth, death, and random mutation.

At steady state the virtual creatures allocated responses exactly like modern matching theory predicts.

The same simple selection process created bias, undermatching, and overmatching without extra assumptions.

One mechanism—evolution by consequences—could explain both the average pattern and the messy deviations seen in live data.

McDowell (2021) took this 2010 engine and dropped it into clinical practice.

He showed how to run the model before treatment to guess the reinforcer ratio that will beat problem behavior.

Morris et al. (2023) pushed further, fitting the code to real self-injury data and getting tighter predictions than older methods.

The 2010 paper does not replace the classic statics; it gives them a living motor.

Baer (1974) and White (1979) mapped the shapes of undermatching and bias—this study reveals how those shapes emerge from selection pressure alone.

You now have a free, open-source tool that turns a matching-law assessment into a dynamic forecast.

Run the simulation with your client’s reinforcer rates to see if the plan will create undermatching or a risky bias.

If the curve drifts, tweak the virtual contingencies first instead of the real child’s session.

Evolution made the behavior; let evolution help you predict what happens next.