Application of the matching law to pitch selection in professional baseball

Five Major League Baseball pitchers were tracked across real games. Researchers counted every pitch they threw and the game situation when they threw it. They tested if the matching law could predict pitch choices in live competition.

The matching law says behavior follows reinforcement rates. In baseball, a fastball might be reinforced by a strike. A curveball might be reinforced by a swing and miss. The study checked if pitchers chose pitches based on these past payoffs.

The matching law fit the data almost perfectly. Pitchers' choices shifted exactly as the equation predicted. When fastballs got more strikes, pitchers threw more fastballs. When curveballs got more outs, they threw more curves.

Game context changed the math. In early innings, pitchers showed strong bias toward fastballs. In late innings with runners on base, they became more sensitive to each pitch's recent success. The same pitcher used different rules in different situations.

Hawley et al. (2004) found similar patterns in women's basketball. Both studies show ABA principles work in elite sports. The basketball study tracked momentum after turnovers. The baseball study tracked pitch choices after each outcome.

Sarber et al. (1983) warns about testing too hard. Their probe trials created false failures. Cox et al. avoided this by using real game data instead of artificial tests. The pitchers weren't being tested - they were just playing baseball.

Thompson et al. (1971) showed shock changes turtle behavior. While different species, both studies confirm that behavior follows reinforcement contingencies. The turtle bit when shocked. The pitcher threw curves when curves worked.

You can use this in your practice today. When a client shows context-specific behavior, check their reinforcement history in each setting. The same skill might have different payoffs at home versus school. Track the actual consequences, not what adults think should work. Then adjust the environment to make the desired behavior the most reinforced option in each context.