The original and power function versions of the matching law and matching theory are mathematical accounts of continuous choice that have been applied to the behavior of many vertebrate species, including humans, and have become influential in a variety of disciplines, such as economics, behavioral pharmacology, and clinical science. Each version can be interpreted either from a response strength perspective or from a purely algebraic perspective. The point of departure of both versions of matching theory is the assertion that all behavior is choice governed by the matching law. This assertion permits sets of equations to be derived that constitute the 2 versions of the theory. Considerable experimental evidence leads to the conclusion that both interpretations of the original version of the matching law and theory are false and that both interpretations of the power function version of the law and theory are tenable. Specifically, the behavior of animal and human subjects conforms exactly to the mathematical forms required by the power function version of the law and theory, and the theoretical assumptions entailed by both interpretations of this version are supported by the empirical evidence. In addition, the theoretical interpretations of all but 1 of the parameters of the power function version's equations are consistent with available evidence. The one problematic parameter is interpreted as the rate at which unmeasured background resources are acquired. The theoretical interpretation of this parameter must be reconciled with observation in order to maintain matching theory's central assertion that all behavior is choice.

No abstract

Caron (2013b) asserted in his comment on my article on the matching law and matching theory that a number of my analyses and conclusions were unsystematic, unreliable, doubtful, inappropriate, or violated theoretical assumptions. These assertions are based on mistakes and conceptual and theoretical misunderstandings. Specifically, my conclusions about Equation 6' are not affected by the parametric constraints applied when fitting it, and I used 2 pooled data sets appropriately as part of my evaluation of Equations 7' and 8'. All the conclusions I reached in my article are valid, apply to individual subjects, and stand as I stated them.