Biological constraints revisited: A critique

Four types of supporting evidence for inferences of biological constraints on conditioning are evaluated: (1) unexpected failures to condition (discriminative leverpress avoidance); (2) rapid or one-trial learning (taste-aversion studies); (3) crossover learning effects (the Garcia-Koelling effect); and (4) unique predictions from biological constraint-based frameworks. According to a current logic model of scientific explanation and prediction, none of these types of evidence is adequately compelling for inferences of biological constraints. The core problem identified is that negative evidence not attributable to an identified causal factor in terms of E/C (experimental/control) design outcome has been used to both reject general-process learning laws and compel inferences of biological constraints. A design solution using a blocking E/C conditioning paradigm is presented.