Davison M
Department of Psychology, University of Auckland Private Bag, Auckland, New Zealand.
Behav Processes. 1990 Jun;21(2-3):189-95. doi: 10.1016/0376-6357(90)90024-A.
Melioration theory suggests that performance in choice situations arises from a process in which subjects allocate more time to alternatives that provide the higher local rate of reinforcers. If this process is unconstrained (as in concurrent aperiodic schedules), melioration predicts a strict equality between molar time-allocation (and, as a result, response-allocation) ratios and molar ratios of overall obtained reinforcers on concurrent variable-interval schedules. The available data, however, suggest that log behavior measures are nearer to indifference (called undermatching that predicted by this theory. The modification of melioration theory suggested here shows that, if animals cannot discriminate local reinforcer-rate differences below a fixed threshold, undermatching is predicted, and the degree of undermatching depends on the absolute size of the threshold. It also predicts the finding that the sensitivity of behavior ratios to changes in reinforcer ratios falls with decreasing overall reinforcer rates.
改进理论表明,在选择情境中的表现源于一个过程,即主体会将更多时间分配给能提供更高局部强化率的选项。如果这个过程不受限制(如在并发非周期性强化程序中),改进理论预测在并发可变间隔强化程序中,整体时间分配(以及由此产生的反应分配)比率与整体获得的强化物的整体比率之间会严格相等。然而,现有数据表明,对数行为测量结果更接近无差异状态(称为该理论预测的欠匹配)。这里提出的对改进理论的修正表明,如果动物无法区分低于固定阈值的局部强化率差异,那么就会预测到欠匹配,并且欠匹配的程度取决于阈值的绝对大小。它还预测了这样一个发现,即行为比率对强化物比率变化的敏感性会随着整体强化率的降低而下降。