Harris Justin A, Kwok Dorothy W S, Andrew Benjamin J
School of Psychology.
J Exp Psychol Anim Learn Cogn. 2014 Jul;40(3):335-54. doi: 10.1037/xan0000023. Epub 2014 Apr 7.
We investigated conditioned inhibition in a magazine approach paradigm. Rats were trained on a feature negative discrimination between an auditory conditioned stimulus (CS) reinforced at one rate versus a compound of that CS and a visual stimulus (L) reinforced at a lower rate. This training established L as a conditioned inhibitor. We then tested the inhibitory strength of L by presenting it in compound with other auditory CSs. L reduced responding when tested with a CS that had been reinforced at a high rate, but had less or even no inhibitory effect when tested with a CS that had been reinforced at a low rate. The inhibitory strength of L was greater if it signaled a decrease in reinforcement from an already low rate than if it signaled an equivalent decrease in reinforcement from a high rate. We conclude that the strength of inhibition is not a linear function of the change in reinforcement that it signals. We discuss the implications of this finding for models of learning (e.g., Rescorla & Wagner, 1972) that identify inhibition with a difference (subtraction) rule.
我们在杂志接近范式中研究了条件性抑制。大鼠接受了一种特征性负辨别训练，即在一种速率下强化的听觉条件刺激（CS）与以较低速率强化的该CS和视觉刺激（L）的复合物之间进行辨别。这种训练将L确立为条件性抑制剂。然后，我们通过将L与其他听觉CS复合呈现来测试其抑制强度。当与以高速率强化的CS一起测试时，L减少了反应，但当与以低速率强化的CS一起测试时，L的抑制作用较小甚至没有抑制作用。如果L表明强化从已经很低的速率下降，那么它的抑制强度比表明从高速率等量下降时更大。我们得出结论，抑制强度不是它所表明的强化变化的线性函数。我们讨论了这一发现对学习模型（例如，Rescorla & Wagner，1972）的影响，这些模型用差异（减法）规则来识别抑制。
