Ramnani Narender
Department of Psychology, Royal Holloway, University of London, Egham, UK.
Prog Brain Res. 2014;210:255-85. doi: 10.1016/B978-0-444-63356-9.00010-8.
During learning, performance changes often involve a transition from controlled processing in which performance is flexible and responsive to ongoing error feedback, but effortful and slow, to a state in which processing becomes swift and automatic. In this state, performance is unencumbered by the requirement to process feedback, but its insensitivity to feedback reduces its flexibility. Many properties of automatic processing are similar to those that one would expect of forward models, and many have suggested that these may be instantiated in cerebellar circuitry. Since hierarchically organized frontal lobe areas can both send and receive commands, I discuss the possibility that they can act both as controllers and controlled objects and that their behaviors can be independently modeled by forward models in cerebellar circuits. Since areas of the prefrontal cortex contribute to this hierarchically organized system and send outputs to the cerebellar cortex, I suggest that the cerebellum is likely to contribute to the automation of cognitive skills, and to the formation of habitual behavior which is resistant to error feedback. An important prerequisite to these ideas is that cerebellar circuitry should have access to higher order error feedback that signals the success or failure of cognitive processing. I have discussed the pathways through which such feedback could arrive via the inferior olive and the dopamine system. Cerebellar outputs inhibit both the inferior olive and the dopamine system. It is possible that learned representations in the cerebellum use this as a mechanism to suppress the processing of feedback in other parts of the nervous system. Thus, cerebellar processes that control automatic performance may be completed without triggering the engagement of controlled processes by prefrontal mechanisms.
在学习过程中，表现的变化通常涉及从受控加工的转变，在受控加工中，表现灵活且对持续的错误反馈有响应，但费力且缓慢，转变到一种加工变得迅速且自动的状态。在这种状态下，表现不受处理反馈要求的阻碍，但其对反馈的不敏感降低了其灵活性。自动加工的许多特性与人们对前馈模型所期望的特性相似，许多人认为这些特性可能在小脑回路中得以体现。由于分层组织的额叶区域既能发送也能接收指令，我讨论了它们既可以充当控制器又可以充当受控对象的可能性，以及它们的行为可以由小脑回路中的前馈模型独立建模的可能性。由于前额叶皮层区域对这个分层组织的系统有贡献并向小脑皮层发送输出，我认为小脑可能有助于认知技能的自动化，以及有助于形成对错误反馈有抗性的习惯性行为。这些观点的一个重要前提是，小脑回路应该能够获得高阶错误反馈，该反馈能表明认知加工的成功或失败。我已经讨论了这种反馈可以通过下橄榄核和多巴胺系统到达的途径。小脑输出抑制下橄榄核和多巴胺系统。有可能小脑里的习得表征将此用作一种机制来抑制神经系统其他部分的反馈处理。因此，控制自动表现的小脑过程可能在不触发前额机制的受控过程参与的情况下完成。
