Champalimaud Research, Champalimaud Center for the Unknown, Avenida Brasilia, 1400-038, Lisbon, Portugal.
Trends Cogn Sci. 2021 Jun;25(6):431-433. doi: 10.1016/j.tics.2021.02.009. Epub 2021 Mar 9.
As research investigating controllability perception gains momentum, the algorithmic definition of controllability must be updated to avoid confusion between controllability and predictability. Reframing controllable environments as environments that allow agents to exert causal influences over state transitions can circumvent the confounding influence of predictability when designing cognitive tasks.
随着研究可控性感知的不断深入,算法定义的可控性必须进行更新,以避免可控性和可预测性之间的混淆。将可控环境重新定义为允许代理对状态转换施加因果影响的环境,可以在设计认知任务时避免可预测性的干扰影响。