Estimation method for higher-order judgment process with psychophysical reverse correlation

反向相关（RC）被认为是探测大脑内部计算的一种有效方法，并已在生理学研究中得到应用。心理物理学研究也采用RC分析来理解观察者判断背后的计算，但其应用仅限于简单任务和刺激的实验。这是因为RC是白噪声分析的一种变体，它强烈地限制了实验刺激。本研究提出了一种方法，使我们能够探索具有任意刺激的心理物理决策过程的非线性计算。该方法提供了内部计算的无偏估计，而不受刺激分布的任何限制。数值实验和心理物理实验证实了该方法的有效性。这种方法有助于揭示高阶认知功能的本质，而高阶认知功能需要呈现结构化的刺激而不是简单的噪声输入。

Reverse correlation (RC) is known as a powerful method for probing internal computations in the brain, and it has been used in physiological researches. Psychophysical studies have also employed the RC analysis for understanding the computations underlying observers' judgments, but its application has been limited to the experiments with simple tasks and stimuli. This is because RC is a variation of white noise analysis, and it strongly restricts experimental stimuli. The present study proposes a method that enables us to explore nonlinear computations underlying psychophysical decision processes with arbitrary stimuli. This method provides unbiased estimates of internal computations without any restrictions on stimulus distributions. Numerical and psychophysical experiments confirmed the validity of the proposed method. This method could contribute to unveil the nature of higher-order cognitive functions which is desirable to present structured stimuli rather than simple noise input.