Researchers have reconstructed and validated a mathematical model developed by physicist Richard Feynman regarding the optimal balance between exploration and exploitation in decision-making.

研究人員重建並驗證了物理學家理查德·費曼（Richard Feynman）開發的一個數學模型，關於決策過程中「探索」與「利用」之間的最佳平衡。

The theoretical framework originated from a 1970s observation by Richard Feynman concerning the selection of menu items at a Thai restaurant. This scenario was conceptualized as a 'stopping problem' within decision theory, a discipline situated at the nexus of psychology and economics. The core objective was to determine the precise moment at which an agent should cease searching for a superior option and instead commit to the highest-quality alternative previously encountered.

這個理論框架源於理查德·費曼在 1970 年代觀察到在泰國餐廳選擇菜單項目的情況。這個場景被概念化為決策理論中的一個「停止問題」，而決策理論是一門處於心理學與經濟學交匯點的學問。核心目標是確定一個代理人在哪個精確時刻應停止搜尋更佳選項，轉而選擇先前遇到過的最高品質替代方案。

Following the decipherment of Feynman's handwritten notes by Tom Griffiths and Brian Christian, the problem was expanded to encompass the selection of dining establishments within a city over a fixed temporal horizon. The model posits that the utility of exploration is inversely proportional to the remaining time; consequently, the quality threshold required to cease exploration declines as the duration of the stay diminishes. This occurs because the prospective utility derived from a high-quality discovery is contingent upon the number of subsequent opportunities to utilize that information.

在 Tom Griffiths 與 Brian Christian 解讀費曼的手寫筆記後，該問題被擴展至在固定時間範圍內於一座城市中選擇餐飲場所。該模型假設探索的效用與剩餘時間成反比；因此，停止探索所需的品質門檻會隨著逗留時間的減少而下降。這是因為從高品質發現中獲得的預期效用，取決於隨後利用該資訊的機會次數。

Further analysis indicated that the optimal strategy is contingent upon the distribution of quality among available options. In environments characterized by a scarcity of high-quality 'gems' amidst a preponderance of inferior options, a higher initial threshold is maintained to justify prolonged exploration. Conversely, in environments with a high density of above-average options, the threshold is lowered. Empirical validation involving 2,520 participants demonstrated that human intuitive behavior closely approximates this mathematical solution, although participants tended to reduce their thresholds linearly rather than following the more complex curve proposed by Feynman.

進一步分析指出，最佳策略取決於可用選項中品質的分佈。在高品質「寶藏」稀少且劣質選項佔主導的環境中，需維持較高的初始門檻以證明長時間探索的合理性。相反，在高品質選項密度較高的環境中，門檻則會降低。涉及 2,520 名參與者的實證驗證顯示，人類的直覺行為與此數學解法非常接近，儘管參與者傾向於線性地降低門檻，而非遵循費曼提出的更複雜曲線。

The study confirms that Feynman's mathematical approach provides an optimal strategy for stopping problems, mirroring the intuitive heuristics employed by humans.

研究確認費曼的數學方法為停止問題提供了一個最佳策略，鏡像了人類所使用的直覺啟發法。