Newsletter No. 406

4 No. 406, 4.11.2012 從數獨遊戲 窺探理性思維 Everyone is a Good Reasoner and a Reveler I n the prologue to his Autobiography , Bertrand Russell describes his life’s passions: ‘I have wished to understand the hearts of men…. And I have tried to comprehend the Pythagorean power by which number holds sway above the flux.’ At the risk of over-simplification, understanding the hearts of men has always been the scholarly pursuit of the humanists, while scientists seek the fundamental and immutable laws of nature or what holds sway above the flux. Situated somewhat in the middle, cognitive scientists till the human mind for the intractable processes of human cognition and reasoning. Through the common Sudoku puzzles that have fascinated millions around the world, Prof. Louis Lee of the Department of Educational Psychology believes he might have found evidence that everybody has deductive abilities. Deduction is the reasoning process by which one draws conclusions that necessarily have to be true if the given premises are true. It can be illustrated by a simple syllogism: All cars are wheeled vehicles. All wheeled vehicles can move. Therefore, all cars can move. Those who stand on the nurture side of the nature- nurture debate would argue that even if it does not take a mathematician or logician to make the above deduction, at least some kind of training is required. Professor Lee’s findings point the other way. When Professor Lee was still a PhD candidate at Princeton University, he noted with special interest the Sudoku puzzles serialized in the Times , whose readership ranged from the age of 8 to 80, regardless of educational level or socio-economic status. Sudoku puzzles are digit placement puzzles said to derive from the Latin squares of the great 18th century mathematician Leonhard Euler. A typical puzzle is a 9 × 9 matrix which itself is further divided into nine smaller matrices of 3 × 3. Numbers appear in some of the cells of the matrix but some cells are empty. The challenge is to fill in the missing numbers in all the empty cells according to the general rule that every number from 1 to 9 must occur exactly once in each row, in each column, and in each of the smaller 3 × 3 matrices. Sudoku puzzles differ in terms of difficulty, from easy, mildly difficult, difficult to fiendish. As Professor Lee sees it, though the task is to find the missing numbers, Sudoku has nothing to do with numbers or computation. Any symbol or object can take the place of the numbers in the rows and columns of the puzzles. Pure deductive reasoning is required for solving Sudoku puzzles. Professor Lee further postulated that even naïve individuals were able to use simple tactics to complete the task. A simple tactic is one that starts with definite digits and that enables individuals to deduce the missing digit from them. For example, if the numbers 1 to 8 already appear in a row, then the missing one must be 9. A series of simple tactics would become the steps in a strategy for finding the solution to a puzzle. Obviously, the difficulty of a particular puzzle would depend on the number of constraints involved, or what is known as relational complexity. For instance, if the missing number has to satisfy the general rule for both a row and a column, more effort and time would be required. In this example, the row 略難 Mildly difficult 1 9 2 7 9 2 5 2 3 3 1 4 2 8 4 1 2 8 5 9 7 1 3 9 4 6 7 5 困難 Difficult 8 7 8 1 7 2 5 1 4 7 2 9 1 5 5 8 7 3 2 3 7 8 4 9 7 3 4 5 艱深 Fiendish 7 5 9 4 6 9 1 3 2 2 6 1 7 8 2 1 3 8 5 3 9 2 4 8 4 3 7 9 在 羅素的《自傳》序言中,這位哲人提到他一生追 求的兩樣東西:「我極期渴望能夠洞悉人心之所 思……我竭力參透數字主宰萬物變化背後的玄機。」 籠統而言,人文學者竭力了解人心;科學家則致力尋求永恆 不變的自然定律。認知科學家則介乎兩者之間,他們最想 破解人類認知及推理的神秘法則。教育心理學系 李雅言 教 授,透過瘋魔全球的「數獨」遊戲,可能已找到每個人都有 演繹推理能力的證據。 演繹法是一種推理過程,在前提是真實的情況下,其結論必 然真實。後天論者認為一般人至少要受過相關訓練,始可 演繹推理。但李教授的研究,正正提出相反的事實。 當李教授仍是美國普林斯頓大學博士生時,他就留意到英 國《泰晤士報》連載的「數獨」遊戲廣受大眾歡迎,從八 歲小孩到八十歲長者,不論何種教育程度或社會階層均愛 不釋手。 「數獨」據說是由十八世紀的偉大數學家萊昂哈德.歐拉 的拉丁方陣衍生出來。典型的「數獨」是一個九乘九的矩 陣,每個矩陣內又有九個三乘三的小矩陣。矩陣部分方格 內有數字,部分留白。玩法是在每一列、每一行及每一小 矩陣的空格內填上1至9的數字,但在每列、每行和每個 小矩陣內,1至9的數字每個只 可出現一次。 李教授認為,雖然這個遊戲是 找出從缺的數字,但「數獨」 本身不涉及任何數理或運算步 驟。光是運用推理,便足以解開 「數獨」的謎底。李教授進一 步假設,沒有受過訓練的人,也 可透過簡單策略去完成任務。 簡單策略就是從已知的確切數字,推斷出從缺的數字。舉 例來說,若1至8各個數字已出現於一列中,那麼餘下的數 字必然是9。一連串的簡單策略漸漸發展成為步驟,讓玩 的人可以偵破難題。 明顯地,「數獨」的難易程度視乎規限條件多少,或所謂關 係複雜度而定。若從缺的數字需要同時通過行和列對數字 的規限,那就得花上更多時間和精力去解難。在這個例子之 中,行與列便是規限條件,關係複雜度是2。最後,對於那 些較難的遊戲來說,簡單策略可能不管用,故要使出進階策 略。這些進階策略有兩個步驟:個人首先推測某些空格中的 可能數字,接着利用這些可能數字去淘汰其他空格中的可 能數字。在每一個步驟中,個人仍需要倚靠簡單策略。 為了測試這些假設,李教授等學者在一個實驗中邀請先 前沒有接觸過「數獨」的中大學生,試玩三種不同難度的 「數獨」( 圖一 )。他們要在十五分鐘內,盡量填上缺少的 數字,也要解釋為何會作出這樣的判斷。研究發現這些首 次接觸「數獨」的學生,在十五分鐘內就能自發地掌握一 些推理策略。正如實驗者所預測,大部分學生採用了簡單 策略,而從他們提出的理由,足證他們意識並可清楚地解 釋選擇這些策略的原因。 在第二個實驗中,一批普林斯頓大學的學生獲邀試玩一系 列關係複雜度由2至5的「數獨」。結果發現「數獨」難度 愈高,解難策略也相應加深,學生需要發展出進階策略, 要紀錄下有可能出現在空格中的數字。換句話說,簡單策 略只能夠應付簡單的遊戲,不足以應付更艱深的遊戲。 在最後一個實驗中,二十名中大學生分為兩組進行進階 策略的測試。第一組學生的遊戲跟第二個實驗的相似,但 部分空格提供了可能出現的數字( 圖二 );第二組學生則 沒有。換句話說,第一組學生不用採取進階策略的第一個 步驟,可以馬上用已有的提示去剔除那些可能性不高的答 案。正如所料,第一組學生可以更快捷、更容易去解開「數 獨」難題。 李教授等研究者總結,「總的來說,『數獨』遊戲可以證明 沒有邏輯訓練的人也有能力去就抽象事物作出推斷,他們 也樂於作這類推斷。」* 這結論質疑那些 認為演繹邏輯能力是需要透過教育和經 驗得來的心理學理論。 這項研究的另一個發現,是部分學生遇上 較難的「數獨」時會調整策略,令我們對 人類的思考過程甚至創意思維了解多一 點。事實上,李教授把他的個人研究興趣 與他培育年輕學子的熱忱結合起來。身為 伍宜孫書院輔導長,他透過簡單卻能挑戰 思考的遊戲來鼓勵學生思考他們是如何思 考的,並分析當中的過程,從而發展出創意的解難方案。以 他任教的「思維心理學及其應用」課程為例,便是藉着跟學 生玩「珠機妙算」遊戲,探討不同假設測試策略的效用。莫 可測向的人類心靈活動,將會繼續是李教授的研究焦點。 圖一:數獨遊戲 Figure 1: Sudoku Puzzles

RkJQdWJsaXNoZXIy NDE2NjYz