1、Improving metacognitive accuracy: How failing to retrieve practice items reduces overconfidence提高元認知的準確性：提取練習工程失敗如何降低過度自信Authors：Tyler M. Miller & Lisa GeraciSource ：Consciousness and CognitionI F：2.235abstractPeople often exhibit inaccurate metacognitive monitoring. For example, overconfidence occu

2、rs when people judge that they will remember more information on a future test then they actually do.人們經常表現出錯誤的元認知監測。以為在未來的測試上，他們會記住更多的信息即比實踐記住的要多，那么過度自信就出現了。The present experiments examined whether a small number of retrieval practice opportunities would improve participants metacognitive accuracy

3、by reducing overconfidence. 當前的實驗研討少數提取練習時機能否會經過降低過度自信提高被試元認知的準確性。1. IntroductionThe term metacognition refers to a persons knowledge about the quality and accuracy of their own cognition. Flavell(1979) formalized the term and investigated the phenomena from a developmental perspective.元認知指一個人知識的質量和

4、認知本身的準確性。Flavell(1979)正式提出這個術語，并從開展的角度研討了這種景象。Studies today show a similar pattern for college-aged participants. They predict that they will earn a much better score on a given test than they actually do (for one example see Hacker, Bol, Horgan, & Rakow, 2000). People are also overconfident in othe

5、r abilities, including their driving ability (Knouse, Bagwell, Barkley, & Murphy, 2005).Ability to complete projects before deadlines (Buehler, Griffin, & Ross, 1994).More accurate metacognition is associated with better academic performance (Everson & Tobias,1998) and memory performance (Thiede et

6、al., 2003, 2005). Other studies have confirmed the link between accurate metacognition and improved performance (e.g., Nelson, Dunlosky, Graf, & Narens, 1994).準確的元認知在教育環境中尤其有利，更準確的元認知和更好的學業成果(Everson & Tobias,1998)和記憶成果(Thiede et al., 2003, 2005)相關。其他的研討也證明準確的元認知和成果的提高之間有關聯(e.g., Nelson, Dunlosky, G

7、raf, & Narens, 1994)。Metacognitive ability is highly resistant to change.The present study examined a different method for improving metacognitive accuracy. Here, we examined whether practice retrieving a single test item (Experiments 1 and 2) or more test items (Experiment 3) could lead to more acc

8、urate metacognitive monitoring.元認知才干很難被改動。當前的研討以一種不同的方法來提高元認知的準確性。作者調查了能否提取練習一個測試工程實驗1和實驗2或更多的測試工程實驗3能導致更準確的元認知監測。1.1. The effects of testing on metacognition Memory advantages of taking a practice test versus restudying material prior to taking a final memory test (for reviews see Dunlosky, Rawson,

9、 Marsh, Nathan, & Willingham, 2021;Roediger,Putnam, & Smith, 2021Testing produces better organization of knowledge (Zaromb & Roediger, 2021) Reduces proactive interference (Szpunar, McDermott, & Roediger,2021).Relationship between testing and metacognition?One, are people aware of the benefits of te

10、sting ? Answer is mixed.Two, does testing improve metacognition?Have little effect on metacognition accuracy (Karpicke & Roediger,2021)主要緣由是強調測試對隨后測試成果的影響而沒有關注元認知，另一個是短少對照組。1.2. The present experimentsThe present experiments were designed to directly examine whether practice retrieving items can lea

11、d to more accurate metacognitive monitoring. 當前的實驗直接調查提取練習工程能否能導致更準確的元認知監測。we examined whether retrieval practice with a single item (Experiment 1) would improve subsequent performance predictions. 實驗1調查提取練習單個工程能否會提高隨后的預測成果。We also examined the role of failure (vs. success) in retrieving a single pr

12、actice item on subsequent judgments (Experiment 2) and how much failure is needed to improve the accuracy of participants predictions (Experiment 3).實驗2調查提取單個工程失敗或勝利對后來判別的作用，實驗3調查失敗多少次才干提高被試預測的準確性。In each of the three experiments, participants were asked to make global performance predictions before

13、 and after engaging in retrieval practice or no retrieval practice.每個實驗中，在被試提取練習之前或提取練習之后或沒有提取中都對總成果進展預測。2. Experiment 12.1. Method and materials2.1.1. ParticipantsEighty young adults (53% male) between the ages of 18 and 21 (M = 19.11, SE = .09) participated in the experiment in return for partial

14、course credit. Participants were undergraduate students with 1216 years of education (M = 13.14, SE = .10). Across conditions, participants did not differ in average age or education (Fs 1.5).2.1.2. DesignExperiment 1 used a 2 Condition (Retrieval Practice and No Retrieval Practice Control) 2 Perfor

15、mance Prediction (Original and Adjusted) mixed randomized repeated model study design. they made two performance predictions. The dependent variable of interest was the accuracy of each performance prediction. 2條件:提取練習，非提取練習控制組 2成果預測：原始的，調整的混合實驗設計。因變量為成果預測的準確性預測成果-回想成果/40。2.1.3. Materials and proced

16、ureWe used a subset of 40 LithuanianEnglish paired associates from norming data provided by Grimaldi, Pyc, and Rawson (2021). Items were chosen based on the probability they were recalled on the memory test during trial one in Grimaldi et al. (2021). The average probability of recall on trial one wa

17、s .23, but items from the entire range were selected (.04.49 probability of recall).根據記憶測試中回想的能夠性選擇了40個取自Grimaldi et al. (2021). 的立陶宛語英語的配對聯想詞對，每個工程回想正確的能夠性為0.23，范圍為0.040.49。ABfinal recall test.predictionadjusted prediction“sesuo _demographic questionnaire2.2. Experiment 1 results2.2.1. Metacognitiv

18、e accuracy analysis 兩種條件在原始預測和調整預測條件下都出現了過度自信。提取練習條件沒有顯著的影響這些預測。反復丈量方差分析結果闡明預測成果的主效應不顯著， F(1,78) = 1.17, MSE = 0.01,p = .28, 2 p = .02 ；條件的主效應也不顯著，F(1,78) = 0.39, MSE = 0.02, p = .53, 2 p = .01.交互作用也不顯著F(1,78) = 0.94, MSE = 0.01, p = .34, 2 p = .01。 2.2.2. Supplementary analysisParticipants in the R

19、etrieval Practice condition recalled the English equivalent of the Lithuanian cue sesuo at a rate higher than was expected based on the previous literature (Grimaldi et al., 2021). That is, 70% of the participants recalled the English equivalent sister. We completed a supplementary analysis to deter

20、mine the influence of retrieval practice success or failure on adjusted performance predictions.提取練習組回想“sesuo 對應詞“sister的能夠性為70%,高于原文獻中的50%(Grimaldi et al., 2021)。分析提取練習的勝利或失敗對調整預測成果的影響。We split the data for the Retrieval Practice condition into two groups: a group that correctly recalled the Englis

21、h equivalent Participants who correctly recalled the practice item did not change their performance predictions (M = .30, SE = .03 to M = .31,SE = .04) and their memory performance was M = .28 (SE = .03).a second group that did not recall the English equivalent.participants who failed to retrieve th

22、e practice item decreased their performance predictions (M = .30, SE = .05 to M = .19, SE = .03; F(1,38) = 13.90, MSE = 0.01, p = .001, 2 p = .27)，M = .21, SE = .04 . In terms of metacognitive monitoring accuracy, retrieval practice failure affected participants monitoring accuracy, whereas retrieva

23、l practice success did not. In particular, retrieval failure prior to the final memory test allowed participants to adjust their predictions. That is, after retrieval failure, participants reduced their predictions and became less overconfident. In Experiment 2, we directly manipulated retrieval suc

24、cess and failure to examine their effects on metacognitive monitoring accuracy. 提取練習失敗影響了被試記憶監測的準確性，而提取勝利不影響。最后測試之前的提取失敗讓被試調整了他們的預測，也就是說提取失敗之后，被試降低了預測，不再過分自信。 實驗2直接支配了提取的勝利與失敗，來調查元認知監測準確性的影響。3. Experiment 2More participants would fail to retrieve the difficult practice item than the easy practice it

25、em .Retrieval practice failure is more beneficial for metacognitive accuracy than retrieval practice success.Hypothesis:3. Experiment 23.1. Method and materials3.1.1. ParticipantsNinety younger adults (52% male) between the ages of 18 and 29 (M = 19.74, SE = .21) participated in return for partial c

26、ourse credit. Participants were undergraduate students with between 12 and 16 years of education (M = 13.74, SE = .11).Across conditions, participants did not differ in average age or education (Fs 1.5).3.1.2. DesignExperiment 2 used a 2 Condition (Easy Retrieval Practice and Difficult Retrieval Pra

27、ctice) 2 Performance Prediction (Original and Adjusted) mixed randomized repeated model. Participants were randomly assigned to either the Easy (n = 45) or the Difficult retrieval practice condition (n = 45) and made two performance predictions. 3.1.3. Materials and procedureWe used the same 40 Lith

28、uanianEnglish paired associates as Experiment 1 from Grimaldi et al. (2021).Easy Retrieval Practice condition : recall the English equivalent of sesuo (sister)(50%).不通知被試他們是在容易提取條件下還是困難提取條件下，程序和實驗1大致類似。Difficult Retrieval Practice condition : recall the English equivalent of muilas (soap)(4%).圖1 實驗1

29、和實驗2中被試在初始預測和調整預測中總的百分比。虛線為每種條件下的平均回想成果。 But results from experiment 2 do not explain why retrieval practice failure is better for metacognitive accuracy than retrieval success. Practice items must be diagnostic of the final memory test (i.e., the diagnosticity assumption; Dunlosky, Rawson, & McDona

30、ld, 2002).提取練習失敗提高了元認知的準確性。單個的提取練習工程失敗對元認知的準確性也是有益的。練習工程能夠是最后記憶測試的診斷診斷假設。 Retrieval failure improved metacognitive accuracy. A single instance of retrieval practice failure was beneficial for metacognitive accuracy.One could infer that the difficult retrieval practice item was diagnostic of the memo

31、ry test; therefore participants had practice with a test item that was representative, or diagnostic, of the final test. Thus, the difficult test condition benefitted participants metacognitive accuracy.一種能夠的推斷是困難提取練習工程是記憶測試的診斷，因此被試用一個典型的診斷性的練習工程來測試。4. Experiment 34.1. Method and materials實驗1和2的結果闡明

32、，提取練習失敗對提高元認知的準確性比提取勝利有更大的益處。預測1，較高的提取失敗的數量對元認知準確性的提高比較低的更有益。預測2，過多的提取失敗將會使被試過度降低預測成果，導致自信心缺乏的元認知錯誤。預測3，根據診斷假設實際 diagnosticity assumption (Dunlosky et al., 2002)，人們不太關懷提取勝利或失敗的總體數目，而關懷提取練習測試上的成果和最后測試成果的類似性。練習工程的成果和最后測試成果越相近就代表被試的元認知準確性越好。實驗3就是梳理這些潛在的假設，為什么提取失敗提高元認知的準確性。Two hundred and seventy-four y

33、ounger adults (26% male) between the ages of 17 and 28 (M = 18.85, SE = .10) participated in return for partial course credit. Participants were undergraduate students with between 12 and 16 years of education (M = 12.50, SE = .05). Participants did not differ in average age or education across cond

34、itions (Fs 1.0).4.1.1. Participants4.1.2. Design Experiment 3 used a 4 Condition (No Retrieval Practice Control, Easy Retrieval Practice, Medium Retrieval Practice, Difficult Retrieval Practice Difficult) 2 Performance Prediction (Original and Adjusted) mixed randomized repeated model. Participants

35、were randomly assigned to either the No Retrieval Practice Control (n = 68), Easy (n = 69), Medium (n = 70), or difficult retrieval practice (n = 67) and made two performance predictions. 4.1.3. Materials and procedure The materials and procedure were nearly identical to the previous experiments wit

36、h the key exceptions being that participants in the retrieval practice conditions attempted to recall four practice items, rather than one practice item (as was the case in Experiments 1 and 2). The four items were all easy items, all difficult items, or a combination of easy and difficulty items to

37、 make up the medium difficulty condition according to norms reported in Grimaldi et al. (2021). 4.2. Experiment 3 results4.2.1. Metacognitive accuracy analysisFirst, retrieval practice performance depended on the condition easy, medium, or difficult (F(3,270) = 124.13,MSE = 0.04, p .001, 2 p = .58).

38、 Participants in the Easy Retrieval practice condition had the most retrieval practice success (M = 0.62) followed by participants in the Medium Retrieval Practice condition (M = 0.41), and finally, participants in the Difficult Retrieval Practice had the least amount of success (M = 0.13). 圖2 實驗3中被試在初