失忆症crowd_完成对象属性的失忆症吗？

失忆症crowd

对Chen＆Wyble的再分析，2015年 (Reanalysis of Chen & Wyble, 2015)

Chen and Wyble published an interesting paper (2015) where they demonstrate that participants cannot report attributes of attended stimuli unless the participants are previously informed that this attribute is important. For instance, you wouldn’t remember the color of the apple if you had had just told someone the shape. I would have expected the opposite, so … cool!

Chen和Wyble发表了一篇有趣的论文（2015年），他们证明参与者无法报告参与的刺激的属性，除非参与者事先被告知该属性很重要。 例如，如果您刚刚告诉某人形状，您将不会记得苹果的颜色。 我本来期望相反，所以……很酷！

After reading the paper (you can check it out at http://wyblelab.com/publications), I became curious whether participants might unconsciously retain some information about these forgotten attributes. Chen and Wyble posted their data to databrary.com (https://nyu.databrary.org/volume/79), so I downloaded the data and did some quick analyses that you see here! I want to commend Chen and Wyble for sharing their data. This is something everyone should start doing (including me).

阅读了这篇论文（您可以在http://wyblelab.com/publications上查看 ）之后，我很好奇参与者是否会不知不觉地保留了有关这些被遗忘属性的信息。 Chen和Wyble将他们的数据发布到databrary.com（ https://nyu.databrary.org/volume/79 ），所以我下载了数据并做了一些快速分析，您在这里可以看到！ 我要赞扬Chen和Wyble共享他们的数据。 这是每个人（包括我）都应该开始做的事情。

Below, I will start by showing I can replicate Chen and Wyble’s analyses, then I will investigate whether there’s a trace of unconscious memory for the “forgotten” features.

下面，我将首先展示我可以复制Chen和Wyble的分析，然后研究“遗忘”功能是否存在无意识记忆的痕迹。

EDIT[12/22/15]: Brad Wyble recently pointed out that I overstated the claim in their paper. They do not claim participants have complete amnesia for unqueried object attributes. Rather, Chen and Wyble focus on the dramatic performance change between the first and second trial following the initial query about an object attribute. This performance change demonstrates amnesia, but not necessarily complete amnesia.

EDIT [12/22/15]：Brad Wyble最近指出，我在他们的论文中夸大了这一说法。 他们不声称参与者对未查询的对象属性具有完全的健忘症。 相反，Chen和Wyble专注于在最初查询对象属性后在第一次和第二次试验之间的巨大性能变化。 这种性能变化表明健忘症，但不一定是完全健忘症。

参考资料 (References)

Chen, H., & Wyble, B. (2015). Amnesia for Object Attributes Failure to Report Attended Information That Had Just Reached Conscious Awareness. Psychological science, 26(2),203-210.

Chen H.，＆Wyble，B.（2015年）。 对象属性的失忆症无法报告刚达到意识的出席信息。 心理科学，26（2），203-210。

Wyble, B. (2014). Amnesia for object attributes: Failure to report attended information that had just reached conscious awareness. Databrary. Retrieved November 22, 2015 from http://doi.org/10.17910/B7G010

Wyble，B.（2014年）。 对象属性的失忆症：无法报告刚达到意识状态的出席信息。 资料库。 2015年11月22日从http://doi.org/10.17910/B7G010检索

加载相关库并编写分析功能 (Load relevant libraries and write analysis functions)

I’ll start by loading the python libraries that I’ll use throughout analyses.

我将从加载将在整个分析中使用的python库开始。

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9

Here are some quick functions I wrote for running different statistical tests and plotting the data. I won’t explain this code, but encourage you to look through it later if you’re wondering how I did any of the analyses.

这是我为运行不同的统计测试和绘制数据而编写的一些快速功能。 我不会解释此代码，但是如果您想知道我是如何进行任何分析的，那么鼓励您稍后仔细阅读。

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 11 11 12 12 13 13 14 14 15 15 16 16 17 17 18 18 19 19 20 20 21 21 22 22 23 23 24 24 25 25 26 26 27 27 28 28 29 29 30 30 31 31 32 32 33 33 34 34 35 35 36 36 37 37 38 38 39 39 40 40 41 41 42 42 43 43 44 44 45 45 46 46 47 47 48 48 49 49 50 50 51 51 52 52 53 53 54 54 55 55 56 56 57 57 58 58 59 59 60 60 61 61 62 62 63 63 64 64 65 65 66 66 67 67 68 68 69 69 70 70 71 71 72 72 73 73 74 74 75 75 76 76 77 77 78 78 79 79 80 80 81 81 82 82 83 83 84 84 85 85 86 86 87 87 88 88 89 89 90 90 91 91 92 92 93 93 94 94 95 95 96 96 97 97 98 98 99 99 100 100 101 101 102 102 103 103 104 104 105 105 106 106 107 107 108 108

实验1 (Experiment 1)

Next, load Experiment 1 data

接下来，加载实验1数据

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 11 11 12 12 13 13

The data is loaded, lets just take a quick look at the data after loading it in.

数据已加载，加载后让我们快速浏览一下数据。

1 1

		Sub#	子＃	Block	块	Trial#	试用＃	TarCol	焦油	Tar_Iden	Tar_Iden	Tar_Loc	Tar_Loc	Col_Resp	Col_Resp	Iden_Resp	Iden_Resp	Loc_Resp	Loc_Resp	Col_Acc	Col_Acc	Iden_Acc	Iden_Acc	Loc_Acc	Loc_Acc
0	0	6	6	1	1个	2	2	1	1个	2	2	4	4	0	0	0	0	2	2	0	0	0	0	0	0
1	1个	6	6	1	1个	3	3	3	3	4	4	3	3	0	0	0	0	3	3	0	0	0	0	1	1个
2	2	6	6	1	1个	4	4	1	1个	3	3	1	1个	0	0	0	0	1	1个	0	0	0	0	1	1个
3	3	6	6	1	1个	5	5	3	3	1	1个	4	4	0	0	0	0	4	4	0	0	0	0	1	1个
4	4	6	6	1	1个	6	6	2	2	2	2	1	1个	0	0	0	0	1	1个	0	0	0	0	1	1个

I want to create a new variable.

我想创建一个新变量。

Before explaining the new variable, I should explain a little about Chen and Wyble’s experiment. Half the participants were instructed to find the letter among numbers and the other half were instructed to find the number among letters. 4 items were briefly flashed on the screen (150 ms) then participants reported where the target item had been. Each of the 4 items was a different color.

在解释新变量之前，我应该先解释一下Chen和Wyble的实验。 一半的参与者被指示在数字中寻找字母，另一半被指示在字母中寻找数字。 屏幕上短暂闪烁了4个项目（150毫秒），然后参与者报告了目标项目的位置。 4个项目中的每一个都是不同的颜色。

Participants reported target location for 155 trials. On the 156th trial, the participants reported the target location then (to their surprise) reported what specific letter/number the target was and what color it was. Even though participants knew where the target was, they had no idea what either the target’s letter/number or color were. They had “forgotten” what the target was (even though they must have known what the target was when they initially located it).

参与者报告了155个试验的目标位置。 在第156次试验中，参与者报告了目标位置，然后（令他们惊讶的是）报告了目标的具体字母/数字和颜色。 即使参与者知道目标在哪里，他们也不知道目标的字母/数字或颜色是什么。 他们“忘记”了目标是什么（即使他们最初定位目标时必须知道目标是什么）。

The new variable will code whether the trial is a “pre-surprise” trial (trials 1-155), a surprise trial (trial 156) or the trial after the surprise (trial 157).

新变量将编码该试验是“惊喜前”试验（1-155试验），突击试验（156试验）还是突击后试验（157试验）。

I’ll call this variable “TrialType”

我将这个变量称为“ TrialType”

TrialType: 0=Presurprise; 1=Surprise Trial; 2=Postsurprise trials.

TrialType：0 =惊喜； 1 =惊喜审判； 2 =后突击试验。

1 1 2 2

实验1：复制Chen＆Wyble（2015）的分析 (Experiment 1: Replicating Chen & Wyble’s (2015) analyses)

Lets just take a quick look at overall accuracy. Make sure everyone is doing ok on the task. Below I plot the mean accuracy of each participant…looks like participant 23 struggled a little. Chen and Wyble (2015) notes that no participants were excluded or replaced.

让我们快速看一下整体准确性。 确保每个人在任务上都做得很好。 下面，我绘制了每个参与者的平均准确度……看起来参与者23有点挣扎。 Chen和Wyble（2015）指出，没有参与者被排除或替代。

I might have replaced participant 23 since his/her accuracy is easily 2.5 standard deviations below the mean accuracy (I print this value below)…seems like participant 23 was doing something different in this task.

我可能已经替换了参与者23，因为他/她的准确度很容易比平均准确度低2.5个标准偏差（我在下面打印此值）……似乎参与者23在此任务中做了一些不同的事情。

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 11 11

mean accuracy
0.89
standard deviation of accuracies
0.07
2.5 standard deviations below mean accuracy
0.71
mean accuracy
0.89
standard deviation of accuracies
0.07
2.5 standard deviations below mean accuracy
0.71
 

Lets look at participants’ performance when asked to identify the target’s location. I will plot performance as mean accuracy in the presurprise,surprise, and postsurprose trials.

让我们看看参与者在确定目标位置时的表现。 我将在惊喜之前，惊喜和惊喜之后的试验中将性能表示为平均准确性。

I will also run some quick statistical tests. For these tests, I take the arcsine of the square root of the accuracies (Rao, 1960) to increase the accuracies’ normality (I use adj. to indiciate that the tested data is transformed). I test whether this worked with a Shapiro-Wilk test of normality. If the p-value of the Shapiro-Wilk test is greater than 0.1, I run a t test to see if the accuracy in the two conditions is significantly different. If the p-value of the Shapiro-Wilk test is less than or equal to 0.1, then I run a Wilcoxon signed rank test since this test does not care about normality.

我还将运行一些快速的统计测试。 对于这些测试，我采用精确度平方根的反正弦（Rao，1960年）来提高精确度的正态性（我使用调式来指示测试数据已转换）。 我测试这是否与Shapiro-Wilk正常性测试一起工作。 如果Shapiro-Wilk检验的p值大于0.1，我会进行测试以查看两种条件下的准确性是否存在显着差异。 如果Shapiro-Wilk检验的p值小于或等于0.1，则我将执行Wilcoxon有符号秩检验，因为该检验不关心正态性。

Loc_Acc = df.pivot_table(values='Loc_Acc', index='Sub#', columns='TrialType', aggfunc=np.mean) CIs = np.array(OneWayConfInterval(Loc_Acc)) Analysis_and_Plot(Loc_Acc, CIs) Loc_Acc = df . pivot_table ( values = 'Loc_Acc' , index = 'Sub#' , columns = 'TrialType' , aggfunc = np . mean ) CIs = np . array ( OneWayConfInterval ( Loc_Acc )) Analysis_and_Plot ( Loc_Acc , CIs )

The y-axis represents percent correct. All graphs in this post will have percent correct on the y-axis.

y轴表示正确百分比。 这篇文章中的所有图表在y轴上的百分比正确。

Replicating Chen and Wyble, participants perform no worse in the surprise and post surprise trials, indicating that they succesfully found the target.

复制品Chen和Wyble，参与者在惊喜和惊喜测试中的表现均不差，表明他们成功找到了目标。

Now lets look at participants’ ability to report the target’s color in the surprise trial and the trial immediately following the surprise test.

现在，让我们看看参与者在意外试验和意外试验后立即进行试验中报告目标颜色的能力。

Below I plot the percent of participants that correctly identified the target’s color in the surprise and post-surprise trials

下面我绘制了在意外和惊喜后试验中正确识别目标颜色的参与者百分比

1 1 2 2 3 3 4 4

---------------------------------
Mean Surprise: 0.3
Mean Postsurprise: 0.7
Postsurprise - Surprise: 0.4
---------------------------------
Postsurprise vs Surprise
     Surprise Test. Comparison to Chance: 17.0 P-value: 0.5899
     After Surprise Test. Comparison to Chance: 33.0 P-value: 0.024
     Chi-Square Comparison: 6.4 P-value: 0.0114
---------------------------------
Mean Surprise: 0.3
Mean Postsurprise: 0.7
Postsurprise - Surprise: 0.4
---------------------------------
Postsurprise vs Surprise
     Surprise Test. Comparison to Chance: 17.0 P-value: 0.5899
     After Surprise Test. Comparison to Chance: 33.0 P-value: 0.024
     Chi-Square Comparison: 6.4 P-value: 0.0114
 

We perfectly replicate Chen and Wyble; participants respond more accurarely in the post-surprise trial than in the surprise trial.

我们完美地复制了Chen和Wyble； 与惊奇试验相比，参与者在意外后试验中的React更为准确。

The next cell examines participants’ ability to report the target’s identity on the surprise trial and the trial immediately following the surprise trial. Remember, the participants locate the target based on whether its a letter or number, so they know the broad category of the target. Nonetheless, they cannot report the target’s identity on the surprise trial

下一个单元格检查参与者在突击试验以及突击试验之后立即进行的试验中报告目标身份的能力。 请记住，参与者根据目标是字母还是数字来定位目标，因此他们知道目标的大致类别。 但是，他们无法在突击试验中报告目标的身份

Trial_Trimmer = df['TrialType2'] > 0 Iden_Acc = df[Trial_Trimmer].pivot_table(values='Iden_Acc', index='Sub#', columns='TrialType2', aggfunc=np.mean) CIs = SimpleComparisonCI(Iden_Acc) Analysis_and_Plot_2(Iden_Acc, CIs) Trial_Trimmer = df [ 'TrialType2' ] > 0 Iden_Acc = df [ Trial_Trimmer ] . pivot_table ( values = 'Iden_Acc' , index = 'Sub#' , columns = 'TrialType2' , aggfunc = np . mean ) CIs = SimpleComparisonCI ( Iden_Acc ) Analysis_and_Plot_2 ( Iden_Acc , CIs )

实验1 –样品间分析 (Experiment 1 – Intertrial analyses)

So far, I’ve perfectly replicated Chen & Wyble (which is good since this is their data).

到目前为止，我已经完美地复制了Chen＆Wyble（这很好，因为这是他们的数据）。

Now I want to see if the target’s color or identity on the previous trial influences the current trial’s performance in the location task. I am only examining presurprise trials, so this should be trials when the participants don’t “remember” the target’s color or identity.

现在，我想看看目标在先前试验中的颜色或身份是否会影响当前试验在定位任务中的性能。 我只是在检查惊喜之前的试验，因此应该是参与者没有“记住”目标物体的颜色或身份的试验。

First I want to make some variables representing whether the target’s color and identity repeat across trials.

首先，我想做一些变量来代表目标的颜色和身份在试验中是否重复。

1 1 2 2 3 3 4 4 5 5

Lets see what happens when the target’s color and identity repeat.

让我们看看当目标的颜色和身份重复出现时会发生什么。

1 1 2 2 3 3 4 4 5 5

---------------------------------
Mean match: 0.918
Mean mismatch: 0.8925
Match - Mismatch: 0.0255
---------------------------------
Match vs Mismatch
     normality test adj. Test value: 0.92 P-value: 0.0821
     Wilcoxon. Test value: 51.0 P-value: 0.04
---------------------------------
Mean match: 0.918
Mean mismatch: 0.8925
Match - Mismatch: 0.0255
---------------------------------
Match vs Mismatch
     normality test adj. Test value: 0.92 P-value: 0.0821
     Wilcoxon. Test value: 51.0 P-value: 0.04
 

Looks like a 2.5% increase in accuracy. Now, this wasn’t really a planned comparison, so please take this result with a grain of salt.

看起来准确性提高了2.5％。 现在，这并不是真正的计划比较，因此请带着一点盐来考虑这个结果。

As a sanity check, lets look at how repetitions in the target’s location (the reported feature) effect performance.

作为健全性检查，让我们看看目标位置（报告的功能）中的重复如何影响性能。

We have to quickly create a new variable coding target location repetitions

我们必须快速创建一个新的变量编码目标位置重复

df['Prev_Loc'] = df['Tar_Loc'].shift(periods=1) df['Prev_Loc_match'] = np.where(df['Prev_Loc']==df['Tar_Loc'], 1, 2) df [ 'Prev_Loc' ] = df [ 'Tar_Loc' ] . shift ( periods = 1 ) df [ 'Prev_Loc_match' ] = np . where ( df [ 'Prev_Loc' ] == df [ 'Tar_Loc' ], 1 , 2 )

Trial_Trimmer = df['TrialType'] == 0 Loc_Acc1 = df[Trial_Trimmer].pivot_table(values='Loc_Acc', index='Sub#', columns='Prev_Loc_match', aggfunc=np.mean) CIs = SimpleComparisonCI(Loc_Acc1) Analysis_and_Plot_3(Loc_Acc1, CIs) Trial_Trimmer = df [ 'TrialType' ] == 0 Loc_Acc1 = df [ Trial_Trimmer ] . pivot_table ( values = 'Loc_Acc' , index = 'Sub#' , columns = 'Prev_Loc_match' , aggfunc = np . mean ) CIs = SimpleComparisonCI ( Loc_Acc1 ) Analysis_and_Plot_3 ( Loc_Acc1 , CIs )

Target location repetitions lead to a 2% increase in performance. Again, this result is robust.

目标位置重复导致性能提高2％。 同样，该结果是可靠的。

It’s a good sign that this effect is about the same size as repetitions in the unreported features.

这是一个好兆头，表明此效果与未报告功能中的重复效果大致相同。

用实验1b复制实验1的内部分析 (Replicate Experiments 1 Intertrial Analyses with Experiment 1b)

Experiment 1 had some evidence that participants unconsciously knew the color and identity of the target, since they performed a little better when the color and identity repeated. The effect was small, so I am not 100% confident that it’s robust.

实验1有一些证据表明参与者无意识地知道了目标的颜色和身份，因为当颜色和身份重复时，他们的表现要好一些。 影响很小，所以我不是100％确信它的强大。

The best way to demonstrate that this effect is real would be to show that it also exists in another similar Experiment. Chen and Wyble provide a replication of Experiment 1. In this experiment, the only difference is the target and distractors appear for longer and are not masked (making them easier to see).

证明此效果是真实的最佳方法是证明它也存在于另一个类似的实验中。 Chen和Wyble提供了实验1的复制品。在此实验中，唯一的不同是目标和干扰物的出现时间更长，并且没有被遮盖（使它们更容易看到）。

If participants response more accurately when the target color and identity repeat in Experiment 1b, then we can be a little more confident that participants are unconsciously aware of the target’s color and identity.

如果参与者在实验1b中重复了目标颜色和身份时能更准确地做出React，那么我们可以更有信心地使参与者无意识地意识到目标的颜色和身份。

1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 11 11 12 12 13 13 14 14 15 15 16 16 17 17

1 1 2 2

---------------------------------
Mean match: 0.9716
Mean mismatch: 0.9644
Match - Mismatch: 0.0072
---------------------------------
Match vs Mismatch
     normality test adj. Test value: 0.93 P-value: 0.1875
     T-test adj. Test value: 2.81 P-value: 0.0112
---------------------------------
Mean match: 0.9716
Mean mismatch: 0.9644
Match - Mismatch: 0.0072
---------------------------------
Match vs Mismatch
     normality test adj. Test value: 0.93 P-value: 0.1875
     T-test adj. Test value: 2.81 P-value: 0.0112
 

Wow. Only a 1% change in accuracy, so again not big. Nonetheless, this result is signficant. So, Some evidence that participants perform a little better when the targets’ color and identity repeat.

哇。 精度只有1％的变化，因此又不是很大。 尽管如此，这个结果还是很重要的。 因此，一些证据表明，当目标的颜色和身份重复时，参与者的表现会更好。

翻译自: https://www.pybloggers.com/2015/12/complete-amnesia-for-object-attributes/

失忆症crowd