《解读别人的想法》

5i52

自闭症儿童可能没有或缺乏所谓的心智（mind)
这也可能是核心缺陷

我手头有一本书，
<<解读别人的想法>>《understanding other minds》2ndED by Simon Baron-Cohen
第一篇："Theory of Mind and Autism: a fifteen-year Review."
我想翻译出来，不过我的英语很烂，可能要很久。

雁冲天

下面引用由5i52发表的内容：

自闭症儿童可能没有或缺乏所谓的心智（mind)
这也可能是核心缺陷

下面这个网站有些信息：
http://www.core.org.cn/OcwWeb/Brain-and-Cognitive...

又多了一位热心且颇具实力的朋友，感谢你为大家做的一切

thankstoyilin

Simon Baron-Cohen是英国剑桥大学的著名自闭症学者。他从心理学、诊断学、神经科学的角度对自闭症谱系的疾病作出了杰出的研究。Understanding Other Minds 好像是由他做主编的研究项目文集。

binfeng2000

关于心智理论,国内介绍的不是很多,但我觉得这对于培养社交能力以及心理素质,要比RDI及其他方法更有直接的指导意义.在这个栏目里,我和caroline也曾翻译过作者的另一本书<<必要的差异>>(The Essential Difference)的两章.我没有他的那本书,但有其他的两本(除前面说的那本书,还有一本是(mindblindness),但是我不大喜欢他过多的学究风格.其实他做过很多的试验,这些试验很有直接的启发.我不知道你的这本书能不能把心智理论讲清楚讲通俗.我也看过这位剑桥大教授的文章,文驺驺的,害的我老翻字典还是稀里糊涂的.

5i52

几篇网上的文章：
http://hmb.utoronto.ca/HMB300H/Lectures%202005/ToM.pdf
http://courses.umass.edu/ling510/spring05/lecture_notes/510Lecture8.pdf
http://icdl.com/publications/pdf/10-15yearfollow-up.pdf
http://www.thorne.com/pdf/journal/7-4/autism.pdf

chenhhellen ：“心灵理论” （theory of mind）TOM测试
http://www.elimautism.org/leadbb ... 451&E=0&p=2

binfeng2000：<<必要的差异---男女两性的头脑以及自闭症之真相>>
http://www.elimautism.org/leadbb ... 940&E=0&p=2

5i52

5i52

参考书：
《儿童的心智》 Janet Wild Astington, 孙中欣译  辽海出版社 1999.12
    其实就是大雪无痕推荐过的《哈佛家庭教育经典译丛》中的一本。

为了更好的理解，可能需要了解一下认知心理学，关于皮亚杰，后皮亚杰之类的学说和研究方法。请参考：
《认知发展心理学》林崇德，沈德力， 浙江人民出版社
《儿童心智》 Child's Mind, Margaret, 汉菊德译  远流出版事业

============================================================

摘自<儿童的心智>

现在我们通常就将常识心理学称为"关于心智的理论"或"心智理论"

常识心理学是有关解释和预测人类行为的一个理论

信念和愿望作为心智呈现的形式,他对我们的行为常常起到调节的作用

心智是做什么的?---心智呈现
我们可以从两个角度来理解呈现这一概念.
其一是指心理活动的状态:如想法,愿望,信念,意图等等.
其二指的是心智活动的过程
所以它既是名词,又是动词;既是过程,又是结果

心智状态,心理状态,精神状态
心智状态是不是就是我们的大脑状态? 存在两种相反的看法.

5i52

张春华

想法解读课堂运用实例：
1、爷爷浇花的卡片：
问：“他在干什么？”
答：“他在浇花。”
问：“他为什么要浇花？”
答：“他想让花长得快一点。”或“他想让花长得更漂亮。”

2、小朋友端一杯热茶，对面坐着一位老奶奶的卡片：
问：“小朋友手里端的什么？”
答：“小朋友手里端了一杯热水。”
问：“他要把热水送给谁啊？”
答：“他要把热水送给奶奶喝。”
问：“奶奶会对他说什么？”
答：“奶奶说‘谢谢’。”或“奶奶说‘你真是好孩子’。”

3、小故事《小熊的手》：
图1：小熊在拍球。
图2：妈妈喊小熊吃饭。
图3：小熊的手很脏。
问：“小熊该怎么做呢？”
答：“小熊要洗洗手。”
看图4小熊洗手的图片。
问：“洗完手以后，小熊的手就会变得怎么样？”
答：“小熊的手就会变干净了。”
看图5小熊的手干净的图片。
问：“然后小熊就可以干什么了？”
答：“小熊就可以吃饭了。”
问：“如果小熊不洗手就吃饭就会怎么样？”
答：“会生病的。”

以上是比较简单的想法解读的项目。

再举例一个有些难度的项目：故事《小熊种豆子》，情节如此：春天来了，小动物们都到地里种豆子，小羊种豆子，小狗种豆子，小熊也种豆子。小熊浇水（图片）。后来，小狗种的豆子长出来了，小羊种的豆子也长出来了，还开出了美丽的花朵。只有小熊种的豆子没有长出来，他很纳闷。小熊就把泥土挖开，发现不是豆子，是一颗纽扣！小熊一掏口袋，原来豆子还在口袋里呢！

当给孩子将这个故事完全讲完也进行了故事梗概的复述的时候，
指着指着小熊浇水的图片，问：“小熊知道不知道这里边是不是豆子？”
指着第一页小羊种豆子的图片，问：“小羊种到土地里的，是不是豆子？”
等等。

5i52

谢谢张老师,

理论要结合实际才有意义,看了张老师的帖子,我才真正明白那些研究的价值.这些是在书本上看不到的.

我们国家的问题是理论和实践脱轨. 就特殊教育来言,搞理论的,不屑走到第一线去接触孩子,而真正在第一线的,又没有机会深入研究理论. 象张老师游刃于理论和实践之间而有余的,真是屈指可数.(是不是有点阿谀奉承的味道).其他行业更是如此。

所以,以研究生为界线,只要上了研究生,就可以每天坐办公室,不用带孩子了.喝茶聊天看报纸之余,搞点闭门造车和抄袭拼凑的"研究成果",和没有完全并没什么两样.

即使身处第一线的，很多也是心在曹营身在汉，这也有情可原，因为在第一线待遇低，money不够用，但是没有高水平，指什么拿高报酬？于是就这样恶性循环下去。特别是这个行业又是高度依赖经验的，留不住人，哪能积累经验？

中国人口这么多，特殊儿童数量巨大，不是没得可搞，关键是如何搞。我们的现实就是，自闭症这个领域，弄得最好的两个女士，一个是外贸系的教师，一个是外语系的教师。

就我们做家长的，说是要再学一个专业，但是谈何容易！自闭症的教育和治疗就像是塔的最高层，就说现有的治疗手段吧，它的下面有神经医学，遗传学，消化系统医学，免疫学，生物化学，生理学，医药学，检验学，还有其他这个学那个学的一大堆，根本无从下手。这个常识心理学（心智理论），也要了解儿童心理学，认知心理学，行为学，不是搞这个专业的，也只能看到些皮毛。

孩子生病了，就后悔当年没有学医，也后悔没有找个学医的老婆。
现在发现孩子自闭，后悔的就更多了。

希望张老师能更多地分享您的知识和经验。

张春华

谢谢5i52表扬！让你说得都快飘飘然了。

在我的实际操作中，我一直觉得在中高功能的自闭症孩子的教学中运用起思想解读的技巧来，比较容易操作，材料比较容易选择，话题也比较广泛，孩子也比较容易将生活中的经验运用到话题上来。说到底，思想解读的基础是孩子的生活经验或学习经验。RDI在任何程度孩子的个训中都可以运用，只是分享的程度和能够深入的程度不同而已。ABA的回合式教学则在较低功能的孩子中比较广泛适用；对中高功能的孩子，将ABA的操作理念贯彻和渗透进去即可。不知道我的想法和做法对不对，还有待于别人的考察和实践的进一步验证。

5i52

朋友们知道我儿子是自闭症，正在接受训练，便很好奇地问我如何训练。怎么向他们解释ABA呢？ 我只好避开术语，告诉他们就像驯兽一样，胡罗卜加大棒，做得好就给点奖励；只是对动物不用讲人性，做的不好可以打骂，对儿童通常却只能忽视。

每次讲到这，朋友们的表情马上从轻松和好奇转为高度同情，我读得懂那同情是真心的。
如果跟他们讲RDI的活动或解读他人思想的训练，他们的反应可能不会这样大。

ABA适合中低功能的孩子。而且ABA被认为有效，我想这也是一个原因，其他的许多宣称有效的方法往往是在中高功能的孩子身上才能看到效果，还不知道这进步是方法的作用还是孩子自己长出来的。

对高功能的孩子，ABA可能会引起他的抗拒，因为他有可能看穿我们的把戏。如果有人用回合式ABA教学的方法训练我，我不扁他才怪呢！顶多也是表面应付他一下。

可能是为了扩大顾客群，RDI一直宣任何程度孩子都可以运用，但是对功能很低的孩子可能需要训练者素质够高才玩得转，很难分享，很难深入，这是没法回避的事实。

其实，所有这些有名有目的方法只不过是提供给我们这些普通人当做菜谱，一步一步按部就班地 follow 的。就像我们看过的武侠小说，真正的高手无门无派，无招无式，一切信手拈来，无影无形，无息无声中打得对手狼狈逃窜。任凭对手千变万化，都能见招拆招。根本就不用苦苦地去寻找九阴真经之类的东西的。

我的这些看法全是空想出来的，没有一点实证，不要太当真。

5i52

第十八章：教自闭症儿童解读别人的想法
                                                                                         [I]John Swettenham [/I]

这一章将讨论最近对训练自闭症儿童解读别人的想法的可能性研究。这种研究应该具有很大的临床意义。

如果答案是肯定的，那么就如Baron-Cohen等宣称的，由于心智缺乏导致的社交障碍，语用学的问题，假想游戏的不足，就可能改善，从而减轻自闭症症状。

Bishop 最近强调，这个研究的意义不但在于实验性的调查，研究技术本身也很有意义。（译者注：请注意下面的逻辑，看有没有问题）我们假设社交等方面的障碍是因为心智缺乏引起的，我们通过一些提高自闭症儿童解读别人的想法能力的训练，结果儿童在社交等方面取得了很大的进步，我们就有很强的证据证明社交等方面的障碍是因为心智缺乏引起的。

但现在为止的研究表明，经过教导，自闭症儿童确实能够学习而通过特定的心智测验。因为通过这些测验的前提是能够判断出人的心智状态，儿童通过了这个测验，说明他确实判断出了人的心智状态，似乎可以作结论了。

但话还不能这么讲， 儿童可以不用判断出人的心智状态就可以通过测试，他们使用另外的原则来判断。

例如，第一章讲到的回答谁知道箱子里有什么的测试，儿童可能记住这个测试的答案就是那个看箱子的人，但他并不知道为什么。（译者注：作者使用的例子第一章没提到，译者换成个第一章提到过的例子，意思是一样的）。

对这种表面的成功还要注意下面一点，自闭症儿童泛化能力较差。即使他通过了某个测试，这个新的知识不能很好地泛化在其他的测试以及日常社会交往中。

测试结果泛化能力差可能是因为
1．        儿童使用的是非心智状态的原则（谁知道箱子里有什么的测试—回答看箱子的那个人）。这种原则不适合没有教过的情形。或者
2．        他们有更多的泛化方面的问题

第2项可以由几个原因产生：过于选择性的关注于测试中的某个细节过程，或者不会处理测试情景与变化的测试情景（或日常的社交生活）中的共同特性。

如果问题出在泛化本身，那么了解自闭症特有的泛化广度就非常重要。

我这里不会涵盖目前存在的各种不同的训练自闭症儿童社交技能的方法。本章的目的是研究提高自闭症儿童解读别人想法的能力，下面几节概括了已经做过的针对不同程度的自闭症儿童的各种训练的研究。                                        --未完待续--

星晴

谢谢5i52，期待着您的继续~~

5i52

关于训练解读他人想法的能力的研究

Ozonoff和Miller是最早研究是否可以训练解读他人想法能力这个课题的（1995）。对象是5个高功能自闭症或广泛发育障碍的青少年。早期的研究发现，少数自闭症患者通过训练可以完成诸如Sally-Anne这类任务的第一级心智测验(译者注：请参考《“雨人的秘密”--解读自闭症之谜》第72页：Sally有个篮子，Anne有个箱子。Sally把球放入篮子中，然后出去散步。 Sally外出期间，Anne从篮子中取出球，放入箱子中。Sally回来后，会到哪里去找球？自闭症儿童倾向回答到箱子里去找，因为他们不会站在Sally的角度想问题)，高功能的甚至能通过第二级测验。由此我们可以乐观地相信，功能相对高一点的自闭症经过训练，有通过更高级的心智测验的潜力。在训练的前后，分别进行四个心智转换的任务测试，它们是：
1。 Smarties测试,“错误内容”（Perner: False contents task）
2。 第二级错误信念,“Mary认为John在想....”(Baron-Cohen)
3。 第二级错误信念，“外套大衣”（Bowler: overcoat task）
4。 第三级错误信念，“虚张声势”(Happe:Dobule bluff story)

教师和家长同时使用社交技巧衡量表（SSRS:Social Skill Rating Scale）对儿童社交技巧进行评估。(译者注：我对这个衡量表非常感兴趣，求之不得）

因为这个干预要经过相当长的时间（四个半月），重要的一点是要判断儿童的进步是否是自然发育的作用。 为此，4个年龄，智商，程度相仿的青少年（CARS打分相同）作为对比组，对比组也在开始和结束时参加同样的测试，只是中间没有参加特殊干预。

(以下部分先放上英文, 有时间再翻)

The first set of seven training sessions focused on teaching interactional and conversation skills such as how to begin and maintain topics of conversation, how to choose topics of interest to others, how to interpret and express non-verbal signals, and how to listen and express interest in others. The second set of sessions focused specifically on teaching theory of mind skills. The main focus of the training was role playing exercises with the emphasis on providing underlying problem-principles that might be applied to different situations. These sessions began with a role playing exercise in which children led a blindfolded trainer around an obstacle course. This was designed to teach the children how to take another person's visual perspective. Next the children were taught that what someone sees or hears indicates what they will know(i.e. the principle that 'perception influences knowledge'). Finally the children took part in a series of role-plays which mirrored the format of second-order false-belief tasks, although the specific content was different from the transfer tasks(e.g. Using different locations). For example, in one role-play children A,B,and C decide to go bowling together. After child A leaves to get changed at home, B and C revise the plan and decide to go to a film instead. B and C then go home separately. On the way home, B stops at A's house to tell him of the new plan. They agree to meet at the cinema and A leaves to go there. After changing, C stops at A's house and find he is not there. The groups members were asked to predict where C would think A had gone. After each of these scenarios the group discussed the underlying principle behind the role-plays. For example, for second-order false-belief role-plays, the principle was that 'since child C had not seen child B speak with A, he could not know that A knew of a revised plan.

Prior to training, the group scored an average of 6 out of a possible composite score of 13, although it it not clear from the results exactly which tasks the children were failing and which they were passing. When the children who had taken part in the training control group. However, the training appeared to have no effect on everyday social skills may have been because the SSRS did not provide a sensitive measure of the sort of skills one might expect to improve as a result of training. For example, the SSRS includes items such as 'is self confident in social situations' and 'is liked by other'. Alternatively, the improvement on the theory of mind task may have been the result of learning a specific non-mental rule which was applicable to some of theory of mind task in the battery, rather than learning to use mental-state terms. This is certainly possible given that the second-order role-plays involved the same scenarios as the second-order transfer tasks. This at least suggests that limited generalisation to different materials and characters can take place, even with a non mental-state rule. It is also possible that some children learned to infer mental states and to generalise this knowledge. A non mental-state rule learned for the second-order role-play scenario could not be used to pass the third-order double bluff task, and at least one child achieved this, since he achieved a maximum score after training. Unfortunately, since only composite score were presented we cannot be sure if any of the other children improved on task other than the second-order false-belief tasks. However, with this small sample of relatively high-functioning adolescents with autism the result were promising.

A group of less able children with autism(CA between none and ten years, VMA between five and six years) took part in training study conducted by Hadwin et al.(1996,1997). In this study the intention was to asses whether it would be possible to teach children with autism in one of three domains; (i)understanding of emotion;(ii)understanding of belief; or (iii)production of pretend play.

Thirty children with autism took part in the study. Prior to the training all the children were tested on tasks from each of the three domain. In each domain there were five tasks which were ordered into 'developmental levels', so that a level 1 task was the simplest and level 5 the most difficult. In the emotion domain the tasks were:(i)photographic face recognition; (ii)schematic face recognition;(iii)situation-based emotion understanding; (iv)desir-based emotion understanding; and (v)belief-based emotion understanding. In the belief domain the task were;(i)simple perspective taking; (ii)complex perspective taking;(iii)seeing leads to knowing; (iv)true belief understanding; and (v)false belief understanding. Finally, the five levels of play behaviour were:(i)sesorimotor paly; (ii)functional play(two or fewer examples); (iii)functional play (more than two examples);(iv)pretend play(two or fewer examples);(v)pretend play(more than two examples). For the pre-training assessment each child was presented with tasks in order from level 1 onwards until two consecutive levels were failed. The pre-training assessment therefore established a developmental level's score for each child in each domain.

Two additional assessments of conversation skills and use of mental-state words were also made before and after training. Children were asked to tell a story from a picture book and the frequencies of:(i)one-word answers; (ii)two or more sentences; (iii)echolalia; and (iv)unclear statements, were recored. Parents were instructed to ask questions and give prompts during the story telling in order to create a conversation-like interaction. The hypothesis was that improved theory of mind might help the child to understand that people know different things and that states of knowledge can be shared, so that theory of mind training would result in children expanding upon their 'conversation'. The frequency of mental-state words produced during this assessment was also recorded.

For the training itself, the children were divided into three groups, and each group was taught in a particular domain--emotion, belief, or pretence. The pre-training assessment indicated at which level teaching should begin for each child. For the emotion and belief groups each child was taugh on tasks in subsequent levels by question-and-answer with corrective feedback, and in addition was given a general principle governing the understanding of the mental state involved in the task. The teaching strategy in the pretence group, by contrast, was spontaneous and unstructured. The aim here was simply to encourage children to produce and participate in pretend play acts with a series of toys related to a theme(e.g shopping or dinner-time), and junk objects, using modelling and verbal guidance. Training took place over eight consecutive days with one half hour session per day.

The result showed that children in the emotion and belief groups improved in their performance on the task they were being trained on. For the children trained in pretence there was no significant improvement in performance. Of course this does not mean that pretending is therefore harder to train than emotion of belief understanding--the children may have failed to produce more pretending because of the materials used or the type of training used, etc. All we can conclude is that the children in this study did not learn to produce more pretend play acts as a result of this particular training technique.

When children from each group were re-tested on the task from all three domains following training, the results revealed that children trained in one domain did not improve in their performance on tasks from another domain. That is, there was no evidence of generalisation between theory of mind domains. The children also showed no significant change in the ability to expand on conversation, and no change in the number of mental-state terms used during conversion. Thus, despite improvement on theory of mind task used during training, there was no evidence of an improvement in conversation skills. However, the children did pass task similar to the ones they had learned to pass during training, but which used different materials. Whilst this suggests that a limited degree of generalisation was possible, it was still not clear what was being generalised--new knowledge about inferring mental state or a non mental-state rule for passing the tasks.

What we cannot tell from either Ozonoff and Miller's or Hadwin et al.'s studies is whether the failure to generalise to novel tasks is a problem specific to autism. It is possible that children without autism, but which the same mental age, would show the same problems in generalising if trained on the same tasks. One way to investigate this possibility is to include control groups who are trained on the same tasks as the children with autism and assessed in the same way.Swettenham attempted to do this by including eight children with autism, eight three-year-olds, and eight children with Down's Syndrome in a theory of mind training study. The aim was was to teach children from each group to pass the Sally-Anne false-belief task with repeated presentations of a computer version of the task. Generalisation was then assessed with five post-training false-belief tasks, all of which had been failed by all children proir to training. Two of these tasks involved the same scenario as the training tasks--a computer version without instruction and the Sally-Anne task with dolls--and were refered to as the close transfer tasks since they required minimal generalisation. The other three transfer tasks also assessed understanding of false belief but used different scenarios to the training task. These were two versions of deceptive-appearance false-belief task based on Perner et al. 1987; and the Tom(theory of mind) task, in which the child is told abouta character's fasle belief, and asked to predict to his behaviour. These tasks were referred to as distant transfer task as they required a greater degree of generalisation. Importantly, it was thought that the distant transfer tasks could not be passed by the use of a non mental-state rule learned during training.

In addition, the children were given four true-belief tasks before and after training. These computer presented tasks involved scenario in which the ball is not transferred from one location to another--for example, Sally hides the ball in location A, Anne removes the ball and then replaces it in the same location(A). All the children passed these true-belief tasks prior to training.

The computer version of the Sally-Anne task was mouse driven and include music,text, and animation, which characters hiding and retrieving a ball in one of two locations as in the conventional presentation. The initial hiding place for the task varied randomly. The child was instructed to click the mouse on characters, or on a door on the screen, in order to move through the sequences of the task. Throughout each presentation prompts appeared on the screen explaining what the characters were thinking. For example, after initially hiding the ball Sally says, "Now I think the ball is in the red/blue box", and, "I must rember to look in the red/blue box if I want my ball". When Anne transfers the ball she explains, and, "Sally hasn't seen me move her ball", "Sally will think that the ball is in the red/blue box". At the end of each presentation of the task the child was asked to select the container where Sally thinks her ball is. If the child is correct the character approaches the container, looks inside and exclaims,"Oh no, the ball is not there anymore." If the child choose incorrectlys,Sally replies,"I think the ball is in the red/blue box because that's where I left it."

Each child received a set number of 48 trials, divided into eight sessions over the course of week. All three groups showed a steady rate of increase in the number of correct trials per session--althrough, surprisingly, it was the group of children with Down's Syndrome who had the lowest mean score throughout training. The children with autism and the three-year-olds consistently passed 5 out of 6 trials per session earlier than the children with Down's Syndrome. Children from all three groups were able to pass the close transfer tasks following training, suggesting that they were all capable of generalising what they had learned to tasks similare to those used for training, but which involved different materials. In contrast, none of the children with autism passed any of the distant transfer tasks, whilst five of the three-year-olds and five of the children with Down's Syndrome passed at least one of the distant transfer tasks following training. The same results were found when the children were re-tested on the transfer tasks three month later.

Since the children with Down's Syndrome and the three-old-years were able to generalise what they had learned to pass tasks which involved different scenarios to the trained task, it seems likely that they had learned to infer mental states during training. Howeverm the children with autism failed to generalise what they had learned to help them pass the distant transfer tasks. One possible reason for this may have been that children with autism learned the simple rule;"Sally always thinks the ball is in the container that is empty'. This rule would alway lead to success on the Sally-Anne false-belief task. However, if this rule was then generalised to a rule-belief task, in which the ball is replaced in its original location, then the rule would lead to an incorrect answer. The results showed that children with autism continued to pass the true-belief tasks following training. This meant that either they had not learned that rule("Sally always think the ball is in the container that is empty") during training, or that they had not applied the rule to the true-belief task. So, this study showed that children with autism could learn to pass the false-belief task, but, unlike normal three-year-olds or children with Down' Syndrime, they were unable to generalise what they learned.

There is another method that tell us whether children with autism can learn to understand false-belief, rather than learn a non mental-state rule, without requiring evidence of generalisation. In order to learn a non mental-state rule during a training study, a child would have to be given the correct solutions to the trials. If children with autism can learn to pass false-belief tasks without being given information on or reinforcement for the correct response, then this would presumably be evidence that the children had learned to infer false belief.

Bowler and Stromm attempted to help children pass the Sally-Anne false-belief task without directly providing the correct answer on each trial. Instead, they provided action and emotional cues to the protagonist's false beliefs(based on Moses and Flavell 1990), and also gave the children the opportunity to experience their own false belief in a similar scenario. All the children in the study initially failed the Sally-Anne false-belief task. All the tasks were presented with real people playing the roles of Sally and Anne. [I](to be continued)[/I]

5i52

The study included children with autism, children with learning difficulties,normally-developing three-year-olds(nearly four years of age) and young three-year-olds(under three years, six months). Children from each group were divided into an experimental or a control condition. In the experimental condition, children received five trials. The first was the action false-belief trial in which the assistant playing the role of Sally return to the room and stood next to the container where the ball had originally been hidden. The second was the surprise false-belief trial in which the assistant playing Sally returned to the room, looked inside the hiding place, showed a surprised expression and said 'Gosh, my(object) isn't here!'. The third was the own false-belief condition in which the child hid an object, left the room, and returned to find it had been moved. The fourth trial was a repeat of the action false-belief trial, and the final trial was a repeat of the surprise false-belief trial. The correct answer was not directly given on any of these trials. Children in the control condition were given the repeated presentations of the standard condition of the false-belief trial and the own false-belief trial.

The result showed that in the experimental condition children with autism as well as some of the normally-developing older three-year-olds, benifited to a significant extent from the enhanced and emotional cues. Younger three-year-olds and children was learning difficulties did not benefit from the cues. In contrast, none of the children in any of the four groups, in the control condition, benefited from simple repeated presentations of the false-belief task. One explanation for the success of the children with autism and older three-old-years in the experimental condition is that they genuinely learned about mental states as a result of the enhanced action and emotional cues. And since no direct indication of the result was given during training it would not be possible to learn a non mental-state rule for the correct answer. Unfortunately a simpler explanation for improve performance can be given. As Bowler and Stromm point out, the children may have been responding to the surface sttructure of the trial. That is there is considerable stimulus enhancement of the correct location in the experimental condition even though it is not specifically described as the correct location for the task. Children may begin to choose the correct location simply because their attention has been drawn to it. This study therefore does not provide cast-iron evidence that children with autism have learned to employ a theory of mind as a result of being provided with action and emotion cues.

Whiten et al. also used real people in a theory of mind training study. In this study, adults with autism and three-year-olds matched for verbal mental age took part in a series of trials, some of which involved hiding a familiar caretaker's coat and a series which involved hiding the caretaker's cup of coffee. The participants were taking part in another experiment when the caretaker entered to see how things were going. The caretaker then announced that s/he had to leave, providing an excuse(e.g. making a phone call), and made a point of leaving his/her coat--"I'll leave my coat on this chair and come back and get it in a minute." Once s/he had left, the experimenter suggested that they play a trick and hide the coat in a box, Having checked that the participant knew where the coat originally was, and where it was now, the experimenter asked, "when (caretaker's name) comers back, where will s/he look for his/her coat?' If the participant was wrong, then the experimenter provide the correct answer; "I think s/he'll look on the chair.' Participant were gived up to six trials of the coat test followed by up to six trials of an equivalent test using a cup left on a table.Trials were terminated once two successive correct responses were made.

Both the adults with autism and the three-year-olds benefited from the Coat series, as fewer trials were need to achieve success on the Cup series. Both groups also showed a significant improvement on the traditional Sally-Anne task which was used as a transfer task and presented as a series of pictures. Generalisation had therefore taken place from a real life scenario to an artificial task using a very similar scenario. This means that even if they had learned a non mental-state rule, the participants must have recognised  that the real life scenario was equivalent to the picture story.

It is not clear then whether any of the training studies described so far have genuinely taught theory of mind understanding to children with autism. It is certainly the case that children with autism can learn to pass theory if mind tasks, but their ability to generalies is limited. One possibility is that they can learn about mental states, but have a general dificit in generalisation. Alternatively, it is possible that children with autism have a specific deficit in understanding mental states so that they focus on non mental-state aspects of training tasks in order to acquire a rule or strategy to pass the tasks. The problem is that the rule or strategy they acquire is not useful for passing other tasks or for reasoning about mental states in everyday interactions.


THOUGHTS ARE LIKE PICTURES IN THE HEAD

An alternative approach then might be to try to teach children with autism to employ a specific strategy, which would be useful in a variety of situations, based on an aspect of their intact abilities. Swettenham et al. attempted to teach children with autism the analogy that people have photos in their heads. This strategy draws on a domain of intact cognition (understanding photographic representations) in an attempt to bypass a cognitive impairment in another domain ( Understanding mental representations). Leslie and Thaiss and Leekam and Perner have shown that whilst children with autism were unable to understand that a person's belief can differ from reality, most could understand that a photograph can differ from reality. That is, they were able to understand that an object's location would remain the same in a photograph, even when this had changed in reality. Of course mental states are not exactly like photos in many respects. For example, mental states are unobservable, they are only held by animate agents, and they may not be visual-image based. And whereas for adults, photos serve only as a resonable metaphor for what belief are like(they are about things, they endure even when reality changes, they can be true or false, etc.), for children with autism they may be the closest approximation to what beliefs are actually like, if they cannot conceive of belief normally.

The aim then, was to teach the metaphor that 'people have picture in their head' in the hope that this would be a useful prosthetic device to compensate for the lack of a 'real' theory of mind (much as Braille is used to enable a blind person to read). The study focused on teaching children the ability to understand false beliefs. The prediction was that if the children were able to learn the photo-in-the-head strategy, they would not only learn to solve a specific task, but would also have learned a powerful generalisable strategy that would enable them to deal with a variety of theory of mind problem. Explicit teaching of the photo strategy was given only in the context of the Sally-Anne false-belief task. However, four different types of transfer tasks were given before and after teaching in order to assess whether children were able to generalise the strategy. These four tests were:(i)the Sally-Anne false-belief task; (ii)a deceptive contents task, the "Smarties" task, (iii) a 'seeing-leads-to-knowing' task; and (iv) an appearance-reality task.

Children with autism were gradually introduced to the photo-in-the-head analogy over the course of five separate sessions. A large manikin's head was used with a slot in the top for inserting photos. Children were taught a set of principles which govern how the photo-in-the-head analogy worked. Teaching consisted of demonstrations and questions to check that the child had learned each principle before moving onto the next. The children were taught that when the manikin(called Sally) looks at something she get a picture of the object in her head. Next children were shown that when Sally is absent and the world changes, the picture remains in Sally's head. Children were then shown the false-belief scenario using the manikin and asked to insert the appropriate photo of an object in a location into Sally's head, and to name what the photo in Sally's head was of when she returned to look for her ball. Finally, the children were taught that by referring to the photo in Sally's head it was possible to tell what Sally was thinking (i.e. 'that the object is in the location as shown in the photo'), and that the photo would also be an indication of how Sally would act (i.e.'that Sally would search for the object in the location shown in the photograph'). Interestingly, children had difficulty with the idea that the photo showed what Sally was thinking. This may have been lack of familiarity with the word 'thinking'.

All the children were able to learn the basic steps of the photo-in-the-head strategy. Although the strategy was not spontaneously used to predict behaviour or mental states, when an explicit link was made(that Sally will refer to the photo in her head to tell her where to look for the hidden object), seven out of eight children were able to pass the Sally-Anne task. Furthermore, there was a significant improvement in performance on the traditional dolls version of the Sally-Anne false-belief task, and an improvement on the seeing-lead-to-knowing task.

This result suggested that the children had learned the photo-in-the-head rule and that this could be generalised to a task which had not beem used during training(seeing-leads-to-knowing). It would appear that the children with autism had used the rules that "when someone sees something they get a picture in their head" and 'the picture in the head is what someone is thinking' to help them choose that the character who sees inside the box is the one who knows what is inside. However, it was also the case that there was no significant improvement on the Deceptive Appearance task or on the Appearance-Reality task. Lack of improvement on the Deceptive Appearance task suggests that the children's ability to generalise remains limited, whilst lack of improvement on the Appearance-Reality task suggests the limitations of the metaphor, since a sponge that looks like a rock will look the same in a photo.

McGregor et al. have used a picture-in-the-head technique in a training study attempting to teach theory of mind to children with autism, and also compared this technique with another teaching method that involved highlighting the protagonist' intentions in the Sally-Anne type false-belief scenario. Two experimental groups, matched for mental age, were given training using both techniques: (i) a group of adults with autism; and (ii) a group of three-years-olds. Two non-intervention control groups were also included to control for natural change over the course ofthe training.

Training using this technique focused on two versions of the Sally-Anne false-belief task and involved slotting pictures of the location of hidden objects in doll's heads. The technique of highlighting also focused on the Sally-Anne scenario and involved emphasising the location that Sally hides her object and her expectation that it will remain there. Transfer tasks were given before and after training for both experimental and non-intervention control groups. These tasks were:(i)the traditional Sally-Anne tasks(without photographs); (ii) a deceptive-appearance task;(iii)the windows deception task; (iv) the coat task( a real life false-belief task in which a character's coat is moved in his/her absence), and (v) an appearance-reality task.

Result showed that the picture-in-the-head technique was substantially more effective for teaching theory of mind compared with the technique of highlighting intention. Both the three-year-old children and the individuals with autism who had been trained were able to pass the standard Sally-Anne fasle-belief task following training. The three-year-olds also showed substantial evidence of generalisation to other false-belief tasks. However, for individuals with autism generalisation was limited to being able to pass a test of their own false belief.

Another representational device that might help compensate or bypass deficits in understanding and employing mental states is 'thought-bubbles'--after all, thought-bubbles are often used to depict a person's thoughts in cartoons and magazines. In face, typically-developing children as young as three years of age appear to have no difficulty in understanding that thoughts can be represented in 'thought-bubbles'. Using thoughts bubbles to teach theory of mind has already been piloted in a single case study in which a child with autism successfully learned how to select thought bubbles which represented a character's thoughts in a false-belief task. Larger scale studies are currently underway to investigate the efficacy of using thought bubbles as an aid to mentalising.

TRAINING STUDIES WITH NORMAL CHILDREN

A number of recent studies have focused on teaching theory of mind to normally-developing three-year-olds Whilst these studies can ultimately only tell us about normal development, they may provide some clues for research with individuals with autism--what works with normal three-year-olds might work with children with autism. Most of the research with normal children has been conducted to test an empiricist view that theory of mind is a consciously constructed theory made up of interrelated concepts. For example, Wellman argue that a concept like 'belief' is understood in relation to other concepts like 'perception', 'action', 'desire', and 'intention'. A number of training studies have shown that emphasising the relation between 'belief' and other mental state concepts within a theory has led to improved performance on false-belief tasks.

Gopnik et al. found an improvement in understanding other's false belief on a deceptive-appearance when children were taught that their own changed visual perspective is like a change in what they thought about an object. When belief was paired with other mental states such as desires and emotions in a deceptive transfer false-belief task, Bartsch also found improved performance in three-year-olds on the false-belief task. Morses also tested three-year-olds' understanding of false belief in the context of intentions. When children were presented with a film showing an actor's surprise or sadness at having failed to complete an intended act, and then questioned about unfulfilled intentions, they performed better on false-belief tasks. Finally, Slaughter and Gopnik found that three-year-olds improved in their performance on a variety of theory of mind tasks, including false-belief understanding when trained on the related concepts of desire and intention.

These results are interesting because they suggest that in normal development it may be possible to teach a concept such as false belief by enhancing understanding of related concepts to help children form a theory. It remains to be seen whether such methods will be succssful in teaching children with autism. Hadwin et al. did attempt to teach children in different related concepts(pretence, emotion, and belief) but found that no benefit in belief understanding was gained by teaching about emotion or pretence. [I](to be continued)[/I]

5i52

CONCLUSIONS

The training studies described in this chapter have all demonstrated that individuals with autism can be taught to pass theory of mind tasks. However, it is not clear what has been learned during these training studies---the ability to infer mental states and predict action from mental states, or a non mental-state rule that can be used to obtain the correct answer. Evidence of generalisation to tasks similare to the ones that had been trained does not help disambiguate these two possibilities. Children may have learned a non mental-state rule to pass the training tasks and used that rules to pass the transfer task. Individuals with autism showed limited generalisation to novel theory of mind tasks, on which they had not been trained, and no significant improvement in everyday social interaction following training. This may be because they learned a non mental-state rule during training, or because they suffer from a general deficit in generalisation.

Two studies assessed whether teaching theory of mind would lead to improvements in social interaction, and found no difference in social interaction after training. Of course this null result dones not mean that social interaction cannot be improved through training theory of mind. It is possible that more extensive training ove a longer period, or the use more sensitive assessment measures, might prove more successful, and there are some groups for optimism. For example, although the training studies found no evidence of transfer from artificial tasks to 'real life' social interaction skills, it is worth noting that several studies showed transfer from 'real life' scenario to artificial tasks. Furthermore, individuals who pass false-belief tasks(without training) score more highly, than those who fail, on the Vineland Adaptive Behaviour Scales, a correlation which suggests that theory of mind ability does relate to social communication skills. So. despite the relative lack of success in training studies so far, it would be unwise to assume that future studies will not be more successful.

Although the training studies described here do not provide a direct test of what causes the theory of mind dificit itself, they may provide clues. For example it is clear that individuals with autism have difficulty in generalising new knowledge. A general deficit in the ability to generalise,which is specific to autism, is one candidate for an underlying primary psychological deficit in autism. However, in order to test this idea further we need to know whether individuals with autism have difficulty learning and generalising concepts or rules which don't involve mental states. If training studies that focus on teaching aspect of psychological functioning other than theory of mind lead to a genuine ability to infer mental states and predict action then this might be good evidence for a primary psychological deficit in autism. For example, if individuals with autism have a primary deficit in selective attention, then focusing teaching on what aspect of a social scenario to attend to and what to ignore may lead to improvement on theory of mind tasks. If there is some other psychological function or behaviour, the disruption of which early in development consequently impairs theory of mind development, then it may be preferable to conduct an early intervention to attempt to alleviate that early occurring impairment. For example, one current early intervention study is attemption to teach eighteen-month-olds with autism to attend to faces and to engage in joint attention as part of an intervention, and it will be interesting to examine the effect this training has on theory of mind development. Similarly, if the perceptual and attentional impairments which seem apparent in autism can be alleviated early in development, then the relationship between these apparently non-social impairments and the development of theory of mind can be examined.

One promising new approach for older individuals with autism has been to provide an analogy that 'thoughts are like picture'. This may lead to a genuine understanding of mental states, or individuals may continue to use the analogy as a useful aid to solving teory of mind problems. Ultimately the success of the picture analogy or any other approach will be measured by how effective it is in helping individuals with autism to understand and predict the behaviour of others.

Note:
1. Theory of mind training studies cannot however reveal much about what cause the theoryof mind deficit itself, only whether a deficit in theory of mind relates to impairments in social communication.

2. It might be argue that this is less likely in the caseof Ozonoff and Miller's study, because the children with autism were high-functioning. However, problem in generalisation(or any psychological process) may result from having any severe developmental disorder.
[I](end of chapter)[/I]